Low-Cost Micromechanically Tunable Optical
Devices: Strained Resonator Engineering,
Technological Implementation and Characterization
Vom Fachbereich Elektrotechnik / Informatik
der Universität Kassel
zur Erlangung der Würde eines
Doctor-Ingenieurs (Dr.-Ing.)
genehmigte
Dissertation
von
Dipl.-Ing. Amer Tarraf
Kassel, im November 2005
Doctoral adviser Prof.Dr. rer. nat. H. Hillmer
Referee Prof.Dr. rer. nat. H. Hillmer
Co-Referee Prof.Dr. -Ing. P. Meißner
Examiners Prof.Dr. rer nat. H. Hillmer
Prof.Dr. -Ing. P. Meißner
Prof.Dr. -Ing. S. Hentschke
Prof.Dr. -Ing A. Zündorf
Day of defense: 30.11.2005

For my mother Mantoura
and my father Antoine
Danksagung
Ich möchte mich ganz herzlich bei Herrn Prof. Dr. rer. nat. Hartmut Hillmer für
die Themenstellung und die Betreuung dieser Arbeit bedanken. Weiterhin bedanke
ich mich bei ihm für sein Engagement, die Schaffung der notwendigen Arbeitsvoraus-
setzungen und sein stets förderndes Interesse.
Herrn Prof. Dr. -Ing. Peter Meißner danke ich für die fruchtbare Kooperation zwi-
schen dem INA und der Technischen Universität Darmstadt, sowie für die Referierung
meiner Arbeit und die wissenschaftliche Unterstützung. Den Mitarbeitern seines Insti-
tutes, Herrn Frank Riemenschneider und Herrn Hubert Halbritter danke ich ebenfalls.
Mein weiterer Dank gilt Prof. Dr. rer. nat. Rainer Kassing für seine wissenschaftlichen
Hilfestellungen und seine persönliche Unterstützung.
Ohne die tatkräftige Unterstützung aus dem Technikbereich und dem Sekretariat des
INA, den Damen Heike Schröter-Hohmann, Ina Kommallein und Ina Wensch, sowie
den Herren Dietmar Gutermuth, Albert Malkomes, Michael Plätzer und Dirk Albert,
wäre die Durchführung meiner Experimente und die Anfertigung dieser Arbeit nicht
möglich gewesen. Ihnen allen gebührt mein besonderer Dank.
Meinen Kollegen und Freunden Dr. -Ing. Jürgen Daleiden, Sören Irmer und Fried-
hard Römer danke ich für zahllose fruchtbare Diskussionen, fachliche Unterstützung
und Hilfestellung bei der Durchführung von Experimenten.
Dr. -Ing. Martin Strassner unterstützte mich durch Überlassung von Probenmaterial
und Einweisung in Fertigungsprozesse.
Weiterhin möchte ich mich bei Herrn Prof. Dr. rer. nat. Josef Salbeck und seiner
Forschungsgruppe für die Zusammenarbeit bedanken. Namentlich möchte ich Herrn
Dr. rer. nat. Fuhrmann-Lieker und Herrn Till Spehr erwähnen.
Mein Dank gilt Herrn Ludwig Söhngen, der mir bei der Aufnahme meines Studiums in
Deutschland und bis heute stets zur Seite stand.
Mein ganz besonderer Dank gilt meiner verstorbenen Mutter Mantoura, die mich im
Verlauf dieser Promotionsarbeit sicher oft vermisst hat, sowie meinem Vater Antoine,
der mir immer zur Seite stand und mich unterstützte.
Abstract
The rapid growth of the optical communication branches and the enormous demand for
more bandwidth require novel networks such as dense wavelength division multiplexing
(DWDM). These networks enable higher bitrate transmission using the existing optical
fibers.
Micromechanically tunable optical microcavity devices like VCSELs, Fabry-Pérot filters
and photodetectors are core components of these novel DWDM systems. Several air-
gap based tunable devices were successfully implemented in the last years. Even though
these concepts are very promising, two main disadvantages are still remaining. On the
one hand, the high fabrication and integration cost and on the other hand the undesired
adverse buckling of the suspended membranes.
This thesis addresses these two problems and consists of two main parts:
• PECVD dielectric material investigation and stress control resulting in mem-
branes shape engineering.
• Implementation and characterization of novel tunable optical devices with tailored
shapes of the suspended membranes.
For this purposes, low-cost PECVD technology is investigated and developed in detail.
The macro- and microstress of silicon nitride and silicon dioxide are controlled over a
wide range.
Furthermore, the effect of stress on the optical and mechanical properties of the sus-
pended membranes and on the microcavities is evaluated. Various membrane shapes
(concave, convex and planar) with several radii of curvature are fabricated. Using
this resonator shape engineering, microcavity devices such as non tunable and tunable
Fabry-Pérot filters, VCSELs and PIN photodetectors are succesfully implemented. The
fabricated Fabry-Pérot filters cover a spectral range of over 200 nm and show resonance
linewidths down to 1.5 nm.
By varying the stress distribution across the vertical direction within a DBR, the shape
and the radius of curvature of the top membrane are explicitely tailored. By adjusting
the incoming light beam waist to the curvature, the fundamental resonant mode is
supported and the higher order ones are suppressed. For instance, a tunable VCSEL
with 26 nm tuning range, 400µW maximal output power, 47 nm free spectral range
and over 57 dB side mode suppresion ratio (SMSR) is demonstrated.
Other technologies, such as introducing light emitting organic materials in microcavit-
ies are also investigated.
Zusammenfassung
Die faseroptische Datenübertragung hat sich als äußerst vielversprechendes Verfahren
erwiesen, um den steigenden Ansprüchen der sich rasant entwickelnden Informations-
technik gerecht zu werden. Dies trifft besonders auf die Übertragungsbandbreite und
-bitrate zu. Die Übertragungsbandbreite des bereits vorhandenen Einkanalsystems
reicht jedoch nicht aus, um den steigenden Bedarf an Datenübertragung zu decken. Die
Wellenlängenmultiplextechnik (engl. dense wavelength division multiplexing, DWDM)
ermöglicht eine Erhöhung der Übertragungskapazität dieser Faserstrecken, ohne dass
zusätzliche Faserverlegungen notwendig sind.
Für neuartige DWDM-Systeme bieten abstimmbare vertikale mikro -opto -elektro -
mechanische (engl. micro-opto-electro-mechanical, MOEM) Bauelemente entscheidende
Vorteile. Zu diesen Bauelementen zählen auf der Sendeseite Vertikalresonator emittie-
rende Luftspalt-Laser (engl. air-gap vertical cavity surface emitting laser, VCSEL)
und empfängerseitig Luftspalt-Filter und Luftspalt-Photodetektoren (engl. air-gap fil-
ter, air-gap receiver).
Herkömmliche kantenemittierende Laserdioden weisen Chip-Faser Koppelverluste und
eine große Divergenz des austretenden Lichts auf. Oberflächenemittierende Mikro-
laser mit vertikalem Resonator (VCSEL) ermöglichen demgegenüber eine verbesserte
Koppeleffizienz in eine Glasfaser durch geringere Divergenz, symmetrische Modenpro-
file und eine flexible laterale Strukturierung der Austrittsfläche des Lasers. Im Einsatz
befindliche DWDM-Systeme benötigen bis zu 80 Laserdioden mit unterschiedlichen
Trägerwellenlängen auf der Sendeseite und ebenso viele Photodetektoren auf der Emp-
fangsseite. Für diese DWDM-Netze werden abstimmbare Bauelemente benötigt, um
auf die Ausfälle einzelner Kanäle flexibel reagieren zu können. Die Abstimmung der
Laserdioden und Filter erfolgt in dem in dieser Arbeit verfolgten Konzept durch eine
Änderung der Resonatorlänge, realisiert durch thermische oder kapazitive mikromecha-
nische Aktuation einer der beiden Resonatorspiegel.
Obwohl dieses Konzept erhebliche Vorteile in sich bringt, ist die Serienfertigung solcher
Bauelemente bislang immer noch eine technologische, wissenschaftliche und finanzielle
Herausforderung. Ein Schlüsselaspekt ist dabei die gezielte Einstellung optischer und
mechanischer Eigenschaften aktuierbarer Bragg-Reflektoren. Dabei wird die Krüm-
mung der beweglichen Membran gezielt durch eine Verspannung der einzelnen Bragg-
Schichten eingestellt. Dies geschieht, ohne die optischen Eigenschaften des Bauele-
Bmentes negativ zu beeinflussen. Die Entwicklung einer kostengünstigen Implementie-
rungstechnologie ist ebenso eine notwendige Voraussetzung für eine erfolgreiche Serien-
fertigung.
Die vorliegende Arbeit befasst sich mit dieser Problematik und bearbeitet systematisch
zwei Themenbereiche:
• Die Untersuchung der Verspannung und deren Einfluss auf die Membrankrüm-
mung und darausfolgend auf die optischen Eigenschaften des Resonators.
• Die Herstellung und Charakterisierung von mikromechanisch abstimmbaren Bau-
elementen durch eine gezielte Stresseinstellung der oberen Membran.
Zunächst untersucht die vorliegende Arbeit das plasmaunterstützte Abscheideverfahren
(engl. plasma enhanced chemical vapor deposition, PECVD) und diskutiert die daraus
resultierenden Ergebnisse. Die Abscheidung beeinflussende Parameter, wie Gasfluss,
Druck, Temperatur, etc. besitzen einen starken Einfluss auf die optischen Eigenschaften
(Brechungsindex und Absorption), jedoch kaum auf die Verspannung.
Durch eine Modulation der Plasmaanregungsfrequenzen des Abscheideverfahrens kann
hingegen eine gezielte Verspannungseinstellung der dielektrischen Bragg-Schichten er-
reicht werden. Die Untersuchung dieser Verspannung erfolgte parallel mit einem Ver-
fahren, das die Verspannung im makroskopischen Bereich erfasst sowie mit einer Meth-
ode, die die Mikroverspannung analysiert.
Die Makroverspannung ergibt sich jeweils aus der Messung der Krümmung vor und
nach der Beschichtung eines Wafers. Daraus resultiert eine Kenngröße, welche über
die gesamte Länge entlang einer Richtung auf dem Wafer gemittelt wird. Diese Meth-
ode erlaubt allerdings keine Ortsauflösung der Verspannung. Aber gerade dies ist
besonders bedeutend für die Mikrokomponenten, die sich an unterschiedlichen Orten
auf dem Wafer befinden.
Die Mikroverspannung wird in dieser Arbeit mit Mikrostrukturen analysiert, welche
durch einen oberflächenmikromechanischen Prozess (engl. surface micromachining)
mit konventionellem Photolack als Opferschicht hergestellt wurden.
Beide Methoden wurden verglichen und auf ihre Relevanz untersucht. Dabei konnte
gezeigt werden, dass die Plasmaanregungsfrequenz eine wesentliche Rolle bei der Ver-
spannungseinstellung von Siliziumnitrid spielt. So wurden mit Siliziumnitrid (300 ◦C)
Makroverspannungen zwischen -300 MPa (Zugverspannung) und +850 MPa (Druck-
verspannung) erreicht. Bei niedrigen Temperaturen (60 ◦C) weist die Verspannung von
Siliziumnitrid ähnliche Zusammenhänge auf, während die Verspannung von Silizium-
dioxid in beiden Temperaturbereichen nur eine sehr schwache Frequenzabhängigkeit
zeigt. Die Untersuchung der Mikroverspannung bestätigte die Ergebnisse der Mak-
roverspannungsuntersuchung.
CWeiterhin ist der Einfluss der Verspannung auf die Krümmung der Membran und
deren Auswirkung auf die Geometrie des Resonators systematisch untersucht worden.
Dank dieser Erkenntnisse ist es gelungen, gezielte Membranformen (konkav, konvex und
planar) mit verschiedenen Krümmungsradien zu entwickeln um damit diverse Resona-
torgeometrien zu realisieren.
Zahlreiche nicht abstimmbare und abstimmbare Bauelemente mit einer angepassten
Resonatorgeometrie konnten im Rahmen dieser Arbeit gezeigt werden. Die realisierten
Fabry-Pérot Filter mit einem Luftspalt und einer gekrümmten Membran weisen eine
spektrale Spannweite von über 200 nm und Linienbreiten bis 1.5 nm auf.
Mit derselben Technologie wurden im Rahmen einer engen Zusammenarbeit mit der
Technischen Universität Darmstadt (Fachgebiet Optische Nachrichtentechnik) und dem
Königlichen Technologischen Institut in Schweden (KTH) neuartige abstimmbare opti-
sche Bauelemente wie Fabry-Pérot Filter, VCSEL und Photodetektoren angefertigt.
Der monolithische Fabry-Pérot Filter ist thermoelektrisch durch mäanderförmigeMikro-
widerstände aus Chrom abstimmbar, welche auf die gekrümmten Membran aufgebracht
sind. Dabei wurde ein Abstimmbereich von 15 nm bei 1 mA Stromfluss und 2 kΩ
Chrom-Dünnfilmwiderstand erzielt.
Durch eine gezielte Verspannungseinstellung und die sich ergebende Krümmung der
oberen Membran eines abstimmbaren VCSELs konnte der fundamentale Mode ge-
zielt unterstützt und so eine hohe Nebenmodenunterdrückung erreicht werden. Der
realisierte Laser besitzt einen Abstimmbereich von über 26 nm und eine Seitenmode-
unterdrückung von 57 dBm.
Durch eine Kombination eines abstimmbaren Fabry-Pérot Filters mit einem Photo-
detektor, ist ein spektral selektiver Empfänger für DWDM Systeme entstanden. Hier
konnte eine Abstimmbarkeit der spezifischen Empfindlichkeit (0.3 A/W) über den freien
Spektralbereich (> 35 nm) hinaus mit einer Einfügedämpfung von 3.5 dB demonstriert
werden.
Die mit dieser Technologie erzielten Ergebnisse dienten als Motivation, um neue Mate-
rialien und Ansätze zu untersuchen. So wurde ein organisch lichtemittierendes Material
(Makromoleküle) in eine Mikrokavität erfolgreich eingesetzt und eine Linienbreitenver-
engung erreicht.
Im Rahmen dieser Arbeit ist es gelungen, Verspannungen gezielt einzustellen, um
beliebige Resonatorgeometrien anzufertigen. Abstimmbare Mikrokavitätsbauelemente
mit optimierten Resonatoren wurden so erfolgreich demonstriert. Die Technologie auf
PECVD-Basis bildet dabei die Grundlage eines kostengünstigen Produktionsprozesses.
DErwähnt sei an dieser Stelle ein redaktioneller Bericht über Teile dieser Arbeit in
"Photonics Spectra" (Juni 2004), der das internationale Interesse an diesemForschungs-
thema dokumentiert.
Weitere Vertiefungen dieser Arbeit könnten zu einer Industrialisierung dieser Techno-
logie führen und dadurch zur Massenproduktion kostengünstiger Mikrokavitätsbauele-
mente wie beispielsweise ein monolithischer abstimmbarer elektrisch gepumpter VC-
SEL. Ein Schwerpunkt weiterer Forschungen könnte die Herstellung von passiven und
aktiven organisch basierenden abstimmbaren Elementen sein, welche in verschiedenen
Bereichen wie der Sensorik, der Medizintechnik und der Unterhaltungstechnologien
eingesetzt werden könnten.
List of own journal publications and
conference papers
1. J. Daleiden, M. Strassner, N. Chitica, J. Pfeiffer, A. Tarraf, F. Römer, B. Ayele,
C. Prott, and H. Hillmer, "Micromachined Multi-Airgap Bragg Mirrors for VC-
SELs," European Semiconductor Laser Workshop, Berlin, September 2000.
2. J. Daleiden, N. Chitica, M. Strassner, C. Prott, F. Römer, A. Tarraf, and
H. Hillmer, "Micromechanically Tunable Optical Devices," Summer School and
European Optical Society Topical Meeting on Semiconductor Microcavity Photon-
ics, Technical Digest, Ascona, October 2000.
3. J. Daleiden, A. Tarraf, E. Ataro, N. Chitica, S. Irmer, C. Prott, F. Römer, M.
Strassner, H. Hillmer, "Micromechanically Tunable Filters for Optical Commu-
nication Systems," Proceedings of the 1st Int. Workshop on Optical MEMS and
Integrated Optics, Dortmund, June 2001.
4. J. Daleiden, S. Irmer, H. Hohmann, D. Feili, A. Tarraf, E. Ataro, F. Römer, C.
Prott, H. Hillmer, "Trockenätzen von phosphidischen III-V Halbleitern," Work-
shop dry etching for III-V semiconductors, Ulm, April 2001.
5. H. Hillmer, C. Prott, F. Römer, A. Tarraf, S. Irmer, M. Strassner, N. Chitica,
and J. Daleiden, "Continuously tunable 1.55 µm vertical air cavity filters and
VCSELs using micromachined electrostatic actuation," European Semiconductor
Laser Workshop, Berlin, September 2001.
6. J. Daleiden, A. Tarraf, F. Römer, N. Chitica, S. Irmer, C. Prott, M. Strassner,
H. Hillmer, "Continuously Tunable Air Gap Devices," Proceedings of the IEEE
/ LEOS International Conference on Optical MEMS, Technical Digest, Okinava,
2001.
7. J. Daleiden, V. Rangelov, S. Irmer, F. Römer, M. Strassner, C. Prott, A. Tarraf,
H. Hillmer, "Record tuning range of InP-based multiple air-gap MOEMS filter,"
Electronic Letters, vol.38, no.21, pp.1270-1271, October 2002.
8. A. Tarraf, S. Irmer, E. Ataro, J. Daleiden, D. Gutermuth, C. Prott, V. Rangelov,
F. Römer, H. Schröter-Hohmann, H. Hillmer, "Bionik: Erfolgsrezepte der Natur-
Anwendungsbeispiele in der Mikrosystem- und Datenübertragungstechnik," Tag
II
des Wissenschaftlichen Nachwuchses, Universität Kassel, 2002 (POSTERPREIS
2002)
9. J. Daleiden, S. Irmer, A. Tarraf, F. Römer, C. Prott, E. Ataro, M. Strassner,
H. Hillmer, " Multiple air-gap InP-based VCSELs and filters with ultra-wide
wavelength tuning - flexibility and shape of the membranes," Proceedings of the
Optical MEMS Conference, Hawai, August 2003.
10. F. Römer, C. Prott, J. Daleiden, S. Irmer, M. Strassner, A. Tarraf, H. Hillmer,
"Micromechanically tunable air gap resonators for long wavelength VCSEL’s,"
IEEE LEOS International Semiconductor Laser Conference, Garmisch, Septem-
ber 2003.
11. A. Tarraf, J. Daleiden, F. Römer, C. Prott, V. Rangelov, S. Irmer, E. Ataro
and H. Hillmer, "A novel low-cost tunable dielectric air-gap filter," Proceedings
of the Optical MEMS Conference, Lugano, August 2003.
12. M. Joodaki, A. Tarraf, M. Salih, D. Albert, H. Schröter-Hohmann, W. Scholz,
G. Kompa, H. Hillmer, and R. Kassing, "Improvements of thermal resistance
and thermal stress in quasi-monolithic integration technology (QMIT) with a
new fabrication process," European Microwave Week, Milan, September 2002.
13. H. Hillmer, J. Daleiden, C. Prott, F. Römer, A. Tarraf, S. Irmer, V. Rangelov,
S. Schüler, M. Strassner, "Ultra-wide continuously tunable 1.55 µm vertical air-
cavity filters and VCSEL’s based on micromachined electrostatic actuation," Pro-
ceedings of the SPIE, series 4646 (Photonics West), San Jose, 2002.
14. H. Hillmer, J. Daleiden, C. Prott, F. Römer, S. Irmer, V. Rangelov,A. Tarraf, S.
Schüler and M. Strassner, "Potential for micromachined actuation of ultra-wide
continuously tunable optoelectronic devices," Applied Physics B, vol. 75, 2002.
15. M. Strassner, N. Chitica, A. Tarraf, "Investigations of growth conditions for
InP suited for Micro Opto Electro Mechanical Systems for data communication,"
Proceedings of the 14th International Conference on InP and related Materials,
Stockholm, 2002.
16. J. Daleiden, S. Irmer, E. Ataro, C. Prott, V. Rangelov, F. Römer, M. Strassner,
A. Tarraf, and H. Hillmer, "Continuously tunable air-gap micro-cavity devices
for optical communication systems," Proceedings of the SPIE, series 4871 (IT-
COM), Boston, 2002.
17. J. Daleiden, S. Irmer, V. Rangelov, F. Römer, A. Tarraf, C. Prott, M. Strassner
and H. Hillmer, "Record wavelength tuning of 127nm for vertical cavity Fabry-
Pérot filter," Proceedings of the Optical MEMS Conference, Lugano, August 2002.
18. M. Strassner, C. Luber, A. Tarraf, N. Chitica, "Widely tunable - constant
bandwidth monolithic Fabry-Pérot filter with a stable cavity design for WDM
III
systems," IEEE Photonics Technology Letters, vol. 14, No. 11, pp. 1548-1550,
2002.
19. H. Hillmer, J. Daleiden, S. Irmer, F. Römer, C. Prott, A. Tarraf, M. Strassner,
E. Ataro and T. Scholz, "Potential of micromachined photonics: miniaturization,
scaling and applications in continuously tunable vertical air-cavity filters," Pro-
ceedings of the SPIE, series 4947 (Photonics Fabrication Europe), pp. 197-211,
2002.
20. S. Irmer, J. Daleiden, V. Rangelov, C. Prott, F. Römer, M. Strassner, A. Tar-
raf, H. Hillmer, "Continuously tunable InP based multiple air-gap MOEMS fil-
ters with ultra wide tuning range", Proceedings of SPIE, Series 4945 (Photonics
Fabrication Europe), 2002.
21. C. Prott, F. Römer, E. Ataro, J. Daleiden, S. Irmer, M. Strassner, A. Tarraf,
H. Hillmer, "Model calculations of vertical cavity air-gap filters and VCSEL’s
for ultra-wide continuous tuning," International Conference on Numerical Sim-
ulation of Semiconductor Optoelectronic Devices (NUSOD), Zürich, September
2002.
22. F. Römer, C. Prott, S. Irmer, J. Daleiden, A. Tarraf, H. Hillmer, "Tuning effi-
ciency and linewidth of electrostatically actuated multiple air-gap filters,"Applied
Physics Letters, vol. 82, no. 2, pp. 176-178, January 2003.
23. H. Hillmer, J. Daleiden, C. Prott, S. Irmer, F. Römer, E. Ataro, A. Tarraf, H.
Rühling, M. Maniak, M. Strassner, "Bionics: Precise color tuning by interference
in nature and technology - applications in surface-micromachined 1.55¸tm vertical
air-cavity filters," Proceedings of the SPIE, series 4983 (Photonics West), pp.
203-214, 2003.
24. S. Irmer, J. Daleiden, V. Rangelov, C. Prott, F. Römer, M. Strassner,A. Tarraf,
H. Hillmer, "Ultra low biased widely continuously tunable Fabry-Pérot Filter,"
IEEE Photonics Technology Letters, vol. 15, no. 3, pp 434-436, March 2003.
25. A. Tarraf, J. Daleiden, S. Irmer, V. Rangelov, F. Römer, C. Prott, E. Ataro,
H. Hillmer, T. Fuhrmann, T. Spher, J. Salbeck, "A novel low-cost and simple
fabrication technology for tunable dielectric active and passive optical air-gap
devices," Proceedings of the SPIE, series 4945 (Photonics fabrication Europe),
pp. 9-20, 2003.
26. J. Daleiden, A. Tarraf, S. Irmer, F. Römer, C. Prott, E. Ataro, M. Strassner,
H. Hillmer, "Wide and continuous wavelength tuning of microcavity devices for
optoelectronic applications," Journal of Microlithographie, Microfabrication, Mi-
crosystems, vol. 2 no. 4, pp. 265-274, October 2003.
27. C. Prott, F. Römer, E. Ataro, J. Daleiden, S. Irmer,A. Tarraf, H. Hillmer, Mod-
eling of Ultra-Widely Tunable Vertical Cavity Air-Gap Filters and VCSEL’s,"
IV
IEEE Journal of Selected Topics in Quantum Electronics, vol. 9, no. 3, pp.
918-928, 2003.
28. H. Hillmer, J. Daleiden, C. Prott, F. Römer, S. Irmer, E. Ataro, A. Tarraf,
D. Gutermuth, I. Kommallein, and M. Strassner, "Ultra-wide continuously tun-
able 1.55 µm vertical air-cavity wavelength-selective elements for VCSELs using
micromachined electrostatic actuation," Proceedings of the SPIE, series 4871 (IT-
COM), Boston 2003.
29. H. Halbritter, M. Aziz, F. Riemenschneider,A. Tarraf, M. Strassner, O.P. Daga,
and P. Meißner, "Performance Evaluation of WDM Components based on Tun-
able Dielectric Membrane Technology", IEE Circuits, Devices and Systems, vol.
150, no. 6, pp. 501-505, December 2003
30. H. Halbritter, C. Dhanavantri, M. Strassner, A. Tarraf, M. Aziz, F. Riemen-
schneider, S. Syguda, B.R. Singh, and P. Meißner, "Tunable and wavelength
selective PIN diodes", Proceedings of SPIE, series 5277 ( Microelectronics, MEMS
and Nanotechnolgoy), pp. 129-137, December 2003
31. H. Halbritter, F. Riemenschneider, M. Strassner, A. Tarraf, I. Sagnes, and P.
Meißner, "Properties of micromechanically tunable VCSEL," Proceedings of the
SPIE, series 5277 (Microelectronics, MEMS, and Nanotechnology), pp. 292-301,
December 2003
32. H. Halbritter, F. Riemenschneider, S. Syguda, C. Dhanavantri, M. Strassner, A.
Tarraf, B.R. Singh, I. Sagnes and P. Meißner, "Tunable and wavelength selective
pin photodiode," Electronics Letters, vol. 40, no. 6, pp. 388-390, March 2004.
33. A. Tarraf, F. Riemenschneider, M. Strassner, J. Daleiden, S. Irmer, H. Hal-
britter, H. Hillmer, and P. Meißner, "Continuously Tunable 1.55 t¸m VCSEL im-
plemented by Precisely Curved Dielectric Top DBR Involving Tailored Stress,"
IEEE Photonics Technology Letters, vol. 16, no. 3, pp. 720-722, March 2004.
34. A. Tarraf, J. Daleiden, S. Irmer, D. Prasai and H. Hillmer, "Stress investigation
of PECVD dielectric layers for advanced optical MEMS," Journal of Micromech-
anics and Microengineering, vol. 14, no. 3, pp. 317-323, 2004.
35. S. Irmer, K. Alex, J. Daleiden, I. Kommallein, M. Oliveira, F. Römer,A. Tarraf,
H. Hillmer, "Surface micromachined optical low-cost all-air-gap filters based on
stress optimized Si3N4 layers," Journal of Micromechics and Microengineering,
vol. 15, no. 4, pp. 867-872, April 2005.
36. H. Halbritter, F. Riemenschneider, B. Kögel, A. Tarraf , M. Strassner, S. Irmer,
H. Hillmer, I. Sagnes, and P. Meißner, "MEMS-Tunable andWavelength Selective
Receiver Front End," Proceedings of the 18th IEEE Conference on Micro Electro
Mechanical Systems, pp. 68-71, Miami, Jannuary 2005
V37. A. Tarraf, M. Nestler, S. Martin, U. Poll, D. Roth, J. Dienelt, H. Neumann,
B. Rauschenbach, S. Irmer, F. Römer, V. Daneker, H. Hillmer, "Dual Ion Beam
Sputter Deposition System for EUVL Masks," XUV Technologies and Applica-
tions workshop, Bad Honnef, June 2004.
38. A. Tarraf, M. Nestler, H.-U. Poll, D. Roth, J. Dienelt, H. Neumann, B. Rauschen-
bach, S. Irmer, F. Römer, H. Hillmer, "A novel dual ion beam sputter deposition
system for implementing low defect density and high-quality EUVL masks," 3rd
International EUVL Symposium, Miyazaki, November 2004.
39. A. Tarraf, M. Nestler, H.-U. Poll, S. Martin, D. Roth, J. Dienelt, H. Neumann,
E. Schubert, B. Rauschenbach, M. Schulze, V. Daneker, S. Irmer, F. Römer, H.
Hillmer, "Innovative concept for implementing particle free EUVLmasks by novel
dual ion beam sputter deposition systems," SPIE Photo Mask Japan, Yokohama,
April 2005.
40. A. Tarraf, "Organic and inorganic tunable photonic micro-cavity devices for
optical communications," Surface Technologies with Plasma and Ion beam Pro-
cesses, Mühlleithen , March 2004.
41. J. Dienelt, H. Neumann, F. Scholze, E. Schubert, B. Rauschenbach, M. Nestler,
A. Tarraf, M. Schulze, "In-situ ellipsometry and beam profile controlled linear
ion beam source - screening for the IBD EUV-mask blank deposition," MNE
2004, Rotterdam, September 2004.
42. A. Tarraf, "Particle free MoSi mask blanks for the EUVL: Implementation and
Metrology," Surface Technologies with Plasma and Ion beam Processes, Mühl-
leithen , March 2005.
43. H. Hillmer, A. Tarraf, S. Irmer, F. Riemenscheider, H. Halbritter, F. Römer,
J. Daleiden, E. Ataro, C. Prott, M. Strassner, A. Hasse, S. Hansmann, and P.
Meißner, "Wide continuously tunable 1.5mm vertical air-cavity wavelength select-
ive elements for filters and VCSELs using micromachined actuation," Proceedings
of the SPIE, Series 5825A (OPTO Irland 2005), April 2005.
44. J. Dienelt, H. Neumann, M. Kramer, E. Schubert, F. Scholze, M. Nestler, A.
Tarraf, M. Schulze, and B. Rauschenbach, "EUV-mask blanks by ion beam
sputter deposition: A novel tool concept and first results," 4th International
EUVL Symposium, San Diego (USA), November 2005.
Patents submissions
1. H. Hillmer, J. Daleiden, C. Prott, J. Daleiden, F. Römer, S. Irmer, D. Gutermuth,
A. Tarraf, E. Ataro, "Optimierung eines mikromechanischen Bauelements mit
Hilfe eines Ätzstern," DE: AZ 103 57 421.2-33
2. S. Irmer, J. Daleiden,A. Tarraf, H. Hillmer "Leitfähige Haltepfosten in luftspalt-
basierenden, optoelektronischen Bauelementen," DE: AZ 10353546.2
VI
3. H. Hillmer, J. Daleiden, C. Prott, F. Römer, A. Tarraf, S. Irmer, S. Schüler, V.
Rangelow "Bauelement mit "chirped" DBR-Spiegeln," PCT: AZ PCT / DE2004
/ 000605, DE: AZ 10318767.7
Contents
1 Background and Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 State of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Brief presentation . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 Highlights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Introduction and content . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Basics of tunable microcavities 8
2.1 Optical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 Bragg reflectors . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Fabry-Pérot filter . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.3 Resonator stability . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.1.4 Active devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.5 Photodetector devices . . . . . . . . . . . . . . . . . . . . . . . 18
2.2 Mechanical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2.1 Micromechanical tuning . . . . . . . . . . . . . . . . . . . . . . 19
2.2.2 Stress in thin films . . . . . . . . . . . . . . . . . . . . . . . . . 21
3 PECVD technology: basics and material properties 25
3.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1 Configuration of the involved PECVD . . . . . . . . . . . . . . 25
3.1.2 Background of PECVD layer deposition . . . . . . . . . . . . . 26
3.2 Material investigation (stress, composition and optical properties) . . . 28
3.2.1 Silicon nitride . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.2 Silicon dioxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.3 Bragg mirrors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4 Technology for air-gap based microcavity devices 36
4.1 Bulk and surface micromachining . . . . . . . . . . . . . . . . . . . . . 36
4.1.1 Photoresist based novel technology . . . . . . . . . . . . . . . . 37
4.1.2 PECVD dielectric membrane . . . . . . . . . . . . . . . . . . . 37
5 PECVD stress engineering 40
5.1 Stress control of PECVD dielectric material . . . . . . . . . . . . . . . 40
5.2 Impact of stress on optical and mechanical layer properties . . . . . . . 46
i
ii Contents
5.2.1 Cavity length . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2.2 Radius of curvature . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.2.3 Filter characteristics . . . . . . . . . . . . . . . . . . . . . . . . 52
6 Results of microcavity devices 56
6.1 Non tunable Fabry-Pérot filter . . . . . . . . . . . . . . . . . . . . . . . 56
6.1.1 Solid stack filters . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.1.2 Air-gap filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.2 Tunable Fabry-Pérot filter . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.3 Non tunable VCSELs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.4 Tunable VCSEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.5 Tunable high end receiver . . . . . . . . . . . . . . . . . . . . . . . . . 68
7 Related applications 71
7.1 Organic microcavities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.1.1 PECVD materials and DBRs . . . . . . . . . . . . . . . . . . . 71
7.1.2 Organic half cavity . . . . . . . . . . . . . . . . . . . . . . . . . 74
8 Conclusion 80
Bibliography 81
A State of the art: detailed description 97
A.1 Fabry-Pérot filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
A.2 Vertical cavity surface emitting laser . . . . . . . . . . . . . . . . . . . 99
B Diagram supplements 103
B.1 PECVD technological investigation . . . . . . . . . . . . . . . . . . . . 103
B.2 Stress investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
B.3 Stress impact on mechanical and optical properties . . . . . . . . . . . 108
C Technological process flow 111
C.1 Air-gap Fabry-Pérot filters . . . . . . . . . . . . . . . . . . . . . . . . . 111
C.1.1 Design IMA2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
C.1.2 Design IMA3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
C.2 Process flow of the tunable Fabry-Pérot filter . . . . . . . . . . . . . . . 115
C.3 Process flow of the non tunable VCSEL . . . . . . . . . . . . . . . . . . 117
D DBRs for the blue wavelength range 118
E Abbreviations 119
F Stress impact on the ROC, Lcav, RW and FWHM 121
Chapter 1
Background and Introduction
1.1 Background
Who has not made this experience before? You just bought a mobile phone a few
weeks ago and the next model with more advanced features is already available on
the market for the same price! This trend can also be observed in other technological
branches. Thus, faster personal computers and internet, novel cars electronics and
optoelectronics, better medical and telecommunications technologies and smaller, more
sophisticated personal electronics as well as highly precise sensors are needed in order
to satisfy the demands of the society in the 21th century. Furthermore, these more
stringent requirements are expected at low-cost.
However, in the last decades, the main focus was on information techology. Thus
at the beginning of the 19thcentury, a worldwide telephone network based on copper
cables was established. The main purpose was the transmission of voice and text data.
To meet the increasing demand for more capacity, high frequency communication (via
coaxial cable) was introduced around 1940. With the first semiconductor computer in
the early 1980s, a new area of semiconductor based communication and data transfer
was ushered in. The rapid development of the semiconductor industry, resulted in
the demand for large data transmission. This leads to the further development of the
local networks (e.g. firstly within a company) and eventually to the a world wide web
(WWW).
Nowadays, with the internet well established as the main communication tool world-
wide (for private as well as for business purposes), a huge capacity for data transmission
is needed. The more people are involved, the more data should be processed and the
more bandwidth is required. In order to transmit the huge volumes of data, optical
communication via glass fibers is used.
The introduction of the erbium doped fiber amplifier (EDFA) at the beginning of
the 1990s enabled the development of the wavelength division multiplexing (WDM) and
dense wavelength division multiplexing (DWDM) technologies, that ensure transmis-
sion rates up to 1 Tbp/s. This technology exploits the bandwidth of the existing fiber
2 1.1 Background
network in a efficient manner without deploying further ones. Thus, this technology
recently became very attractive for long distance data transmission. An introduction
to the DWDM technology can be found elsewhere [1].
The boom of the optical networking [2] market in 2000 lead to overcapacity which
consequently cooled down this market at the end of the year [3]. Due to this enormous
overcapacity of the telecommunication industry, the infrastructure investment shrank
worldwide [4]. A recent surge in optical communication was seen in the metropolitan
area networks (MAN), where capacity on demand and dynamic network flexibility play
major roles. Deploying more fiber is the most economical solution for transmission
distances under 20 km, whereas the metro DWDM represents the optimum economic
solution1 above 50 km [5] (The metropolitan network is considered to be up to 150 km
[6]).
Many market analyses forecast a growth in the investments in the metro DWDM
systems. For instance, Pioneer Consulting predicts an increase in market opportunities
from 1.4 billions dollars in 2001 to 12.6 billions by 2006. According to a market study by
ElectroniCast Corp. [7], which presents a forecast for the consumption of major DWDM
components and optical add/drop multiplexers, the growth will reach the amount of
34.6 billions dollars in 2009 (54% North America, 21% Europe, 23% Japan and Pacific
Rim and 2% the rest of the world). According to this study, the consumption of
components was in the order of 2.85 billions dollars in the year 2000 ( 17% Europe,
13% Japan and Pacific Rim, 70% North America and 1% the rest of the world).
The report by KMI Corp. [8] predicts a compound annual growth rate (CAGR)2 of
43% in 2005 for DWDM systems, which corresponds to a market volume of 54 billions
of dollars. However, by 2005, the short distance investments will represent only 18%
of the mentioned market.
Further development of the market is conditional on the implementation of low-cost
solutions [6]. The cost of the components is essential in deciding which kind of systems
should be used. The adoption of coarse wavelength division multiplexing (CWDM) or
DWDM, for instance, for metropolitan networks depends mainly on the costs of optical
filters [6]. Even though that DWDM filters are key components in the novel DWDM
systems, their per-channel cost is lower than those of the transmitters and receivers.
Stephen Montgomery, ElectroniCast president, noticed ”while the per-channel cost
of transmitter/receiver (T/Rs) pairs at a fixed wavelength will drop more rapidly than
filter cost, there will be a strong countering trend toward deployment of higher data
rate, more expensive T/Rs” [7]. The consumption forecast of DWDM components in
the ElectroniCast study shows that 62% of all the DWDM components in 2000 are
1Best solution for both, overall costs and costs per wavelength.
2Rate at which a given present value would grow to a given future value within a difined period
of time. CAGR =
¡
FV
PV
¢ 1
ny − 1, where FV is the future value, PV is the present value and ny is the
time period in years
1 Background and Introduction 3
T/Rs and integrated optoelectronics (wavelength converter 6%, integrated optoelec-
tronics smaller than 1%, multiwavelength transmitter 50%, DWDM filter module 13%,
optical amplifier 19% and multichannel receiver 12%). In the year 2009, these modules
will reach 85% of the needed DWDM components (wavelength converter 1%, integ-
rated optoelectronics 37%, multiwavelength transmitter 39%, DWDM filter module
8%, optical amplifier 5% and multichannel receiver 9%).
Micro-electro-mechanical-systems (MEMS) and micro-opto-electro-mechanical sys-
tems (MOEMS) are potential key components of an effective low-cost solution for the
metropolitan networks. However, these devices need to satisfy severe requirements set
by the international telecommunication union (ITU). In contrast to the amplifiers and
the receiver photodiodes, the transmitter and filter modules of the DWDM systems
should operate very precisely according to the stringent requirements [7]. Thus, the
transmitter and filter sets should provide several operating wavelengths with precise
channel spacing and very narrow spectral linewidth. A major challenge worldwide is
the development and implementation of low-cost components fulfilling these tough re-
quirements. One solution for reducing the overall costs is to utilize tunable devices,
allowing a probable wavelength shift to be spectrally corrected without replacing these
components. This solution avoids the expensive costs incurred by having to store the
same modules operating with different wavelengths. Another approach reducing the
costs is to establish a low-cost technology for implementing these tunable devices.
1.2 State of the art
1.2.1 Brief presentation
The simplest microcavity device, presented during this thesis, is a Fabry-Pérot filter
(FPF). It consists of a microcavity (solid or air) embedded between two distributed
Bragg reflectors (DBRs). Each DBR is commonly fabricated by alternating two ma-
terials (the optical thickness is a quarter wavelength) having high and low refractive
indices (nH and nL). A DBR consists of several periods and each period comprises
two different layers (Figure 1.1). In this case, an in-phase construction of reflections at
the consecutive interfaces between the layers ensures a strong reflection at the desired
wavelength and a stop band can be formed.
A Fabry-Pérot resonator is shown in Figure 1.2 (a). It consists of two mirrors
and a cavity. In this example, three modes (standing waves) can oscillate within the
resonator. Figure 1.2 (b) shows the modes as spectral lines. Combining high reflective
DBRs (instead of normal mirrors) with a cavity (of length Lcav), both spectra (DBR
and Fabry-Pérot modes) overlap resulting in a single mode selection. Thus an optical
Fabry-Pérot filter is formed. A detailed theoretical study can be found in chapter
2. Figure 1.3 shows a cross section of the optically used part of such a filter as well
as several common lateral designs, which have been frequently used in the literature.
Introducing a light emitting or a light absorbing material in the cavity, VCSELs and
4 1.2 State of the art
photodetectors could be fabricated. By changing the cavity length Lcav, a spectral shift
of the selected mode occurs. This effect is called spectral tuning.
1.2.2 Highlights
Tunable MEMS and MOEMS microcavity devices (e.g. vertical cavity surface emitting
lasers (VCSELs), Fabry-Pérot filters (FPFs) and photodetectors) are the core of new
dynamic dense wavelength division multiplexing (DWDM) systems. Over the past
years, countless works dealing with tunable active and passive microcavities have been
published. Relevant devices are those combining the advantages of the mechanical
tuning with the optimized optical properties.
Concerning tunable Fabry-Pérot filters, the devices differs by the tuning concept and
the technological implementation. Based on the InP material system, Irmer et. al. [9]
present recently a continuously widely tunable InP/air based filter with 140 nm tuning
range. The tuning (electrostatic) voltage is 3.2V, whereas the FWHM is between 3 nm
and 5 nm. A filter with a GaAs/AlAs bottom DBR and an Au/SiNxHy/GaAs top
movable membrane is presented by Larson et. al. [10], and exhibits a tunability of
32 nm by a 14V tuning voltage at a central wavelength of 932 nm. The membrane is
actuated electrostatically and the FWHM of the filter resonance is 3 nm. A hybrid
concept using a semiconductor cavity and dielectric mirrors ist presented by Hohlfeld
et. al. [11]. The filter, based on the thermooptical effect, consists of a silicon cavity
and two Si3N4/SiO2 DBRs and shows a tunability of 3.5 nm and a FWHM of 1.19 nm.
Low-cost all dielectric air-gap tunable filters are rare and difficult to implement due to
the stress induced buckling of the top membrane. A novel 1.55µm Fabry-Pérot filter
with a stable half symmetric cavity design [12], shows a tunability of 70 nm by a tuning
voltage of 14V. The filter consists of two SiO2/TiO2 DBRs separated by an air-gap
cavity. The radius of curvature of the top membrane is around 310µm. The FWHM
of the filter characteristics is 0.27 nm.
Using the same concept, Vakhshoori et. al. [13] presented a 1.55µm tunable cw
VCSEL with 2mW output power and 50 nm tuning range. The novelty of this work
was the implementation of a stable resonator by a curved top DBR with a radius of
curvature (ROC) of 300µm. The tuning occurs electrostatically. However, the authors
did not reveal either the material system nor the technological implementation details of
the curved mirrors. Recently, electrically pumped tunable VCSEL have been presented:
Maute et. al. [14] produced an electrically tunable VCSEL, whose active region has
already been described in [15]. The bottom mirror consists of CaF2/a-Si multilayer
and gold, the top movable mirror is a doped GaAs/AlGaAs DBR. The tuning of the
two-chip based device [16] is done electrothermally. The device shows a single mode
operation across 30 nm tuning with a SMSR better than 30 dB and a maximum output
power of 76µW. By the same technological implementation process, Riemenschneider
et. al. [17] increased the tunability to 40 nm and the output power to 100µW.
1 Background and Introduction 5
Stop band
Wavelength λ
R
ef
le
ct
iv
ity
dH
dL
nH
nL
Period
dH = 0.25λ / nH dL = 0.25 λ / nL
Distributed Bragg reflector
R
ef
le
ct
iv
ity
Figure 1.1: A distributed Bragg reflector (DBR) consists of alternating layers of high and
low refractive indices. The optical thickness of each layer is a quarter wavelength. Due to the
constructive interference, a stop band with high reflection values can be formed.
i ii iii iiiiii
λ
Top mirror
Bottom mirror
Cavity length Lcav
(a) (b)
(c)
Cavity modes
λ
Selected cavity
mode
Fabry-Pérot resonator
Fabry-Pérot resonator with DBR mirrors
Reflection 
spectrum
Top DBR
Bottom DBR
(d)
Cavity
Figure 1.2: (a) Fabry-Pérot resonator with common mirrors and schematically standing
waves (modes) within the cavity, (b) Fabry-Pérot modes (spectral lines), (c) Fabry-Pérot
filter with two DBRs. Due to the DBR spectral stop band, one mode can be selected (overlap
of DBR stop band and Fabry-Pérot resonator modes).
6 1.3 Introduction and content
X
Y A
B B
B B
C A
A
B B
B
C
DBR mirror 1
DBR mirror 2
Cavity (solid or air) A
Optically used 
part of the filters
X
Z
Reflected wavelengths
Transmitted wavelength
B
Supporting posts
C
Suspensions
Figure 1.3: The optically used part of the Fabry - Pérot filter (Z-axis) consists of the cavity
and two DBRs. This direction is called ”vertical direction” during this thesis. The optical
pass (light input and light output) occurs in this direction. In the XY plane (top view), the
lateral design of the filter unit is shown. Several frequently published designs are illustrated.
The optically used part ”A” is positioned by suspensions ”C”. The suspensions are fixed to
the substrate by supporting posts ”B”. The parts ”A” and ”C” may be suspended or attached
to the substrate.
1.3 Introduction and content
Even though the progress in microcavity devices seems to meet soon the stringent
optical communications requirements, two main problems remain. The first one is
the immense costs resulting from the implementation process and the non economic
concepts used, and the second one is the control of the stress induced buckling of the
membranes leading often to unproper operation of these devices.
These two aspects are the core of this thesis. Thus, the aim is to develop a low-cost
technology enabeling the fabrication of such devices in an economic way with high yield
and to control and taylor the buckling of the released membranes. Seven chapters of
this thesis deal with these issues.
In chapter 1, a survey of market analysis is presented to evoke the importance and
the necessity of low-cost microcavity devices. It is shown that these devices claim a
considerable part of the market. However, the economic yield of these devices can be
reached only when a low-cost solution is available.
Chapter 2 presents the theoretical basics of the microcavity devices. In this chapter,
the existing theory is extended and is explained in respect to the devices described in
this thesis.
1 Background and Introduction 7
PECVD technology is investigated in chapter 3, with the goal to indentify a suitable
technology enabling the implementation of novel low-cost devices, able to contribute
to the economic boom of the optical communications. For this a common parallel plate
reactor is used. The dependence of the mechanical (e.g. bulk stress) and the optical
(e.g. refractive index) properties on the reactor parameters (e.g. gas flow, pressure,
temperature, etc.) is investigated. It is shown that the optical properties can be
readily affected by varying these parameters, whereas the mechanical properties (e.g.
bulk stress) are less affected.
The stress control (especially the microstress) requires the development of a novel
microscale technology, presented in chapter 4. Here, the deposition of dielectric PECVD
layers on top of a common photoresist is demonstrated for the first time.
Chapter 5 presents the results of the dependence of macro- and microstress on
plasma excitation frequency. For this purpose, different measurement techniques are
used, and stress control in a wide range is achieved. Furthermore, the effect of stress
on the mechanical properties of the suspended dielectric membranes is discussed.
Thus, differently shaped membranes (planar, convex and concave) with several radii of
curvature are implemented. A wide range of cavity length variation is demonstrated.
In the second part of the thesis, the knowledge acquired during the investigations
is used for the implementation of several microcavity devices. Thus, in chapter 6,
different microdevices like non tunable and tunable vertical cavity surface emitting
lasers (VCSELs), non tunable and tunable Fabry-Pérot filters (FPFs) as well as tunable
PIN photodiodes are presented. Furthermore, based on the results in chapter 6, the
implementation of light emitting organic materials in microcavities is addressed in
chapter 7. First promising results are presented and discussed. Finally, the whole
work of this thesis is concluded in chapter 8.
Chapter 2
Basics of tunable microcavities
2.1 Optical properties
2.1.1 Bragg reflectors
Micromechanically tunable vertical cavity devices for the DWDM systems (e.g. FPFs,
VCSELs and photodetectors) consist among other parts of an air-cavity embedded
between two DBRs. The complete theory of these devices can be found elsewhere
[18—22].
A distributed Bragg reflector consists of a periodic alternating layers of a high (nH)
and low (nL) refractive index material. In analogy to the diffraction of X-ray light at
atomic planes of a solids, the Bragg condition for a periodic multilayer is given by
Λ =
λB
Neff
· Θbr
2
(2.1)
where Λ is the grating period, λB the wavelength in vacuum, Neff the effective re-
fractive index of the alternating materials and Θbr the Bragg order. Constructive
interference occurs when Θbr takes an odd integer values (1, 3, 5, ...). By consider-
ing the physical thickness of the high and low refractive index materials, dH and dL,
respectively, equation 2.1 can be rewritten as
Θbr · λB
4
= dH · nH = dL · nL (2.2)
A frequent method for calculating the reflectivity of a DBR is the transmission or
transfer matrix method. This considers the electric (E) and magnetic (H) fields of a
transverse electromagnetic wave propagating in a layer (see Figure2.1).
By using the Maxwell´s equations, the following transmission matrix for the layer
b can be found [19]
µ
E (0)
ηoH (0)
¶
=
Ã
cos (kbdb + iαbdb) inb sin (kbdb + iαbdb)
inb sin (kbdb + iαbdb) cos (kbdb + iαbdb)
!
·
µ
E (db)
ηoH (db)
¶
(2.3)
2 Basics of tunable microcavities 9
Layer a Layer b Layer c
Ex
Hy
0 db z
Figure 2.1: A layer b is embedded between two layers, a and c, respectively. The layer has a
physical length db and a refractive index nb. The electromagnetic fields at positions 0 and db
are E(0), H(0), E(db) and H(db), respectively. The direction Z is often described as vertical
direction during this thesis.
E (0) is the electric field at position 0
H (0) is the magnetic field at position 0
ηo is the impedance of the free space
nb is the refractive index of layer b
db is the physical thickness of the layer b
αb is the amplitude absorption coefficient in layer b
kb = 2π/λ0nb is the phase propagation constant in the layer
λ0 the wavelength in the free space
For a DBR with p periods and 2p layers and a total length LDBR, a total transmis-
sion matrix Mtotal can be introduced. Thus, we can writeµ
E (0)
H (0)
¶
=Mtotal ·
µ
E (LDBR)
H (LDBR)
¶
(2.4)
Mtotal =
Y
Mlayers =
µ
m11 m12
m21 m22
¶
(2.5)
With Y0 and Ys the wave admittances of the ambient and substrate media respect-
ively, the reflection coefficient can be written as
r =
Y0m11 + Y0m12 −m21 − Ysm22
Y0m11 + Y0m12 +m21 + Ysm22
(2.6)
Frequently, the DBR used in such devices is mostly implemented by using quarter
wavelength (Θbr = 1) thick layers. The incident electromagnetic wave is reflected and
transmitted at each interface. The partial reflection (r) and transmission (t) coefficients
of the wave at the interfaces are given respectively as
r =
nH − nL
nH + nL
(2.7)
10 2.1 Optical properties
t =
2nL
nH + nL
(2.8)
The waves are not phase shifted by passing the interfaces of the layers with high
refractive indices to the layers with the lower ones. However a phase shift π occurs in
the other case. In this case, the partially reflected waves from the interfaces recom-
bine constructively resulting in high reflection within a wavelength range (stop band).
Beyond this range, a destructive interference ensures a low reflectivity.
The reflectivity of a DBR comprising p periods of lossless quarter wavelength layers
(short: lossless p period quarter wavelength DBR) can be concluded from the reflectiv-
ity at the central wavelength. At this wavelength, the amplitude of the reflection
coefficient [20] is given by
r2p =
1− nS
n0
³
n1
n2
´2p
1 + ns
n0
³
n1
n2
´2p (2.9)
Where n0, nS, n1 and n2 are the refractive index of the ambient, of the substrate,
and the two different DBR materials, respectively. For the case of an additional layer,
that means for a 2p+ 1 periodic DBR, the reflection coefficient can be written as
r2p+1 =
1− n
2
1
n0·nS
³
n1
n2
´2p
1 +
n21
n0·nS
³
n1
n2
´2p (2.10)
The reflectivity at the central wavelength can be calculated by R = |r|2. Thus,
the reflectivity of a lossless p period quarter wavelength DBR can be rewritten, by
substituting from 2.9, as
R = tanh2
∙
1
2
ln
µ
n0
nS
¶
+ p ln
µ
n1
n2
¶¸
(2.11)
By applying the equations 2.9 and 2.10, and by disregarding the term (nL/nH)4,
the following approximations for the reflectivity could be found [20] for n1 = nL and
n1 = nH , alternatively and respectively.
R2p ≈ 1− 4
nS
n0
µ
nL
nH
¶2p
(2.12)
R2p ≈ 1− 4
n0
nS
µ
nL
nH
¶2p
(2.13)
R2p+1 ≈ 1− 4
n2L
n0nS
µ
nL
nH
¶2p
(2.14)
2 Basics of tunable microcavities 11
R2p+1 ≈ 1− 4
n0nS
n2H
µ
nL
nH
¶2p
(2.15)
In the case of lossy DBRs [19], by assuming that αidi < 1, and by neglecting the
the second and higher order terms in αidi which leads to coshαidi = sinhαidi ≈ 1 (i
indicates the layer number), we obtain the reflectivity R by multiplying matrices as in
2.3. Let γ = n2/n1, we have
R = tanh2
⎧
⎨
⎩
1
2
ln
⎡
⎣
⎛
⎝(α2d2n1 + α1d1n2) (γ
1−p − γp+1) + nSγ−p (1− γ2)³
α2d2
n1
+ α1d1
n2
´
(γ1−p − γp+1) + n0γp (1− γ2)
⎞
⎠
⎤
⎦
⎫
⎬
⎭ (2.16)
For Θbr = 1, large p and small constant α values in all layers, the reflectivity
becomes
R = 1− 2αλ0 (n
2
1 + n
2
2)
2n0 |n21 − n22|
(2.17)
However, due to different phenomena like material growth uncertainties and re-
fractive index changes, a wavelength deviation δλ from the Bragg wavelength λB may
occur. A deviation in the wavelength implies a change in the refractive index in form
of dispersion δn1 and δn2. By considering the equations 2.3 and 2.5, a new matrix can
be adopted
M = (Ma ·Mb)p (2.18)
Ma =
⎛
⎝ −
πδλ
2λB
³
1− δn1
n1
λB
´
i
n1
in1 − πδλ2λB
³
1− δn1
n1
λB
´ ⎞⎠
Mb =
⎛
⎝ −
πδλ
2λB
³
1− δn2
n2
λB
´
i
n2
in2 − πδλ2λB
³
1− δn2
n2
λB
´ ⎞⎠
Using equation 2.9 and neglecting the second and higher terms of δλ, the reflectivity
R for a large number p of periods is given by
R = ±1± i πδλ
λBn0
· (n2 + nn)n2n1 − λB (n
2
1δn2 + n
2
2δn1)
|n22 − n21|
(2.19)
A high reflectivity of a DBR can be achieved by increasing either the number of peri-
ods p or the contrast between the refractive indices of the DBR layer pairs. However,
absorption limits the reflectivity even when the number of periods p is increased [23,24].
12 2.1 Optical properties
The effect of absorption on the reflectivity for a given number p of periods can be re-
duced by choosing alternating material layers with high refractive index contrast. This
is related to the optical penetration depth of the wave into the DBR. The optical
penetration depth [20] Pe in quarter wavelength λ/4n is given by
Pe =
λ
4n
·
nL
nH
∙
1 +
n2L
n2S
³
n0
nH
´2p−1¸ ∙
1−
³
n0
nH
´2p¸
1− n0
nH
∙µ
1 +
n4L
n2Hn
2
S
³
n0
nH
´4p−2¶¸ (2.20)
The shorter the penetration depth, the lower is the total absorption within the
optical path of the wave. On the other hand, an increase in the contrast between the
refractive indices contributes to a larger bandwidth of the DBR. This can be deduced
from the following equation, where ∆ω and ω0 are the frequency bandwidth and the
central frequency of the DBR stop band, respectively.
∆ω =
4ω0
π
arcsin
µ
nH − nL
nH + nL
¶
(2.21)
2.1.2 Fabry-Pérot filter
A simple Fabry-Pérot filter consists of a single or multiple optical half wavelength cavity
embedded between two planar DBRs, oriented parallel to each other. The destructive
and constructive interference of the waves in this cavity cause the filter characteristic.
Thus, the reflectivity in the stop band tends towards zero for a few wavelengths, which
leads to a high transmission. A deep theoretical study of FPF can be found in [18,22].
Beginning with the FP resonator with planar mirrors, the essential characteristics will
be summarized. The free spectral range (FSR), which is important for the tuning later
on, is the distance separating two neighboring filter characteristics. The FSR of a
Fabry-Pérot filter can be written, in terms of the cavity length Lcav, the light velocity
c and a refractive index ncav, as
FSR =
c
2 · ncav · Lcav =
λ · ν
2 · ncav · Lcav (2.22)
The finesse of a Fabry-Pérot filter, a figure of merit of the optical performance,
indicates the ability to transmit several channels without serious interference among
them. The finesse F can be expressed as a function of the reflectivity of both mirrors
(R1, R2) and the absorption αcav of the cavity, by
F =
π 2
√
R1R2 · e−2αcavLcav
1− (R1R2 · e−2αcavLcav) (2.23)
The term R1 · R2 · e−2αcavLcav is a factor correlated to the intensity variation for a
round trip at the resonance frequency. The FWHM of a filter characteristic is affected
by the cavity losses, the absorption in the material and the geometry of the resonator.
2 Basics of tunable microcavities 13
Thus the scattering of waves within a resonator is directly related to its geometry.
Furthermore, the transversal mode in the cavity should match with the filter aperture
in order to avoid diffraction losses.
FWHM =
FSR
F
=
c · £1− ¡R1R2 · e−2αcavLcav¢¤
(2 · ncav · Lcav) ·
¡
π 2
√
R1R2 · e−2αcavLcav
¢ (2.24)
The maximum transmission Tmax and the residual reflectivity Rres at resonance are
given, respectively, by
Tmax = −
T1T2
2
√
R1R2
2
√
R1 ·R2 · e−2αcavLcav
1− (R1 ·R2 · e−2αcavLcav) (2.25)
Rres = R1 −
T 21
R1
¡
R1 ·R2 · e−2αcavLcav
¢
1− (R1 ·R2 · e−2αcavLcav) (2.26)
For low loss optical resonators, we introduce the δ symbol to represent small devi-
ations from unity at high reflectivity of the DBRs. In this case, we can writeR1 = 1−δ1.
A general definition for high and low reflective mirrors is given by
R ≡ e−δ (2.27)
The total cavity loss δcav is given by the round trip loss δ0 and the mirror losses δ1
and δ2
δcav = δ0 + δ1 + δ2 = 4 · α · Lcav + ln
µ
1
R1R2
¶
(2.28)
This formula can also be rewritten to take into account the negative loss −δact
(gain) of an active medium
δcav = δ0 + δ1 + δ2 − δact (2.29)
Referring to 2.28, the finesse can be written as
F =
2π
δcav
(2.30)
The quality factor Qcav of the cavity is a relation between the stored power in the
resonator and the dissipated one per cycle of light.
Qcav =
4πLcav
λδcav
(2.31)
14 2.1 Optical properties
2.1.3 Resonator stability
A highly important criteria for the fibre-filter coupling is the stability of the filter
resonator. This stability is related to the radius of curvature of the bent membrane,
to the cavity length as well as to the geometric alignment of the curved DBR1. Three
main resonator configurations are possible, with a stable resonator being desirable. A
resonator is stable when the distribution of the electric field remains unchanged after
one round trip, which requires the curvature of the phase front of the electric field to
agree with the radius of curvature of the mirror membrane. This can only occur in
resonators fulfilling the equation 2.32 [25].
0 <
µ
1− Lcav
ROC1
¶
·
µ
1− Lcav
ROC2
¶
< 1 (2.32)
Unstable half symmetric resonators are shown in Figures 2.2(e) and 2.2(f). In the
first case, the radius of curvature of the DBR2 is infinite (ROC2 = ∞) whereas the
radius of curvature of the DBR1 is smaller than the cavity length Lcav (ROC1 < Lcav).
In this case, equation 2.32 is clearly not satisfied. In the second case, equation 2.32 is
also not satisfied since ROC2 =∞ and ROC1 < 0. A stable half symmetric resonator
satsfying equation 2.32 can be found in Figure 2.2(d). Here, we have ROC1 > Lcav and
theROC2 =∞. Figure 2.2(b) and Figures 2.2(a and c) show symmetric resonators that
correspond to equations 2.33 and 2.34, respectively. These resonators are described as
”nearly stable”, since the smallest change in their curvature could lead to a unstable
operation. µ
1− Lcav
ROC1
¶
·
µ
1− Lcav
ROC2
¶
= 0 (2.33)
µ
1− Lcav
ROC1
¶
·
µ
1− Lcav
ROC2
¶
= 1 (2.34)
Thus, the symmetric concentric resonator (Figure 2.2(a)) with ROC1 = ROC2 =
Lcav/2 satisfies the condition of equation 2.34. The symmetric confocal resonator (Fig-
ure 2.2(b)) with ROC1 = ROC2 = Lcav corresponds to equation 2.33. The plane-plane
symmetric resonator described in Figure 2.2(c) with ROC1 = ROC2 =∞ also satisfies
equation 2.34.
In a stable half symmetric resonator (Figure 2.2(d)), the frequency of the resonant
cavity mode [26] can be calculated by
ωres =
c0
2Lcav
Ã
ml +
2mr +ma + 1
π
arccos 2
r
1− Lcav
ROC1
!
(2.35)
1A curved DBR is possible when we consider a FPF with an air-gap cavity. This will be the subject
of through study in this thesis.
2 Basics of tunable microcavities 15
DBR2
DBR1
DBR2
DBR1
Lcav
ROC1 ROC2
(a) (b) (c)
ROC1 ROC2
Lcav
Lcav
LcavROC1 LcavROC1 Lcav
ROC1
(d) (e) (f)
Figure 2.2: Different geometrical configurations of resonators with partially or fully curved
mirrors: (a) symmetric concentrical nearly stable resonator with ROC1 = ROC2 = Lcav/2;
(b) symmetric confocal nearly stable resonator with ROC1 = ROC2 = Lcav; (c) plane nearly
stable resonator; (d) stable half symetric resonator with ROC1 > Lcav; (e) non stable reson-
ator with ROC1 < Lcav; (f) unstable resonator with negative ROC1 values.
whereml,mr andma are the longitudinal, the radial and the angular mode numbers.
For a symmetric geometry of the bent DBR1 and a given fundamental mode number
ml, numerous side modes with higher and lower resonance frequencies are available. In
this case, no fundamental mode operation can occur and an arbitrary excitation leads
to a multi mode operation. However, a fundamental mode operation occurs if the
beam waist of the incoming Gaussian light beam (outcoming from the fibre) satisfies
the condition
B2w =
λ
π
2
p
Lcav (ROC1 − Lcav) (2.36)
Angular modes with mode numbersma 6= 0 occur by a geometrically non symmetric
bent DBR1 or by a non symmetric profile of the incoming light beam. In other words,
in order to reach fundamental mode operation in a half symmetric stable resonator,
either an appropriate fibre collimator should be used or a symmetric tailored radius of
curvature of the bent membrane should be technologically implemented.
If the spectral spacing between several neighboring wavelengths is smaller than the
central wavelength λ0, we can convert the frequency spacing into wavelength spacing
by using
∆λ =
λ20
c
∆ω (2.37)
For a symmetric geometry of the bent DBR1 and a tailored light beam profile,
and by assuming that ROC1 À Lcav (the case for most of the devices technologically
implemented in this thesis), a fundamental mode operation can occur (mr = ma = 0).
16 2.1 Optical properties
The wavelength spacing of the fundamental modes ∆λfund is given by incrementing
the fundamental mode number ml (in our case ml = 1) by using equation 2.35 as
∆λfund =
λ20
2Lcav
(2.38)
For the case where the symmetric incoming beam profile matches with the symmet-
ric geometry of the bent DBR1 membrane (but where the conditions of equation 2.36
are not fulfilled), a radial mode operation beside the fundamental mode occurs. The
spacing ∆λside of neighboring side modes is calculated by incrementing mr at a fixed
mode number ml and for ma = 0. In this case , the radial mode spacing ∆λradial , is
∆λradial = 2∆λside (2.39)
Thus the spacing of the neighboring side modes is given by
∆λside =
λ20
2πLcav
arccos 2
r
1− Lcav
ROC1
(2.40)
However, equations 2.35 and 2.40 are not valid for active devices like VCSELs. Here,
bulk material (active and non active) as well as quantumwells could be included. In this
case, several factors such as the group refractive index (dependent on the wavelength)
for example should be considered.
2.1.4 Active devices
The major difference between passive (e.g. Fabry-Pérot filter) and active devices (e.g.
VCSEL) is, in a first estimation, the active light emitting material within the reson-
ator. In this case, most of the relations concerning modes spacing in the Fabry-Pérot
resonator are not applicable. However, the relations between the incoming beam waist,
the geometry of the top DBR as well as the modes propagation remain valid in this
case. The emphasis of this thesis is put on devices with curved top DBRs. The interest
in the active region play in this case a minor role. A detailed study of active regions
in lasers can be found elsewhere [19—21,27]. Nevertheless, the fundamental properties
of the VCSEL will be summarized.
A VCSEL with a bulk active region of gain g is considered. The VCSEL is embed-
ded between two mirrors of reflectivities R1 and R2, respectively. The photon density
ρ, which is induced by an electromagnetic wave in the active material, increases pro-
portionally with the penetration depth and the total round trip of the wave in the
resonator. By considering the gain distance dgain and the loss factor αtotal, a distance
related photon density ρ(dgain) can be defined [19] as follows
ρ(dgain) = ρ0e
(g−αtotal)dgain (2.41)
2 Basics of tunable microcavities 17
The threshold gain gth is the gain level which overcomes the overall losses αtotal
in the VCSEL. It is expressed as a function of the cavity length Lcav, the transverse
confinement factor Γt, R1 and R2 [19].
gth = αtotal +
1
2LcavΓt
ln
1
R1R2
(2.42)
The threshold current Ith of a VCSEL is given by [27]
Ith = π
µ
D
2
¶2
Jth =
eV Nth
ηiτ s
∼= eV Beff
ηiηspon
(2.43)
D is the diameter of the circular active region (the most common shape)
Jth is the threshold current density
e is the electron charge
V is the volume of the active region
Nth is the threshold carrier density
ηi is the injection efficiency
ηspon is the spontaneous emission efficiency
τ s is the recombination lifetime
Beff is the effective recombination coefficient
where
Nth = Nt +
αa + αd + α1,2
A¯0Γopt
(2.44)
Nt is the transparent carrier density
αa is the absorption loss coefficient averaged per unit length
αd is the diffraction coefficient averaged per unit length
α1,2 is the DBRs loss coefficient
A¯0 is the gain coefficient expressing the differential gain G˙ = dg˙/dN
g˙ is the optical gain per cm
Γopt is the optical energy confinement factor, Γopt = ΓtΓl
Γl is the longitudinal confinement factor
The optical output power Pout [20] of a VCSEL can be expressed in terms of the
band gap energy Eg, the spontaneous emission factor S´, the driving current I, the
threshold current Ith, the differential quantum efficiency ηd, the spontaneous emission
efficiency ηspon and the injection efficiency ηi, as follows
For I ≤ Ith
Pout = ηd · ηspon · S´ · Eg · I (2.45)
For I ≥ Ith
Pout = ηd ·Eg · (I − Ith) + ηd · ηspon · S´ ·Eg · Ith (2.46)
18 2.1 Optical properties
where ηd is the differential quantum efficiency
ηd =
α1,2
αa + αd + α1,2
= ηi
³
1
Lcav
´
ln
³
1
R1
´
(αa + αd) +
³
1
Lcav
´
ln
³
1
2√R1R2
´ (2.47)
The power conversion efficiency ηp [20] for I À Ith is given in terms of the output
power, the driving current and the bias voltage Vb, as
ηp =
Pout
VbI
=
⎡
⎣ηi
³
1
Lcav
´
ln
³
1
R1
´
(αa + αd) +
³
1
Lcav
´
ln
³
1
2√R1R2
´
⎤
⎦ · Eg
Vb
·
µ
1− Ith
I
¶
(2.48)
2.1.5 Photodetector devices
An important use for photodiodes is in combination with tunable filters. By using these
novel devices, we can lower the overall costs of the DWDM systems. Several approaches
have been suggested and discussed in the literature [28—31]. In particular, the two-
chip concept described in [16] allows independent optimization of the various parts
and parameters of the system (e.g. buckling of the membranes, wavelength tuning,
selectivity and responsivity, etc.). Thus, an optimum functionality of the devices can
be reached.
The main characteristics of a PIN photodiode are the responsivity, the junction
capacitance and its influence on the demodulation bandwidth and the signal to noise
ratio. The responsivity Rresp is the ratio of the generated current, Iout, to the optical
input power, Pin. The responsivity can be written as a function of the quantum
efficiency η, the elementary charge e, Planck´s constant h and the frequency w.
Rresp = η
e
hw
=
Iout
Pin
(2.49)
The frequency w = c0/λ0 is inversely correlated to the wavelength. Therefore, for
λ0 = 1550 and η = 1, the maximum responsivity, which can be reached, is Rresp,max =
1.25 A /W. However, this theoretical value, can never be practically reached due
to the light backscattering (between the layers), material absorption and absorption
mechanisms like recombinations on the layer surfaces.
The junction capacity is given as
Cj = A
2
s
εeND
2(UD − U)
(2.50)
Where A is the area of the diode, ε is the dielectric number, ND is the donator
concentration and (UD − U) is the potential difference applied to the diode. If we
2 Basics of tunable microcavities 19
consider a diode under potential, the capacity can be written as a function of the
depletion region lzone and ε0 and εr, the dielectric constants for air and the pin diode
material, respectively.
Cj =
ε0εrA
lzone
(2.51)
The bandwidth of the pin photodiodes is given as
BW3dB =
1
2πRsCj
(2.52)
where, Rs is the serial conductance of the diode. Substituting from equation 2.51,
the bandwidth can be written as follows
BW3dB =
lzone
2πε0εrARs
(2.53)
Generally, the signal to noise ratio is given as
SNRpin =
I2out
I2noise
(2.54)
where Inoise, the noise current of the pin photodiode, is a function of the quantum
noise of the photocurrent Iout and the dark current Idark.
Inoise =
2
p
2eBW3dB(Iout + Idark) (2.55)
If we also consider the thermal noise current Ith in a pin photodiode, the overall
signal to noise ratio can be written as follows
SNRpin =
I2out
I2noise + I
2
th
(2.56)
with Ith given by
Ith =
2
r
4kTBW3dB
Rs
(2.57)
where k is the Boltzmann constant and T is the temperature.
2.2 Mechanical properties
2.2.1 Micromechanical tuning
Tunable microdevices for the optical communication technology are often associated
with active devices like lasers; in the last decades, several tunable laser sources have
been commercially available on the market. These sources are used in medical engineer-
ing, sensing, spectroscopy, and material diagnostics. A thorough study of such devices
20 2.2 Mechanical properties
can be found in [32]. Nowadays, as the devices became smaller and more complicated,
their tuning occurs often by micromechanical actuation (especially vertical cavity based
devices). Thus, MEMS and MOEMS devices like micromirror arrays, sensors, align-
ment systems, as well as active and passive devices for optical communication [33,34]
are emerging.
Micromechanical tuning of the devices investigated in this thesis, namely FP filters,
VCSELs and photodiodes occurs either electrostatically or electrothermally. Electro-
static tuning is carried out by a potential difference between two conducting or semi-
conducting layers, whereas thermal tuning is based on the expansion of a material
through heating by a current. Mechanical actuation of a layer automatically implies
that the layer should be movable. Commonly, the top DBR of such devices is under-
etched and encloses an air-gap cavity. Figure 2.3 shows the tuning principle in such
devices.
DBR2
DBR1
Lcav
DBR2
DBR1
Lcav- ∆ Lcav
DBR2
DBR1
Lcav+ ∆ Lcav
(c)
(b)
(a)
1400 1500 1600 1700 1800
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
1,1
1,2
Tr
an
sm
is
si
on
Wavelength / nm
 Filter characteristic for L
cav
 Filter characteristic for Lcav+ ∆ Lcav
 Filter characteristic for L
cav
- ∆ L
cav
Figure 2.3: Tuning principle of a Fabry-P érot filter with physical cavity length Lcav (a). A
physical shift of the cavity length (b and c) ±∆Lcav implies a spectral shift in the wavelength
by ±∆λ.
A physical change of the cavity length results in spectral displacement of the central
resonance wavelength of the filter. However, the sensitivity of the spectral wavelength
displacement strongly depends on the cavity length and the geometry of the resonator.
This is captured by the tuning efficiency [35]
T =
∆λ
∆Lcav
(2.58)
Although the technological implementation of tunable devices by bulk or surface
micromachining has been well investigated and advanced, the performance of these
2 Basics of tunable microcavities 21
devices is still significantly restricted by fabrication issues, such the adverse effects of
residual stress. In particular, devices involving air-gaps and suspended membranes
(FPFs,VCSELs and photodiodes) require a precise stress control for a successful im-
plementation [36, 37]. Due to the high sensitivity of these devices to stress, stress en-
gineering is indispensable. Undesired buckling and bending of released micromachined
structures (e.g. top DBRs) may be caused by compressive stress, whereas cracking
may occur if the tensile stress is too high [38]. Several approaches reducing the effect
of the overall stress on the released membranes have been proposed and successfully
demonstrated [39—42]. They use additional stress compensation layers and suitable
mechanical designs. However, these approaches do not directly deal with the origin of
the residual stress and may be unsuitable for a wide range of optical devices. Thus,
additional stress compensation layers in released distributed Bragg reflectors (DBRs)
may induce phase shifts in the light path, thus disturbing functionality of the device.
Other approaches using an appropriate mechanical design suitable for stress reduc-
tion in such devices, could be incompatible with the optical requirements, thus leading
to undesired side and angular modes in the cavities. Consequently, the main goal of
stress engineering is to control the stress in optical air-gap devices without modifying
the mechanical or optical designs.
2.2.2 Stress in thin films
The distinction between stress and strain is clearly explained in the literature [43—46].
Strain is defined as the change in dimensions of a body under an external force and is
expressed as the extension per unit length. Stress is the internal pressure of a body
when submitted to external forces. When a body is placed under stress, strain results.
Thus the strain ξ and the stress σ are related by the biaxial modulus Y [47].
σ = Y · ξ (2.59)
When the stress is homogeneous and biaxial, Y can be expressed in terms of the
Youngs modulus E and the Poisson ratio ν of a material.
Y =
E
1− ν (2.60)
When the body is no longer subjected to external forces, it tends to relax and
dissipate the internal stress. In the case of thin semiconductor and dielectric films, the
overall stress is mostly described as residual stress. The residual stress, which is highly
undesirable, is the tension or compression pre-existing in a bulk material (when the
external force is zero). The origins of the residual stress in such films are mostly process
related phenomena like defects, mismatching of the materials during the growth, ions
implantation, thermal effects and boundary conditions. In this thesis the residual
stress is devided into an intrinsic and an extrinsic stress. The intrinsic stress is due
to the morphology and the dynamic composition of the material during the process,
22 2.2 Mechanical properties
whereas the extrinsic stress occurs after the dynamic process is accomplished. However,
the extrinsic stress is not decoupled from the intrinsic one. The best example of an
extrinsic stress is the thermal stress which results from the temperature change between
deposition and characterization and the difference in thermal expansion coefficients of
the film and the substrate [48—51]. The intrinsic stress, which is strongly related to the
process parameters, influences the thermal expansion coefficients. In order to avoid
obscurities, this thesis handles two different kinds of stress: the overall lateral bulk
stress (described as homogeneous stress in several publications) and the vertical spatial
stress variation (less precise but short: gradient stress). The overall lateral bulk stress
σbulk,total consists of the intrinsic stress σint and the thermal stress σth.
σbulk,total = σint + σth (2.61)
In contrast to the intrinsic stress, the thermal stress can be calculated by the thermo
dynamical equation
σth =
Ef
1− νf
· (αs − αf) ·∆T (2.62)
in terms of the Young´s modulus Ef , the Poisson ratio νf and the thermal expansion
coefficient αf of the film, the thermal expansion coefficient of the substrate αs and the
difference in temperature ∆T between deposition and characterization.
The vertical spatial stress, referred to in some publications as gradient stress or
strain, is the vertical variation of the internal force potentials which causes the de-
flection of the released cantilevers. Note that the overall bulk stress may also cause a
deflection of structures that are fixed at all their extremities. In this case, it is difficult
to find out which kind of stress is responsible for the deflection.
During the investigation within this thesis, the overall bulk stress is characterized
at two levels: macroscopically averaged stress and microscopically detected stress [52].
The first quantity is calculated from the measured curvature of the wafer, and results
in global values. The macroscopic stress measurement set-up is based on the optical
imaging method [53]. It consists mainly of a HeNe laser source, a diffraction grid
and a detector array [54]. The collimated laser beam travels through a telescope and
a transmission grating and generates a spatial diffraction pattern on the wafer (one
dimensional array). If the wafer is bent, these light spots are then reflected under
slightly different angles and are observed on a distant screen. The position of the spots
on the screen are evaluated and compared to their positions originating from the blank
wafer before the deposition. If the wafer bends due to compressive thin film stress,
the wafer acts as a convex mirror and the reflected laser beam diverges. Thus, in this
case, bending of the substrate increases the distance between the separated diffraction
structures. In case of a tensile stress, the wafer acts as a concave mirror. The total
stress of the deposited dielectric layers can be calculated, based on the radii of curvature
before and after the deposition, by Stoney´s equation as
2 Basics of tunable microcavities 23
la
δi
di
l i
supporting post
actuator beam
indicators
sacrificial layer (a)
δcant
cantilever of length lcant
(b)
Figure 2.4: MEMS structures for stress detection: (a) top view of a microstructure for
microscopic stress detection, (b) cross section of a microstructure (cantilever) for detecting
the vertical variation (short: gradient stress) of the stress.
σbulk,total =
Es
6 (1− νs)
d2s
df
µ
1
ROC1
− 1
ROC2
¶
(2.63)
where Es and νs are the Young modulus and the Poisson ratio of the substrate, ds
and df the thickness of the substrate and the film, ROC1 and ROC2 are the radii of
curvature before and after the deposition, respectively.
The microscopically detected stress is monitored by MEMS structures [52] and
describes the stress topography and thus delivers precise local values. These structures
were first proposed by Ericson et. al. [55]. In the literature, several MEMS approaches
have been proposed for stress diagnostics [36,55—61] as well as for the characterization
of the physical parameters of the films (e.g. thermal expansion coefficient, Young
modulus, elastic modulus and fracture strength) [62—66]. The MEMS structure used
for the bulk stress evaluation during this thesis is described in Figure 2.4(a).
The technological implementation of these structures is described in detail in chapter
4. The structure is completely underetched except under the supporting posts at the
end of the actuator beams. After releasing the structure, the thin films tend to relax
and can expand or shrink according to the nature of the stress. The magnitude of the
stress is proportional to the rotation of the indicators. The direction of the rotation
indicates whether the local stress in the lateral direction is compressive or tensile: the
indicators rotate clockwise if the stress is compressive, and counter clockwise if tensile.
In the case of stress free thin films, no rotation of the indicator beams occurs. The
local stress can be calculated according to equation 2.64.
σbulk,total =
Ef
6 (1− νf)
di
2lila
δi (2.64)
24 2.2 Mechanical properties
The vertical spatial stress of a given film, can be calculated by evaluating the
deflection of a released cantilever (2.4(b)) according to equation
∂σ
∂dcant
=
Ef
(1− νf)
2
l2cant
δcant (2.65)
where σ is the internal stress of the film, dcant and lcant the thickness and the length
of the cantilever and δcant is the deflection of the cantilever. Positive values of the
gradient are associated with upwards deflection whereas negative values correspond
to downwards deflection. Figure 2.5(a) shows a white light interferometer picture of
released cantilevers with positive gradient value. A negative gradient is shown in Figure
2.5(b).
(a) (b)
Figure 2.5: White light interferometer pictures of: (a) released cantilevers with a positive
stress gradient, (b) released cantilevers with a negative stress gradient.
Chapter 3
PECVD technology: basics and
material properties
3.1 Basics
3.1.1 Configuration of the involved PECVD
In this thesis, a plasma enhanced chemical vapor deposition technology (PECVD) is
used for implementing dielectric layers. The system is a standard simple capacitively
coupled parallel plate reactor (Plasma Lab 80 plus) from Oxford Instruments. Figure
3.1 shows a schematic layout of the system. The deposition chamber consists of a paral-
lel plate reactor driven by a low (130 kHz) and a high (13.56MHz) frequency generator.
The high frequency generator is matched to the chamber trough an automatic match-
ing unit (AMU). The duty cycle1 of the plasma excitation frequency is varied during
the process by an automatic switching unit (SWI). The substrate holder is grounded
and can be heated up to 400 ◦C. Three mass flow controllers (MFC) ensure precise
proportioning of the reactive gases. The radicals and ions that do not contribute to
the deposition are eliminated by a root and a dual stage rotary pump.
Amorphous silicon nitride, silicon dioxide, silicon oxynitride and silicon can be
deposited by the system. Two percent silane diluted in nitrogen and ammonia are
used for silicon nitride whereas silicon dioxide results from the same silane and nitrous
oxide. Other process gases like nitrogen (N2), Argon (Ar), Oxygen (O2) and a cleaning
gas like carbon tetrafluoride with oxygen (CF4+O2) are also available. In this thesis
we concentrate on the materials silicon nitride and silicon dioxide. Table C.7 shows
the standard process parameters used for the deposition of silicon nitride and silicon
dioxide at 300 ◦C. In order to study the dependence of the material properties, namely
deposition rate, refractive index and stress, on the process parameters, one parameter
is varied at a time while keeping the remaining parameters unchanged.
1The frequency duty cycle is defined in this thesis as: The time ratio between the high and the low
frequency during the deposition process.
26 3.1 Basics
Table 3.1: Standard PECVD process parameters for silicon nitride and silicon dioxide de-
posited at 300 ◦C.
Process parameters Si3N4 : Hx SiO2
2% SiH4−N2 flow / sccm 1000 430
NH3 flow / sccm 20 0
N2O flow / sccm 0 710
HF-power / W 20 20
LF-power / W 20 20
Duty cycle Ψ 0.538 1
Temperature / ◦C 300 300
Pressure / torr 0.65 1
3.1.2 Background of PECVD layer deposition
Plasma enhanced chemical vapor deposition is a very common, simple and reliable
technique for implementing dielectric layers. Yet even by using the same technological
process, different reactor configurations result in different material properties. Thus in
the literature, different results for the same deposition process could be found. The
main advantage of the PECVD technology is its ability to deposit dielectric materi-
als at low temperatures (250 ◦C−400 ◦C), compared to the LPCVD. Hence, PECVD
technology is used in a wide range of applications. However, this advantage negatively
affects the properties of the materials. Thus silicon nitride deposited at lower temper-
atures may not be stoichiometric and may contain a high2 concentration of hydrogen.
The incorporation of hydrogen in silicon dioxide is not a main issue in the literature.
An overview dealing with PECVD deposition and material properties can be found
in [67,68].
Based on several publications, Habracken et. al. [67] reviewed the mechanical and
optical properties of silicon nitride. They found that the properties of the plasma
nitrides depend on the Si/N concentration ratio, on the hydrogen content as well as
the concentration ratio of the Si-H/N-H bonds [69—71]. Furthermore, the N/Si ra-
tio decreases by increasing the deposition temperature and the process pressure [72].
However, the N/Si ratio is proportional by the square root to the precursors ratio of
NH3/SiH4 [73]. By increasing the deposition temperature, the hydrogen concentration
decreases [72, 74]. The incorporation of hydrogen in silicon nitride derived from SiH4
and NH3 precursors can be as high as 39% [75]. This hydrogen is mostly bonded to
the silicon and nitrogen as Si-H and N-H compounds. Thus, the N/Si ratio and con-
sequently the N-H/Si-H ratio defines the dominant hydrogen compound in the films. A
strong concentration of N-H bonds can be found in nitrogen rich films, otherwise Si-H
bonds dominate [71,72]. The N/Si ratio as well as the hydrogen concentration strongly
affect the refractive index and the stress of the films. However, statements on the mor-
phology of thin films deduced from measuring the refractive index are mostly incorrect
since different film compositions may result in the same refractive index value [70].
2The concentration depends on several factors, e.g. the deposition temperature and the plasma
conditions.
3 PECVD technology: basics and material properties 27
Plasma
SWI AMU
HF-Generator
13.56 MHz
LF-Generator
130 KHz
R
O
O
TS
R
O
TA
R
Y
Gas-Pod
MFC
N2, Ar, O2
CF4+O2
2% SiH4-N2
NH3
N2O 
R
O
O
TS
R
O
TA
R
Y
R
O
O
TS
R
O
TA
R
Y
Figure 3.1: Configuration of the plasma enhanced chemical vapor deposition (PECVD) ma-
chine used (Plasma Lab 80 plus, Oxford Instruments).
For a given hydrogen content, the variation of the refractive index is proportional
to the N/Si ratio. By increasing the nitrogen content, the refractive index decreases.
Incorporating larger amounts of Si in the films lead to a higher refractive index. On
the other hand, for a constant N/Si ratio, the variation in the refractive index value is
inversely proportional to the hydrogen concentration. Thus by increasing the hydrogen
content in the films, the refractive index decreases. The refractive index is strongly
related to the dynamic process conditions and parameters like deposition temperature,
process gas flow and pressure.
Concerning the mechanical properties of thin films, different correlations between
the stress and the films composition is made in the literature [74, 76]. However, the
stress seems to be strongly related to the dynamic of the deposition process (ion en-
ergy and the atomic architecture of the films). The stress control by tuning the layer
morphology (via the plasma frequency) is thoroughly discussed in this thesis (chapter
5).
These correlations will be used to comment on the results found during the PECVD
investigation in section 3.2. However, while discussing the results, I will not refer again
to these references.
28 3.2 Material investigation (stress, composition and optical properties)
3.2 Material investigation (stress, composition and
optical properties)
For the PECVD material investigation, the process parameters described in table C.7
are used. The behavior of the refractive index, the deposition rate and the stress3 of
the silicon nitride and silicon dioxide films as a function of various process parameters
is studied. For this, a single parameter is varied at a time whereas the rest is kept
unchanged.
3.2.1 Silicon nitride
Throughout this series of experiments, both the high and low frequency power are fixed
at 20W. The frequency duty cycle Ψ is fixed at 0.538. Figure 3.2 exhibits the behavior
of silicon nitride for varying the process temperature. The deposition temperature
varies between 230 ◦C and 320 ◦C. The silane and ammonia flows are 1000 sccm and
20 sccm, respectively. The deposition occurs at a pressure of 0.65 torr. The refractive
index increases (1.928 to 1.984) with increasing the deposition temperature. This is
due to the decrease of the hydrogen concentration in the film. In this case, the density
became higher, and consequently the deposition rate lower since higher matter per
unit volume is reached. The effect of the temperature on the total bulk stress variation
is rather weak. In this case, the stress is tensile for the whole temperature range.
Stress values between −11MPa and −100MPa were found. However by increasing the
deposition temperature, the stress tend to get less tensile. This is probably due to the
decrease of the hydrogen content, and the increase of the density. The break down of
the refractive index trend at 300 ◦C is also been noticed in [68].
By varying the silane flow from 800 sccm to 1200 sccm (Figure 3.3) which corres-
ponds to a variation of the N/Si ratio between 0.025 and 0.0166, a strong change in the
refractive index is observed (1.93 to 2.0). The refractive index is strongly related to
the N/Si ratio. The ammonia flow is 20 sccm. The deposition occurs at 0.65 torr.The
tensile stress behavior (−12MPa to −130MPa) and the deposition rate show no con-
stant corrolation with the N/Si ratio variation.
Furthermore, a decrease of the refractive index (2.01 to 1.91) is observed by in-
creasing the ammonia flow (Figure 3.4). Since the temperature has a fixed value of
300 ◦C, the refractive index depends in this case on the N-H/Si-H ratio only. Thus by
varying the ammonia flow range between 15 sccm and 30 sccm, the N/Si ratio varies
between 0.015 and 0.03. A higher N/Si ratio means a higher concentration of nitrogen
and thus a lower refractive index. In this experiment, the silane flow is 1000 sccm.
The deposition occurs at a pressure of 0.65 torr. The stress varies from −60MPa to
−200MPa, which can be explained by assuming that decreasing the Si content in the
3In this chapter, the measured bulk stress is reported as a result of varying the process parameters.
The stress control by varying the duty cycle of the PECVD plasma excitation frequencies is reported
in chapter 5.
3 PECVD technology: basics and material properties 29
220 242 264 286 308 330
-120
-90
-60
-30
0
30
60
90
 
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Deposition temperature / ° C
2% SiH
4
-N
2
             1000 sccm
NH3                         20 sccm
HF & LF power       20 W
Pressure                 0.65 Torr
Duty cycle Ψ           0.538
1,9 0
1,9 2
1,9 4
1,9 6
1,9 8
2 ,0 0
Te
ns
ile
C
om
pr
es
si
ve
 
 
R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
12
14
16
 
Figure 3.2: The dependence of the bulk stress σtotal, the refractive index nSi3N4 and the
deposition rate of silicon nitride on the PECVD process temperature. The values of the data
plotted in the diagram can be found in table B.2.
700 800 900 1000 1100 1200 1300
-150
-100
-50
0
50
100
 
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Silane (2% SiH4-N2) flow / sccm
1, 9 3
1, 9 5
1, 9 6
1, 9 8
1, 9 9
2 , 0 1
 
14,42
14,45
14,48
14,52
14,55
14,59NH3                         20 sccm
HF & LF power       20 W
Pressure                 0.65 Torr
Duty cycle Ψ           0.538
Temperature           300 ° C
Te
ns
ile
C
om
pr
es
si
ve
R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
Figure 3.3: The dependence of the bulk stress σtotal, the refractive index nSi3N4 and the
deposition rate of silicon nitride on the silane flow. The values of the data plotted in this
diagram can be found in the table B.3.
30 3.2 Material investigation (stress, composition and optical properties)
films leads to higher N-H bonds and thus less dense films. This fact forces the films
to contract causing tensile stress. Higher N-H bonds concentration leads normally to
higher tensile stress.
12 16 20 24 28 32
-200
-150
-100
-50
0
50
100
Te
ns
ile
 
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Ammonia (NH3) flow / sccm
1,9 0
1,9 3
1,9 5
1,9 7
2 ,0 0
2 ,0 2
co
m
pr
es
si
ve 2% SiH4-N2             1000 sccm
HF & LF power       20 W
Pressure                 0.65 Torr
Duty cycle Ψ            0.538
Temperature           300 ° C
 
R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
13,65
13,80
13,95
14,10
 
Figure 3.4: The dependence of the bulk stress σtotal, the refractive index nSi3N4 and the
deposition rate of silicon nitride on the ammonia flow. The values of the data plotted in this
diagram can be found in the table B.4.
Figure 3.5 shows the dependence of the optical and mechanical properties of silicon
nitride on the process pressure. The temperature is fixed at 300 ◦C. The silane and
ammonia flows are 1000 sccm and 20 sccm, respectively. The pressure varies between
0.5 torr and 1.1 torr. The refractive index is observed to increase (1.9 to 2.0) with
increasing the process pressure. Since the ammonia and silane flows as well as the
temperature are constant, and the NH3/SiH4 ratio is 0.02 and proportional to the N/Si,
the refractive index should not vary based on the previous assumption. However, it
seems that the retention time of N and Si behave differently at different pressure, which
leads to a variation of the N/Si ratio and thus the refractive index. The stress and the
deposition rate varies arbitrarily with the pressure.
3.2.2 Silicon dioxide
Concerning the deposition of silicon dioxide, the incorporation of hydrogen in the layers
is rarely described in the literature. When silane is cracked in the plasma, hydrogen
ions and radicals are present so that in the absence of oxygen (N2O in this case),
hydrogenized amorphous silicon films will result. On the other hand, in the presence
of oxygen, any stoichiometry from hydrogenated to nearly pure silicon dioxide can be
deposited. This flexibility is a real process challenge. Good control of the gas mixtures
is required to deposit pure, hydrogen free silicon dioxide. An excess of silane leads to
Si-H bonds in the films, whereas too much oxygen delivers Si-OH compounds [77].
3 PECVD technology: basics and material properties 31
0,38 0,57 0,76 0,95 1,14 1,33 1,52
-400
-300
-200
-100
0
100
200
300
400
 
 
 
 R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
Process pressure / Torr
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
1,92
1,95
1,98
2,01
2,04
2% SiH4-N2             1000 sccm
NH3                         20 sccm
HF & LF power       20 W
Duty cycle Ψ            0.538
Temperature           300 ° C
 
11,40
12,35
13,30
14,25
15,20
Te
ns
ile
C
om
pr
es
si
ve
 
Figure 3.5: The dependence of the bulk stress σtotal, the refractive index nSi3N4 and the
deposition rate of silicon nitride on the process pressure. The values of the data plotted in
this diagram can be found in B.5.
The refractive index and the deposition rate vary with changing process parameters.
The bulk stress of silicon dioxide, on the other hand, does not seem to depend on the
process parameters and exhibits compressive values with rather low variation (Figures
3.6, 3.7, 3.8 and 3.9). This effect is most probably due to the strong Si-O bonds, that
seems to be stronger then the tension in the films.
Figure 3.6 shows the dependence of the refractive index, the bulk stress and depos-
ition rate of silicon dioxide deposited by changing the deposition temperature between
230 ◦C and 320 ◦C. The silane and nitrous oxide gas flows are 430 sccm and 710 sccm,
respectively. The high and low frequencies power are 20W, whereas the process pres-
sure is 1 torr. The frequencies duty cycle Ψ is 1 in this case. The stress is compressive
and varies between +232MPa and +135MPa. The refractive index increases weakly
from 1.464 to 1.467 with increasing temperature. The determining factor here is the
hydrogen concentration in the films; the higher the temperature, the lower the hydro-
gen content and thus the higher the refractive index. The very small variation in the
refractive index (33.3E−6 ◦C−1) is due to the low hydrogen content in the films for these
process parameters. The deposition rate increases from 63.4 nm /min to 69 nm /min.
By varying the silane flow between 400 sccm and 520 sccm only (Figure 3.7), the
stress shows no strong dependence (compressive, +164MPa to +83MPa). However,
the refractive index increases from 1.468 to 1.487 by increasing the silane flow. This is
due to the higher Si content in the films. The deposition rate increases from 66 nm /min
to 70 nm /min.
Figure 3.8 exhibits the dependence of the bulk stress, refractive index and the
deposition rate on the nitrous oxide flow (650 sccm and 850 sccm).
32 3.2 Material investigation (stress, composition and optical properties)
220 242 264 286 308 330
-100
-50
0
50
100
150
200
250
 
 
 
Deposition temperature / ° C
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
1,463
1,464
1,465
1,466
1,467
1,468
Te
ns
ile
C
om
pr
es
si
ve
2% SiH4-N2             430 sccm
N2O                        710 sccm
HF power               20 W
Pressure                1 Torr
Duty cycle Ψ           1
 R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
 
 
63,0
64,5
66,0
67,5
69,0
 
Figure 3.6: The dependence of the bulk stress σtotal, the refractive index nSiO2 and the
deposition rate of silicon dioxide on the process temperature. The values of the data plotted
in this diagram can be found in the table B.6.
The refractive index and the deposition rate are inversely proportional to the nitrous
oxide flow. The higher the nitrous oxide flow, the higher the oxygen content and the
lower the refractive index. However, by varying the process pressure (Figure 3.9), the
refractive index and deposition rate seem to vary arbitrarily. This observation can
be explained only if we consider the dwell time of the reactive ions in the deposition
chamber.
3.2.3 Bragg mirrors
The main task in this thesis is to establish a PECVD technology enabling the imple-
mentation of differently stressed DBR without affecting the optical properties. This
is rather complicated task since the optical and mechanical properties are correlated.
However, based on this investigation and the stress study later on (see chapter 5), the
mechanical and optical properties have been tuned independently. Thus I tailor the
DBRs precisely to fulfil the requirements for several applications. Figure 3.10 shows the
shematic cross section of a PECVD DBR implemented by differently stressed silicon
nitride and silicon dioxide layers. The first 3 periods consist of compressively stressed
silicon nitride (+850MPa) and silicon dioxide (+517MPa). The rest of the DBR is
implemented by nearly stress free silicon nitride (+20MPa) and compressively stressed
silicon dioxide (+517MPa).
The DBR is designed to be implemented as a bent membrane (after bulk releasing)
in a two chip VCSEL (see chapter 6). Therefore, a precise tailoring of the stress across
the vertical structure of the DBR is needed. Thus, a defined radius of curvature as
well as a desired cavity length could be obtained. However, this mechanical tailoring
3 PECVD technology: basics and material properties 33
360 390 420 450 480 510 540
-100
-50
0
50
100
150
200
250
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Silane (2% SiH4-N2) flow / sccm
 
1,4 6 5
1,4 70
1,4 75
1,4 8 0
1,4 8 5
1,4 9 0
N2O                         710 sccm
HF power                20 W
Pressure                 1 Torr
Duty cycle Ψ            1
Temperature           300 ° C
Te
ns
ile
C
om
pr
es
si
ve
 R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
 
66,3
68,0
69,7
71,4
 
Figure 3.7: The dependence of the bulk stress σtotal, the refractive index nSiO2 and the
deposition rate of silicon dioxide on the silane flow. The values of the data plotted in this
diagram can be found in the table B.7.
550 600 650 700 750 800 850 900
-100
-50
0
50
100
150
200
250
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Nitrous oxide (N2O) flow / sccm
 
1,4 6 8
1,4 70
1,4 72
1,4 74
1,4 76
1,4 78
1,4 8 0
Te
ns
ile
C
om
pr
es
si
ve
 R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
 
66,5
67,2
67,9
68,6
2% SiH4-N2             430 sccm
HF power                20 W
Pressure                 1 Torr
Duty cycle Ψ            1
Temperature           300 ° C
Figure 3.8: The dependence of the bulk stress σtotal, the refractive index nSiO2 and the
deposition rate of silicon dioxide on the nitrous oxide. The values of the data plotted in this
diagram can be found in the table B.8.
34 3.2 Material investigation (stress, composition and optical properties)
0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4
-100
-50
0
50
100
150
200
250
300
Process pressure / Torr
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
 
1,38
1,40
1,42
1,44
1,46
1,48
2% SiH4-N2             430 sccm
N2O                        710 sccm
HF power               20 W
Duty cycle Ψ           1
Temperature          300 ° C
Te
ns
ile
C
om
pr
es
si
ve
 R
ef
ra
ct
iv
e 
in
de
x 
&
 D
ep
os
iti
on
 ra
te
 / 
nm
 x
 m
in
-1
 
51,3
54,0
56,7
59,4
62,1
64,8
67,5
 
Figure 3.9: The dependence of the bulk stress σtotal, the refractive index nSiO2 and the
deposition rate of silicon dioxide on the process pressure. The values of the data plotted in
this diagram can be found in the table B.9.
.
.
.
Si3N4 (1, 3, 5)
[+850MPa]
SiO2 (2, 4,..., 24)
[+517MPa]
Si3N4 (7, ..., 25)
[+20MPa]
GaAs-substrate
Figure 3.10: 12.5 periods PECVD DBR implemented by differentely stressed silicon nitride
and silicon dioxide layers.
3 PECVD technology: basics and material properties 35
900 1000 1100 1200 1300 1400 1500 1600 1700
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
1,1
R
ef
le
ct
iv
ity
Wavelength / nm
 Tsn032IVB 
 Tsn034I 
980 nm 1300 nm
R
ef
le
ct
iv
ity
R
ef
le
ct
iv
ity
Figure 3.11: Optical spectra of two DBRs with the same mechanical properties and different
optical requirements.
should not affect the optical properties of this DBR. By considering both, the optical
investigation in this chapter and the mechanical study (see chapter 5), all requirements
have been fulfilled. Figure 3.11 shows as an example the optical spectra of two DBRs,
which have the same structure described in Figure 3.10. However, the first DBR
(Tsn032IVB) is designed for an optically pumped VCSEL with a central wavelength
of 1610 nm, whereas the second DBR (Tsn034I) is intended for the use in a VCSEL
emitting at 1550 nm. Even though the stop band spectra are shifted to fulfill the
optical requirements, the reflectivity of both DBRs at the pumping laser wavelengths
(980 nm & 1300 nm) is very low, enabling a good coupling of the pumping light in
the VCSEL. Furthermore, an optical pumping at 980 nm is also possible due to the
low reflectivity of the DBR at this wavelength. After release, both membranes exhibit
the same mechanical properties. This example, among several others, demonstrate the
ability to control the optical and mechanical properties of the DBRs independently.
Chapter 4
Technology for air-gap based
microcavity devices
4.1 Bulk and surface micromachining
For the implementation of MEMS and MOEMS, two different technological approaches
are relevant: bulk micromachining and surface micromachining [78]. In the first case,
e.g., backside etching of the substrate is necessary to form microstructures and free
standing membranes. MEMS for optical communication systems are implemented
based on this technology [79]. However, using bulk micromachining technology, only a
single free standing membrane or a multi layer stack can be fabricated. In order to im-
plement multiple free standing membranes (e.g. multi air-gap Fabry-Pérot filter [24]),
surface micromachining, a totally different concept, is needed. For this, sacrificial lay-
ers are used to form microstructures from the deposited or grown thin films on the
substrate. This involves several technological and fabrication steps [80]. Fabrication
steps like lithography, dry etching, deposition, annealing, sputtering are well studied
and controlled. Nevertheless, surface micromachining still presents a challenge since
crucial techniques like wet chemical etching of the sacrificial layer, stiction and stress
engineering of the released membranes have not been properly controlled up to now.
Even though the sacrificial layer is removed during the structuring process, it plays a
major role in determining the quality of the resulting microstructures. Thus, the sacri-
ficial layer defines the geometrical and structural as well as the optical properties of the
released membranes. In crystalline semiconductor systems, the sacrificial layer should
be lattice matched to the grown films in order to avoid strain and fractures. Several
microstructures involving sacrificial layers were implemented based on the III-V com-
pound system [81—83]. For microstructures based on amorphous thin films, different
materials can be used for the sacrificial layers [84—94]. After the release of the thin
films by wet chemical etching of the sacrificial layer, two main technological phenomena
lead to a malfunctioning of the structures: stiction and buckling. Stiction occurs due
to capillary, electrostatic and Van der Waals forces as well as hydrogen bridging [95],
whereas buckling occurs due to compressive intrinsic stress. Stiction can be avoided by
using a CO2 critical point drying techniques, which are commercially available. Buck-
4 Technology for air-gap based microcavity devices 37
ling control is the subject of active research and is thoroughly investigated later on in
this thesis.
4.1.1 Photoresist based novel technology
For further investigations in this work, both bulk and surface micromachining are
used. Applying surface micromachining, MEMS structures designed for stress evalu-
ation (Figure 2.4) and process control are implemented. For this, the technology steps
described in Figure 4.1 (a-d) are used: the wafer is first coated with a photoresist act-
ing as sacrificial layer (Figure 4.1a). A PECVD process for a low temperature (60◦C)
deposition on the photoresist is developed. After the deposition of a dielectric layer
at low temperatures (PECVD) (Figure 4.1b) on top of the photoresist, the lateral
structure is lithographically defined. A reactive ion etching process using triflurometh-
ane (CHF3) and argon (Ar) is used to perform the vertical structuring (Figure 4.1c).
Afterwards, the membrane is chemically underetched, wet or dry (Figure 4.1d), and
remains fixed by the residual sacrificial layer parts underneath the larger supporting
posts. The underetching process is isotropic and can be controlled by the etch time.
The wet chemical underetching involves a solution of isopropanol and acetone followed
by a CO2 critical point drying process to avoid stiction of the suspended membrane.
Dry chemical underetching is obtained by a microwave barrel O2-plasma process.
Another technological approach, where the sacrificial layer can be completely re-
moved is presented in Figure 4.1 (e-h). This technology is used for the tunable and
non tunable optical devices presented in this work. After coating the substrate with
an arbitrary photoresist (here AZ1518), an optical lithography step is used to define
structures in this sacrificial layer (Figure 4.1e). In this case, the dielectric layer is de-
posited simultaneously, both on the photoresist and the substrate. Since the deposition
is isotropic, the side walls of the photoresist are also coated with dielectric material
(Figure 4.1f). The lateral and vertical structure definition (Figure 4.1g) is implemen-
ted by the same lithographic and dry etching steps used in the first approach. Finally,
the photoresist can be removed completely since the membrane is still fixed by the
coated holes defined by the first lithographic step (Figure 4.1h). The advantage of the
second approach compared to the first one is that rapid thermal annealing can be used
afterwards to reduce the material absorption. Furthermore, this approach has a higher
process feasibility since it does not depend on the underetching time.
4.1.2 PECVD dielectric membrane
Based on the surface micromachining technologies described in Figure 4.1, free stand-
ing dielectric membranes are implemented. For this, different technological silicon
nitride and silicon dioxide PECVD processes enabling a low temperature plasma en-
hanced deposition in a parallel plate reactor on common photoresist are developed.
This deposition is crucial since the plasma and the ion bombardment could damage
the photoresist. However, the processes were optimized enabling a smooth deposition.
38 4.1 Bulk and surface micromachining
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
substrate
photoresist
PECVD dielectric film
Figure 4.1: Two technological approaches for implementing optical devices and stress eval-
uation MEMS structures. Both are based on surface micromachining. (a-d) the sacrificial
layer is partially removed, (e-h) the sacrificial layer is completely removed.
Figure 4.2(a) shows a silicon nitride membrane implemented according to the tech-
nological steps described in Figure 4.1 (a-d). The square membrane is positioned by
four suspensions connected to supporting posts. The posts are fixed to the substrate
by the residual photoresist sacrificial layer after the membrane release. A membrane
implemented by the technology described in Figure 4.1 (e-h) is shown in Figure 4.2(b).
The circular membrane and the suspensions are fixed by supporting posts with post
holders. This technology enables the fabrication of suspended membranes with surface
micromachining and a complete removal of the sacrificial layers by using a dry or wet
chemical sacrificial layer etching.
supporting posts post holder
suspension
sacrificial layer
(a) (b)
air-gap
circular membranesquare membrane
Figure 4.2: Scanning electron micrographs of - (a) free standing silicon nitride membrane
implemented by the technology described in figure 4.1(a-d). The supporting posts of this
membrane are fixed by the residual sacrificial layer. (b) Silicon nitride membrane implemented
by the second approach 4.1(e-h). The supporting posts are fixed by the post holders.
These two scanning electron micrographs show that in both cases, stiction can be
4 Technology for air-gap based microcavity devices 39
avoided. However, the stress related buckling due to the intrinsic stress is observed
in both cases. This buckling affects the mechanical (tuning) and the optical (cavity
length, wave propagation, etc.) properties of the tunable optical devices. Therefore,
complete control of the buckling is highly desired. This control problem is a main focus
in this thesis (see chapters 5 and 6).
Chapter 5
PECVD stress engineering
5.1 Stress control of PECVD dielectric material
The material properties of dielectric material were described in chapter 3.2. Stress,
composition, and optical properties of thin films were investigated by varying the com-
mon PECVD parameters such as temperature, pressure, and gas flow.
This chapter deals with the specific material stress engineering and the influence
on the device relevant properties (cavity length, radius of curvature and filter charac-
teristics). The stress contol is achieved by tuning the plasma excitation frequency.
The strain in crystalline semiconductors results from a mismatch in the lattice
constant of two different materials. Thus, by varying the alloy composition of the
crystalline material, the strain induced stress can be controlled. Unlike in semicon-
ductor materials, the stress of amorphous materials is quite difficult to control, since
no defined internal lattice structure is involved. Furthermore, the influence of the pro-
cess parameters (e.g. gas flow, pressure, temperature, etc.) on the stress is negligible
(see chapter 3). Generally, bulk stress control in dielectric amorphous layers can be
achieved by tuning either the thermal or the intrinsic component. Changing the depos-
ition temperature provides considerable composition changes in the stoichiometry of
the layers. Therefore, it is highly preferable to tune the intrinsic part without changing
the deposition temperature.
In this case, the plasma excitation frequency, which influences the ion bombardment
energy, is the key parameter in controlling the material intrinsic stress [96, 97]. The
influence of the frequency on the intrinsic stress of thin silicon nitride and silicon
dioxide films was described in earlier studies [98]. It has been shown that at high
frequencies (13.56MHz) the stress of silicon nitride is tensile whereas at low frequencies
(50 kHz) the stress is compressive. The lower the excitation frequency the more ions in
the plasma are able to oscillate according to the alternating electric field and, hence,
transfer energy to the growing silicon nitride film, causing a densification. Due to both,
the “ion peening” process and the implementation of high energetic H+ ions with low
atomic mass into spaces in the growing film, the film will tend to stretch out with
5 PECVD stress engineering 41
respect to the substrate, and causes compressive stress. At high frequencies, not all
the ions can follow the alternating field; there is less significant ion energy transfer,
and the film is less dense and contains an amount of microvoids and hydrogen atoms.
The resulting Si-N-H bonds tend to contract causing a tensile stress. On the other
hand, studies showed that the stress of silicon dioxide is not strongly affected by the
RF frequencies.
For the macroscopically averaged stress investigation, a PECVD is used for imple-
menting thin silicon nitride and silicon dioxide films at low (60 ◦C) and high (300 ◦C)
temperatures on silicon wafers. The homogeneous intrinsic bulk stress is adjusted by
varying the duty cycle Ψ of the low (130 kHz) and the high (13.56MHz) frequencies.
The duty cycle Ψ is given in equation 5.1, where tHF and tLF represent the time periods
of the high and low plasma excitation frequencies of one cycle during the deposition
process, respectively.
Ψ =
(tHF − tLF )
tHF + tLF
(5.1)
All other parameters are kept constant throughout the investigation. Table 5.1
shows the deposition parameters of silicon dioxide and silicon nitride used for this
investigation. The thermal stress of the deposited silicon dioxide and silicon nitride is
calculated according to equation 2.62. For ESi3N4 = 304 GPa, νSi3N4 = 0.24, αSi3N4 =
3.3 · 10−6 ◦C−1, ESiO2= 73.1 GPa, νSiO2 = 0.17, αSiO2 = 0.5 · 10−6 ◦C−1 and αSi = 2.6 ·
10−6 ◦C−1, the thermal stress of silicon nitride deposited at 300 ◦C and 60 ◦C is found
to be −78.4MPa (tensile) and −11.2MPa (tensile), respectively. For silicon dioxide,
the thermal stress is +51.78MPa (compressive) at 300 ◦C and −7.4MPa (tensile) at
60 ◦C.
Table 5.1: PECVD process parameters for silicon nitride and silicon dioxide deposited at
300 ◦C and 60 ◦C for the stress investigation.
Process parameters Si3N4 :Hx SiO2
2% SiH4−N2 flow / sccm 1000 1500 430 430
NH3 flow / sccm 20 5 0 0
N2O flow / sccm 0 0 710 710
HF-power / W 20 20 20 20
LF-power / W 20 20 20 20
Duty cycle Ψ Variable
Temperature / ◦C 300 60 300 60
Pressure / torr 0.65 0.65 1 1
Using the set-up described in 2.2.2, the stress is measured along three directions
on the wafer. For this, the one dimensional array of the reflected spot pattern is
aligned along different directions (< 010 >, < 001 >, < 011 >) which refer to the
42 5.1 Stress control of PECVD dielectric material
crystal structure of the substrate. The three resulting stress values are averaged and
the standard deviations are taken as error bounds.
Figure 5.1 shows the total macroscopically averaged bulk stress of silicon nitride and
silicon dioxide deposited at 300 ◦C. In this diagram, the stress is plotted as a function
of the duty cycle of the plasma excitation frequencies. Three main regions of this
diagram are interesting and correspond to the compressive (positive values), tensile
(negative values) and low (±25MPa) stress region. The total bulk stress of silicon
nitride deposited at 300 ◦C is varied in a wide range between +850MPa (Ψ = −1)
compressive and −300MPa tensile (Ψ = 1). Applications requiring stress free silicon
nitride layers, corresponds to the region between Ψ = 0.43 and Ψ = 0.54. The total
macroscopically averaged bulk stress of silicon dioxide shows, as expected, no strong
dependence on the plasma excitation frequencies. The stress values of silicon dioxide
deposited at 300 ◦C are for the most Ψ range nearly +200MPa. This seems to be the
result of the composition and the bonds morphology of the material. Regarding the
thermal stress values, the intrinsic stress of silicon nitride deposited at 300 ◦C varies
between +930MPa and −220MPa, whereas the intrinsic stress of silicon dioxide is
approximately +150MPa.
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 0,50 0,75 1,00 1,25
-400
-200
0
200
400
600
800
1000
Te
ns
ile
C
om
pr
es
si
ve
Silicon nitride :        σth= -78.4 MPa (calculated)
Silicon dioxide :       σth= 51.78 MPa (calculated)
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Frequencies duty cycle: Ψ = (tHF-tLF) /  (tHF+tLF)
 Silicon nitride (Si3N4) deposited @ 300 ° C)
 Silicon dioxide (SiO2) deposited @ 300 ° C)
Figure 5.1: The dependence of the total macroscopically averaged bulk stress of silicon nitride
and silicon dioxide deposited at 300 ◦C on the PECVD plasma excitation frequencies (duty
cycle Ψ).
Similar stress measurements are performed on silicon nitride and silicon dioxide
layers deposited at 60 ◦C. The total macroscopically averaged bulk stress is investigated
in a range of 0.55 < Ψ < 0.61, and varies from +121MPa compressive (Ψ = 0.55) to
−36MPa tensile (Ψ = 0.61) (Figure 5.2). At this temperature, silicon dioxide shows
a low thermal stress value of +7.4MPa. Thus, an average value of +17MPa can be
deduced for the intrinsic stress.
5 PECVD stress engineering 43
0,40 0,45 0,50 0,55 0,60 0,65 0,70
-150
-100
-50
0
50
100
150
200
Te
ns
ile
C
om
pr
es
si
ve
Si3N4 deposited @ 60 ° C       σth= -11.2 MPa (calculated)
SiO2 deposited @ 60 ° C         σth= 7.4 MPa (calculated)
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Frequencies duty cycle: Ψ = (tHF-tLF) /  (tHF+tLF)
Figure 5.2: The dependence of the total macroscopically averaged bulk stress of silicon nitride
and silicon dioxide deposited at 60 ◦C on the PECVD plasma excitation frequencies (duty cycle
Ψ).
Macroscopically averaged stress results provide an overview of the stress in PECVD
dielectric layers and enable coarse stress control in a wide range. However, this method
of stress evaluation does not meet the requirements for microscale applications, since
several crucial issues in the implementation of microdevices cannot be addressed by
considering only the macroscopically averaged stress measurement. For example, in-
homogeneities within the layers, as well as wafer border effects contributing to the
averaged stress values, can not be distinguished. The effects of the layer interfaces
(e.g. the sacrificial medium) and of the process steps on the stress are crucial in mi-
cromachined fabrication. Therefore, an in-process technology for monitoring the stress
as a function of the lateral position with high spatial resolution (microscopically de-
tected stress) is required.
For this purpose, the structures described in 2.2.2 are used. The microdetected
stress is evaluated using equation 2.64, with ESiO2 = 73.1 GPa, ESi3N4 = 304 GPa,
νSiO2 = 0.17, νSi3N4 = 0.24, di = 10µm (the distance between the anchor of the
actuator beam and the center of rotation), la (value to be measured (approximately
500µm), the length of the actuator beams), li = 228µm (the length of the indicator
beams) and δi (the deflection between the two indicators). The lateral extension of the
MEMS structure is given by 2la and 2li.
Using these microstructures, the microdetected stress can be adjusted in a range of
interest. Several structures were implemented by differently stressed silicon nitride and
silicon dioxide layers deposited at low temperatures. Figure 5.3(a) shows a microscope
image of such a MEMS structure fabricated from stress free silicon dioxide deposited
at 60 ◦C. In this case, the perfectly facing two indicator beams indicate very low bulk
stress in the lateral direction. The released actuators tend to neither expand nor to
contract and thus, the indicators do not rotate. This result is in agreement with
44 5.1 Stress control of PECVD dielectric material
macroscopically averaged stress measurements of silicon dioxide (Figure 5.2) showing
low stress values.
However, in some cases, unexpectedly stressed structures implemented by the same
material are observed. These structures were located close to the wafer border where
the layer deposition is laterally inhomogeneous because of the “border effect”. Fig-
ures 5.3(b) and 5.3(c) show such stressed silicon dioxide layers, with +70MPa and
+100MPa, respectively. The released actuators expand due to the compressive bulk
stress and thus rotate the indicators in clockwise direction.
(b) σ = ~ 70 MPa
(a) σ = ~ 0 MPa
(c) σ = ~ 100 MPa
Figure 5.3: MEMS structures for detecting silicon dioxide microstress at different lateral
positions on the wafer. (a) silicon dioxide layer deposited at 60 ◦C with very low bulk stress.
(b) and (c) show stressed layers with +70MPa and +100MPa, respectively.
The microstress of silicon nitride deposited at 60 ◦C is also studied. Figures 5.4(a),
(b) and (c) show stress detection microstructures implemented by silicon nitride lay-
ers deposited using different duty cycle values of Ψ = −1 (+130MPa), Ψ = 0.567
(+715MPa) and Ψ = 0.588 (+980MPa), respectively.
For certain duty cycles Ψ, different local stress values are observed at different
positions on the wafer. This stress variation can reach 15% for silicon nitride and
30% for silicon dioxide. A relation between the position of the MEMS structures on
the wafer and the related stress is observed. However, this correlation is not due to
physical effects such as crystal orientation and lattice mismatch; stability and reliability
of the process, as well as homogeneity of the dielectric layers (thickness and density)
strongly affect the lateral stress variation across the wafer. The sacrificial layer also
plays a major role. These stress fluctuations (15% for silicon nitride and 30% for silicon
dioxide) are deduced by studying several process runs. These deviations, which are
detected by the microstructures, remain undetectable by the macroscopically averaged
5 PECVD stress engineering 45
(b) σ = ~ 320 MPa
(c) σ = ~ 890 MPa
(a) σ = ~ 130 MPa
Figure 5.4: MEMS structures for detecting silicon nitride microstress at different lateral
positions on the wafer. (a) silicon nitride layer deposited at 60 ◦C with +130MPa bulk
stress. (b) and (c) show stressed layers with +320MPa and +890MPa, respectively.
stress measurement techniques. Therefore, the local stress evaluation is of utmost
importance when implementing microdevices.
The detected microstress and the averaged macrostress are correlated, in the sense
that the average of the sum of all microstress values along one direction should agree
with the measured macrostress value (for this direction). In order to compare these
two methods, a weakly stressed silicon dioxide layer on silicon substrate (Ψ = 0.481)
is considered, since relative stress fluctuations can be detected more precisely at small
absolute stress values. The macroscopically averaged stress is measured along three dir-
ections on the wafer (< 100 >,< 111 >,< 110 >) corresponding to the defined angles
of the measurement set-up (0◦, 45◦and 90◦), respectively (Figure 5.5). The standard
deviation of these three measurements are included as error bounds. The microscop-
ically detected stress is measured on different closely neighboring points on the wafer.
The stress average value of points within the narrow area defining one direction is com-
pared to the corresponding macrostress. In this case, error bounds are formed from
standard deviation of many points within this area and stress values of microstructures
located close to the wafer border are considered.
Figure 5.5 shows the results of this comparison. It can be seen that the microscop-
ically detected and macroscopically averaged stress values do not differ significantly,
and thus macroscopically averaged stress method will be sufficient for stress estimation.
Microscopic stress detection, on the other hand, seems to be a reliable low cost method
to generate process control structures for precise stress evaluation and for detecting
the inhomogeneity in various defined positions on the wafer close to the microdevices.
Finally, fabrication tolerances and requirements of the micromachining applications
dictate the choise of micro- or macrostress detection measurement techniques.
46 5.2 Impact of stress on optical and mechanical layer properties
0 15 30 45 60 75 90
-200
-150
-100
-50
0
50
100
150
200
Spatial diffraction pattern
(one dimentional array of laser spots)
Wafer
90°
45°
0°
C
om
pr
es
si
ve
Te
ns
ile
Silicon dioxide layer deposited @ 60 °C   Ψ= 0.481
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Macro stress measurement orientation / degree
 microscopically detected stress
 macroscopically estimated stress
Figure 5.5: Evaluation of the correlation between the microstress detection and the mac-
rostress averaging by considering a low stressed silicon dioxide layer deposited at 60 ◦C.
5.2 Impact of stress on optical and mechanical layer
properties
The impact of stress on released membranes in air-gap microcavities is highly significant
and may affect the functionality of the devices. Thus, optical characteristics in air-
gap microcavity devices (Fabry-Pérot filters, VCSELs, detectors, etc.), such as tuning
efficiency and light coupling factor in the cavity, are strongly affected by the mechanical
behavior of the suspended membranes. Sophisticated novel devices, such as tunable air-
gap VCSELs, involve several complex correlations and require careful adjustments of
their optical properties (e.g. gain profile, wave propagation, light coupling and tuning
efficiency) and their mechanical properties (e.g. cavity length, shape and ROC). For
instance, the optical excitation of these devices is strongly dominated by the shape of
the suspended membrane [99]. Thus, successful implementation of such devices requires
precise tailoring to achieve the best agreement on all these properties. The geometry
and the stress in the microstructures (membranes) influence several highly relevant
device characteristics, such as Lcav, ROC, FWHM as well as the resonance frequency
(channel selection). These device characteristics can thus be adjusted by varying the
geometry and the stress in the microstructures.
For this purpose, Fabry-Pérot air-gap filter devices1 consisting of 5 periods of
Si3N4−H/SiO2 bottom DBR and 5.5 periods Si3N4−H / SiO2 suspended DBR mem-
brane are implemented. The effect of the design of these devices on the mechanical
and optical properties is investigated using two geometries (Figure 5.6). The membrane
described in Figure 5.6(a) is implemented by the technology previously described in
Figure 4.1(e-h), whereas the membrane in Figure 5.6(b) is fabricated by the technolo-
1The technological implementation of these devices is shown in details in appendix B.
5 PECVD stress engineering 47
IMA3
(b)
40 µm
150 µm
10 µm
l
IMA2
(a)
supporting posts underetch holes
Figure 5.6: Shematic top view of the two membrane designs. (a) by using the mask set
IMA2 (technology described in figure 4.1(e-h)) and (b) by using the set IMA3 (technology
described in figure 4.1(a-d)).
gical approach in Figure 4.1(a-d). Two mask sets are therefore necessary: IMA2 [100]
and IMA3 [101], which correspond to the membranes in Figures 5.6(a) and 5.6(b),
respectively. The circular membrane (40µm diameter) is fixed (in the lateral plane) by
four suspensions of 10µm lateral width. The length of the suspensions varies between
10µm and 80µm for the mask set IMA2 and between 10µm and 60µm for IMA3.
In the case of IMA2, the suspensions are connected to four square shaped support-
ing posts of 150 · 150µm2. Figure 5.7 shows a three dimensional view of a membrane
implemented by using the mask set IMA2.
D = 40
 µm
l Lcav
Si3N4
SiO2
Figure 5.7: Three dimensional view of the Fabry-Pérot filter membrane implemented by the
mask set IMA2.
48 5.2 Impact of stress on optical and mechanical layer properties
5.2.1 Cavity length
Cavity length control is achieved in this case by varying the geometrical design (e.g.
suspension lengths) and the stress gradient in the layers for both mask sets. Since the
stress of silicon dioxide is weakly affected by the duty cycle Ψ, only the stress of silicon
nitride is varied. The variation of Ψ throughout the investigation in this chapter is
related to the duty cycle variation during the silicon nitride deposition. Using different
combinations of these parameters, a wide range of passive cavity tuning is achieved.
The term passive cavity tuning is defined in this thesis as the tuning by means of
different parameters for several devices (i.e. not by micromechanical actuation).
Figure 5.8 shows the dependence of the cavity length Lcav on the suspension length
l at different Ψ values for the membrane design IMA2. The cavity length, measured
using the white light interferometry, can be varied in a wide range between 360 nm
(l=10µm and Ψ = −1) and 12.8µm (l=80µm and Ψ = 0.25). Normally, for a positive
stress gradient, the cavity length should increase with increasing suspension length;
consequently, a decrease in the cavity length with increase in suspension length should
be observed if the stress gradient is negative. However, it is very difficult in our case
to correlate the cavity length only with the stress gradient, because the geometry and
the motion dynamics of the membrane when this is released play a major role.
0 10 20 30 40 50 60 70 80 98
0,00
2,00
4,00
6,00
8,00
10,00
12,00
14,00
C
av
ity
 le
ng
th
 (L
ca
v) 
/ µ
m
Suspension length (l) / µm
Duty cycle
 Ψ = -1       Ψ = -0.75
 Ψ = -0.5    Ψ = -0.25
 Ψ = 0        Ψ = 0.25
IMA 2
Figure 5.8: The dependence of the cavity length Lcav on the suspension length l at different
duty cycle (Ψ) values of silicon nitride. The filters are implemented using the design IMA2
and consist of a cavity embedded between two bottom and top DBRs of 5 and 5.5 periods
(Si3N4 / SiO2), respectively.
The motion dynamics during the release of a membrane are extremely complicated
and require highly sophisticated theoretical models in order to be understood. These
theoretical model based calculations are not within the scope of this thesis. Never-
theless, these dynamics can be explained empirically by studying a relatively simple
membrane. Figure 5.9 shows such a released membrane implemented by the design
5 PECVD stress engineering 49
IMA2 and consisting of three supporting posts and suspensions. A break in symmetry
is introduced in order to study the effect of the geometry on the released membrane.
The membrane is held by three suspensions that are attached to the three supporting
posts (1,2 and 3). While two suspensions are symmetrically attached to the corners of
supporting posts (2) and (3), the third suspension is connected to the middle of the
edge of supporting post (1).
supporting post 3
supporting post 1
supporting post 2membrane
ba
c
Figure 5.9: A released membrane implemented by the mask set IMA2. The membrane is
unsymmetrically supported by three suspensions connected to edge a and corners b and c.
When the membrane is in a state of static equilibrium (Figure 5.9), two main
observations can be made. First, the membrane is bent upwards, and second, the
stress gradient of the supporting posts (1, 2 and 3) is negative and all edges (except
the ones which are connected to the suspensions (2 and 3)) tend to pull down towards
the substrate. Corners (b) and (c) show a positive gradient, whereas the gradient of
border edge (a) is negative. In this case, it is difficult to find out whether the structure
exhibits a positive or a negative gradient.
When the membrane is released, independent of the geometry of the membrane,
two scenarios are possible for the stress distribution along the vertical structure of the
membrane:
• a distribution with a negative gradient
• a distribution with a positive gradient.
In the first scenario, where the gradient is negative, the released areas of the sup-
porting posts of the membrane tend to pull down towards the substrate. However,
due to the length of the suspensions, it may happen that the membrane reaches the
substrate without finding a relaxation optimum, which forces the membrane to deflect
upwards, searching for an equilibrium. In this case, the corners (b) and (c), where the
deflection forces are weak due to the small area, deflect upwards. Border edge (a) of
the supporting post (1) still deflects downwards because of the enormous forces applied
to it. Figure 5.9 shows the equilibrium status in this case. Here, it is important to
note that, depending on the stress distribution, several equilibriums are possible.
In the second scenario, the stress gradient is positive and the deflection is upwards.
Corners (b) and (c) and the membrane deflect upwards after the release. Because the
50 5.2 Impact of stress on optical and mechanical layer properties
suspensions are not long enough for the system to reach an equilibrium, the suspension
connected to border edge (a) pulls down forcing this border downwards. An equilibrium
is then reached. However, this scenario is less probable due to two main facts. First,
the other edges and corners should exhibit a positive gradient and pull upwards and
second, corners (b) and (c) should pull downwards first due to their small mechanical
resistance. This is indeed not the case, and thus it can be concluded that the gradient
is negative.
Figure 5.10 shows the relationship between the suspension lengths and the cavity
length at different Ψ values for the design IMA3. For this design, the absolute cavity
length variations are smaller than for IMA2. The smaller absolute passive tuning of
the cavity length is due to the dimension of the supporting posts. For design IMA3,
the area of the supporting posts is bigger and thus, they are less influenced by the
stress than are those of IMA2.
0 10 20 30 40 50 60 98
0,00
2,00
3,00
4,00
5,00
6,00
7,00
C
av
ity
 le
ng
th
 (L
ca
v) 
/ µ
m
Suspension length (l) /µm
Duty cycle
 Ψ = -1      
 Ψ = -0,75
 Ψ = -0,5    
 Ψ = -0,25
 Ψ = 0       
 Ψ = 0,25
IMA 3
Figure 5.10: The dependence of the cavity length Lcav on the suspension length l at different
duty cycle (Ψ) values of silicon nitride. The filters are implemented using the design IMA3
and consist of a cavity embedded between two bottom and top DBRs of 5 and 5.5 periods
(Si3N4 / SiO2), respectively.
Even though these results demonstrate the capability of a passive cavity length
tuning and control over a wide range, accurate predictions can only be made using
theoretical dynamic model calculations. The dependence of the cavity length of the
Fabry-Pérot filters implemented by both designs, IMA2 and IMA3, on the silicon nitride
duty cycle Ψ at different suspensions length l can be found in appendix F.
5.2.2 Radius of curvature
The stability of a Fabry-Pérot resonator was discussed in detail in chapter 2. It was
shown that the radius of curvature of the membranes is very important for the function-
5 PECVD stress engineering 51
ality of the devices. Furthermore, the shape of the membrane is extremely important
at a certain ROC absolute value. Thus, adjusting the ROC and the shape simultan-
eously, is a main goal of novel optoelectronics. To my best knowledge, no publications
dealing with the passive tuning of the shape and the ROC of the membranes exist.
By applying the same technology and designs as in the investigation of the cavity
length, the shape and the ROC are tuned passively in a wide range. Figures 5.11 and
5.12 show the dependence of the ROC on the suspension length and on the duty cycle
Ψ for the designs IMA2 and IMA3. Three possible membrane shapes are included in
the Figures:
• flat membranes
• concave membranes
• convex membranes.
Flatness of the membrane can approximated by either low concavity or low convex-
ity. The design IMA3 (Figure 5.12) enables a wider ROC range tuning than the IMA2
design. However, control of the ROC in the design IMA2 is more precise. Here, the
ROC is controlled between −0.994mm and +2.718mm, whereas +110µm convex and
−548µm concave membranes are reached. In the case of IMA3 design, the ROC is con-
trolled between −8.746mm and +15.17mm reaching +183µm convex and −1.8mm
concave membranes.
The dependence of the radius of curvature of the filter membranes for the designs
IMA2 and IMA3 on the duty cycle Ψ at different suspensions length can be found in
appendix F.
0 10 20 30 40 50 60 70 80 98
-1,50
-0,75
0,00
0,75
1,50
2,25
3,00
Fl
at
Fl
at
R
ad
iu
s 
of
 c
ur
va
tu
re
 (R
O
C
) /
 m
m
Suspension length (l) / µm
Duty cycle
 Ψ = -1       Ψ = -0.75
 Ψ = -0.5    Ψ = -0.25
 Ψ = 0        Ψ = 0.25
IMA 2
C
on
ve
x
C
on
ca
ve
Figure 5.11: The dependence of the radius of curvature ROC on the suspension length l at
different duty cycle (Ψ) values of silicon nitride. The membrane is implemented using the
design IMA2 and consists of 5.5 periods (Si3N4 / SiO2).
52 5.2 Impact of stress on optical and mechanical layer properties
0 10 20 30 40 50 60 98
-15,00
-10,00
-5,00
0,00
5,00
16,00 IMA 3
Fl
at
Fl
at
 
R
ad
iu
s 
of
 c
ur
va
tu
re
 (R
O
C
) /
 m
m
C
on
ca
ve
C
on
ve
x
Suspension length (l) / µm
Duty cycle
 Ψ = -1       Ψ = -0.75
 Ψ = -0.5    Ψ = -0.25
 Ψ = 0        Ψ = 0.25
Figure 5.12: The dependence of the radius of curvature ROC on the suspension length l at
different duty cycle (Ψ) values of silicon nitride. The membrane is implemented using the
design IMA3 and consists of 5.5 periods (Si3N4 / SiO2).
5.2.3 Filter characteristics
The full width at half maximum (FWHM) is an essential parameter when evaluating a
Fabry-Pérot filter. Lower values of the FWHM are needed for precise channel selection
in DWDM systems, while higher values are required for other applications. From the
industrial and commercial point of view, it is desirable to implement filters for several
wavelengths with different FWHM using the least technological investments, costs and
time. These points are investigated during this thesis. The first results show that by
varying the design (here, IMA2 and IMA3), the suspension length (l) and the stress
(here by using Ψ), a wide range of the FWHM and the resonance wavelengths can be
achieved by a single batch process.
Figures 5.13 and 5.14 show the dependence of the FWHM on the suspension length
and stress (by varying the duty cycle Ψ) for the designs IMA2 and IMA3. For the
design IMA2, the FWHM varies between 1.5 nm and 64.25 nm. The design IMA3
enables a FWHM variation between 2.75 nm and 44.25 nm. However, these FWHM
are not deduced from resonances of fixed wavelengths. The dependence of the FWHM
of the filters implemented by the designs IMA2 and IMA3 on the duty cycle Ψ for
different suspension lengths l can be found in appendix F.
The variation of the resonance wavelength for both designs, IMA2 and IMA3, is seen
in Figures 5.15 and 5.16, respectively. For the first design, the resonance wavelength
is varied between 1455 nm and 1595.75 nm, whereas the variation in the second case
is between 1450 nm and 1649 nm. The wavelength span is ∆λIMA2 = 140.25 nm for
the design IMA2 and ∆λIMA3 = 199nm for the design IMA3. The dependence of the
5 PECVD stress engineering 53
0 10 20 30 40 50 60 70 80 98
0
10
20
30
40
50
60
70
98
Fu
ll 
w
id
th
 a
t h
al
f m
ax
im
um
 (F
W
H
M
) /
 n
m
Suspension length (l) / µm
Duty cycle
 Ψ = -1
 Ψ = -0.75
 Ψ = -0.5
 Ψ = -0.25
 Ψ = 0
 Ψ = 0.25
IMA2
Figure 5.13: The dependence of the FWHM of the filters on the suspension length l at
different duty cycle (Ψ) values of silicon nitride. The filters are implemented using the design
IMA2 and consist of a cavity and two bottom and top DBRs of 5 and 5.5 periods (Si3N4 /
SiO2), respectively.
0 10 20 30 40 50 60 98
0
5
10
15
20
25
30
35
40
45
Fu
ll 
w
id
th
 a
t h
al
f m
ax
im
um
 (F
W
H
M
) /
 n
m
Suspension length (l) / µm
Duty cycle
 Ψ = -1
 Ψ = -0.75
 Ψ = -0.5
 Ψ = -0.25
 Ψ = 0
 Ψ = 0.25
IMA3
Figure 5.14: The dependence of the FWHM of the filters on the suspension length l at
different duty cycle (Ψ) values of silicon nitride. The filters are implemented using the design
IMA3 and consist of a cavity and two bottom and top DBRs of 5 and 5.5 periods (Si3N4 /
SiO2), respectively.
54 5.2 Impact of stress on optical and mechanical layer properties
0 10 20 30 40 50 60 70 80 98
1428
1479
1530
1581
1632
1683 IMA2
R
es
on
an
ce
 w
av
el
en
gt
h 
/ n
m
Suspension length / µm
Duty cycle
 Ψ = -1
 Ψ = -0,75
 Ψ = -0,5
 Ψ = -0,25
 Ψ = 0
 Ψ = 0,25
Figure 5.15: The dependence of the resonant wavelengths of the filters on the suspension
length l at different duty cycle (Ψ) values of silicon nitride. The filters are implemented using
the design IMA2 and consist of a cavity and two bottom and top DBRs of 5 and 5.5 periods
(Si3N4 / SiO2), respectively.
0 10 20 30 40 50 60 98
1428
1479
1530
1581
1632
1683
R
es
on
an
ce
 w
av
el
en
gt
h 
/ n
m
Suspension length / µm
Duty cycle
 Ψ = -1
 Ψ = -0,75
 Ψ = -0,5
 Ψ = -0,25
 Ψ = 0
 Ψ = 0,25
IMA3
Figure 5.16: The dependence of filters resonant wavelength on the suspension length l at
different duty cycle (Ψ) values of silicon nitride. The filters are implemented using the design
IMA3 and consist of a cavity and two bottom and top DBRs of 5 and 5.5 periods (Si3N4 /
SiO2), respectively.
5 PECVD stress engineering 55
resonance wavelength of the filters for the designs IMA2 and IMA3 on the duty cycle
Ψ at different suspensions length l can be found in appendix F.
Figure 5.17 shows white light interferometer pictures of differently shaped mem-
branes. A concave membrane shape with ROC = −0.31mm is shown in Figure 5.17
(a), a flat membrane with ROC = −184.71mm in Figure 5.17 (b) and a convex one
with ROC = 0.19mm is shown in Figure 5.17 (c).
Concave ROC = -0.31mm
ROC = -184.71 mmFlat
ROC = 0.19 mmConvex
(a)
(c)
(b)
Figure 5.17: White light interferometer pictures of a concave (a), flat (b) and convex (c)
suspended dielectric DBR membrane.
Chapter 6
Results of microcavity devices
6.1 Non tunable Fabry-Pérot filter
6.1.1 Solid stack filters
Figure 6.1 shows the optical spectrum of a solid dielectric filter. The Fabry-Pérot
filter is implemented by the PECVD at a temperature of 300 ◦C. It consists of a Si3N4
cavity (1λ optical length) embedded between two Si3N4/SiO2 DBRs. The bottom DBR
consists of 7.5 periods of λ/4 (optical length) alternating layers, whereas the top DBR
comprises 8 periods of the same material. The technological data of the filter can be
found in appendix B.
1200 1300 1400 1500 1600 1700
0,0
0,2
0,4
0,6
0,8
1,0
SiN/SiO2-filter demonstrator
on Si-Substrate:
Top DBR:
8 periods 0.243λ SiO2 /  Si3N4
cavity:
1λ Si3N4
Bottom DBR:
7.5 periods  0.243λ SiO2 / Si3N4
Central wavelength = 1570 nm
monochromator 
filter bandwidth: 10 nm
 Reflection measurement
 Theoretical model
          calculations
Wavelength / nm
R
ef
le
ct
iv
ity
Figure 6.1: Optical measurements and theoretical model calculations of a solid dielectric
Fabry-Pérot filter deposited at 300 ◦C. The filter consists of a λ Si3N4 cavity, a bottom
and top DBR of 7.5 and 8 periods, respectively. The design wavelength is 1550 nm. The
monochromator resolution is 10 nm.
The filter exhibits resonance at 1544 nm. At the resonance wavelength, contrary to
the theoretical prediction, the reflectivity seems to be only attenuated from 1 to 0.63.
6 Results of microcavity devices 57
This effect is due to the monochromator resolution of the optical spectrum analyzer,
which is only 10 nm in this case. Thus, the filter line is not fully resolved.
Figure 6.2 shows the characteristics of the filter measured at another position of
the chip using a better monochromator resolution. Theoretical model calculations have
been used to fit the filter optical spectrum; by using 0.243λ instead of 0.25λ for the
optical length of the DBR layers, a good match between the measured data and the
theoretical values has been obtained. This shift in the thicknesses is due to the PECVD
process shift and can be corrected by using an in-situ ellipsometer, for example.
1520 1525 1530 1535 1540 1545 1550 1555 1560
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
1,1
monochromator filter bandwidth: 0,5 nm
R
ef
le
ct
iv
ity
Wavelength / nm
Figure 6.2: Fabry-Pérot filter characteristics measured by using a high resolution (0.5 nm)
of the optical spectrum analyzer monochromator.
6.1.2 Air-gap filter
Based on the technology described in chapter 4 (Figure 4.1(a-d)), and using the design
IMA3, an air-gap dielectric filter is implemented. The technological process used for
the implementation of the air-gap filter is described in appendix B. The filter consists
of two DBRs embedding an air-gap cavity. The bottom DBR is implemented at 300
◦C and consists of 5 periods of Si3N4 / SiO2. The top DBR is implemented by 5.5
periods of Si3N4 / SiO2 deposited on a photoresist sacrificial layer (TI35ES) at 60 ◦C.
After releasing the top membrane, the cavity length increases due to the stress induced
buckling.
Figure 6.3 shows the spectrum of a filter (Tsn074II4) measured in transmission
by the measurement set-up described in [102]. The filter membrane is positioned by
four suspensions of 10µm length. In this case, the cavity length is 3.54µm (including
the initial 2.1µm thickness of the sacrificial layer) and the ROC is 1.92mm. The
transmission value of the filter is expressed in dB and the insertion loss of the filter is
−0.19 dB. Reflection measurements and the calculated transmission (1-reflection) of
two other filters (Tsn072II1V30X4Y3 and Tsn072II1V40X3Y2), also plotted in Figure
58 6.1 Non tunable Fabry-Pérot filter
1400 1450 1500 1550 1600 1650 1700
-30
-27
-24
-21
-18
-15
-12
-9
-6
-3
0
Insertion loss = -0.19 dB
Tr
an
sm
is
si
on
 / 
dB
Wavelength / nm
 tsn074II4
           measured in transmission
 Tsn072II1 V30 X4 Y3
            transmission calculated from the reflexion measurement
 Tsn072II1 V40 X3 Y2
            transmission calculated from the reflexion measurement
Figure 6.3: Optical spectra of air-gap Fabry-Pérot filters measured in transmission
(Tsn074II4) and reflection (Tsn072II1V30X4Y3 and Tsn072II1V40X3Y2). The insertion
loss of the filter Tsn074II4 is −0.19 dB.
6.3, show that the absorption is very low. Figure 6.4 shows the transmission and
reflection characterization of the filter. The FWHM of the filter is 5mm.
1450 1500 1550 1600 1650
0,0
0,2
0,4
0,6
0,8
1,0
Tsn074II4
5 periods bottom DBR
(300 ° C, ΨSi3N4 = 0.538, ΨSiO2 = -1)
2.2 µm Ti35 (sacrificial layer)
5.5 periods top DBR
(60 ° C, ΨSi3N4 = 0.538, ΨSiO2 = 0.48)
R
ef
le
ct
iv
ity
Wavelength / nm
 Transmission
 Reflexion
Figure 6.4: Measured optical spectra of the filter Tsn074II4. The transmission and reflection
of the filter are compared. The filter exhibits a FWHM of 5 nm.
6 Results of microcavity devices 59
6.2 Tunable Fabry-Pérot filter
The technological approach used for the implementation of the tunable device is de-
scribed earlier in chapter 4 (see Figure 2.4 (e-h)). Tuning of the device is achieved by
an electrothermal actuation of the top membrane using microheaters placed on the top
of the membrane. The vertical structure of the dielectric filters is similar to those de-
scribed in 6.1.2. The upper DBR consists of 5.5 periods of λ/4 layers of Si3N4 and SiO2,
while the lower one comprises 5 periods. A detailed description of the technological
implementation process can be found in appendix B. Figure 6.5(a) shows a micro-
scopic top view picture of the filter. The membrane is supported by four suspensions
of 40µm length. A 100 nm Chromium thin film is sputtered on the top DBR and is
wet chemically etched to define the meander like heaters (Figure 6.5(b)). After release
of the top DBR using an O2 plasma, the membrane bends upwards and increases the
air-gap cavity length to 5.12µm (Figure 6.5(c)). The radius of curvature of the circular
membrane is 1.94mm. The gradient stress of the whole upper DBR is calculated to
be 307MPa /µm, as a first approximation, by analyzing the upwards bent cantilevers
(test structures near to the device) according to equation 2.65.
(a) (b)
(c)
micro
heaters
Figure 6.5: (a) Microscopic top view picture of the tunable air-gap Fabry-Pérot filter
(IMA2), (b) 100 nm thick meander like chromium microheaters on the top of the released
DBR, (c) white light interferometry picture of the air-gap filter.
Figures 6.6(a) and 6.6(b) show SEMmicrographs of the tunable filter and the micro
Chromium heaters on the suspensions and on the membrane borders, respectively. The
filter exhibits a tunability of 15 nm /mA at a microheater resistance of 2 kΩ. Due to
the low number of DBR periods, the filter exhibits a FWHM of 8 nm. The FWHM
can be improved by increasing the number of DBR periods [103].
The spectral reflectivity (Figure 6.7 (a)) of the filter was measured using a single
mode fibre set-up [102], an optical spectrum analyzer and an Erbium doped fibre
amplifier (spectral range: 1.45µm to 1.65µm). In the spectral range studied, the
60 6.2 Tunable Fabry-Pérot filter
(a) (b)
heating micro
resistance
membrane
cavity
Figure 6.6: Scanning electron micrographs of: (a) the tunable Fabry-Pérot filter, (b) the
released top DBR with the meander like chromium microheaters.
wavelength varies linearly with the heating current (Figure 6.7 (b)) and thus deliver a
linear tuning characteristic.
1472 1495 1518 1541 1564
0,0
0,2
0,4
0,6
0,8
1,0
su
sp
en
si
on
   
l =
 6
0 
µm
m
em
br
an
e 
di
am
et
er
 =
 4
0 
µm
  SiO2 / Si3N4
 tunable filter
Tsp042IID M22 V60
R
ef
le
ct
iv
ity
Wavelength / nm
 0.0 mA
 0.5 mA
 0.7 mA
 1.0 mA
1512 1519 1526
0,0
0,2
0,4
0,6
0,8
1,0
R
micro-heaters
 = 2 KΩ
Tsp042IID M22 V60
 H
ea
tin
g 
cu
rr
en
t (
I) 
/ m
A
Wavelength / nm
su
sp
en
si
on
   
l =
 6
0 
µm
m
em
br
an
e 
di
am
et
er
 =
 4
0 
µm
R
ef
le
ct
iv
ity
 H
ea
tin
g 
cu
rr
en
t (
I) 
/ m
A
Figure 6.7: (a) Spectral characteristics of the tunable dielectric air-gap filter (measured in
reflection), (b) the linear tuning characteristic.
6 Results of microcavity devices 61
6.3 Non tunable VCSELs
The non tunable VCSEL structure consists of an active region embedded between two
PECVD dielectric DBRs. The DBRs are deposited at 300 ◦C using the parameters
listed in table C.7. The bottom DBR, starting with SiO2, comprises 12.5 periods of
Si3N4 / SiO2. The top DBR consists of 12 periods and is closed by a Si3N4 layer,
enabling a high refractive index contrast to the air. The technological fabrication
process of the VCSEL is described in Figure 6.8. The monolithic III-V (InP/InGaAsP)
semiconductor structure is grown by metal organic vapor phase epitaxy in an Aixtron
200/4 reactor using Trimethylindium (TMIn ), Trimethylgallium (TMGa), Phosphine
(PH3), and Arsine (AsH3) as precursors. The growth temperature is 680 ◦C and the
pressure is 100 mtorr. The structure consists of an InP substrate, a 200 nm thick
InGaAs etch stop layer, and the active region (Figure 6.8(a)). On top of the active
region, a 12.5 periods DBR of alternating optical λ/4 Si3N4 and SiO2 layers is deposited
(Figure 6.8(b)). In the next step, the whole structure is coated by 50 nm /300 nm
Ti/Au respectively and is bonded up side down (using Indium) to a copper heat sink
(Figure 6.8(c)). The Indium bonding of the structure to the heat sink allows good
heat dissipation, thus improving the functionality of the device. The InP substrate
and the InGaAs are then removed selectively by wet chemical etching (Figure 6.8(d)
and 6.8(e)). The last technological step is the deposition of a 12 periods DBR with the
same dielectric materials as the bottom one (Figure 6.8(f)). The whole technological
implementation process is described in detail in appendix B.
(a)
(b)
(c)
(d)
(e)
(f)
InP substrateInGaAs
Active region
Dielectric
DBR
Indium
Cu heat
sink
Figure 6.8: Technological implementation of the non tunable VCSEL: (a) MOVPE growth
of the active region and the etch stop layer, (b) PECVD deposition of the bottom dielectric
Si3N4/SiO2 DBR, (c) backside Indium bonding of the half VCSEL on a cupper heat sink, (d)
bulk micromachining removal of the InP substrate, (e) removing of the etch stop layer, (f)
PECVD deposition of the top dielectric Si3N4/SiO2 DBR.
The active region (Figure 6.9) is formed of three half-wave periods embedded
62 6.3 Non tunable VCSELs
between two InP spacers. Each single period of the active region contains a strain
compensated package of two Ga0.21In0.79As0.75P0.25 quantum wells (0.9 % compressive
strain) and three Ga0.47In0.53As0.75P0.25 barriers. The two quantum wells packages are
surrounded by two Ga0.29In0.71As0.63P0.37 cladding layers to position them at the an-
tinode of the resonant field. The quantum wells, with a maximum room temperature
photoluminescence (PL) at 1545 nm, are populated with photo induced carriers gen-
erated by absorption of the pump radiation (λpump˜980 nm) mainly in the cladding
layers. The top and bottom InP spacer layers have an optical thickness of λ/4 in order
to enhance the heat spreading of the active region.
Ga0.29In0.71As0.63P0.37 98.45 nm Confinement layer
Ga0.29In0.71As0.63P0.37 98.45 nm Confinement layer
InP 123.8 nm Spacer
Ga0.47In0.53As0.75P0.25 6 nm Barrier
Ga0.21In0.79As0.75P0.25 8 nm Well
Ga0.47In0.53As0.75P0.25 6 nm Barrier
Ga0.21In0.79As0.75P0.25 8 nm Well
Ga0.47In0.53As0.75P0.25 6 nm Barrier
1 
pe
rio
d
3 
pe
rio
ds
.
.
.
InP 123.8 nm Spacer
1 
pe
rio
d
3 
pe
rio
ds
1 
pe
rio
d
3 
pe
rio
ds
Figure 6.9: Structure of the active region of the non tunable VCSEL, grown by MOVPE.
Figure 6.10 shows the optical spectra of the dielectric DBRs, the half cavity res-
onance (bottom DBR and the active region) as well as the normalized VCSEL output
power characteristic. The half cavity resonance peak is positioned at the photolumin-
escence maximum of the quantum wells. However, there is an offset between the design
(1545 nm+5nm offset) and the lasing wavelength (1556.7 nm). This wavelength shift
at the lasing operation is due to several mechanism related either to the design or to
the material properties. Concerning the material properties, we have to distinguish
between the passive and the active ones. While the passive material exhibits linear de-
pendence on the temperature, the active material shows different linear and non linear
effects. Thus by increasing the temperature, the passive material expands, resulting
in a cavity length extension. In this case, the resonance characteristic of the cavity
shifts to longer wavelengths. Regarding the active material used in this device, two
operations related to the wavelength shifting are highly relevant: the continuous and
the pulsed optical or electrical pumping. The effect of both electrical pump opera-
tions on the wavelength shift is intensively studied in [104]. It has been shown that
for a continuous pumping and by increasing the injection current, a blue and a red
wavelength shift occur below and upon the threshold, respectively. In this case, the
blue shift is due to the strong increase of the carrier density, whereas the red shift is
related to the increase of the temperature in the device. For the pulsed operation case,
a blue wavelength shift below and upon the threshold with different slopes is observed.
6 Results of microcavity devices 63
The blue wavelength shift upon the threshold is attributed to the low temperature
operation of the device.
In order to ascertain whether the material properties or the design are responsible
for the wavelength shift in the VCSEL, the dependence of the emission characteristic
on the temperature should be considered.
1100 1200 1300 1400 1500 1600 1700
0,2
0,4
0,6
0,8
1,0
R
ef
le
ct
iv
ity
Wavelength / nm
 DBR reflection spectrum
 Half cavity spectrum
          measured in reflection
-80
-70
-60
-50
-40
 V
C
SE
L 
ou
tp
ut
 p
ow
er
 / 
dB
m
Figure 6.10: Optical spectra of the dielectric DBRs, the half cavity resonance (Bottom DBR
and the active region) as well as the normalized VCSEL output power characteristic.
The VCSEL is optically pumped by a pulsed 980 nm laser. The pulse repetition
interval (PRI) is 70µs, whereas the pulse width is 4µs. Figure 6.11 shows the spectral
emission of the VCSEL at room temperature. The VCSEL emits at 1556.7 nm and
exhibits a FWHM< 0.1 nm and a SMSR of 25 dBm. In the inset of this Figure, the
dependence of the VCSEL emission wavelength on the pumping power can be seen.
Since the VCSEL is pumped at room temperature, a red shift of the wavelength is
expected with increasing pumping power. However, a blue shift with a slop∆λ/∆ppump
0.03 nm /mW is observed. This blue shift is due to the pulsed operation and the heat
dissipation concept used for the VCSEL.
Assuming that the VCSEL is operated under low temperatures leading to a blue
shift, we can exclude the possibility of passive material expansion. Thus, the red
shift between the design and the implementation results is not related to the material
properties but rather to design problems occurring due to the PECVD inhomogeneities
during the deposition of the DBRs. It seems that parts of the DBRs contribute optically
to an expansion of the optical length of the cavity.
Figure 6.12 shows the pump power - output power (P-P) curve of the optically
pumped VCSEL. The VCSEL exhibits an output power of 0.5µW at room temperature
and has a threshold pump power of 35mW. At 6 ◦C, the pump threshold power is
30mW and the maximum output power is over 2µW.
64 6.3 Non tunable VCSELs
1564,50 1566,25 1568,00 1569,75 1571,50
-80
-70
-60
-50
-40 Pump laser power:
   61 mW
   67 mW
   72 mW
VC
SE
L 
ou
tp
ut
 p
ow
er
 / 
dB
m
Wavelength / nm
60 65 70 75
1566,6
1566,7
1566,8
1566,9
1567,0
 V
C
SE
L 
em
is
si
on
 W
L 
/ n
m
 Pump laser power / mW
Figure 6.11: Optical spectrum of the VCSEL during the optical excitation. The blue
wavelength shift is due to the pulsed operation.
20 25 30 35 40 45 50
0,0
0,5
1,0
1,5
2,0
Pe
ak
 o
ut
pu
t p
ow
er
 / 
µW
Laser pump power / mW
Pump laser:
WL = 980nm 
PRI = 70µs PW=35µs
 20 ° C
 15 ° C
 9 ° C
 6 ° C
Figure 6.12: Optical pumping power-output power dependence of the non tunable VCSEL
at different operating temperatures.
6 Results of microcavity devices 65
6.4 Tunable VCSEL
The tunable VCSEL is implemented by the two chip technological approach [16] and
is described in previous work [105]. The two main chips of the device are the lower
InP-based part1 and the bent dielectric top DBR membrane2. The two chips were
assembled3 to form the tunable active device. A schematic cross section of the VCSEL
is shown in Figure 6.13. The VCSEL consists mainly of an active region and an air-gap
cavity embedded between two DBRs of different material systems. The total cavity
length Lcav is defined by the length of the active region and the length of the air-gap
L0 between the curved dielectric mirror membrane and the solid part of the cavity.
InP-substrateFixed bottom
DBR
Active region:
GaInAsP QW‘s
Bulk micromachined
substrate
Curved suspended
Si3N4/SiO2 top DBR
Air-gap (L‘) GaAs
substrate
Figure 6.13: Schematic cross section of the two-chip tunable VCSEL: the lower semicon-
ductor chip comprises the bottom DBR and the active region, the upper part consists of a bent
DBR with tailored intrinsic stress.
The lower chip, comprising the bottom DBR and the active region, is grown mono-
lithically by a low-pressure MOVPE from TMIn, TMGa, PH3, and AsH3 precursors.
The bottom DBR is designed for a central wavelength (CWL) of 1575 nm and consists
of 48 InP/InGaAsP pairs yielding 99.8% nominal reflectivity.
The semiconductor part of the resonator has an optical thickness of 4.5λ and con-
sists of a periodic gain active region optimized for photopumping. The active region
is similar to the one described in section 6.3 and is made up of three half-wave peri-
ods embedded between two InP layers (Figure 6.14). A strain compensated pack-
age of two Ga0.21In0.79As0.75P0.25 quantum wells (0.9% compressive strain) and three
Ga0.47In0.53As0.75P0.25 barriers form one period. Two Ga0.29In0.71As0.63P0.37 cladding
layers embed the two quantum wells packages. The quantum wells, with a maximum
1The lower part is delivered by the Royal Institute of Technology, Kista, Sweden.
2The top DBR is implemented by the Institute of Nanostructure Technology and Analytics (INA)
at the University of Kassel, Germany.
3The assembly is done by the Technical University of Darmstadt (department of optical commu-
nications), Germany.
66 6.4 Tunable VCSEL
room temperature photoluminescence (PL) at 1545 nm, are populated with photo-
induced carriers generated by absorption of the pump radiation (λpump ∼ 980 nm)
mainly in the cladding layers. The top InP layer has an optical thickness of 2.25λ
to enhance the heat spreading [7] whereas the lower has a thickness of 0.75 λ. In
this case, the InP spacers are thicker then those of the non tunable VCSEL, since the
thermal conductivity of the dielectric DBRs are considered to be better than those of
the semiconductor ones. The thermal shift of the gain during lasing is compensated
by a tailored spectral detuning of 5 nm−30 nm between the PL and the CWL of the
laser (5 nm detuning in the non tuned state and 30 nm in the maximum tuned state).
Ga0.29In0.71As0.63P0.37 Confinement layer
Ga0.29In0.71As0.63P0.37 Confinement layer
InP 2.25 λ Spacer
Ga0.47In0.53As0.75P0.25 Barrier
Ga0.21In0.79As0.75P0.25 Well
Ga0.47In0.53As0.75P0.25 Barrier
Ga0.21In0.79As0.75P0.25 Well
Ga0.47In0.53As0.75P0.25 Barrier1 
pe
rio
d 
= 
0.
5 
λ
3 
pe
rio
ds
 =
 1
.5
 λ
.
.
.
InP 0.75 λ Spacer
1 
pe
rio
d 
= 
0.
5 
λ
3 
pe
rio
ds
 =
 1
.5
 λ
1 
pe
rio
d 
= 
0.
5 
λ
3 
pe
rio
ds
 =
 1
.5
 λ
Figure 6.14: Structure of the active region of the tunable VCSEL, grown by MOVPE.
The vertical design of the dielectric top DBR is given in Figure 6.15. It consists of
12.5 periods of silicon nitride (Si3N4) and silicon dioxide (SiO2) layers with an optical
length of λ/4 for the designed lasing wavelength (1550 nm). However, the air-gap length
Lcav, the membrane´s curvature as well as the deflection dynamics of the membrane
are adjusted by precisely tailoring the intrinsic stress of the dielectric layers across the
DBR. Relying on the PECVD stress investigation in chapter 5 (Figure 5.1), the stress
is varied in the vertical direction across the top DBR from −150MPa tensile (top layer
Si3N4 (25)) to +400MPa compressive (bottom layer Si3N4 (1)). Figure 5.1 is updated
by inserting the values used for the DBR layers in the diagram (Figure 6.16).
After the bulk micromachined backside etching of the GaAs substrate, the mem-
brane bends towards the GaAs substrate. Thus a radius of curvature of 4.5mm and
an air-gap length of 16µm are obtained (Figure 6.17). According to equation 2.32 and
Figure 2.2, this VCSEL is characterized by a half symmetric stable resonator. The
circular membrane has a diameter of 300µm and is fixed by 4 suspensions (600µm
length and 70µm width) to the substrate.
The injection of a small heating current through a thin metallic layer on the top of
the flexible suspensions enables micromechanical actuation of the membrane during the
pumping. Thus spectral tuning can be achieved. The metallic layer acts as a heating
resistor, converting the dissipated power into thermal heat which slightly increases the
length of the suspension beams.
6 Results of microcavity devices 67
Si3N4 (1)
[+400MPa]
Si3N4 (25)
[-150MPa]
SiO2 (2, 4,..., 24)
[+517MPa]
Si3N4 (3, 5,..., 23)
[+20MPa].
.
.
GaAs-substrate
Figure 6.15: Top dielectric Si3N4/SiO2 DBR (upper chip) with tailored intrinsic stress. The
stress of silicon nitride varies in the vertical direction during the deposition.
-1,0 -0,5 0,0 0,5 1,0
-400
-200
0
200
400
600
800
1000
Si3N4 (3, 5,..., 23)
Si3N4 (25)
Si3N4 (1)
SiO2 (2, 4,..., 24)
Te
ns
ile
C
om
pr
es
si
ve
To
ta
l b
ul
k 
st
re
ss
 (σ
to
ta
l) 
/ M
Pa
Frequency duty cycle Ψ = (tHF-tLF)/(tHF+tLF)
 Si3N4 deposited @ 300 °C
 SiO2 deposited @ 300 °C
Figure 6.16: Modified PECVD stress diagram (Figure 5.1) by including the stress values of
Si3N4 and SiO2 used along the bent top DBR.
Suspension
Suspended membrane:
differently stressed dielectric
layers in the vertical direction
2D – profile of the membrane
Cross view
Figure 6.17: White light interferometer picture of the bent top DBR.
68 6.5 Tunable high end receiver
The optical characterization of the tunable VCSEL is carried out by a fiber-to-fiber
transmission measurement set-up. The light of a pump laser (980 nm) is coupled into a
lensed fiber, delivering a Gaussian beam with a waist of 20µm diameter at the position
of the active region. The beam waist is well adapted to the ROC and the cavity length
according to equation 2.36, thus guaranteeing effective excitation of the fundamental
mode and suppression of higher order modes. The output laser light is coupled into
a second lensed fiber connected to an optical spectrum analyzer. The pump light is
absorbed in the active region and the quaternary material of the bottom DBR so that
there is no need to separate the pump light from the VCSEL output.
Figure 6.18 shows the optical spectrum of the tunable VCSEL under optical excit-
ation (980 nm, 30 − 50mW) in CW-operation at room temperature. The continuous
tuning range is 26 nm. The output power of the VCSEL from the bottom side is
300− 400µW. The device has a free spectral range of 47 nm and a side mode suppres-
sion ratio of 57dBm. The laser peak has a line width below the 0.1 nm resolution of
the optical spectrum analyzer. The relation between wavelength tuning and heating
power due to thermal actuation is exactly linear with a sensitivity of 7 nm/mW.
1515 1530 1545 1560 1575 1590
-70
-60
-50
-40
-30
-20
-10
0
Pump power:
30-50 mW cw
Output power:  Front side     100 µW
                         Back side     300-400 µW
Tuning range:
26 nm
O
ut
pu
t p
ow
er
 / 
dB
m
Wavelength / nm
Figure 6.18: Optical spectrum of the tunable VCSEL. A tuning range of 26 nm is achieved.
6.5 Tunable high end receiver
The two-chip concept described in [16] is used for the implementation of a wavelength
selective PIN photodiode. The results of this diodes have been recently published
[106—108]. Similar to the device presented in section 6.4, the wavelength selective
6 Results of microcavity devices 69
diode consists of two main parts. The lower chip comprises the bottom dielectric DBR
and the InP based PIN region. The upper part consists of a tailored bent dielectric
DBR (Figure 6.19). The bent top DBR comprises 8.5 pairs of λ/4 alternating Si3N4
and SiO2 layers deposited by the PECVD at 300 ◦C. After a bulk micromachining
etch process of the InP substrate, the top DBR bends due to the tailored intrinsic
stress in the layers. The radius of curvature is then 36mm. The lower chip consists
of a mesa structured PIN photodiode with top p-contacts and a 9 periods PECVD
dielectric (Si3N4 and SiO2) DBR, which is deposited on the backside of the thinned
InP substrate. An antireflection coating is deposited by the PECVD on top of the
PIN photodiode in order to avoid the back scattering of the non absorbed light in the
absorbing layer. The antireflection coating simultaneously serves as a passivation layer.
The electrical contacts are 50Ω coplanar pads. The tuning occurs by thermomechanical
actuation of the top membrane, similar to the device in section 6.4.
Bulk micromachined
substrate
Curved suspended
Si3N4/SiO2 top DBR
Air-gap (Lcav) InP
substrate
Tuning 
contact pads Thinned InP substrate
Dielectric 
DBR
Antireflection coating
PIN
P++ InGaAs
P-Contacts n-Contacts
n++ InGaAs
Figure 6.19: Schematic cross section of the tunable two-chip PIN photodiode.
Assembling the two chips results in a cavity length of Lcav = 32µm. Taking into
account the cavity length and the radius of curvature of the top membrane, the beam
waist of the incoming light can be calculated according to equation 2.36. A tunable
laser, sweeping over the wavelength range of interest in 0.05 nm steps, is used to record
the responsivity spectrum of the PIN photodiode. The responsivity of the PIN photodi-
ode is tuned thermally (by actuating the top DBR) over the spectral wavelength range
and is shown in Figure 6.20. A detailed consideration of the measurement set-up can
be found in [109]. The photodiode shows a FSR of 35 nm, a FWHM below 0.15 nm, and
a finesse exceeding 35. However, the tuning range is higher than the FSR. The tuning
characteristic between the selected wavelength and the dissipated electrical power in
the membrane suspensions is linear and has the value of 0.33 nm /mW. Furthermore,
the device shows a peak responsivity (Rpeak) of 0.30A /W and an insertion loss of
around 7 dB (3.5 dB in the optical domain). The photocurrent crosstalk from neigh-
boring channels is calculated to be−40 dB over the whole tuning range, when assuming
70 6.5 Tunable high end receiver
a channel spacing of 0.8 nm. The side modes, which can be seen in Figure 6.20 are due
to the mismatch between the incoming excitation light beam (a non Gaussian profile
in this case) of the fibre and the resonator geometry.
1530 1540 1550 1560 1570
-80
-70
-60
-50
-40
-30
-20
-10
0
78mA
25mA
0mA78mA 60mA50mA 25mA0mA
R
es
po
ns
iv
ity
 / 
dB
Wavelength / nm
Figure 6.20: Dependence of the responsivity of the tunable PIN photodiode on the wavelength
by several tuning currents.
Chapter 7
Related applications
7.1 Organic microcavities
7.1.1 PECVD materials and DBRs
Based on the results in chapters 5 and 6, we investigate the possibility of integrating
organic light emitting macromolecules in microcavity devices. The goal is to establish
a basic technology for novel organic microdevices like VCSELs and photodiodes. The
work presented in this section should be considered as a basic investigation yielding
promising first results.
In chapter 3, a low temperature PECVD deposition process was developed and
presented. In this case, the mechanical properties of the deposited material were an
important consideration as we optimized the technological aspect. However, regarding
macromolecules in microcavities, the optical characteristics of the deposited material
are predominant. For this reason, the absorption (by mean of the extinction coeffi-
cient k) of PECVD Si3N4 layers deposited at low temperatures (60 ◦C) is investigated
by varying the duty cycle (Ψ). Silicon dioxide is not considered since the hydrogen
bonds are, to our best knowledge, the reason for the absorption at a low temperature
deposition (see chapter 3). Using spectroscopic ellipsometry, several refractive indeces
and extinction coefficient dispersions are measured for different Ψ values. It has been
shown that for a duty cycle Ψ = −1, the extinction coefficient (proportional to the
absorption) of Si3N4 exhibits a minimum (Figure 7.2). Unfortunately, at this Ψ value,
the refractive index is also at a minimum (Figure 7.1). Furthermore, the stress is highly
compressive. However, at this point, the goal is to investigate the optical behavior of
the macromolecules in the cavities and therefore, this duty cycle is considered for all
the layers Si3N4 within this investigation. The effect of the stress on the mechanical
properties of the light emitting material is investigated elsewhere [110,111].
Spectroscopic ellipsometer measurements show difficulties fitting the Psi and Delta
data using a Cauchy model in the non absorbing wavelength range, leading to undula-
tions in the extinction coefficient dispersion curves in Figure 7.2. Several effects, such
as hydrogen absorption peaks, inhomogeneities the vertical and lateral directions in
72 7.1 Organic microcavities
the film structures as well as measurement artefacts, may be responsible for these un-
dulations. Hydrogen absorption peaks and measurement artefacts can be excluded as
direct undulation source. Furthermore, the amorphous nature of the material implies
the absence of birefringence. Thus, the most probable explanation for the deviation of
the Cauchy model causing the undulations are vertical inhomogeneities in the refractive
index of the films, resulting from the deposition conditions.
400 600 800 1000 1200 1400 1600
1,80
1,85
1,90
1,95
2,00
2,05
2,10
2,15
2,20
2,25
2%SiH4-N2          1500 sccm
NH3                     5 sccm
HF/LF power      20 W
Pressure             0.65 Torr
Temperature       60 ° C
R
ef
ra
ct
iv
e 
in
de
x
Wavelength / nm
 TSn067I, Ψ = -1
 TSn067II, Ψ = -0.75
 TSn067III, Ψ = -0.5
 TSn067IV, Ψ = -0.25
 TSn068I, Ψ = 0
 TSn068II, Ψ = 0.25
 TSn068III, Ψ = 0.5
 TSn068IV, Ψ = 0.75
 TSn069I, Ψ = 1
ψ = -1
R
ef
ra
ct
iv
e 
in
de
x
R
ef
ra
ct
iv
e 
in
de
x
Figure 7.1: The dependence of the refractive index dispersion of silicon nitride (deposited
at 60 ◦C) on the PECVD duty cycle Ψ. The data have been obtained using a spectroscopic
ellipsometer.
Based on the optical properties of the investigated materials, several Bragg mirrors
are implemented for the short wavelength range. Figure 7.3 shows the spectra of two
such DBRs deposited at a temperature of 300 ◦C and designed for the wavelength
range 350 nm−500 nm. The technological implementation process can be found in
appendix D. The two DBRs are measured by two different techniques, the spectroscopic
ellipsometer (TQ067, incident angle = 15◦) and the reflection measurement set-up
(TQ069, incident angle = 0◦). At this wavelength range, different incident angles
result in a spectral shifting and an attenuation of the reflectivity.
The spectral shift is due to different optical paths at different incident angles. On
the other hand, the reflection attenuation is due to the high absorption in the layers
at this wavelength. Normally, the reflectivity of a DBR implemented by low absorbing
material shows nearly no dependence on the measurement incident angle. For the case
of a DBR with materials of high absorption coefficient, it can be assumed that the
longer the optical path of the light, the higher is the absorption of the light. The
dependence of the reflectivity on the incident angle is shown in Figure 7.4. Thus, the
reflectivity of different DBRs decreases by increasing the incident angle. The slop of
the reflectivity-incident angle curves seems to be the same for all the measured DBRs.
Figure 7.5 shows the spectra of DBRs implemented at high and low temperatures
7 Related applications 73
250 300 350 400 450 500 550 600 650 700 750
0,00
0,02
0,04
0,06
0,08
0,10
0,12
0,14
0,16
0,18
2%SiH4-N2          1500 sccm
NH3                     5 sccm
HF/LF power       20 W
Pressure             0.65 Torr
Temperature       60 ° C
Ex
tin
ct
io
n 
co
ef
fic
ie
nt
Wavelength / nm
 TSn067I, Ψ = -1
 TSn067II, Ψ = -0.75
 TSn067III, Ψ = -0.5
 TSn067IV, Ψ = -0.25
 TSn068I, Ψ = 0
 TSn068II, Ψ = 0.25
 TSn068III, Ψ = 0.5
 TSn068IV, Ψ = 0.75
 TSn069I, Ψ = 1
ψ = -1
Ex
tin
ct
io
n 
co
ef
fic
ie
nt
Ex
tin
ct
io
n 
co
ef
fic
ie
nt
Figure 7.2: The dependence of the extinction coefficient of silicon nitride (deposited at 60
◦C) on the duty cycle Ψ. The data have been obtained using a spectroscopic ellipsometer.
400 450 500 550 600 650 700
0,0
0,2
0,4
0,6
0,8
1,0
 TQ067  Spectroscopic ellipsometer
          (incident angle = 15 degree)
 TQ069 Reflection measurement set-up
          (incident angle = 0 degree) 
R
ef
le
ct
iv
ity
Wavelength / nm
Figure 7.3: PECVD distributed Bragg reflectors deposited at high temperatures (300 ◦C) for
the short wavelength range.
74 7.1 Organic microcavities
15 20 25 30 35 40 45
0,65
0,70
0,75
0,80
0,85
0,90
 Incident angle / degree
R
ef
le
ct
iv
ity
 TQ050, 15 periods DBR, 60 °C
 TQ055, 15.5 periods DBR, 60 °C
 TQ067, 12.5 periods DBR, 300 °C
Figure 7.4: Dependence of the DBR reflectivity at a fixed wavelength on the light incident
angle.
and measured ellipsometrically at an incident angle of 15◦. The DBRs TQ067 and
Tsn033I are deposited at 300 ◦C on glass and silicon substrates, respectively. It can
be seen that the DBR deposited on the silicon substrate exhibits higher reflectivity.
This is due to the deposition temperature coupling since the substrate absorbs the light
at this wavelength. At 60 ◦C, the spectrum of the DBR TQ055 shows high material
absorption, which affects the FWHM and the reflectivity of the DBR.
Several rapid thermal annealing processes are incorporated in order to enhance the
reflectivity of such DBRs deposited at low temperatures. Even though the reflectivities
are noticeably increased, the technological relevance of such step is very weak since the
DBRs should be deposited on organic materials.
7.1.2 Organic half cavity
An organic solid state light emitting material should exhibit a high morphological sta-
bility, a high fluorescence quantum yield and a low reabsorption. These requirements
are met by the class of Spiro linked oligophenyles, of which Spiro-Sexiphenyl (Spiro-6f)
has already shown excellent results in amplified spontaneous emission (ASE) exper-
iments performed on spin coated films [112]. The rigid molecular structure of these
materials leads to amorphous glasses with high glass transition temperatures above
200 ◦C. The Spiro concept was extended by adding further oligophenyl chains, linked
to the core molecule via additional Spiro junctions [113]. The molecular structure of
the newly synthesized 4-Spiro2 used in this investigation is shown in Figure 7.6.
Four biphenyl and two sexiphenyl units are incorporated in the chemical structure
and thus form a stable amorphous glass with a glass transition temperature of 273
◦C. The moderate molecular weight of 4-Spiro2 when compared to polymeric materials
7 Related applications 75
300 350 400 450 500 550 600 650 700
0,0
0,2
0,4
0,6
0,8
1,0
R
ef
le
ct
iv
ity
Wavelength / nm
 TQ067, 12.5 periods DBR
           300°C, TT021016
           p-polarisation, 15° meas. ang.
 TQ055, 15.5 periods DBR
           60°C, AT020720
           p-polarisation, 15° meas. ang.
 Tsn033I, 12.5 periods DBR
           300°C, TT021016
           p-polarisation, 15° meas. ang.
Figure 7.5: Optical spectra of DBRs implemented by the PECVD at different temperatures
and on different substrates.
Physical properties Optical properties
Tg=273°C Tm=447°C λabs=353nm λem=429nm
4- Spiro2: 2,2’,7,7’-Tetra-(9,9’-spirobifluoren-2-yl)-9,9’-spirobifluorene
Figure 7.6: Chemical structure of the light emitting material 4-Spiro2. The glass and melt
temperatures are 273 ◦C and 447 ◦C, respectively. The absorption maximum is located at
353 nm, whereas the emission occurs at 429 nm. With kind permission of the mmCmm group
at the University of Kassel, Germany.
76 7.1 Organic microcavities
allows the preparation of thin films by vacuum vapor deposition. The optical proper-
ties of 4-Spiro2 are shown in Figure 7.7. The maximum intensity of the spontaneous
emission is located at 429 nm, whereas the absorption has a maximum at 353 nm. The
spectral shift of the absorption and emission characteristics ensures an optical pumping
with a nitrogen laser at 337 nm, far away from the emission wavelength.
300 350 400 450 500 550 600
0,0
0,2
0,4
0,6
0,8
1,0
No
rm
al
is
ed
 a
bs
or
pt
io
n 
an
d 
em
is
si
on
Wavelength / nm
4-Spiro2
 Absorption
 Emission
Figure 7.7: Absorption and emission spectra of 4-Spiro2. The results in this diagram are
courtesy of the mmCmm (with kind permission).
At a low pumping power density threshold of 3.2µJ / cm2, an amplified spontaneous
emission (ASE) spectrum was observed [114] for a 106 nm thick film of 4-Spiro2. Figure
7.8 shows the transition from the normal fluorescence spectrum to ASE for 4-Spiro2
by varying the pumping power density. At a pumping energy of 10µJ /cm2, the ASE
peak at 428 nm has a FWHM of 3.2 nm.
Using this material, a half cavity resonance is demonstrated. For this, a 15.5 periods
Si3N4/SiO2 DBR is deposited by the PECVD at 60 ◦C on a glass substrate. The
nominal reflectivity of the DBR is approximately 80%. The cavity is implemented by
92.22 nm of 4−Spiro2 and 56 nm Si3N4 serving as a cap layer and prohibiting the light
emitting material from oxidizing. Figure 7.9 shows the optical spectra of the lower DBR
and the half cavity resonance. However, the half cavity resonance is shifted to a higher
wavelengths and matches with a reflectivity of 60% of the lower DBR. Furthermore, the
half cavity resonance is located at approximately 450 nm, far away from the emission
wavelength of the organic material (4−Spiro2). In this case a lasing in a cavity is not
possible.
It is extremely difficult to design and characterize an organic material in a mi-
crocavity since the wavelength used in the measurements set-ups strongly affects the
optical and mechanical properties of these materials. Despite that, based on these
7 Related applications 77
300 350 400 450 500 550 600
0,0
0,2
0,4
0,6
0,8
1,0 Amplified Spontaneous
Emission (ASE) in 4-Spiro2
Pump laser power density
No
rm
al
is
ed
 in
te
ns
ity
Wavelength / nm
 119.4 µJ/cm2
 4.2 µJ/cm2
 1.9 µJ/cm2
Figure 7.8: Amplified spontaneous emission spectrum of 4-Spiro2. The results in this dia-
gram are courtesy of the mmCmm (with kind permission).
375 400 425 450 475 500 525 550
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
Spiro2
emission wavelength
λspiro2, em = 429 nm
 15.5 periods DBR (TQ055)
         + 92.22 nm Spiro2
         + 56 nm Si3N4
 TQ055 15.5 periods DBR
R
ef
le
ct
iv
ity
Wavelength / nm
Figure 7.9: Organic half cavity structure consisting of a 60 ◦C PECVD DBR (TQ055),
92.22 nm 4-Spiro2 and 56 nm Si3N4 cap layer.
78 7.1 Organic microcavities
results, a microcavity is implemented showing an ASE resonance. For this we used an
organic light emitting material (Small-Spiro-Octo) similar to the 4-Spiro2. The Small-
Spiro-Octo (2,2’,4,4’,7,7’-Hexaphenyl-9,9’-spirobifluoren, C61H40) shown in Figure 7.10
exhibits a glass and a melt temperatures of 181 ◦C and 291 ◦C, respectively.
Physical properties Tg=181°C Tm=291°C
Small-Spiro-Octo: 2,2’,4,4’,7,7’-Hexaphenyl-9,9’-spirobifluorene
Figure 7.10: Chemical structure of the Small-Spiro-Octo light emitting organic material.
With kind permission of the mmCmm.
The microcavity consists of a bottom PECVD DBR deposited at 120 ◦C, a cavity of
Small-Spiro-Octo and a top DBR. Figure 7.11 shows the optical spectra of the DBRs,
the half cavity resonance, the spontaneous emission of the Small-Spiro-Octo and the
ASE resonance in the cavity. The half cavity resonance, the second emission maxima
(Small-Spiro-Octo) and a reflectivity over 90% of the DBRs, coincide to narrow the
emission at 412.5 nm. The FWHM of the ASE resonance is 2.8 nm.
375 380 385 390 395 400 405 410 415 420 425
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
N
or
m
al
is
ed
 in
te
ns
ity
Wavelength / nm
 PECVD DBR (120 degree)
 Half cavity resonance by using
          Small Spiro-Octo
 Emission spectrum of Small Spiro-Octo
 ASE in the micro cavity
Figure 7.11: Optical spectrum showing the reflectivity of the DBRs, the half cavity resonance
and the ASE in a microcavity. As well as the spontaneous emission of the Small-Spiro-Octo.
7 Related applications 79
These results demonstrate the possibility of integrating organic light emitting ma-
terials in microcavities. This work was not pushed further in the context of this thesis.
However, extensive research is in progress as a cooperation between the Institute of
Nanostructure Technology and Analytics (INA) and the Institute for Macromolecu-
lar Chemistry and Molecular Materials (mmCmm), both at the University of Kassel,
Germany.
Chapter 8
Conclusion
The aim of this thesis was to establish a low-cost technology enabling the effective
integration of microcavity devices, e.g. into DWDM systems. Both parts, the tech-
nological optimization and the microcavity device implementation, are investigated
within the scope of this research. For this purpose, a standard parallel plate PECVD
reactor was employed. Besides the optical properties like the refractive index and the
absorption, the bulk macrostress of silicon nitride and silicon dioxide (deposited at
different temperatures) have been extensively studied. The bulk macrostress of silicon
nitride has been explicitly controlled, by varying the plasma excitation frequency, in a
wide range between +850MPa (compressive) and −300MPa (tensile) for 300 ◦C. The
stress of silicon dioxide is quite independent of the frequency. Thus, several DBRs
fulfilling specific optical and mechanical requirements were deposited.
Furthermore, a low-cost micromachining technology enabling the fabrication of mi-
crodevices by using a standard photoresist as a sacrificial layer has been developed.
For this purpose, low temperature (60 ◦C) silicon nitride and silicon dioxide processes
have been optimized. Based on this technology, MEMS structures for detecting the
microstress on different neighboring positions along the wafer have been implemented
and successfully applied.
Using differently designed air-gap devices (e.g. Fabry-Pérot filters) which are im-
plemented by this low-cost technology, the effect of the stress on these membranes
was extensively studied. For the first time, a batch process enabling the low-cost in-
tegration of differently shaped membranes (convex, concave and planar) with various
radii of curvatures and enclosing different cavity lengths was established. Fabry-Pérot
filters covering a spectral range over 190 nm and with several full width at half max-
ima between 1.5 nm and 64 nm have been demonstrated. The reliability of this novel
low-cost technology was demonstrated with an electrothermally tunable Fabry-Pérot
filter. The filter consists of an air-gap cavity embedded between a bottom and a bent
top dielectric DBRs, respectively. The filter exhibits a tuning range of 15 nm.
These novel technological achievements are used to fabricate microcavity devices
with tailored bendings of the top membranes. Thus, an optically pumped tunable
VCSEL with an optimized bent suspended top DBR was fabricated for enhancing the
8 Conclusion 81
fundamental mode and for suppressing the higher order modes. The desired bending
(radius of curvature, shape and cavity length) of the DBR is achieved by a specific
selective silicon nitride stress distribution across the vertical direction. The top DBR
is actuated electrothermally to achieve a tunability of 26 nm.
Based on the results of the tunable filter and the VCSEL, a tunable PIN photo-
detector was fabricated by using appropriate bent and non bent dielectric top and
bottom DBRs, respectively. The air-gap resonator of the PIN photodiode with a bent
membrane is tuned electrothermally and exhibits a responsivity tuning range of more
than the 35 nm free spectral range.
Based on these results, the integration of light emitting organic materials in mi-
crocavities was investigated. Distributed Bragg reflectors for the short wavelength
range (350 nm - 500 nm) have been optimized and deposited. Using two different or-
ganic light emitting materials within a microcavity, a half cavity resonance and an
amplified spontaneous emission were observed, respectively.
Throughout this work, the crucial stress issue in tunable microcavity devices is
adressed and studied. Tailored bent dielectric membranes with different shapes are
fabricated. Using this, microcavity devices with organic and inorganic materials are
demonstrated. On this basis, a considerable step is done in research towards the
commercialisation of low-cost microcavity devices.
A further investigation of this technology, leading for instance to a monolithic integ-
ration of a tailored bent dielectric membrane with an electrically pumped VCSEL, is
of high relevance for optical communications. Besides this, a main focus in subsequent
works could be the integration of active and passive polymers in microcavity devices.
Thus, organic microdevices such as VCSELs, photodetectors and filters, with integrated
organic driver chips may be an alternative innovative future technology. Furthermore,
these organic and inorganic devices could be applied in sensor and medical technologies
as well as in the automotive industry.
Bibliography
[1] S.V. Kartalopoulos, Introduction to DWDM Technology - Data in a Rainbow,
IEEE Press, 2000.
[2] R. Ramaswami and K.N. Sivarajan, Optical Networks - A Practical Perspective,
Academic Press, second edition, 2002.
[3] Richard Cuningham, “Cooling DWDM market should heat up in 2002,”
www.reed-electronics.com, elctronics news, July 2001.
[4] WinterGreen Reaserch Inc., “DWDM market opportunities, strategies, and fore-
casts, 2003 to 2008,” www.gii.co.jp, electronics news, November 2003.
[5] Pioneer Consulting, “Worldwide metro DWDM market to reach nearly $12.6
billion by 2006; slower growth expected through 2002, according to pioneer con-
sulting,” www.pioneerconsulting.com, March 2002.
[6] E. Mounier, “Optical MEMS and MOEMS for telecommunications,” Proceedings
of SPIE, vol. 4945, 2003.
[7] ElectroniCast Corp., “DWDM component market to reach $34.6 billion by 2009,”
www.photonics.com, May 2001.
[8] Photonics.Com, “$54 billion market forecast for DWDM, reports KMI,”
www.pho-tonics.com, electronics news, February 2001.
[9] S. Irmer, J. Daleiden, V. Rangelov, C. Prott, F. Römer, M. Strassner, A. Tarraf,
and H. Hillmer, “Ultralow biased widely continuously tunable fabry-pérot filter,”
IEEE Photonics Technology Letters, vol. 15, no. 3, pp. 434—436, March 2003.
[10] M.C. Larson, B. Pezeshki, and J.S. Harris, “Vertical coupled-cavity microint-
erferometer on GaAs with deformable-membrane top mirror,” IEEE Photonics
Technology Letters, vol. 7, no. 4, pp. 382—384, April 1995.
[11] D. Hohlfeld and H. Zappe, “An all-dielectric tunable optical filter based on the
thermo-optic effect,” IOP, Journal of Optics A: Pure and Applied Optics, vol. 6,
no. 6, pp. 504—511, June 2004.
[12] P. Tayebati, P. Wang, M. Azimi, L. Maflah, and D. Vakhshoori, “Microelec-
tromechanical tunable filter with stable half symmetric cavity,” Electronics Let-
ters, vol. 34, no. 20, pp. 1967—1968, October 1998.
82
Bibliography 83
[13] D. Vakhshoori, P. Tayebati, C.-C. Lu, M. Azimi, P. Wang, J.-H. Zhou, and
E. Canoglu, “2mWCW singlemode operation of a tunable1550nm vertical cavity
surface emitting laser with 50nm tuning range,” Electronics Letters, vol. 35, no.
11, pp. 1—2, May 1999.
[14] M. Maute, F. Riemenschneider, G. Boehm, H. Halbritter, M. Ortsiefer, P. Meiss-
ner, and M.-C. Amann, “Micro-mechanically tunable long wavelength VCSEL
with a burried tunnel junction,” Electronics Letters, vol. 40, pp. 430—431, April
2004.
[15] M. Ortsiefer, R. Shau, G. Boehm, F. Koehler, and M.-C. Amann, “Low-threshold
index-guided 1.5 µm long-wavelength vertical-cavity surface-emitting laser with
high efficiency,” Applied Physics Letters, vol. 76, no. 16, pp. 2179—2181, April
2000.
[16] F. Riemenschneider, M. Aziz, H. Halbritter, I. Sagnes, and P. Meissner, “Low-
cost electrothermally tunable optical microcavities based on GaAs,” IEEE
Photonics Technology Letters, vol. 14, no. 11, pp. 1566—1568, November 2002.
[17] F. Riemenschneider, M. Maute, H. Halbritter, G. Boehm, M.-C. Amann, and
P. Meissner, “Continuously tunable long-wavelength MEMS-VCSEL with over
40-nm tuning range,” IEEE Photonics Technology Letters, vol. 16, no. 10, pp.
2212—2214, October 2004.
[18] H.A. Macleod, Thin-Films Optical Filters, Institute of Physics IOP, third edition,
2001.
[19] T.E. Sale, Vertical Cavity Surface Emitting Lasers, Research Studies Press LTD.,
1995.
[20] H. Li and K. Iga, Vertical-Cavity Surface-Emitting Laser Devices, Spinger, 2002.
[21] M.-C. Amann and J. Buus, Tunable Laser Diodes, Artech House, Inc., 1998.
[22] C.K. Madsen and J.H. Zhao, Optical Filter Design and Analysis- A Signal Pro-
cessing Approach, Wiley-Interscience, 1999.
[23] H. Hillmer, J. Daleiden, C. Prott, F. Römer, S. Irmer, V. Rangelov, A. Tarraf,
S. Schüler, and M. Strassner, “Potential for micromachined actuation of ultra-
wide continuously tunable optical devices,” Applied Physics B, vol. 75, pp. 3—13,
2002.
[24] S. Irmer, J. Daleiden, V. Rangelov, C. Prott, F. Römer, M. Strassner, A. Tarraf,
and H. Hillmer, “Continuously tunable InP based multiple air-gap MOEMS
filters with ultra wide tuning range,” SPIE Proceedings, vol. 4945, pp. 21—29,
2002.
84 Bibliography
[25] J. Peerlings, Mikromechanisch Abstimmbare Zweichip-Fabry-Pérot-Filter und -
Empfänger Für Wellenlängenmultiplex-Systeme, Ph.D. thesis, Technical Univer-
sity of Darmstadt, 1999.
[26] M. Aziz, Mikromechanisch Abstimmbare WDM-Filter mit Halbsymmetrischem
Resonator und Ihre Systemintegration, Ph.D. thesis, Technical University of
Darmstadt, 2002.
[27] H. Soda, K. Iga, C. Kitahara, and Y. Suematsu, “GaInAs/InP surface emitting
injection lasers,” Japanese Journal of Applied Physics, vol. 18, no. 12, pp. 2329—
2330, 1979.
[28] G.S. Li, W. Yuen, and C.J. Chang-Hasnain, “Wide and continuously tunable
(30 nm) detector with uniform characteristics over tuning range,” Electronics
Letters, vol. 33, pp. 1122—1123, 1997.
[29] F. Sugihwo, C.C. Lin, L.A. Eyres, M.M. Feyer, and J.S. Harris, “Broadly-
tunable narrow-linewidth micromachined Laser/Photodetector and phototran-
sistor,” Technical Digest of the Electron Devices Meeting (IEDM), pp. 665—668,
December 1998.
[30] D. Zhou, M. Garrigues, J.L. Leclercq, J. Peng, and P. Viktorovitch, “A phase-
matching and pseudoresonance-enhanced design for tunable micromachined pho-
todetectors,” IEEE Photonics Technology Letters, vol. 15, no. 10, pp. 1443—1445,
October 2003.
[31] M.S. Wu, E.C. Vail, G.S. Yuen, and C.J. Chang-Hasnain, “Widely and continu-
ously tunable micromachined resonant cavity detector with wavelength tracking,”
IEEE Photonics Technology Letters, vol. 8, no. 1, pp. 98—100, January 1996.
[32] F.J. Duarte, Tunable Lasers Handbook, Optics and Photonics, 1995.
[33] Y. Bäcklung, “Micromechanics in optical microsystems-with focus on telecom
systems,” IOP, Journal of Micromechanics and Microengineering, vol. 7, no. 3,
pp. 93—98, September 1997.
[34] J.A. Walker, “The future of MEMS in telecommunications networks,” IOP,
Journal of Micromechanics and Microengineering, vol. 10, no. 3, pp. R1—R7,
September 2000.
[35] C. Prott, F. Römer, E. Ataro, J. Daleiden, S. Irmer, A. Tarraf, and H. Hillmer,
“Modeling of ultrawidely tunable vertical cavity air-gap filters and VCSELs,”
IEEE Journal of Selected Topics in Quantum Electronics, vol. 9, no. 3, pp. 918—
928, May/June 2003.
[36] H. Guckel, D. Burnst, C. Rutigliano, E. Lovell, and B. Choi, “Diagnostic micro-
structures for the measurement of intrinsic strain in thin films,” IOP, Journal of
Micromechanics and Microengineering, vol. 2, no. 2, pp. 86—95, June 1992.
Bibliography 85
[37] N. Chitica, M. Strassner, and J. Daleiden, “Quantitative evaluation of growth-
induced residual stress in InP epitaxial micromechanical structures,” Applied
Physics Letters, vol. 77, no. 2, pp. 202—204, July 2000.
[38] L. Elbrecht, U. Storm, R. Catanescu, and J. Binder, “Comparison of stress meas-
urement techniques in surface micromachining,” IOP, Journal of Micromechanics
and Microengineering, vol. 7, no. 3, pp. 151—154, September 1997.
[39] M.T.-K. Hou, K.-M. Liao, H.-Z. Yeh, B.-W Cheng, P.-Y. Hong, and R. Chen,
“Fabrication of micromachined focusing mirrors with seamless reflective surface,”
Proceedings of SPIE, vol. 4983, pp. 359—366, 2003.
[40] K. Cao, W. Liu, and J.J. Talghader, “Curvature compensation in micromir-
rors with high-reflectivity optical coatings,” Journal of Microelectromechanical
Systems, vol. 10, no. 3, pp. 409—417, September 2001.
[41] F.R. Gass, D.J. Dagel, D.P. Adams, G.D. Grossetete, O.B. Spahn, S.A. Kemme,
S.S. Mani, and K.J. Malloy, “Stress and curvature in MEMS mirrors,” Proceed-
ings of SPIE, vol. 4983, pp. 87—93, 2003.
[42] J.J. Talghader, “Shape control and heat transfer in optical MEMS,” IEEE/LEOS
Newsletter, vol. 16, pp. 3—8, August 2002.
[43] W. Becker and D. Gross, Mechanik Elastischer Körper und Strukturen, Springer,
2002.
[44] H. Göldner and F. Holzweißig, Leitfaden der Technischen Mechanik, Dr. Dietrich
Steinkopff Verlag-Darmstadt, fifth edition, 1976.
[45] F. Ziegler, Mechanics of Solids and Fluids, Springer-Verlag, second edition, 1995.
[46] J. Wittenburg and E. Pestel, Festigkeitslehre, Ein Lehr- und Arbeitsbuch,
Springer, third edition, 2001.
[47] A. Degen, Investigation of Stress in Silicon Membranes, Ph.D. thesis, University
of Kassel, Germany, 2001.
[48] D.L. Smith, A.S. Alimonda, and F.J. Von Preissig, “Mechanism of SiNxHy
deposition from N2 − SiH4 plasma,” Journal of Vacuum Science Technology B,
vol. 8, no. 3, pp. 551—557, May/June 1990.
[49] H. Windischmann, “An intrinsic stress scaling law for polycrystalline thin films
prepared by ion beam sputtering,” Journal of Applied Physics, vol. 62, no. 5, pp.
1800—1807, September 1987.
[50] C.A. Davis, “A simple model for the formation of compressive stress in thin films
by ion bombardement,” Thin Solid Films, vol. 226, pp. 30—34, 1993.
86 Bibliography
[51] C.L.G. Alzar, “Mechanical stress reduction in PECVD a-Si:H thin films,” Ma-
terials Science and Engineering B, vol. 65, pp. 123—126, 1999.
[52] A. Tarraf, J. Daleiden, D. Prasai, and H. Hillmer, “Stress investigation of PECVD
dielectric layers for advanced optical MEMS,” IOP, Journal of Micromechanics
and Microengineering, vol. 14, no. 3, pp. 317—323, March 2003.
[53] K.H. Yang, “An optical imaging methode for wafer warpage measurements,”
Journal of Electrochemical Society, Solid-State Science and Technology, vol. 132,
no. 5, pp. 1214—1218, May 1985.
[54] A. Degen, J. Voigt, M. Kratzenberg, F. Shi, J. Butschke, H. Löscher, R. Kaes-
maier, A. Ehrmann, and I.W. Rangelow, “Stress engineering of SOI silicon stencil
masks by boron doping concentration,” SPIE Proceedings, vol. 3997, 2000.
[55] F. Ericson, S. Greek, J. Söderkvist, and J.-A. Schweitz, “High-sensitivity surface
micromachined structures for internal stress and stress gradient evaluation,” IOP,
Journal of Micromechanics and Microengineering, vol. 7, no. 1, pp. 30—36, March
1997.
[56] L. Nicu, P. Temple-Boyer, C. Bergaud, E. Scheid, and A. Martinez, “Energy
study of buckled micromachined beams for thin-film stress measurements applied
to SiO2,” IOP, Journal of Micromechanics and Microengineering, vol. 9, no. 4,
pp. 414—421, December 1999.
[57] W. Fang, C.-H. Lee, and H.-H. Hu, “On the bucling behavior of micromachined
beams,” IOP, Journal of Micromechanics and Microengineering, vol. 9, no. 3,
pp. 236—244, September 1999.
[58] C. Tsou and W. Fang, “The effect of residual stresses on the deformation of
semi-circular micromachined beams,” IOP, Journal of Micromechanics and Mi-
croengineering, vol. 10, no. 1, pp. 34—41, March 2000.
[59] Y.-H. Min and Y.-K. Kim, “In situ measurement of residual stress in microma-
chined thin films using a specimen with composite-layered cantilevers,” IOP,
Journal of Micromechanics and Microengineering, vol. 10, no. 3, pp. 314—321,
September 2000.
[60] E.I. Bromley, J.N. Randall, D.C. Flanders, and R.W. Mountain, “A technique for
the determination of stress in thin films,” Journal of Vacuum Science Technology
B, vol. 1, no. 4, pp. 1364—1366, December 1983.
[61] W. Fang and J.A. Wickert, “Determining mean and gradient residual stresses
in thin films using micromachined cantilevers,” IOP, Journal of Micromechanics
and Microengineering, vol. 6, no. 3, pp. 301—309, September 1996.
[62] C.H. Pan, “A simple method for determining linear thermal expansion coefficients
of thin films,” IOP, Journal of Micromechanics and Microengineering, vol. 12,
no. 5, pp. 548—555, September 2002.
Bibliography 87
[63] C.-L. Dai, “In situ electrostatic microactuators for measuring the Youngs modu-
lus of CMOS thin films,” IOP, Journal of Micromechanics and Microengineering,
vol. 13, no. 5, pp. 563—567, September 2003.
[64] W. Fang, “Determination of the elastic modulus of thin film materials using
self-deformed micromachined cantilevers,” IOP, Journal of Micromechanics and
Microengineering, vol. 9, no. 3, pp. 230—235, September 1999.
[65] M.W. Donhoff, “Ameasurement of Youngs modulus and residual stress in MEMS
bridges using a surface profiler,” IOP, Journal of Micromechanics and Microen-
gineering, vol. 13, no. 5, pp. 686—692, September 2003.
[66] S. Greek, F. Ericson, S. Johanson, M. Fürtsch, and A. Rump, “Mechanical char-
acterization of thick polysilicon films: Youngs modulus and fracture strength
evaluated with microstructures,” IOP, Journal of Micromechanics and Microen-
gineering, vol. 9, no. 3, pp. 245—251, September 1999.
[67] F.H.P.M. Habraken and A.E.T. Kuiper, “Silicon nitride and oxynitride films,”
Materials Science and Engineering, vol. R12, no. 3, pp. 123—170, July 1994.
[68] R. E. Williams, Modern GaAs Processing Methods, Artech House Inc., second
edition, 1990.
[69] T. Yoshimi, H. Sakai, and K. Tanaka, “Analysis of hydrogen content in plasma
silicon nitride film,” Journal of the Electrochemical Society: Solid-State Science
and Technology, vol. 127, pp. 1853—1854, August 1980.
[70] G.M. Samuelson and K.M. Mar, “The correlations between physical and elec-
trical properties of PECVD SiN with their composition ratios,” Journal of the
Electrochemical Society: Solid-State Science and Technology, vol. 129, no. 8, pp.
1773—1778, August 1982.
[71] W.A.P. Classen, W.G.J.N. Valkenburg, F.H.P.M. Habraken, and Y. Tamminga,
“Characterization of plasma silicon nitride layers,” Journal of the Electrochemical
Society: Solid-State Science and Technology, vol. 130, no. 12, pp. 2419—2423,
December 1983.
[72] W.A.P. Claassen, W.G.J.N. Valkenburg, M.F.C. Willemsen, and W.M.V.D.
Wijgert, “Influence of deposition temperature, gas pressure, gas phase compos-
ition, and RF frequency on composition and mechanical stress of plasma silicon
nitride layers,” Journal of the Electrochemical Society: Solid-State Science and
Technology, vol. 132, no. 4, pp. 893—898, April 1985.
[73] E. Bustarret, M. Bensouda, M.C. Habrad, J.C. Bruyère, S. Poulin, and S.C.
Gujrathi, “Configurational statistics in a−SixNyHz alloys: A quantitative bond-
ing analysis,” Physical Review B, vol. 38, no. 12, pp. 8171—8182, October 1988.
88 Bibliography
[74] T.J. Cotler and J. Chapple-Sokol, “High quality plasma-enhanced chemical va-
pour deposition silicon nitride films,” Journal of the Electrochemical Society, vol.
140, no. 7, pp. 2071—2075, July 1993.
[75] W.A. Lanford andM.J. Rand, “The hydrogen content of plasma-deposited silicon
nitride,” Journal of Applied Physics, vol. 49, no. 4, pp. 2473—2477, April 1978.
[76] J.A. Taylor, “The mechanical properties and microstructure of plasma enhanced
chemical vapour deposited silicon nitride films,” Journal of Vacuum Science and
Technology A, vol. 9, no. 4, pp. 2464—2468, August 1991.
[77] CVD Inc., “Fundamentals of chemical vapor deposition: Silicon dioxide, proper-
ties and applications,” Electronic Publishing, TimeDomain CVD Inc.
[78] P.J. French and P.M. Sarro, “Surface versus bulk micromachining: The contest
for suitable applications,” IOP, Journal of Micromechanics and Microengineer-
ing, vol. 8, no. 2, pp. 45—53, June 1998.
[79] M. Hoffmann and E. Voges, “Bulk silicon micromachining for MEMS in optical
communication systems,” IOP, Journal of Micromechanics and Microengineer-
ing, vol. 12, no. 4, pp. 349—360, July 2002.
[80] C. Linder, L. Paratte, M.-A. Grétillat, V.P. Jaecklin, and N.F. de Rooij, “Surface
micromachining,” IOP, Journal of Micromechanics and Microengineering, vol.
2, no. 3, pp. 122—132, September 1992.
[81] K. Hjort, “Sacrificial etching of III-V compounds for micromechanical devices,”
IOP, Journal of Micromechanics and Microengineering, vol. 6, no. 4, pp. 370—375,
December 1996.
[82] J.-L. Leclercq, M. Garrigues, X. Letartre, C. Seassal, and P. Viktorovitch, “InP-
based MOEMS and related topics,” IOP, Journal of Micromechanics and Mi-
croengineering, vol. 10, no. 2, pp. 287—292, June 2000.
[83] C. Seassal, J.L. Leclercq, and P. Viktorovitch, “Fabrication of InP-based freest-
anding microstructures by surface micromachining,” IOP, Journal of Micromech-
anics and Microengineering, vol. 6, no. 2, pp. 261—265, June 1996.
[84] Z. Cui and R.A. Lawes, “A new sacrificial layer process for the fabrication of mi-
cromechanical systems,” IOP, Journal of Micromechanics and Microengineering,
vol. 7, no. 3, pp. 128—130, September 1997.
[85] M. Bartek and R.F. Wolffenbuttel, “Dry release of metal structures in oxygen
plasma: Process characterization and optimization,” IOP, Journal of Micromech-
anics and Microengineering, vol. 8, no. 2, pp. 91—94, June 1998.
[86] K.J. Winchester and J.M. Dell, “Tunable fabry-pérot cavities fabricated from
PECVD silicon nitride employing zinc sulphide as the sacrificial layer,” IOP,
Bibliography 89
Journal of Micromechanics and Microengineering, vol. 11, no. 5, pp. 589—594,
September 2001.
[87] J. Bühler, F.-P. Steiner, and H. Baltes, “Silicon dioxide sacrificial layer etching
in surface micromachining,” IOP, Journal of Micromechanics and Microengin-
eering, vol. 7, no. 1, pp. R1—R13, March 1997.
[88] T.E. Bell, P.T.J. Gennissen, D. DeMunter, and M. Kuhl, “Porous silicon as a
sacrificial material,” IOP, Journal of Micromechanics and Microengineering, vol.
6, no. 4, pp. 361—369, December 1996.
[89] S. Deladi, M.J. de Boer, G. Krijnen, D. Rosén, andM.C. Elwenspoek, “Innovative
process development for a new micro-tribosensor using surface micromachining,”
IOP, Journal of Micromechanics and Microengineering, vol. 13, no. 4, pp. S17—
S22, July 2003.
[90] Z. Li, G. Zhang, W. Wang, Y. Hao, T. Li, and G. Wu, “Study on the application
of silicide in surface micromachining,” IOP, Journal of Micromechanics and
Microengineering, vol. 12, no. 2, pp. 162—167, March 2002.
[91] M.A. Benitez, J.A. Plaza, S.Q. Sheng, and J. Esteve, “A new process for re-
leasing micromechanical structures in surface micromachining,” IOP, Journal of
Micromechanics and Microengineering, vol. 6, no. 1, pp. 36—39, March 1996.
[92] D. Westberg, O. Paul, G. Andersson, and H. Baltes, “Surface micromachining by
sacrificial aluminium etching,” IOP, Journal of Micromechanics and Microen-
gineering, vol. 6, no. 4, pp. 376—384, December 1996.
[93] Z. Xiao, Y. Hao, T. Li, G. Zhang, S. Liu, and G. Wu, “A new release process
for polysilicon surface micromachining using sacrificial polysilicon anchor and
photolithography after sacrificial etching,” IOP, Journal of Micromechanics and
Microengineering, vol. 9, no. 4, pp. 300—304, December 1999.
[94] A. Bagolini, L. Pakula, T.L.M. Scholtes, H.T.M. Pham, P.J. French, and P.M.
Sarro, “Polyimide sacrificial layer and novel materials for post-processing surface
micromachining,” IOP, Journal of Micromechanics and Microengineering, vol.
12, no. 4, pp. 385—389, July 2002.
[95] N. Tas, T. Sonnenberg, H. Jansen, R. Legtenberg, and M. Elwenspoek, “Stiction
in surface micromachining,” IOP, Journal of Micromechanics and Microengin-
eering, vol. 6, no. 4, pp. 385—397, December 1996.
[96] T. Takagi, K. Takechi, Y. Watabe, and S. Nishida, “High rate deposition of
a-Si:H and a−SiNx : H by VHF PECVD,” Vacuum, vol. 51, no. 4, pp. 751—755,
1998.
[97] R. Etemadi, C. Godet, M. Kildemo, J.E. Bourée, R. Brenot, and B. Drévillon,
“Dual-mode radio frequency/microwave plasma deposition of amorphous silicon
oxide thin films,” Journal of non-Crystalline Solids, vol. 187, pp. 70—74, 1995.
90 Bibliography
[98] D.W. Hess, “Plasma-enhanced CVD: Oxides, nitrides, transition metals, and
transition metal silicides,” Journal of Vacuum Science Technology A, vol. 2, no.
2, pp. 244—252, April/June 1984.
[99] M. Aziz, J. Pfeiffer, and P. Meissner, “Modal behaviour of passive, stable mi-
crocavities,” Physica Status Solidi A, vol. 188, no. 9, pp. 79—82, 2001.
[100] S. Schüler, “Technologische Realisierung optischer Filter basierend auf Multi-
airgapstrukturen in Indium-Phosphid,” Diplomarbeit an der Universität Kassel,
2001.
[101] P.-U. Archaow, “Optimization and fabrication of tunable airgap InP based fil-
ters,” Diplomarbeit an der Universität Kassel, April 2003.
[102] F. Römer, “Charakterisierung und simulation optischer eigenschaften von mik-
romechanisch abstimmbaren filterbauelementen,” PhD thesis, Universität Kas-
sel, Germany, 2005.
[103] A. Tarraf, J. Daleiden, F. Römer, C. Prott, S. Irmer, E. Ataro, and H. Hillmer,
“A novel low-cost tunable dielectric air-gap filter,” IEEE/LEOS proceedings of
the international Conference on Optical MEMS, pp. 175—176, 2002.
[104] H. Hillmer, S. Hansmann, H. Bukhard, H. Walter, A. Krost, and D. Bimberg,
“Study of wavelength shift in InGaAs/InAlGaAs QW DFB lasers based on laser
parameters from a comparison of experiment and theory,” IEEE Journal of
Quantum Electronics, vol. 30, no. 10, pp. 2251—2261, October 1994.
[105] A. Tarraf, F. Riemenschneider, M. Strassner, J. Daleiden, S. Irmer, H. Hal-
britter, H. Hillmer, and P. Meissner, “Continuously tunable 1.55 µm VCSEL
implemented by precisely curved dielectric top DBR involving tailored stress,”
IEEE Photonics Technology Letters, vol. 16, no. 3, pp. 720—722, March 2004.
[106] H. Halbritter, F. Riemenschneider, S. Syguda, C. Dhanavantri, M. Strassner,
A. Tarraf, B.R. Singh, I. Sagnes, and P. Meissner, “Tunable and wavelength
selective pin photodiode,” Electronics Letters, vol. 40, no. 6, pp. 388—390, 2004.
[107] H. Halbritter, C. Dhanavantri, M. Strassner, A. Tarraf, M. Aziz,
F. Riemenschneider, S. Syguda, B.R. Singh, I. Sagnes, and P. Meissner, “Tunable
and wavelength selective pin diodes,” SPIE proceedings, vol. 5277, pp. 129—127,
december 2003.
[108] H. Halbritter, F. Riemenschneider, B. Kögel, A. Tarraf, M. Strassner, S. Irmer,
H. Hillmer, I. Sagnes, and P. Meissner, “MEMS-tunable and wavelength selective
receiver front end,” Proceedings of the 18th international IEEE conference on
micro-electro-mechanical systems (MEMS), pp. 68—71, 2005.
[109] H. Halbritter, “Entwicklung und Charakterisierung mikromechanisch ab-
stimmbarer Komponenten für optische Kommunikationssysteme,” PhD thesis,
Technische Universität Darmstadt, Germany, 2005.
Bibliography 91
[110] M. Müller-Wiegand, G. Georgiev, E. Oesterschulze, T. Fuhrmann, and J. Salbeck,
“Spinodal patterning in organic-inorganic hybrid laser systems,” Applied Physics
Letters, vol. 81, no. 26, pp. 4940, 2002.
[111] M. Müller-Wiegand, “Spontane Strukturierung in Mehrschichtsystemen unter
Verwendung molekularer Gläser,” PhD Thesis, University of Kassel, October
2003.
[112] N. Johansson, J. Salbeck, J. Bauer, F. Weissoertel, P. Broems, A. Andersson,
and W.R. Salaneck, “Solid-state amplified spontaneous emission in some spiro-
type molecules: A new concept for the design of solid state lasing molecules,”
Advanced Materials, vol. 10, pp. 1136, 1998.
[113] K.H. Weinfurtner, F. Weissörtel, G. Harmgarth, and J. Salbeck, “Novel amorph-
ous molecular materials for organic light-emitting devices,” Proceedings of SPIE,
vol. 3476, pp. 40—48, 1998.
[114] J. Salbeck, M. Schörner, and T. Fuhrmann, “Optical amplification in spiro-type
molecular glasses,” Thin solid films, vol. 417, pp. 20—25, 2002.
[115] J. Bryzek, K. Petersen, and W. McCulley, “Micromachines on the march,” IEEE
Spectrum, pp. 20—31, May 1999.
[116] R. Bischoffberger, M. Cucinelli, and G. Staufert, “New ways to make microma-
chines,” Vacuum Solutions, pp. 15—18, 1997.
[117] E.C. Vail, M.S. Wu, G.S. Li, L. Eng, and C.J. Chang-Hasnain, “GaAs microma-
chined widely tunable fabry-pérot filters,” Electronics Letters, vol. 31, no. 3, pp.
228—229, February 1995.
[118] A.T.T.D. Tran, Y.H. Lo, Z.H. Zhu, D. Haronian, and E. Mozdy, “Surface mi-
cromachined fabry-pérot tunable filter,” IEEE Photonics Technology Letters, vol.
8, no. 3, pp. 393—395, March 1996.
[119] J. Peerlings, A. Dehe, A. Vogt, M. Tilsch, C. Hebeler, F. Langenhan, P. Meissner,
and H.L. Hartnagel, “Long resonator micromachined tunable GaAs-AlAs fabry-
pérot filter,” IEEE Photonics Technology Letters, vol. 9, no. 9, pp. 1235—1237,
September 1997.
[120] P. Tayebati, P.D. Wang, D.Vakhshoori, and R.N. Sacks, “Widely tunable fabry-
pérot filter using Ga(Al)As-AlOx deformable mirrors,” IEEE Photonics Techno-
logy Letters, vol. 10, no. 3, pp. 394—396, March 1998.
[121] A. Spisser, R. Ledantec, C. Seassal, J.L. Leclerq, T. Benyattou, D. Rondi,
R. Blondeau, G. Guillot, and P. Viktorovitch, “Highly selective and widely
tunable 1.55-µm InP/Air-gap micromachined fabry-pérot filter for optical com-
munications,” IEEE Photonics Technology Letters, vol. 10, no. 9, pp. 1259—1261,
September 1998.
92 Bibliography
[122] N. Chitica, J. Daleiden, M. Strassner, and K. Streubel, “Monolithic, InP based
tunable filter with 10 nm bandwidth for optical data interconnects in the 1550
nm band,” IEEE Photonics Technology Letters, vol. 11, no. 5, pp. 584—586, May
1999.
[123] M. Strassner, J. Daleiden, N. Chitica, D. Keiper, B. Stalnacke, S. Greek, and
K. Hjort, “III-V semiconductor material for tunable fabry-pérot filters for coarse
and dense WDM systems,” Sensors and Actuators, vol. 85, pp. 249—255, 2000.
[124] M. Aziz, J. Pfeiffer, M. Wohlfarth, K. Lubers, S. Wu, and P. Meissner, “A new
and simple tunable two-chip microcavity for filter applications inWDM systems,”
IEEE Photonics Technology Letters, vol. 12, no. 11, pp. 1522—1524, November
2000.
[125] C.F.R. Mateus, C.-H. Chang, L. Chrostowski, S. Yang, D. Sun AndR. Pathak,
and C. J. Chang-Hasnain, “Widely tunable torsional optical filter,” IEEE
Photonics Technology Letters, vol. 14, no. 6, pp. 819—821, June 2002.
[126] H. Halbritter, M. Aziz, F. Riemenschneider, and P. Meissner, “Electrothermally
tunable two-chip optical filter with very low-cost and simple concept,” Electronics
Letters, vol. 38, no. 20, pp. 1201—1202, September 2002.
[127] J. Daleiden, V. Rangelov, S. Irmer, F. Römer, M. Strassner, C. Prott, A. Tarraf,
and H. Hillmer, “Record tuning of InP-based multiple air-gap MOEMS filter,”
Electronics Letters, vol. 21, no. 21, October 2002.
[128] M. Strassner, C. Luber, A. Tarraf, and N. Chitica, “Widely tunable-constant
bandwidth monolithic fabry-pérot filter with a stable cavity design for WDM
systems,” IEEE Photonics Technology Letters, vol. 14, no. 11, pp. 1548—1550,
November 2002.
[129] N. Kanbara, K. Suzuki, T. Watanabe, and H. Iwaoka, “Precisely tunable fabry-
pérot filter for optical communications,” IEEE/LEOS proceedings of the Inter-
national Conference on Optical MEMS, pp. 173—174, 2002.
[130] S. Irmer, K. Alex, J. Daleiden, I. Kommallein, M. Oliveira, F. Römer, A. Tar-
raf, and H. Hillmer, “Surface micromachined optical low-cost all-air-gap filters
based on stress-optimized Si3N4 layers,” IOP, Journal of Micromechanics and
Microengineering, vol. 15, no. 4, pp. 867—872, April 2005.
[131] S.-S. Yun and J.-H. Lee, “A micromachined in-plane tunable optical filter using
the thermo-optic effect of crystalline silicon,” IOP, Journal of Micromechanics
and Microengineering, vol. 13, no. 5, pp. 721—725, June 2003.
[132] F. Solomonsson, Processing Technology For Long-Wavelength Vertical-Cavity
Lasers, Ph.D. thesis, Royal Institute of Technology, Sweden, 2001.
Bibliography 93
[133] H.W. Song, D.S. Song, I.Y. Han, C.K. Kim, H.Y. Ryu, and Y.H. Lee, “1.5µm
vertical-cavity surface-emitting lasers utilising low loss AlxOy/GaAs mirrors,”
Electronics Letters, vol. 35, no. 4, February 1999.
[134] J.-H. Baek, W.S. Han, H.K. Cho, B. Lee, and I.H. Choi, “Optically pumped
continuous-wave operation of InAlGaAs/InAlAs/InP based 1.55µm vertical-
cavity surface-emitting laser with SiO2/TiO2 dielectric mirror,” Electronics Let-
ters, vol. 35, no. 10, May 1999.
[135] A. Keating, A. Black, A. Karim, Y.-J Chiu, P. Abraham, C. Harder, E. Hu, and
J. Bowers, “High-temperature optically pumped 1.55-µm VCSEL operating at 6
Gb/S,” IEEE Photonics Technology Letters, vol. 12, no. 2, pp. 116—118, February
2000.
[136] N. Chitica and M. Strassner, “Room-temperature operation of photopumped
monolithic InP vertical-cavity laser with two air-gap bragg reflectors,” Applied
Physics Letters, vol. 78, no. 25, pp. 3935—3937, June 2001.
[137] J. Geske, D. Leonard, M. H. MacDougal, B. Barnes, and J.E. Bowers, “CWDM
vertical-cavity surface emitting laser array spanning 140 nm of the C, S, and L
fiber transmission bands,” IEEE Photonics Technology Letters, vol. 16, no. 5,
pp. 1227—1229, May 2004.
[138] D.I. Babic, J.J. Dudley, K. Streubel, R.P. Mirin, J.E. Bowers, and E.L. Hu,
“Double-fused 1.52-µm vertical-cavity lasers,” Applied Physics Letters, vol. 66,
no. 9, pp. 1030—1032, February 1995.
[139] D.I. Babic, K. Streubel, R.P. Mirin, N.M. Margalit, J.E. Bowers, E.L. Hu, D.E.
Mars, L. Yang, and K. Carey, “Room-temperature continuous-wave operation of
1.54-µm vertical-cavity lasers,” IEEE Photonics Technology Letters, vol. 7, no.
11, pp. 1225—1227, November 1995.
[140] N.M. Margalit, J. Piprek, S. Zhang, D.I. Babic, K. Streubel, R.P. Mirin, J.R.
Wesselmann, J.E. Bowers, and E.L. Hu, “64 ◦C continuous-wave operation of
1.5-µm vertical-cavity laser,” Journal of Selected Topics in Quantum Electronics,
vol. 3, no. 2, pp. 359—365, April 1997.
[141] A. Syrbu, A. Mircea, A. Mereuta, A. Caliman, C.-A. Berseth, G. Suruceanu,
V. Iakovlev, M. Achtenhagen, R. Rudra, and E. Kapon, “1.5-mW single mode
operation of wafer-fused 1550-nm VCSELs,” IEEE Photonics Technology Letters,
vol. 16, no. 5, pp. 1230—1232, May 2004.
[142] A. Karim, A. Black, P. Abraham, D. Lofgreen, Y.J. Chiu, J. Piprek, and
J.E. Bowers, “Superlattice barrier 1528-nm vertical-cavity laser with 85 ◦C
continuous-wave operation,” IEEE Photonics Technology Letters, vol. 12, no.
11, pp. 1438—1440, November 2000.
94 Bibliography
[143] K. Streubel, S. Rapp, J. Andre, and J. Wallin, “Room-temperature pulsed oper-
ation of 1.5-µm vertical cavity lasers with an InP-based bragg reflector,” IEEE
Photonics Technology Letters, vol. 8, no. 8, pp. 1121—1123, September 1996.
[144] Y. Ohiso, H. Okamoto, R. Iga, K. Kishi, K. Tateno, and C. Amano, “High
performance of 1.55-µm buried-heterostructure vertical-cavity surface-emitting
lasers,” IEEE Photonics Technology Letters, vol. 13, no. 9, pp. 918—920, Septem-
ber 2001.
[145] D. Sun, W. Fan, P. Kner, T. Kageyama, R. Pathak, D. Zhang, and W. Yuen,
“Sub-mA threshold 1.5-µm VCSELs with epitaxial and dielectric DBR mirrors,”
IEEE Photonics Technology Letters, vol. 15, no. 12, pp. 1677—1679, December
2003.
[146] C. Kazmierski, J.P. Debray, R. Madani, I. Sagnes, A. Ougazzaden, N. Bouadma,
J. Etrillard, F. Alexandre, and M. Quillec, “+55 ◦C pulse lasing at 1.56µm of
all-monolithic InGaAlAs/InP vertical cavity lasers,” Electronics Letters, vol. 35,
no. 10, May 1999.
[147] J. Boucart, C. Stark, F. Gaborit, A. Plais, N. Bouche, J.C. Remy, J. Bonnet-
Gamard, L. Goldstein, C. Fortin, D. Carpentier, P. Salet, F. Brillouet, and J. Jac-
quet, “Metamorphic DBR and tunnel-junction injection: A CW RT monolithic
long-wavelength VCSEL,” IEEE Journal of Selected Topics in Quantum Elec-
tronics, vol. 5, no. 3, pp. 520—529, May/June 1999.
[148] J. Boucart, C. Starck, F. Gaborit, A. Plais, N. Bouche, E. Derouin, L. Goldstein,
C. Fortin, D. Carpentier, and J. Jacquet, “1-mW CW-RT monolithic VCSEL at
1.55 µm,” IEEE Photonics Technology Letters, vol. 11, no. 6, pp. 629—631, June
1999.
[149] W. Yuen, G.S. Li, R.F. Nabiev, J. Boucart, P. Kner, R.J. Stone, D. Zhang,
M. Beaudoin, T. Zheng, C. He, K. Yu, M. Jansen, D.P. Worland, and C.J.
Chang-Hasnain, “High-performance 1.6µm single-epitaxy top-emitting VCSEL,”
Electronics Letters, vol. 36, no. 13, pp. 1121—1123, June 2000.
[150] E. Hall, S. Nakagawa, G. Almuneau, J.K. Kim, and L.A. Coldren, “Room-
temperature, CW operation of lattice-matched long-wavelength VCSELs,” Elec-
tronics Letters, vol. 36, no. 17, August 2000.
[151] O.-K. Kwon, B.-S. Yoo, J.-H. Shin, J.-H. Baek, and B. Lee, “Pulse operation and
threshold characteristics of 1.55-µm InAlGaAs-InAlAs VCSELs,” IEEE Photon-
ics Technology Letters, vol. 12, no. 9, pp. 1132—1134, September 2000.
[152] A. Syrbu, V. Iakovlev, G. Suruceanu, A. Caliman, A. Rudra, A. Mircea,
A. Mereuta, S. Tadeoni, C. Berseth, M. Achtenhagen, J. Boucart, and E. Ka-
pon, “1.55-µm optically pumped wafer-fused tunable VCSELs with 32-nm tun-
ing range,” IEEE Photonics Technology Letters, vol. 16, no. 9, pp. 1991—1993,
September 2004.
Bibliography 95
[153] J. Boucart, R. Pathak, D. Zhang, M. Beaudoin, P. Kner, D. Sun, R. Stone, R.F.
Nabiev, and W. Yuen, “Long wavelength MEMS tunable VCSEL with InP-
InAlGaAs bottom DBR,” IEEE Photonics Technology Letters, vol. 15, no. 9, pp.
1186—1188, September 2003.
[154] P.Kner, T. Kageyama, J. Boucart, R. Stone, D. Sun, R.F. Nabiev, R Pathak, and
W. Yuen, “A long-wavelength MEMS tunable VCSEL incorporating a tunnel
junction,” IEEE Photonics Technology Letters, vol. 15, no. 9, pp. 1183—1185,
September 2003.
[155] M.Y. Li, W. Yuen, G.S. Li, and C.J. Chang-Hasnain, “Top-emitting micromech-
anical VCSELwith a 31.6-nm tuning range,” IEEE Photonics Technology Letters,
vol. 10, no. 1, pp. 18—20, January 1998.
[156] D. Sun, W. Fan, P. Kner, J. Boucart, T. Kageyama, D. Zhang, R. Phatak,
R. Nabiev, and W. Yuen, “Long wavelength-tunable VCSELs with optimized
MEMS bridge tuning structure,” IEEE Photonics Technology Letters, vol. 16,
no. 3, pp. 714—716, March 2004.
[157] C.J. Chang-Hasnain, “Tunable VCSEL,” IEEE Journal on Selected Topics in
Quantum Electronics, vol. 6, no. 6, pp. 978—987, November-December 2000.

Appendix A
State of the art: detailed
description
Microcavity devices for the optical communications combine the advantages for mech-
anics and optics resulting from the miniaturization. These advantages are difficult to
understand if we consider our macroworld only. In this macroworld we are living in,
machine dimensions span from the sub-mm range (parts in precision mechanisms) to
the km range (large electron accelerating rings). We can readily appreciate the advant-
ages of miniaturization in silicon based microelectronics (e.g. microchips and electronic
devices), which are involved in high speed personal computers, novel digital electronics
(e.g. television, hi-fi electronics, digital cameras, etc.). The smaller the devices, the
faster the computers and the smaller the equipment. In this case, the microcomponents
are part of a system whose main functionality remains unchanged. A digital camera
still delivers pictures, like the one 30 years ago did. However, if we consider the quality,
the price, the size, and the impact on our daily life quality, we appreciate the added
value and performance.
Passive consumer acceptance is then challenged when the miniaturization is applied
to a mechanical system. Thus, it is difficult to imagine that somebody can drive to work
with a car of 2µm length. Nowadays, micromechanics are used as microcomponents
in macrosystems (e.g. air-bag sensors, environmental gas detectors and metrology).
However, micromechanics exhibit huge advantages or even some novel functionality
when compared to the well known macromechanics. Thus, downscaling the dimensions
linearly results in a nonlinear changing of the forces impacting the system [23]. To see
this, consider a microbridge of 100µm length, 0.5µm thickness and 10µm width. In
micromechanics, this bridge is considered to be a robust free hanging system. Scal-
ing up this system to the macroworld dimensions (e.g. meters), a bridge of 1000m
length, 5m thickness, and 100m width is obtained. Such a bridge is unstable and will
collapse. Therefore, many actuated systems (e.g. tunable VCSEL and FPF filters)
can be implemented only by using micromechanics and microcomponents. Thus, in
recent years, MEMS and MOEMS became very interesting for a wide range of systems
integration [115, 116]. On one hand, they reveal the same advantages as microelec-
tronics (e.g. low-cost mass production, low power consumption and small size) and
98 A.1 Fabry-Pérot filter
on the other hand, they benefit from the novel effects of the miniaturization. In this
thesis, the focus is on implementing micromechanically tunable microdevices such as
VCSELs, FPFs and PIN photodetectors. These components are the core of the new
dynamic DWDM systems. In the following, a summary of the devices which have been
implemented in the recent years will be presented.
A.1 Fabry-Pérot filter
Several approaches dealing with non tunable and tunable Fabry-Pérot filters have been
previously published. Even though the filter is a simple device, its tunability is still a
major technological challenge. However, tunable filters are attractive since they present
a low-cost solution for dynamic DWDM systems.
Based on a GaAs/AlAs bottom DBR and an Au/SiNxHy/GaAs top movable mem-
brane, Larson et. al. [10] presented a tunability of 32 nm by a 14V tuning voltage at
a central wavelength of 932 nm. The membrane is actuated electrostatically and the
FWHM of the filter resonance is 3 nm. Vail et. al. [117] show a tunability of 70 nm by
a 5V tuning voltage. The top movable DBR is implemented by a cantilever, whereas
the central wavelength is 950 nm. A tunable filter, based on Si/SiO2 DBRs with 60 nm
tuning range is demonstrated by Tran et. al. [118] for a 1.517µm central wavelength.
The FWHM of the filter characteristic is 3.2 nm. A GaAs/AlAs two-chip concept filter
with a FSR of 46.7 nm, a FWHM of 1.2 nm, an electrostatic tuning range of 103 nm
(by a tuning voltage of 35V), and a 1.517µm central wavelength has been presented
by Peerlings et. al. [119]. Tayebati et. al. [120] show a GaAlAs/AlOx air-gap based
filter with 59 nm tuning range at 50V tuning voltage (FWHM=0.5 nm) and 83 nm
tuning range at 25V tuning voltage (FWHM=2.0 nm). However, multimode opera-
tion is observed in this case. Using a Fabry-Pérot filter with a stable half symmetric
cavity design [12], the same author shows a tunability of 70 nm by a tuning voltage
of 14V. The filter consists of two SiO2/TiO2 DBRs separated by an air-gap cavity.
The radius of curvature of the top membrane is around 310µm. The FWHM of the
filter characteristics is 0.27 nm. Spisser et. al. [121] demonstrate for the first time a
tunable filter based on InP/air DBRs. The filter operates at 1.55µm and exhibits a
tunability of 62 nm by a 14V tuning voltage. A FWHM of 0.4 nm is observed over
40 nm tuning range. A hybrid Si/SiO2 and InGaAsP based tunable filter is presen-
ted by Chitica et. al. [122]. A tuning range of 40 nm has been reached by applying
40V tuning voltage. In this case, the FWHM is 10 nm. A tunability of 0.3Å/V2 of
a InP/air based filter is demonstrated by Strassner et. al. [123]. Here, the mechan-
ical properties (like stress) have been intensively studied. Aziz et. al. [124] present
a tunability of over 20 nm by a tuning voltage of 100V. The two chip based filter
consists of two Si3N4/SiO2 dielectric DBRs and the FWHM of the filter characteristic
is below 0.1 nm. An AlGaAs cantilever based filter with a tuning range of over 100 nm
by a 18V-20V tuning voltage is demonstrated by Mateus et. al. [125]. Halbritter et.
al. [126] show an electrothermally tuned filter with a tuning range of over 34 nm by
using a Si3N4/SiO2 top DBR membrane. The filter exhibits a FWHM of 0.16 nm. An
A State of the art: detailed description 99
InP/air-gap based filter with 112 nm tuning range and 3.5 nm FWHM is achieved by
Daleiden et. al. [127]. The tuning voltage is only 5V. Strassner et. al. [128] demon-
strate an electrostatic tuning to longer wavelengths by an InP/air based filter. The
filter characteristic remains unchanged over the 65 nm tuning range. The FWHM is
4.3 nm, whereas the tuning voltage is 21V. The author of this thesis [103] presents a
low-cost solution of an electrothermally tuned filter with Si3N4/SiO2 dielectric DBRs.
The tunability is 15 nm /mA with a tuning range of 15 nm. The FWHM as predicted
by theoretical model calculation 8 nm. A silicon based filter presented by Kanbara et.
al. [129] exhibits a FWHM of 0.5 nm and an FSR of 35 nm. In this case, the tuning
range of 30 nm is reached by a tuning voltage of 29V. Irmer et. al. [9] present a con-
tinuously widely tunable InP/air based filter with 140 nm tuning range. The tuning
voltage is 3.2V, whereas the FWHM is between 3 nm and 5 nm. The same author
recently presented [130] a non tunable low-cost all dielectric air-gap filter with 1.5 nm
FWHM. Yun et. al. [131] present a tunable in-plane Si/air filter with 9 nm thermoop-
tical tuning range by a tuning temperature of 392K. A FWHM of 1.1 nm is achieved.
Hohlfeld et. al. [11] present a tunable filter based on the thermooptical effect. The
filter consists of a silicon cavity and two Si3N4/SiO2 DBRs and shows a tunability of
3.5 nm and a FWHM of 1.19 nm.
A.2 Vertical cavity surface emitting laser
Simply spoken, a vertical cavity surface emitting laser device is a light emitting active
region embedded vertically between two DBRs. However, this approximation is in
fact extremely simplistic when we consider all the optical and the mechanical aspects
leading to a lasing in a VCSEL. The main feature distinguishing a VCSEL from an edge
emitting laser is the fact that the laser oscillation as well as the laser light emission
occur in a direction perpendicular to the wafer surface. This feature enables a better
design of the emission pattern (usually a circular shape) resulting in low light divergence
and efficient fiber coupling. Further advantages of a VCSEL are the wafer scale batch
fabrication, the in-plane characterization, the simple packaging and the high yield of
the device. Thus, compared to the edge emitting lasers, VCSELs present a low-cost
solution for optical communications.
In this case, a single longitudinal mode operation is needed. This can be easily
obtained due to the small overall cavity length (some few µm), typically of the same
order of the emission wavelength. This fact results in a small gain region leading
to smaller drive currents. Even though higher differential resistance is mostly the
consequence of a voltage drop over the DBRs, the wall-plug efficiency of a VCSEL is
much higher than that of any other semiconductor laser device1 [132]. However, due
to the short active region length, the stimulated light can only benefit from a small
1When discussing the efficiency of the whole laser, the expression wall-plug efficiency is often used.
The wall-plug efficiency equals the output light energy divided by the electrical energy into the wall
plug.
100 A.2 Vertical cavity surface emitting laser
gain path. Therefore, very high reflective DBRs are needed and stringent optical and
mechanical requirements for the active material should be achieved.
Since the first demonstration of a VCSEL operation by Soda et al. in 1979 [27],
countless publications have documented the successful research of several groups. Ab-
solutely dominating are VCSELs for the short wavelength range below 1µm. As the in-
terest in DWDM systems increases, VCSELs for the long haul wavelength range (1.3µm
and 1.5µm) became highly desired. However, the technological implementation of such
devices is still a major challenge due to the material properties at this wavelengths.
Until now, several optically and electrically pumped, non tunable and tunable VCSELs
for the 1.55µm range have been reported. Note, that optically pumped non tunable
devices are easier to implement compared to the electrically pumped tunable ones,
since no micromechanics and no electrical confinements are used.
Song et. al. [133] present a wafer bonded VCSEL with a InGaAs/InGaAsP QWs
system. The top and bottom DBRs consist of a GaAs/AlxOy and GaAs/AlAs multilay-
ers, respectively. An optical threshold pump power of 1.4mW at RT corresponding
to a threshold current of 1.2mA for a 5µm aperture device, leads to a cw operation
with a line width below 0.2 nm. An optically cw pumped VCSEL with hybrid DBRs
is presented by Baek et. al. [134]. The device consists mainly of a periodic gain struc-
ture with 27 InGaAs/InAlGaAs QWs located in an active region embedded between
a InAlAs/InAlGaAs bottom DBR and a dielectric TiO2/SiO2 top DBR. A minimum
threshold pump power of 17mW at 220K is observed. This corresponds to a calculated
threshold power density of 3.4 kW/cm2. The line width of the laser characteristic at
1553 nm is below 0.6 nm and a maximum power of above 100µW is achieved. Keating
et. al. fabricate a wafer fused VCSEL with InGaAsP based QWs and GaAs/AlAs
DBRs [135]. The device exhibits an output power of 1.95mW at RT and at 1547 nm
emission wavelength with a SMSR of 61dB. A novel all air-gap monolithically grown
InP based VCSEL is presented by Chitica et. al. [136]. The cavity comprising the
GaInAsP QWs is embedded between to InP/air DBRs. The threshold is reached at
RT by an optical pump power density of 370 W / cm2. A maximum output power of
110µW and a cw operation up to 32 ◦C are reached. Recently, Geske et. al. presented
a VCSEL array with 140 nm spanning emission wavelength range with 20 nm channel
spacing [137]. The wafer bonded device is optically pumped and shows a cw operation
above 65◦C and a 36 dB SMSR as well as an average output power of -3 dB.
Many approaches towards electrically pumped non tunable wafer fused VCSELs
have been published. Babic et. al. [138] demonstrate a wafer fused InGaAsP based
VCSEL with two AlAs/GaAs DBRs. The threshold current density is in the order of
3 kA / cm2 at 1.52µm emission wavelength. A 12mA threshold current characterizes
laser sizes from 9 to 60µm, whereby a single mode operation could be observed for the
9µm VCSEL only. The same author later presents a double fused VCSEL operating
continuously at RT and emitting at 1.54µm [139]. The active region containing the
InGaAsP is located between a top AlGaAs/GaAs DBR and a bottom AlAs/GaAs
one. Devices with aperture sizes between 8 and 20µm are fabricated and operate
A State of the art: detailed description 101
continuously at RT. The 8µm device possesses the lowest threshold current of 2.3mA
and a maximum output power of approximately 10µW. A double fused VCSEL, with
two AlGaAs/GaAs DBRs and an InGaAsP/InP active region is reported by Margalit
et. al. [140]. The device operates in a cw and pulsed mode up to 64 ◦C and 100 ◦C,
respectively. The lowest threshold current at RT is measured to be 0.8mA and the
maximum output power is 1mW at 15 ◦C. Syrbu et. al. [141] demonstrate a single mode
wafer fused 1.55µm VCSEL exhibiting 1.5mW output power at RT and a SMSR of 30
dB. A double fused 1.528µm VCSEL with cw operation up to 85 ◦C and a maximum
output power of 0.65mW at RT is presented by Karim et. al. [142].
Electrically pumped non tunable VCSEL with hybrid DBRs are reported by sev-
eral groups. K. Streubel and co-workers [143] present a 1.55µm VCSEL with a bottom
GaInAsP/InP DBR and a Si/SiO2 top DBR. The GaInAsP based device shows a pulsed
operation within a temperature range between −160 ◦C to +43 ◦C, while a cw opera-
tion was observed up to −25 ◦C only. Ortsiefer et. al. [15] fabricate an InGaAlAs based
1.55µmVCSEL with 1.6mW output power. The bottom and top DBRs are implemen-
ted by a InGaAlAs/InAlAs and MgF2/a-Si multilayers, respectively. A low resistance
of 70 Ω assures a threshold current of 4mA. A double fused 1.55µm VCSEL with In-
GaAsP QWs between a bottom GaAs/AlAs and a combined SiO2/TiO2-InP/InGaAsP
top DBRs is presented by Ohiso et. al. [144]. Using different device aperture sizes, a
maximum output power of 2.1mW and a low threshold current of 380µA have been
demonstrated in a cw operation at RT. Sun et. al. [145] implement a cw 1.55µm VC-
SEL device with a minimum threshold current of 0.8mA at RT. The device consists of
an InAlGaAs QWs stack and an InAlGaAs/InAlAs bottom DBR as well as TiO2/SiO2
top DBR.
Monolithically grown electrically pumped VCSELs for commercial purposes have
gained importance since they enable a technological batch processing and an in-plane
characterization. Furthermore, such devices are electrically driven and thus suitable for
system integration. Over the past years, several such devices were presented by many
groups. Kazmierski et. al. [146] present a monolithic grown InGaAlAs based VCSEL
with InGaAlAs/InAlAs DBRs. A pulse operation at 1.56µm and up to +55 ◦C is
obtained. A device with a metamorphic bottom InP/InGaAsP DBR and GaAs/AlAs
top one as well as 1mW cw output power at RT is presented by Boucart et. al.
[147,148]. Yuen et. al. [149] demonstrated a cw operation up to +55 ◦C with 0.45mW
output power at RT. The active region containing the InGaAs QWs is sandwiched
between the InAlGaAs/InAlAs bottom DBR and the metamorphic AlGaAs/GaAs top
DBR. A tunnel junction VCSEL with InGaAs QWs and AlGaAsSb/AlAsSb DBRs has
been fabricated by Hall et. al. [150]. A maximum output power of 100µW and a
current threshold of 6.2mA (1.97 kA / cm2 current density) at RT is reached. Kwon
et. al. [151] implemented a fully monolithic VCSEL with InAlGaAs/InAlAs DBRs. A
minimum threshold current density of 2.93 kA / cm2 at 1.57µm is observed.
However, tunable optically and electrically pumped micromachined VCSELs for the
long haul wavelength range are very rare, due to the undesirable side effects generated
102 A.2 Vertical cavity surface emitting laser
from the micromachining process (e.g. stress and buckling). Vakhshoori et. al. [13]
presented a 1.55µm tunable cw VCSEL with 2mW output power and 50 nm tuning
range. The novelty of this work was the implementation of a stable resonator by a
curved top DBR with a ROC of 300µm. The tuning occurs electrostatically. However,
the authors did not reveal either the material system nor the technological implementa-
tion details of the curved mirrors. The author of this thesis [105] published an optically
tunable two chip 1.55µm VCSEL with a curved top dielectric DBR. The lower part
of the device consists of a GaInAsP based active region and an InP/InGaAsP bottom
DBR. The curvature design of the SiNx/SiO2 top DBR is described in detail. A tuning
range of 26 nm, a maximum output power of 400µW, a SMSR of 57 dBm and a FSR
of 47 nm are reached. The tuning occurs by a thermal actuation of the micromachined
top DBR. A wafer fused optically pumped 1.55µm tunable InAlGaAs based VCSEL
with 2mW maximum output power with a power variation less than 1.5 dB across the
32 nm tuning and a SMSR higher than 30 dB is demonstrated by Syrbu et. al. [152].
The top and Bottom DBRs are fabricated using multilayers of AlGaAs and GaAs.
Recently, a few publications reporting progress in electrically pumped tunable VC-
SEL technology have appeared. Boucart et. al. [153] implemented a monolithically
InAlGaAs based 1.55µm VCSEL with InP/InAlGaAs and a movable GaAs/AlGaAs
bottom and top DBRs, respectively. The device operates up to temperatures exceeding
+75 ◦C and shows 0.9mW output power at RT. The tuning range is 17 nm, while the
SMSR exceeds 30 dB. A tunable version of the VCSEL structure described in [149] is
fabricated by Kner et. al. [154]. The electrostatically tunable top DBR is implemented
by a cantilever according to [155]. The VCSEL shows an output power higher than
0.28mW over 10 nm tuning range and a threshold current lower than 1mA. The elec-
trically pumped tunable VCSEL described by Sun et. al. [156] consists of InAlGaAs
QWs between a InAlGaAs bottom DBR and a movable AlGaAs/GaAs top one. The
electrostatic actuation of the top membrane delivers 22 nm tuning range and 1.3mW
output power in cw operation at +15 ◦C. Maute et. al. [14] produced an electrically
tunable VCSEL, whose active region has already been described in [15]. The bottom
mirror consists of CaF2/a-Si multilayer and gold, the top movable mirror is a doped
GaAs/AlGaAs DBR. The tuning of the two-chip [16] based device is done electrotherm-
ally. The device shows a single mode operation across 30 nm tuning with a SMSR better
than 30 dB and a maximum output power of 76µW. By the same technological im-
plementation process, Riemenschneider et. al. [17] increased the tunability to 40 nm
and the output power to 100µW. Furthermore, an overview of VCSEL tunability is
presented in [157].
Appendix B
Diagram supplements
B.1 PECVD technological investigation
Table B.1: The used abbreviation
Abbreviation
Tdep
n
SiH4
NH3
pdep
DR
σ
σer
N2O
Parameter Unit
Deposition temperature ◦C
Refractive index
2% SiH4 −N2 flow sccm
NH3 flow sccm
Deposition pressure E−3 torr
Deposition rate nmmin−1
Bulk stress MPa
Bulk stress standard deviation MPa
N2O flow sccm
Table B.2: The dependence of the silicon nitride refractive index, deposition rate and stress
on the process temperature
Tdep n DR σ σer
230 1.928 15.64 −100.8 31
290 1.968 14.43 −79.9 20
300 1.963 14.25 −11.3 39
320 1.984 13.94 −80 20
Table B.3: The dependence of the silicon nitride refractive index, deposition rate and stress
on the silane flow
SiH4 n DR σ σer
800 1.936 14.58 −94.9 31.6
900 1.958 14.47 −85 16.6
1000 1.974 14.51 −12.3 5.6
1100 1.99 14.5 −100.9 12.7
1200 2.004 14.41 −129 36.5
104 B.1 PECVD technological investigation
Table B.4: The dependence of the silicon nitride refractive index, deposition rate and stress
on the amonia flow
NH3 n DR σ σer
15 2.018 13.86 −61 28.9
20 1.966 14.12 −76.7 11.5
25 1.932 13.6 −77.2 6.5
30 1.911 13.94 −198.6 0.66
Table B.5: The dependence of the silicon nitride refractive index, deposition rate and stress
on the process pressure
pdep n DR σ σer
0.5 1.928 12.52 − −
0.65 1.966 14.12 −76, 7 11, 5
0.8 1.989 15.34 − −
1.1 1.988 12.04 307, 36 38
1.3 2.024 14.13 −302, 4 1, 3
Table B.6: The dependence of the silicon dioxide refractive index, deposition rate and stress
on the process temperature
Tdep n DR σ σer
230 1.464 63.40 232.31 13.22
290 1.466 66.52 184.32 12.27
300 1.467 67.97 176.5 13.65
320 1.467 69.07 135.32 12.2
Table B.7: The dependence of the silicon dioxide refractive index, deposition rate and stress
on the silane flow
NH3 n DR σ σer
400 1.468 66.82 164.7 15.4
430 1.472 67.52 197.4 16.3
460 1.472 68.94 166.2 14.6
520 1.487 70.72 83 2
Table B.8: The dependence of the silicon dioxide refractive index, deposition rate and stress
on the nitrous oxide flow
N2O n DR σ σer
630 1.479 68.32 194.2 11.93
670 1.469 68.8 140.4 16.51
750 1.473 67.5 163.7 20.9
850 1.471 66.30 187.9 1.2
B Diagram supplements 105
Table B.9: The dependence of the silicon dioxide refractive index, deposition rate and stress
on the process pressure
pdep n DR σ σer
0.5 1.381 51.96 238.7 10.5
0.7 1.472 60.97 187.1 20
1 1.472 67.33 124 15.33
1.5 1.468 62.76 179.5 16.4
2 1.466 57.70 105.5 85.78
106 B.2 Stress investigations
B.2 Stress investigations
Table B.10: The dependence of the silicon nitride and silicon dioxide stress on the PECVD
duty cycle of the plasma excitation frequencies (process temperature : 300 ◦C)
Si3N4
Ψ σ σer
−1 856.56 31.11
0.25 187 19.16
0.428 59.33 48.19
0.454 −26.85 19.98
0.478 −52.74 24.97
0.5 −30.22 16.61
0.52 −20.72 26.21
0.538 −39.48 24.83
0.66 −124.72 7.689
1 −297.44 30.66
SiO2
Ψ σ σer
−1 461.98 56.51
−0.09 222.40 19.56
0.25 284.52 12.24
0.538 258.80 45.11
0.612 309.15 39.09
0.666 249.17 0
0.707 252.90 15.88
0.818 175.98 34.60
1 517.94 21.02
B Diagram supplements 107
Table B.11: The dependence of the silicon nitride and silicon dioxide stress on the PECVD
duty cycle of the plasma excitation frequencies (process temperature : 60 ◦C)
Si3N4
Ψ σ σer
0.555 121.11 36.24
0.571 78.66 29.34
0.586 47.54 58.49
0.6 49.24 27.65
0.612 −36.25 77.23
SiO2
Ψ σ σer
0.428 30.33 20
0.481 20.93 37
0.538 42.05 4.9
0.578 12.24 9.7
0.6 46.51 11.8
0.621 08.06 13.2
0.666 14.50 43.6
108 B.3 Stress impact on mechanical and optical properties
B.3 Stress impact on mechanical and optical prop-
erties
Table B.12: The dependence of the cavity length of the FPFs on the suspension length l at
different PECVD duty cycles ψ (Design IMA2)
Ψ = −1 Ψ = −0.75 Ψ = −0.5
l Lcav Lcav,er Lcav Lcav,er Lcav Lcav,er
10 3.62367 0.04216 3.62367 0.04216 3.62367 0.04216
20 4.22467 0.0905 5.05875 0.93395 0.27775 0.07133
30 4.95933 0.07438 5.6844 0.4708 4.6685 0.06096
40 0.43033 0.02656 6.33233 0.0368 4.5305 1.87521
60 0.4315 0.02294 6.84526 0.6107 6.62425 0
80 0.4815 0.09938 8.38806 0.2261 7.39092 0.35823
Ψ = −0.25 Ψ = 0 Ψ = 0.25
l Lcav Lcav,er Lcav Lcav,er Lcav Lcav,er
10 0.13267 0.01358 0.09733 0.119 − −
20 0.30767 0.0725 0.415 0 − −
30 5.552 0.24747 7.20067 0.4135 9.597 0
40 5.98133 0.16359 7.63167 0.22719 10.248 0.12834
60 6.72758 0.03236 8.68478 0.08834 11.1 0.25795
80 7.94625 0.16996 9.86178 0.07094 12.837 0.70139
Table B.13: The dependence of the cavity length of the FPFs on the suspension length l at
different PECVD duty cycles ψ (Design IMA3)
Ψ = −1 Ψ = −0.75 Ψ = −0.5
l Lcav Lcav,er Lcav Lcav,er Lcav Lcav,er
10 2.63933 0.12568 3.092 0.3225 3.02433 0.22813
20 2.48367 0.04464 2.86775 0.08136 3.02267 0.04479
30 2.35 0.0444 2.83125 0.01735 3.334 0.1465
40 2.262 0.08586 2.86975 0.01709 3.81433 0.0935
60 2.31767 0.04588 2.8455 0.03225 3.53833 0.37051
Ψ = −0.25 Ψ = 0 Ψ = 0.25
l Lcav Lcav,er Lcav Lcav,er Lcav Lcav,er
10 3.28967 0.09084 3.07367 0.0216 4.52467 1.09449
20 3.116 0.02762 3.11633 0.2308 3.55 0.05444
30 3.27467 0.05152 3.254 0.0115 5.25167 1.43386
40 3.418 0.01735 3.93267 0.017 5.76267 0.2092
60 3.18167 0.14272 4.15867 0.4896 5.43567 0.57239
B Diagram supplements 109
Table B.14: The dependence of the radius of curvature (ROC) of the FPF membranes on
the suspensions length l at different PECVD duty cycles ψ (Design IMA2)
Ψ = −1 Ψ = −0.75 Ψ = −0.5
l ROC ROCer ROC ROCer ROC ROCer
10 2.718 0.03033 1.77 0.47765 −0.548 0.03271
20 1.454 0.05225 0.95 0.10794 −0.662 0.03114
30 1.148 0.0249 0.81 0.03808 1.042 0.07294
40 −0.758 0.01304 0.76 0.02916 0.876 0.19191
60 −0.93 0.02121 0.74333 0.00577 0.846 0.13465
80 −0.994 0.02302 0.824 0.12321 0.81 0.06892
Ψ = −0.25 Ψ = 0 Ψ = 0.25
l ROC ROCer ROC ROCer ROC ROCer
10 −0, 734 0, 07765 −0, 95 0, 19391 − −
20 −1, 124 0, 06656 −0, 87 0 − −
30 0, 442 0, 0228 0, 28 0, 05745 0, 16 0
40 0, 432 0, 00837 0, 294 0, 03209 0, 11 0, 01732
60 0, 478 0, 02775 0, 304 0, 05459 0, 12333 0, 02887
80 0, 516 0, 0251 0, 302 0, 06611 0, 13 0, 03606
Table B.15: The dependence of the radius of curvature (ROC) of the FPF membranes on
the suspensions length l at different PECVD duty cycles ψ (Design IMA3)
Ψ = −1 Ψ = −0.75 Ψ = −0.5
l ROC ROCer ROC ROCer ROC ROCer
10 −2.85 0.16325 15.17667 − 2.83 0.43841
20 −2.226 0.46079 −1.964 6.80512 4.675 1.91626
30 −2.122 0.59116 −8.746 3.43705 2.445 0.22576
40 −1.802 0.48111 −3.468 1.339 0.2075 2.7831
60 −2.46 0.58991 −3.79143 3.41535 2.68667 0.27209
Ψ = −0.25 Ψ = 0 Ψ = 0.25
l ROC ROCer ROC ROCer ROC ROCer
10 3.87333 1.29063 1.288 0.16193 0.18333 0.07506
20 7.25667 1.36676 2.05 0.34547 0.32 0.04
30 5.70333 0.41004 1.144 0.19616 0.22667 0.04042
40 1.99333 0.05508 0.67 0.09849 0.29 0.07
60 1.90333 0.13051 0.608 0.00837 0.22667 0.00577
110 B.3 Stress impact on mechanical and optical properties
Table B.16: The dependence of the FWHM of the FPFs on the suspensions length l at
different PECVD duty cycles ψ (Design IMA2)
Ψ = −1 Ψ = −0.75 Ψ = −0.5 Ψ = −0.25 Ψ = 0 Ψ = 0.25
l FWHM FWHM FWHM FWHM FWHM FWHM
10 16 14.25 45 45.75 27.75 −
20 17 23.25 33 56.25 31.5 −
30 25.75 27.75 34.25 12.5 27 64.25
40 30.25 17.75 17.75 30.75 5.25 5.75
60 25 12 8.75 20.5 2.25 28.75
80 62.25 9.5 30.75 20 1.5 24.25
Table B.17: The dependence of the FWHM of the FPFs on the suspensions length l at
different PECVD duty cycles ψ (Design IMA3)
Ψ = −1 Ψ = −0.75 Ψ = −0.5 Ψ = −0.25 Ψ = 0 Ψ = 0.25
l FWHM FWHM FWHM FWHM FWHM FWHM
10 − 21.5 14.75 7.75 6.75 33.75
20 − 17.5 9.25 24.75 5.25 44.25
30 − 5 10.25 14.75 5.25 33
40 − 25.5 5.5 10.5 8.25 36.25
60 23.5 22 17 15.25 2.75 26.5
Table B.18: The dependence of the RW of the FPFs on the suspensions length l at different
PECVD duty cycles ψ (Design IMA2 and IMA3)
IMA2 Ψ = −1 Ψ = −0.75 Ψ = −0.5 Ψ = −0.25 Ψ = 0 Ψ = 0.25
l RW RW RW RW RW RW
10 1468.5 1503.75 1573.25 1550.5 1589.5 −
20 1504.5 1584 1535.25 1539.75 1595.75 −
30 1463.5 1468.25 1546.25 1506.75 1549 1504.5
40 1476 1487.5 1529 1566.25 1580.5 1521
60 1455.5 1463 1502 1496.75 1457.75 1521.25
80 1490.75 1524 1550.5 1535.25 1519 1517
IMA3 Ψ = −1 Ψ = −0.75 Ψ = −0.5 Ψ = −0.25 Ψ = 0 Ψ = 0.25
l RW RW RW RW RW RW
10 1556.75 1587.25 1544 1464 1464.25 1562.5
20 1450 1536.75 1488.5 1578.25 1546.75 1561.5
30 1649 1451.25 1570.25 1536.5 1556.25 1559.5
40 1573 1514.25 1481.75 1473.75 1572.75 1509.5
60 1476.5 1566.75 1576.25 1484.5 1546.25 1549.25
Appendix C
Technological process flow
C.1 Air-gap Fabry-Pérot filters
C.1.1 Design IMA2
Table C.1: Process technology data of the air-gap filter implemented using the design IMA2
Process step Parameter Time
1 Substrate cleaning
RCA H2SO4 : H2O2 (3:1), T = 90◦C 30min
RCA NH3 : H2O2 : H2O (1:1:5), T = 60◦C 30min
RCA HCl : H2O2 : H2O (1:1:5), T = 60◦C 30min
2 Deposition of the bottom DBR, 5 periods
SiO2 2% SiH4-N2 (430 sccm)
N2O (710 sccm)
pressure (1E−3 torr)
HF power (20W)
Ψ (1)
T (300◦C)
0.25λ @ 1450 nm, nSiO2 = 1.469
Si3N4 2% SiH4-N2 (1000 sccm) 10min 17 s
NH3 (20 sccm)
pressure (0.65E−3 torr)
HF & LFpower (20W)
Ψ (0.538)
T (300◦C)
0.25λ @ 1450 nm, nSi3N4 = 1.933
112 C.1 Air-gap Fabry-Pérot filters
Table C.2: Continuation
3 Sacrificial layer
”Metall” TI35ES / 4000rpm 2min
pre-bake, T = 90◦C 2min
Exposure 35 s
Waiting 10min
Reversal bake, T = 130◦C 4min
Flood exposure 65 s
Development, AZ826 40 s
O2 ashing pressure (0.7mbar), power (250W) 1min
Long bake T = 130◦C 8 h
4 Deposition of the top DBR, 5.5 periods
SiO2 2% SiH4-N2 (430 sccm)
N2O (710 sccm)
pressure (1E−3 torr)
HF & LF power (20W)
Ψ (0.481)
T (60◦C)
0.25λ @ 1450 nm, nSiO2 = 1.407
Si3N4 2% SiH4-N2 (1500 sccm)
NH3 (5 sccm)
pressure (0.65E−3 torr)
LF power (20W)
Ψ (variable)
T (60◦C)
0.25λ @ 1450 nm, nSi3N4 = 1.832
Table C.3: Continuation
5 Mesa etching
”Mesa” AZ1518 / 4000rpm 2min
pre-bake, T = 90◦C 5min
Exposure 8.5 s
Development, KOH 0.8% 45 s
O2 ashing pressure (0.7mbar), power (50W) 2min
RIE CHF3 (10.5 sccm)
Si3N4 Ar (15.3 sccm)
Power (140W)
pressure (0.035E−3 torr)
T (12◦C)
6 Removal of the sacrificial layer
Underetching pressure (0.7mbar), power (250W) 38min
C Technological process flow 113
C.1.2 Design IMA3
Table C.4: Process technology data of the air-gap filter implemented using the design IMA3
Process step Parameter Time
1 Substrate cleaning
RCA H2SO4 : H2O2 (3:1), T = 90◦C 30min
RCA NH3 : H2O2 : H2O (1:1:5), T = 60◦C 30min
RCA HCl : H2O2 : H2O (1:1:5), T = 60◦C 30min
2 Deposition of the bottom DBR, 5 periods
SiO2 2% SiH4-N2 (430 sccm)
N2O (710 sccm)
pressure (1E−3 torr)
HF power (20W)
Ψ (1)
T (300◦C)
0.25λ @ 1450 nm, nSiO2 = 1.469
Si3N4 2% SiH4-N2 (1000 sccm) 10min 17 s
NH3 (20 sccm)
pressure (0.65E−3 torr)
HF & LFpower (20W)
Ψ (0.538)
T (300◦C)
0.25λ @ 1450 nm, nSi3N4 = 1.933
114 C.1 Air-gap Fabry-Pérot filters
Table C.5: Continuation
3 Sacrificial layer
”Protection” TI35ES / 4000rpm 2min
pre-bake, T = 90◦C 2min
Exposure 35 s
Waiting 10min
Reversal bake, T = 130◦C 4min
Flood exposure 65 s
Development, AZ826 40 s
O2 ashing pressure (0.7mbar), power (250W) 1min
Long bake T = 130◦C 8 h
4 Deposition of the top DBR, 5.5 periods
SiO2 2% SiH4-N2 (430 sccm)
N2O (710 sccm)
pressure (1E−3 torr)
HF & LF power (20W)
Ψ (0.481)
T (60◦C)
0.25λ @ 1450 nm, nSiO2 = 1.407
Si3N4 2% SiH4-N2 (1500 sccm)
NH3 (5 sccm)
pressure (0.65E−3 torr)
LF power (20W)
Ψ (variable)
T (60◦C)
0.25λ @ 1450 nm, nSi3N4 = 1.832
Table C.6: Continuation
5 Mesa etching
”Mesa” AZ1518 / 4000rpm 2min
pre-bake, T = 90◦C 5min
Exposure 8.5 s
Development, KOH 0.8% 45 s
O2 ashing pressure (0.7mbar), power (50W) 2min
RIE CHF3 (10.5 sccm)
Si3N4 Ar (15.3 sccm)
Power (140W)
pressure (0.035E−3 torr)
T (12◦C)
6 Removal of the sacrificial layer
Underetching pressure (0.7mbar), power (250W) 38min
C Technological process flow 115
C.2 Process flow of the tunable Fabry-Pérot filter
Table C.7: Process technology data of the tunable air-gap filter implemented using the design
IMA2
Process step Parameter Time
1 Substrate cleaning
RCA H2SO4 : H2O2 (3:1), T = 90◦C 30min
RCA NH3 : H2O2 : H2O (1:1:5), T = 60◦C 30min
RCA HCl : H2O2 : H2O (1:1:5), T = 60◦C 30min
2 Deposition of the bottom DBR, 5 periods
SiO2 2% SiH4-N2 (430 sccm) 6min 20 s
N2O (710 sccm)
pressure (1E−3 torr)
HF & LF power (20W)
Ψ (0.481)
T (60◦C)
0.25λ @ 1400 nm, nSiO2 = 1.3
Si3N4 2% SiH4-N2 (1500 sccm) 10min 17 s
NH3 (5 sccm)
pressure (0.65E−3 torr)
HF power (20W)
Ψ (1)
T (60◦C)
0.25λ @ 1400 nm, nSi3N4 = 1.7
3 Sacrificial layer
”contacts” AZ5214e / 3000rpm 1min
pre-bake, T = 90◦C 1min
Exposure 12 s
Reversal bake, T = 120◦C 2min
Flood exposure 12 s
KOH, 0.8% 40 s
O2 clean pressure (0.7mbar), power (250W) 1min
116 C.2 Process flow of the tunable Fabry-Pérot filter
Table C.8: Continuation
Process step Parameter Time
4 Deposition of the top DBR, 5.5 periods
SiO2 2% SiH4-N2 (430 sccm) 6min 20 s
N2O (710 sccm)
pressure (1E−3 torr)
HF & LF power (20W)
Ψ (0.481)
T (60◦C)
0.25λ @ 1400 nm, nSiO2 = 1.3
Si3N4 2% SiH4-N2 (1500 sccm) 10min 17 s
NH3 (5 sccm)
pressure (0.65E−3 torr)
HF power (20W)
Ψ (1)
T (60◦C)
0.25λ @ 1400 nm, nSi3N4 = 1.7
5 Micro heaters
Cr 100 nm
”Heat” AZ1518 / 4000rpm 1min
pre-bake, T = 90◦C 5min
Exposure 5.5 s
KOH, 0.8% 40 s
Cr etching Cr etchant (7334) : H2O (1:2) 1min 33 s
PR removal Aceton 2min
”Test structure” AZ1518 / 4000rpm 1min
pre-bake, T = 90◦C 5min
Exposure 7 s
KOH, 0.8% 40 s
Cr etching Cr etchant (7334) : H2O (1:2) 2min 30 s
Table C.9: Continuation
6 MESA etching
Si3N4 and SiO2 Ar (5.1 sccm) 330min
CHF3 (3.5 sccm)
pressure (50E−3 torr)
DC bias (210V)
RF level (30%)
T (12◦C)
7 Membrane release
Polymer removal pressure (0.7mbar), power (50W) 10min
Underetching pressure (0.7mbar), power (250W) 15min
C Technological process flow 117
C.3 Process flow of the non tunable VCSEL
Table C.10: Process parameters of the non tunable VCSEL
Process step Parameter Time
2 Deposition of the bottom DBR, 12 periods
SiO2 2% SiH4-N2 (430 sccm)
N2O (710 sccm)
pressure (1E−3 torr)
HF power (20W)
Ψ (1)
T (300◦C)
0.25λ @ 1450 nm, nSiO2 = 1.469
Si3N4 2% SiH4-N2 (1000 sccm)
NH3 (20 sccm)
pressure (0.65E−3 torr)
HF & LF power (20W)
Ψ (0.538)
T (300◦C)
0.25λ @ 1450 nm, nSi3N4 = 1.933
Ti/Au on the top of the DBR
Ti/Au 50 nm /300 nm
Bonding of the structure up side down by Indium
Coating of sample borders with AZ4562
InP etching HCl (37%) 42min
InGaAs etching FeCl3 (50 g+100ml H2O): H2O [1 : 2] 30 s
PR removal O2 plasma, 400W, 1.3mbar, 0.7 sccm 30min
Deposition of the top DBR, 12 periods, similar to the bottom DBR
Appendix D
DBRs for the blue wavelength
range
Table D.1: Process flow of the short wavelength DBRs (TQ067 and TQ069)
Process step Parameter Time
2 Deposition of DBR, 12.5 periods
SiO2 2% SiH4-N2 (430 sccm)
N2O (710 sccm)
pressure (1E−3 torr)
HF power (20W)
Ψ (1)
T (300◦C)
0.25λ @ 1450 nm, nSiO2 = 1.49
Si3N4 2% SiH4-N2 (1000 sccm)
NH3 (20 sccm)
pressure (0.65E−3 torr)
HF & LFpower (20W)
Ψ (0.538)
T (300◦C)
0.25λ @ 1450 nm, nSi3N4 = 2.03
Appendix E
Abbreviations
Table E.1: Abbreviations of the most frequently used expres-
sions
Abbreviation Description
ASE Amplified Spontaneous Emission
AMU Automatic Matching Unit
AsH3 Arsine
CAGR Compound Annual Growth Rate
CDWDM Coarse Dense Wavelength Division Multiplexing
CPD Critical Point Drying
CWL Central Wavelength
CW Continuous Wave
DWDM Dense Wavelength Division Multiplexing
DBR Distributed Bragg Reflector
EDFA Erbium Doped Fiber Amplifier
FPF Fabry-Pérot Filter
FSR Free Spectral Range
FWHM Full Width at Half Maximum
GaAs Gallium Arsenid
HF High Frequency
ITU International Telecommunications Union
InGaAsP Indium Gallium Arsenid Phosphid
InP Indium Phosphid
LF Low Frequency
LPCVD Liquid Phase Chemical Vapor Deposition
MAN Metropolitan Area Networks
MEMS Micro-Electro-Mechanical-System
MOEMS Micro-Opto-Electro-Mechanical-System
MOVPE Metal-Organic Vapor Phase Epitaxy
MFC Mass Flow Controller
PECVD Plasma Enhanced Chemical Vapor Deposition
120
Table E.1: Continuation
Abbreviation Description
PH3 Phosphine
PR Photoresist
PL Photoluminescence
PRI Pulse Repetition Interval
QW Quantum Well
ROC Radius Of Curvature
RIE Reactive Ion Etching
RT Room Temperature
RW Resonance Wavelength
SEM Scanning Electron Microscope
SMSR Side Modes Suppression Ratio
SiH4 Silane
Si3N4 Silicon nitride
SiO2 Silicon dioxide
Si Silicon
T/R Transmitter/Receiver
TMIn Trimethylindium
TMGa Trimethylgallium
VCSEL Vertical Cavity Surface Emitting Laser
WDM Wavelength Division Multiplexing
WWW World Wide Web
WLI White Light Interferometry
Appendix F
Stress impact on the ROC, Lcav,
RW and FWHM
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
0
10
20
30
40
50
60
70
98
Fu
ll 
w
id
th
 a
t h
al
f m
ax
im
um
 (F
W
H
M
) /
 n
m
Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF - tLF)
Suspension length
 l = 10       l = 20
 l = 30       l = 40
 l = 60       l = 80
IMA2
Figure F.1: Dependence of the FWHM of the Fabry-Pérot filters on the duty cycle Ψ at
different suspension lengths l for the design IMA2.
122
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
0
5
10
15
20
25
30
35
40
45
 Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF - tLF)
Fu
ll 
w
id
th
 a
t h
al
f m
ax
im
um
 (F
W
H
M
) /
 n
m Suspension length (l)
 l = 10
 l = 20
 l = 30
 l = 40
 l = 60
IMA3
Figure F.2: Dependence of the FWHM of the Fabry-Pérot filters on the duty cycle Ψ at
different suspension lengths l for the design IMA3.
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
-1,50
-0,75
0,00
0,75
1,50
2,25
3,00 IMA 2
Fl
at
Fl
at
 
R
ad
iu
s 
of
 c
ur
va
tu
re
 (R
O
C
) /
 m
m
C
on
ca
ve
C
on
ve
x
Suspension length
 l = 10
 l = 20
 l = 30
 l = 40
 l = 60
 l = 80
Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF + tLF)
Figure F.3: Dependence of the radius of curvature of the Fabry-Pérot filter membrane on
the duty cycle Ψ at different suspension lengths l for the design IMA2.
F Stress impact on the ROC, Lcav, RW and FWHM 123
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
-10,00
-5,00
0,00
5,00
16,00
 
IMA 3
 
Fl
at
Fl
at
C
on
ca
ve
C
on
ve
x
R
ad
iu
s 
of
 c
ur
va
tu
re
 (R
O
C
) /
 m
m
Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF + tLF)
Suspension length
 l = 10
 l = 20
 l = 30
 l = 40
 l = 60
Figure F.4: Dependence of the radius of curvature of the Fabry-Pérot filter membrane on
the duty cycle Ψ at different suspension lengths l for the design IMA3.
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
0,00
2,00
4,00
6,00
8,00
10,00
12,00
14,00
C
av
ity
 le
ng
th
 (l
) /
 µ
m
Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF + tLF)
Suspension length
 l = 10
 l = 20
 l = 30
 l = 40
 l = 60
 l = 80
IMA 2
Figure F.5: Dependence of the cavity length of the Fabry-Pérot filters on the duty cycle Ψ
at different suspension lengths l for the design IMA2.
124
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
0,00
2,00
3,00
4,00
5,00
6,00
7,00
 Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF + tLF)
C
av
ity
 le
ng
th
 (L
) /
 µ
m
Suspension length
 l = 10
 l = 20
 l = 30
 l = 40
 l = 60
IMA 3
Figure F.6: Dependence of the cavity length of the Fabry-Pérot filters on the duty cycle Ψ
at different suspension lengths l for the design IMA3.
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
1428
1479
1530
1581
1632
1683
 Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF + tLF)
R
es
on
an
ce
 w
av
el
en
gt
h 
/ n
m
Suspension length
 l = 10
 l = 20
 l = 30
 l = 40
 l = 60
 l = 80
IMA2
Figure F.7: Dependence of the resonance wavelength of the Fabry-Pérot filters on the duty
cycle Ψ at different suspension lengths l for the design IMA2.
F Stress impact on the ROC, Lcav, RW and FWHM 125
-1,25 -1,00 -0,75 -0,50 -0,25 0,00 0,25 1,2
1428
1479
1530
1581
1632
1683
R
es
on
an
ce
 w
av
el
en
gt
h 
/ n
m
Frequencies duty cycle: Ψ = (tHF - tLF) / (tHF + tLF)
Suspension length
 l = 10
 l = 20
 l = 30
 l = 40
 l = 60
IMA3
Figure F.8: Dependence of the resonance wavelength of the Fabry-Pérot filters on the duty
cycle Ψ at different suspension lengths l for the design IMA3.