Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
VERTICAL-CAVITY SURFACE-EMITTING LASER
Document Type and Number:
WIPO Patent Application WO/2018/130462
Kind Code:
A1
Abstract:
An embodiment of the invention relates to a vertical cavity surface emitting laser comprising a first reflector, a second reflector comprising a layer stack of semiconductor or isolating layers, an active region arranged between the first and second reflectors, and an additional layer on top of the layer stack at the light output side, said additional layer forming an output interface of the laser, wherein the refractive index of the additional layer is smaller, equal to or larger than the smallest refractive index of the refractive indices of said layer stack.

Inventors:
BIMBERG DIETER (DE)
LARISCH GUNTER (DE)
LOTT JAMES A (DE)
Application Number:
PCT/EP2018/050238
Publication Date:
July 19, 2018
Filing Date:
January 05, 2018
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
UNIV BERLIN TECH (DE)
International Classes:
H01S5/183
Foreign References:
US20160134083A12016-05-12
US5966399A1999-10-12
US6144682A2000-11-07
Other References:
P. WOLF; P. MOSER; G. LARISCH; D. H. BIMBERG: "High-Speed and Temperature-Stable oxide-confined 980-nm VCSELs for Optical Interconnects", IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, vol. 19, 2013, pages 1 - 7, XP011508664, DOI: doi:10.1109/JSTQE.2013.2246773
A. MUTIG; J. A. LOTT; S. A. BLOKHIN; P. MOSER; P. WOLF; W. HOFMANN ET AL.: "Highly temperature-stable modulation characteristics of multioxide-aperture high-speed 980 nm vertical cavity surface emitting lasers", APPLIED PHYSICS LETTERS, vol. 97, 2010, pages 151101 - 151113
D. ELLAFI; V. IAKOVLEV; A. SIRBU; G. SURUCEANU; Z. MICKOVIC; A. CALIMAN ET AL.: "Control of cavity lifetime of 1.5-Dm wafer-fused VCSELs by digital mirror trimming", OPTICS EXPRESS, vol. 22, 2014, pages 32180
P. WESTBERGH; J. S. GUSTAVSSON; B. KOGEL; A. HAGLUND; A. LARSSON: "Impact of Photon Lifetime on High-Speed VCSEL Performance", IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, vol. 17, November 2011 (2011-11-01), pages 1603 - 1613, XP011477872, DOI: doi:10.1109/JSTQE.2011.2114642
M. KOHLER; A. WIEGAND: "Etching in microsystem technology", 1999, WILEY-VCH
C. A. B. G. C. DESALVO; J. L. EBEL; D. C. LOOK; J. P. BARRETTE; C. L. A. CERNY; R. W. DETTMER; J. K. GILLESPIE; C. K. HAVASY; T. J: "Wet Chemical Digital Etching of GaAs at Room Temperature", J. ELECTROCHEM. SOC., vol. 143, 1996
A. LARSSON: "Advances in VCSELs for Communication and Sensing", IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, vol. 17, 2011, pages 1552 - 1567, XP011459291, DOI: doi:10.1109/JSTQE.2011.2119469
L. A. COLDREN; S. W. CORZINE: "Diode lasers and photonic integrated circuits", 1995, WILEY
G. LARISCH; P. MOSER; J. A. LOTT; D. BIMBERG: "Impact of Photon Lifetime on the Temperature Stability of 50 Gb/s 980 nm VCSELs", IEEE PHOTONICS TECHNOLOGY LETTERS, vol. 28, 2016, pages 2327 - 2330, XP011624216, DOI: doi:10.1109/LPT.2016.2592985
G. LARISCH; P. MOSER; J. A. LOTT; D. BIMBERG: "Correlation of photon lifetime and maximum bit rate for 55 Gbit/s energy-efficient 980 nm VCSELs", 2016 IEEE OPTICAL INTERCONNECTS CONFERENCE, 2016, pages 16 - 17, XP032907849, DOI: doi:10.1109/OIC.2016.7482995
Attorney, Agent or Firm:
FISCHER, Uwe (DE)
Download PDF:
Claims:
Claims

1. Vertical cavity surface emitting laser comprising

- a first reflector,

- a second reflector comprising a layer stack of semiconductor or isolating layers,

- an active region arranged between the first and second re¬ flectors, and

- an additional layer on top of the layer stack at the light output side, said additional layer forming an output in¬ terface of the laser,

- wherein the refractive index of the additional layer is smaller than the smallest refractive index of the refrac¬ tive indices of said layer stack,

wherein the thickness of the additional layer is chosen such that a design parameter H(fr) is in the range between 2.5 and 3.2 dB, and

wherein the design parameter H(fr) is determined by

- applying a voltage that generates a predetermined bias

current and a small sinusoidal current signal through the laser,

- sweeping the frequency of the sinusoidal signal in a given frequency range,

- measuring the small signal response of the radiation that leaves the laser at the output interface,

- fitting the measured small signal response to the follow¬ ing transmission function H(f): and determining the parameters fR, γ , A and fP, wherein fR describes a relaxation resonance frequency, γ the damping parameter, A an offset and fP the 3 dB fre¬ quency of the laser's parasitics, and

- calculating the design parameter H(fR) by calculating the value of the transmission function H(f) at the relaxation resonance frequency fR based on the parameters that were determined as explained above. 2. Laser according to claim 1 wherein

the refractive index of the additional layer is at least 25% smaller than the smallest refractive index of the refractive indices of said layer stack. 3. Laser according to any of the preceding claims wherein an electrical contact is arranged on at least one section of a highly doped layer of said layer stack, the doping concentration of said highly doped layer exceeding 5*1018cm~3. 4. Laser according to claim 3 wherein

the additional layer partly covers the electrical contact.

5. Laser according to any of the preceding claims wherein the additional layer is an isolating layer.

6. Laser according to any of the preceding claims wherein the additional layer is a silicon nitride or a silicon oxide layer or a titanium oxide layer. 7. Laser according to any of the preceding claims wherein the additional layer is a conductive layer.

8. Laser according to any of the preceding claims wherein the additional layer is an indium tin oxide or a GaAs layer.

9. Laser according to any of the preceding claims wherein the thickness of the additional layer is between (0.02*λ and 0,24*λ) + η*0.5*λ or between (0.26*λ and 0,48*λ) + η*0.5*λ where n is an integer.

10. Method of fabricating a vertical cavity surface emitting laser, the method comprising the steps of

- fabricating a first reflector,

- fabricating an active region on top of the first reflec¬ tor,

- fabricating a second reflector on top of the active re¬ gion, the second reflector comprising a layer stack of semiconductor layers or isolating layers, and

- depositing an additional layer on top of the layer stack,

- wherein the refractive index of the additional layer is smaller than the smallest refractive index of the refrac¬ tive indices of said layer stack.

11. Method of claim 10 wherein

the step of depositing the additional layer is completed when a design parameter H(fR) is in the range between 2.5 and 3.2 dB wherein the design parameter H(fR) is determined by

- applying a voltage that generates a predetermined bias

current and a small sinusoidal current signal through the laser,

- sweeping the frequency of the sinusoidal signal in a given frequency range,

- measuring the small signal response of the radiation that leaves the laser at the output interface, - fitting the measured small signal response to the follow¬ ing transmission function H(f): and determining the parameters fR, γ , A and fP,

wherein fR describes a relaxation resonance frequency, γ the damping parameter, A an offset and fP the 3 dB fre¬ quency of the laser's parasitics, and

- calculating the design parameter H(fR) by calculating the value of the transmission function H(f) at the relaxation resonance frequency fR based on the parameters that were determined in the previous step. 12. Method of any of the preceding claims 10-11

- wherein the laser is fabricated by depositing the addi¬ tional layer with a suitable layer thickness that provides a design parameter H(fR) in the range between 2.5 and 3.2 dB, and

- wherein the suitable layer thickness has been determined beforehand based on a plurality of lasers with varying thicknesses of the additional layer and therefore varying design parameters H(fR). 13. Method of any of the preceding claims 10-12 wherein the deposition of the additional layer comprises one or more deposition steps, each deposition step comprising:

(a) depositing a thin layer of isolating or conducting material,

(b) applying a voltage that generates a predetermined bias current and a small sinusoidal current signal through the laser structure, (c) sweeping the frequency of the sinusoidal signal in a given frequency range,

(d) measuring the small signal response of the radiation that leaves the laser structure at the output interface,

(e) fitting the measured small signal response to the follow¬ ing transmission function H(f): and determining the parameters fR, γ , A and fP,

wherein fR describes a relaxation resonance frequency, γ the damping parameter, A an offset and fP the 3 dB fre¬ quency of the laser's parasitics, and

(f) calculating the design parameter H(fR) by calculating the value of the transmission function H(f) at the relaxation resonance frequency fR based on the parameters determined in step (e) ,

(g) if the design parameter H(fR) is in the range between 2.5 and 3.2 dB, finishing the deposition of layer material, and

(h) if the design parameter H(fR) is outside said range between 2.5 and 3.2 dB, repeating above steps (a) - (h) .

14. Method of any of the preceding claims 10-13 further comprising the steps of:

- depositing an electric contact layer on a section of a

highly doped layer of the second reflector, the doping concentration of said highly doped layer exceeding

5*1018cm"3, and

- depositing an isolating material to planarize the wafer surface .

15. Method of any of the preceding claims 10-14 wherein the additional layer is deposited after the deposition of the contact layer. 16. Method of any of the preceding claims 10-15 wherein

- a contact layer is deposited on a section of a highly

doped layer of said layer stack, the doping concentration of said highly doped layer exceeding 5*1018cm~3,

- the additional layer is deposited after the deposition of the contact layer and

- said voltage that generates the predetermined bias current and the small sinusoidal current signal, is applied to the contact layer after the deposition of the additional layer .

17. Method of any of the preceding claims 10-16 wherein

- a contact layer is deposited on a section of a highly

doped layer of said layer stack, the doping concentration of said highly doped layer exceeding 5*1018cm~3,

- the additional layer is deposited after the deposition of the contact layer and

- said voltage that generates the predetermined bias current and the small sinusoidal current signal, is applied to the contact layer after the deposition of the additional layer.

18. Method of any of the preceding claims 10-17 wherein the additional layer is an isolating layer. 19. Method of any of the preceding claims 10-18 wherein the additional layer is a silicon nitride layer, a silicon oxide layer, a titanium oxide layer, or an indium tin oxide layer .

Description:
Description

Vertical-cavity surface-emitting laser Background of the invention

Fundamental goals of research and development in the field of optical data communication are energy-efficiency and error- free data transmission across maximum distances at largest bit rates. For conventional data transmission systems based on multiple-mode optical fiber lengths of less than 2 km, as for supercomputers or between servers of data centers, high modulation bandwidth VCSELs are indispensable. In order to optimize the VCSEL bandwidth, the differential gain dgldN is maximized, the active volume B is minimized (for example by reducing the optical cavity length to ¾ ) , and large pump currents are achieved by minimizing the resistance and the thermal conductivity [1, 2]. An additional degree-of-freedom, that has been largely overlooked until now, is the optimiza ¬ tion of the cavity photon lifetime, essentially controlled by mirror reflectivity. Here we describe a low cost method to adjust the mirror reflectivity to optimize the cavity photon lifetime .

Previous approaches to adjust the cavity photon lifetime used dry etching of the top mirror surface [3, 4] . Very low etching rates and very shallow depths as well as an extremely precise control of the etching progress and the homogeneity across the surface are necessary for this purpose. However, the etching rate during dry etching is not constant in time. As an alternative, several wet chemical etching procedures have been investigated. A frequently used wet chemical solution exposed GaAs layers to a mixture of an acid (usually sulfuric acid H 2 SO 4 , phos ¬ phoric acid H 3 PO 4 , or citric acid C 6 ¾07) and hydrogen perox ¬ ide (H 2 O 2 ) , along with purified water (¾0) to dilute the so- lution and thus reduce the etch rate. It is also possible to use hydrochloric acid (HC1) along with nitric acid (HNO 3 ) , as the oxidizer. Hydrogen peroxide or nitric acid have the task of oxidizing the semiconductor surface. The oxides are etched away by the acids. Depending on the concentration of the so- lution, a desired etching rate can be achieved. However, the etching rate changes very sensitively with the concentration and the temperature of the solution, with its pH value (po ¬ tential or power of hydrogen) and with its movements. Typi ¬ cally the etch rates are time dependent and nonuniform across the surface, unless a large volume of the solution is con ¬ tinuously mixed, to achieve a constant value at the etched surface [5] . Wet etching is also negatively impacted by the surface features of the VCSEL, such as the top metal ring contact and by any surface defects, thus generally resulting in unsatisfactory results being not well suited for high volume manufacturing.

A particularly precise control of the etch depth is claimed by the use of digital etching [6] . Here, oxidation and the etching of the oxides are separated from each other by two alternating repeatable steps. The surface is oxidized with hydrogen peroxide without the presence of an acid. This proc ¬ ess is limited by diffusion and results in a precise oxide thickness on the GaAs surface for VCSELs using GaAs-based top mirrors, or any other surface for other types of VCSELs. For a very broad time window (for example between 5 s and 120 s) of exposure to the hydrogen peroxide an oxidation depth limited to about 15 nm is achieved [6] . The oxide is then re- moved by the acid in the absence of the hydrogen peroxide. The acid removes the oxidized GaAs leaving a fresh GaAs sur ¬ face which may then be again oxidized. The treatment with the acid must be strictly separated from the treatment with the hydrogen peroxide, such that the removal of oxidized layers of GaAs proceeds in steps. To ensure this, the structures are rinsed with ultrapure deionized water after each step and dried using a wafer spinner. The use of HC1 can lead to prob ¬ lems with the metal contacts. The use of CfjRgOi is believed to be more gentle compared to the other acids. The temperature of H 2 S0 4 :H 2 0 2 : H 2 0 must be monitored since the initial mixture results in an exothermic reaction (temperature rise) . This digital etching method has been reported to result in reli ¬ able etching depths. However, any surface roughness, surface defect, or geometric proximity effect renders this technique unreliable. The exposed GaAs mirror surface is strongly at ¬ tacked, irrespective of the acid or etching solution used and the method of application. Figure 1 as an example, shows this very clearly.

More specifically, Figure 1 shows scanning electron micro ¬ graphs of the GaAs mirror surface of completely processed VCSELs after digital wet etching using the technique de ¬ scribed in [6] . Within the top metal ring contact, the GaAs is pitted, the etching is nonuniform, and surface defects lead to serious cracking. Additionally, residual photoresist results in obvious unwanted features near the inner edges of the metal rings. The digital technique is unstable and not suitable for volume production.

Altering the cavity photon lifetime by wet or dry etching is thus at least complex and/or uncontrollable, typically re- suiting in unintended destruction of the VCSEL surface lead ¬ ing to reliability problems.

The invention as described hereinafter proposes a simple and nondestructive method to address the problems discussed above .

Objective of the present invention

An objective of the present invention is to provide a verti- cal cavity surface emitting laser with an improved performance compared to prior art devices.

A further objective of the present invention is to provide a method for fabricating a vertical cavity surface emitting la- ser that exhibits an improved performance compared to prior art devices.

Brief summary of the invention

An exemplary embodiment of the invention relates to a verti- cal cavity surface emitting laser comprising

- a first reflector,

- a second reflector comprising a layer stack of semiconductor or isolating layers,

- an active region arranged between the first and second re- flectors, and

- an additional layer on top of the layer stack at the light output side, said additional layer forming an output in ¬ terface of the laser,

- wherein the refractive index of the additional layer is smaller than the smallest refractive index of the refrac ¬ tive indices of said layer stack. Depending on the design parameters, the exemplary embodiment may show improved laser characteristics, in particular with respect to the small-signal modulation bandwidth, the energy consumption for data transmission at a given bit rate, and/or the achievable laser diode output power.

The embodiment as described above may have, but does not need to have, one or more of the following features, which are considered to further improve the laser's performance, but are not mandatory:

The transmission of the additional layer at the emission wavelength of the laser is preferably larger than 99%. According to a preferred embodiment, the thickness of the ad ¬ ditional layer is chosen such that a design parameter H(f r ) is in the range between 2.5 and 3.2 dB, wherein the design parameter H(f r ) is determined by

- applying a voltage that generates a predetermined bias

current and a small sinusoidal current signal through the laser,

- sweeping the frequency of the sinusoidal signal in a given frequency range,

- measuring the small signal response of the radiation that leaves the laser at the output interface,

- fitting the measured small signal response to the follow ¬ ing transmission function H(f): and determining the parameters f R , γ , A and f P ,

wherein f R describes the relaxation resonance frequency, γ the damping parameter, A an offset and f P the 3 dB fre ¬ quency of the laser's parasitics, and

- calculating the design parameter H(f R ) by calculating the value of the transmission function H(f) at the relaxation resonance frequency f R based on the parameters that were determined as explained above.

An electrical contact is preferably arranged on at least one section of a highly doped layer of said layer stack. The dop ing concentration of said highly doped layer preferably exceeds 5*10 18 cm "3 .

The additional layer may partly cover the electrical contact. The additional layer may be an isolating layer (e.g. a di ¬ electric layer) or a conductive layer.

Preferably, the additional layer is a silicon nitride layer, a silicon oxide layer, a titanium oxide layer, an indium tin oxide layer or a GaAs layer.

The thickness of the additional layer is preferably between (0.02*λ and 0,24*λ) + η*0.5*λ or between (0.26*λ and 0,48*λ)+ η*0.5*λ, where n is an integer. In other words, the thickness d of the additional layer preferably lies in one of the fol ¬ lowing ranges:

0.02*λ + η*0.5*λ ≤ d ≤ 0,24*λ + η*0.5*λ or 0.26*λ + η*0.5*λ ≤ d ≤ 0,48*λ + η*0.5*λ. λ is the wavelength of the emitted radiation in the given material .

A further exemplary embodiment of the invention relates to a method of fabricating a vertical cavity surface emitting la ¬ ser, the method comprising the steps of

- fabricating a first reflector,

- fabricating an active region on top of the first reflec ¬ tor,

- fabricating a second reflector on top of the active re ¬ gion, the second reflector comprising a layer stack of semiconductor layers or isolating layers, and

- depositing an additional layer on top of the layer stack,

- wherein the refractive index of the additional layer is smaller than the smallest refractive index of the refrac ¬ tive indices of said layer stack.

The method as described above may comprise, but does not need to comprise, one or more of the following features which are considered to further improve the laser's performance, but are not mandatory:

The step of depositing the additional layer is preferably completed when a design parameter H(f R ) is in the range be- tween 2.5 and 3.2 dB wherein the design parameter H(f R ) is determined by

- applying a voltage that generates a predetermined bias

current and a small sinusoidal current signal through the laser,

- sweeping the frequency of the sinusoidal signal in a given frequency range, measuring the small signal response of the radiation that leaves the laser at the output interface,

fitting the measured small signal response to the follow ¬ ing transmission function H(f): and determining the parameters f R , γ , A and f P ,

wherein f R describes a relaxation resonance frequency, γ the damping parameter, A an offset and f P the 3 dB fre ¬ quency of the laser's parasitics, and

- calculating the design parameter H(f R ) by calculating the value of the transmission function H(f) at the relaxation resonance frequency f R based on the parameters that were determined in the previous step.

The laser is preferably fabricated by depositing the addi ¬ tional layer with a suitable layer thickness that provides a design parameter H(f R ) in the range between 2.5 and 3.2 dB .

The suitable layer thickness may have been determined before ¬ hand based on a plurality of lasers with varying thicknesses of the additional layer and therefore varying design parame ¬ ters H (f R ) .

The deposition of the additional layer may also comprise one or more deposition steps, each deposition step comprising: (a) depositing a thin layer of isolating or conducting material,

(b) applying a voltage that generates a predetermined bias current and a small sinusoidal current signal through the laser structure, (c) sweeping the frequency of the sinusoidal signal in a given frequency range,

(d) measuring the small signal response of the radiation that leaves the laser structure at the output interface,

(e) fitting the measured small signal response to the follow ¬ ing transmission function H(f): and determining the parameters f R , γ , A and f P ,

wherein f R describes the relaxation resonance frequency, Y the damping parameter, A an offset and f P the 3 dB fre ¬ quency of the laser's parasitics, and

(f) calculating the design parameter H(f R ) by calculating the value of the transmission function H(f) at the relaxation resonance frequency f R , based on the parameters deter ¬ mined in step (e) ,

(g) if the design parameter H(f R ) is in the range between 2.5 and 3.2 dB, finishing the deposition of layer material, and

(h) if the design parameter H(f R ) is outside said range between 2.5 and 3.2 dB, repeating above steps (a) - (h) .

Further, an electric contact layer may be deposited on a sec ¬ tion of a highly doped layer of the second reflector. The doping concentration of said highly doped layer preferably exceeds 5*10 18 cm "3 .

An isolating material is preferably deposited to planarize (flatten or level) the wafer surface. The additional layer may be deposited after the deposition of the contact layer.

According to a preferred embodiment, the contact layer is de- posited on a section of a highly doped layer of said layer stack. The additional layer is deposited after the deposition of the contact layer and said voltage that generates the pre ¬ determined bias current and the small sinusoidal current sig ¬ nal, is applied to the contact layer after the deposition of the additional layer.

The additional layer may be an isolating layer or a conduc ¬ tive layer. Preferably, the additional layer is a silicon ni ¬ tride layer, a silicon oxide layer, a titanium oxide layer, an indium tin oxide layer or a GaAs layer.

Brief description of the drawings

In order that the manner, in which the above-recited and other advantages of the invention are obtained, will be read- ily understood, a more particular description of the invention briefly described above will be rendered by reference to specific embodiments thereof, which are illustrated in the appended figures. Understanding that these figures depict only typical embodiments of the invention and are therefore not to be considered to be limiting of its scope, the inven ¬ tion will be described and explained with additional speci ¬ ficity and detail by the use of the accompanying drawings in which Figure 1 shows scanning electron micrographs of a GaAs

mirror surface of completely processed VCSELs af ¬ ter digital wet etching using the technique de ¬ scribed in [ 6 ] . shows how an added Si x N y or an etched mirror surface influences the bandwidth of a laser:

Fig.2a) : A simulation that shows the change of power reflectivity by changing the out-coupling mirror. Fig.2b) : In parallel to the reflectance the cavity photon lifetime changes with the same change of the out-coupling mirror as in Fig. 2a. Please note the periodicity in thickness of both, the reflectivity and the cavity photon lifetime. Fig.2c) : Small signal responses for one VCSEL un ¬ der the same driving conditions at two different cavity photon lifetimes. The lifetime decreases from the circles to the triangles. shows theoretical small-signal modulation impulse responses of a VCSEL at different forward bias currents [8]. The relaxation resonance frequency £ R is marked for all H(f R ) functions.

shows the ratio as a function of the parame ¬ ter #(R) , based on a calculation where VCSEL parasitics are neglected. The two vertical lines shows the range of (/R) , where the changes are largest . shows a cross-section of a fully processed VCSEL before depositing an additional layer. shows a cross-section through of a fully processed VCSEL after depositing an additional layer. Figure 7 shows the simulated ratio of bandwidth and re ¬ laxation resonance frequency as a function of log for one VCSEL with two di fferent mirror re- flectivities (round/square symbols: lar ¬ ger/smaller reflectivity, respectively ) for in ¬ creasing current. The dashed and full straight lines connecting the symbols present the varia ¬ tion of f 3(iB /f R for decreasing current (left right) for smaller (dashed) and a larger (full) values of the damping parameter γ, respectively.

Decreasing damping corresponds to a shift along the faintly dashed curves. The two vertical lines display the window for maximum bandwidth. Detailed description of the preferred embodiments

The preferred embodiments of the present invention will be best understood by reference to the drawings, wherein identi ¬ cal or comparable parts are designated by the same reference signs throughout.

It will be readily understood that the parameters of the em ¬ bodiments of the present invention, as generally described herein, could vary in a wide range. Thus, the following more detailed description of exemplary embodiments of the present invention, is not intended to limit the scope of the inven ¬ tion, but is merely representative of presently preferred em ¬ bodiments of the invention.

Instead of controlling the thickness of the top mirror (re- flector) layer by subsequent reduction of the layer thickness by etching (as discussed above with respect to Figure 1), ex ¬ emplary embodiments of the present invention as described hereinafter propose the deposition of an additional layer (e.g. an isolating layer). Again, the following explanations are directed to exemplary embodiments only, and are not in ¬ tended to limit the scope of the claimed invention. The mirror reflectivity (see upper (second) reflector 30 in Figures 5 and 6) can be changed by adding an additional layer (see reference numeral 80 in Figures 5 and 6) onto the sur ¬ face in a controlled manner. A variety of materials are available for this purpose, which are nonconducting or con- ducting, but transparent at the VCSEL emission wavelength.

Such a material can be deposited by sputtering, electron beam evaporation, or plasma-enhanced chemical vapor deposition (PECVD) on the VCSEL surface simply and rapidly in a produc ¬ tion environment. In addition hybrid layers of multiple mate- rials can be deposited in the same manner.

In an electron beam (e-beam) evaporator, electrons are released by a heated cathode and accelerated toward a crucible at a voltage of several thousand volts. The material in the crucible is thereby evaporated and condenses on the surface of a sample placed next to the crucible. The deposition rate can be measured in situ via an oscillating quartz sensor. The deposition rate can be controlled via the heating current of the cathode. The advantages of this process are the in situ control and the large variety of materials that can be depos ¬ ited .

Sputtering is quite similar: A noble gas plasma evaporates material from a target, which condenses on the surface of a sample placed next to the target.

Plasma-enhanced chemical vapor deposition (PECVD) is a coat ¬ ing process in which the molecules of the processing gases are broken. Some of the resulting ions and radicals form a solid amorphous layer on the target. The cracking of the molecules is not caused by external heat, but by the acceler ¬ ated electrons of the plasma.

The advantage of all methods is that materials can be depos ¬ ited at relatively low temperatures on the surface of the VCSEL, thus avoiding damage to the VCSELs .

With e-beam, sputtering and PECVD, it is possible to produce high reflectance DBRs by depositing one or two materials of different refractive indices as an extension of the semicon- ductor DBR. With all three thin-film deposition methods, changes of the photon lifetime in VCSELs with top DBRs can be achieved .

The addition of a top layer allows to make a change in the cavity photon lifetime. A series of very thin identical iso ¬ lating coatings has been tested. For practical reasons, the use of the above methods is favored, as these methods allow routine, rapid, and accurate deposition of thin layers . In general many different sorts of layers and deposition methods could be used.

For the investigation of the stepwise change of the cavity photon lifetime, layers are preferably deposited on VCSELs, which have already been completely processed.

BCB preferably used for VCSEL planarization cannot be heated to more than 200 °C because of damage. The PECVD process pro ¬ vides the possibility of coating with both silicon nitride and silicon dioxide at low (well below 200°C) temperatures. A coating recipe for Si x N y deposition at 100°C may be imple ¬ mented. For accurate layer thickness control, it is advanta ¬ geous to carry out a test coating on a test wafer before the actual coating. From the test coating the resulting deposited thickness and index of refraction is determined by means of ellipsometry . It is most easy to coat a second dummy wafer in parallel with the actual sample. From this dummy wafer the layer thickness can be determined.

Since the PECVD coating is of the entire sample surface, the resulting isolating layer covers the top DBR but also the metal contact pads. The insulating layer on the contact pads should therefore be opened in window regions in order to per- form on-wafer probe testing or later in order to place wire- bonds for VCSEL packaging.

Separation of contacts and mirrors:

In a manner, similar to the patterning of the VCSEL surface with metal contacts and pads, a thin-film lift-off process can be used for parts of the topmost dielectric layer. For some coating materials, this lift-off method is independent of the deposited material and of the deposition technique em ¬ ployed. The solvents used to dissolve the photoresist may also attack the BCB used for VCSEL planarization. Therefore, the thickness of the deposited material is preferably sig ¬ nificantly less than the thickness of the negative photore ¬ sist used for the lift-off process. In order to protect the BCB in the case of the deposition of several coatings, the entire sample surface can be coated with an additional layer. In the following, it is assumed that the additional layer is an insulating layer, however, other materials can also be used as already pointed out above. After the coating, only the additional material that is above the contact pads is removed by etching, employing a material- dependent etching recipe. In order to determine the optimum cavity photon lifetime for a given VCSEL the impact of the cavity photon lifetime on the small-signal bandwidth is determined.

Optimization of bandwidth and energy efficiency

In order to achieve the highest possible bandwidth f 3dB for a vertical-cavity surface-emitting laser, by changing the reflectivity and reflectivity phase of the top mirror via the photon life r p , a compromise between r p , attenuation γ, and relaxation resonance frequency f R must be found [4] . A reduc- tion of Tp leads to a reduction of γ and generally to an in ¬ crease of ¾ d B by increasing the ratio f: i(i - B /f v ,. At the same time, however, there is a reduction of f R , that can have a negative influence on the bandwidth. Furthermore the thresh ¬ old current and the optical power increase for decreasing re- flectivity/photon lifetime/damping. A larger threshold current leads to an increase of energy consumption per bit. This increase can overcompensated by the increased bandwidth, which eventually might decrease the consumed energy per bit. A larger output power enables using smaller aperture diame- ters, leading to an additional decrease of consumed energy. The optimization presented here therefore enables increasing the bandwidth in concert with reduced energy consumption per bit and increased optical output power. Figure 2 shows that the change of thickness of an additional Si x N y layer influences the power reflectance of the out- coupling mirror (Fig. 2(a)) and the cavity photon lifetime (Fig. 2(b)) . A reduction of the cavity photon lifetime de- creases the damping. This reduction results in a larger

"overshoot" (an increase of the maximum value of the modula ¬ tion response), but not necessarily a larger bandwidth.

The small-signal modulation bandwidth of a laser is the fre- quency at which the magnitude of the output power, when driven by an electrical (sinusoidal) signal, is reduced to half the value at low frequencies. The change in the output power of the transmitted signal as a function of the fre ¬ quency f is described by the following transmission function [7] : where f R and γ are the relaxation resonance frequency and at- tenuation. The parameter A is an offset and f P is the 3 dB frequency of the VCSEL parasitics. The parameters in (1) are determined from the transmission function measured e.g. with a vector network analyzer. Figure 3 shows simulations of VCSEL impulse responses for in ¬ creasing forward bias currents. The resonance frequencies f R are marked and can be easily estimated from the plotted pulse responses. The attenuation, on the other hand, can be ob ¬ tained by means of fitting the measured data to the transmis- sion function (1) . The relaxation resonance frequency is an independent variable, whereas the attenuation is a function of f R .

H(f R ) is the transfer function value at the resonance fre- quency f R . The -3 dB bandwidth f 3dB is the frequency at which the magnitude of H(f) drops by a factor of 2 from its initial low-frequency value. The small signal modulation response can be measured by using a vector network analyzer (VNA) . A sinusoidal signal with very small amplitude (-25 dBm) is generated by the VNA, modu- lates the laser output and is transmitted from the laser across a fiber, demodulated by a photodiode and sent back to the VNA. The VNA describes this linear network by using the power based scattering parameters (Sll power reflection VNA port 1, S22 power reflection port 2, S12 power transmission from port 2 to port 1, S21 power transmission from port 1 to port 2) . The small signal modulation response is the trans ¬ mission through this network expressed as the scattering parameter S21, measured by the VNA. The small signal modulation response of the system has to be corrected for the response curve of the photodetector to get the small signal modulation response of only the laser (Figure 3) . H(f R ) can be easily determined from the plotted curve.

Both A and the low-pass filter term of the parasite fp in

Equ. (1) are set to 1 in the following analysis, since only the intrinsic transfer function needs to be considered here, leading to the following equati . The transfer function H(f) and thus the small-signal modu ¬ lation bandwidth frequency is depending on f R and γ. The damping γ is given by:

(3) which is thus depending on f R . The term K is given

and

' th (6)

The damping y and the relaxation resonance frequency f R vary strongly with the differential gain the group velocity v s , and the charge carrier transport factor X . f R is ad- ditionally dependent on the optical confinement factor Γ and the active volume vs . The damping γ depends on the gain com ¬ pression factor £ and, of course, on the cavity photon life ¬ time r p . These are all basic laser properties, strongly re ¬ lated to the design of the active region. It is not possible to simply predict the optimum VCSEL transfer function H(f), as it varies with the particular VCSEL optical transmission system. The damping parameter γ is also unsuitable to serve as a VCSEL parameter to optimize. Given the relaxation resonance frequency f R , the bandwidth can be written as: consider the relationship in (7) more closely, the trans function can be limited to its intrinsic value:

A value # ( R) can thus be assigned to the resonance frequency in an unambiguous manner. This value of #(/R) is found from the plot in Figure 3. The damping at fm is:

2mf R

Y =

(9)

The transfer function thus becomes:

The bandwidth f^ dB is defined as:

Then the ratio of the bandwidth to the relaxation resonant frequency is :

(12)

Using it is possible to describe the bandwidth (7) solely by the "form" of the transmission function, whereby the resonant frequency and the value of the function can be estimated at the location of the relaxation resonance fre ¬ quency. The change of the output mirror reflectivity by add- ing a topmost dielectric layer to the VCSEL will directly im- pact the intrinsic bandwidth term /R .

The term fs can now be changed in a limited range by chang- ing the output coupling mirror reflectivity. This change in

reflectivit converges to a maximum value of A as goes to infinity, given by:

3dB

lim = J 1 + 10 '20 ftf l.SSBiJI 1 +T2

(13)

Figure 4 shows a plot of equation (12) A change in the mirror reflectivity and thus a change in causes a large change of R in the H(f R ) interval between -12 dB to 3 dB indicated by the vertical dashed lines. For real lasers, values of

#(R) > _ 2 are typically observed. In the range ^(fs.) < 3 the

value of f& increases. For W(/R) > 3 it a bandwidth increase

(by increasing k ) is essentially compensated by a bandwidth reduction (by reducing /R ) . The threshold current can be arbitrarily increased and the corresponding relaxation reso- nance frequency according to (5) will become arbitrarily small .

The general optimization parameter is This parameter is nominally independent of the laser type, shape, and size. The optimal operating point is just beyond the 3 dB cut-off fre ¬ quency, which can be found by varying the mirror reflectivity. Note: Two VCSELs of the same epitaxial and geometrical struc ¬ ture but having different oxide aperture diameters, will re ¬ quire different coupling mirror reflectivity for optimum op- eration, but # ( R) should be about -3 dB for each. If a laser is to be operated at different operating points, an optimum mirror reflectivity exists for each operating point. However, (/R) is close to -3 dB (between -2.5 dB and -3.2 dB) at all operating points.

Technology for optimizing VCSELs

Antypical embodiment of a method for fabricating a VCSEL will be explained below in further detail. The starting point may be an arbitrary, fully processed VCSEL as shown in Figure 5. Figure 5 shows a cross-section through the top-mesa of a fully processed vertical cavity surface emitting laser,

VCSEL, 10. The VCSEL 10 comprises a first (lower or buried) reflector 20 and a second (upper) reflector 30. The second reflector 20 comprises a layer stack 31 of semiconductor lay- ers 32. An active region 40, that is capable of generating radiation in response to an electrical current, is arranged between the first and second reflectors. The first and second reflectors 20 and 30 are preferably Distributed Bragg Reflec ¬ tors (DBRs) . Planarization material 50 (e. g. Benzocyclobu- tene (BCB) ) is deposited to level or flatten the laser's sur ¬ face. The VCSEL 10 further comprises upper metal contacts 60 and a lower metal contact 70. The lower metal contact 70 may be separated from the first reflector 20 by a substrate which is not depicted in Figure 5.

Figure 6 shows the VCSEL 10 after covering the second reflec ¬ tor 30, parts of the upper contacts 60 and the planarization material 50 with an additional layer 80. The additional layer 80 is deposited on top of the layer stack 31 at the light output side of the VCSEL 10. The additional layer 80 forms an output interface 100 where the radiation R leaves the VCSEL 10.

The refractive index of the additional layer 80 is smaller, equal to or larger than the smallest refractive index of the refractive indices of the layers 32 in the layer stack 31. The transmission of the additional layer 80 at the emission wavelength of the laser is preferably larger than 99%.

In order to characterize the VCSEL 10 and in order to determine whether the thickness of the additional layer 70 lies in a preferred range with respect to the laser's performance, the following steps can be carried out:

An operating point (the forward direct-current bias point at which the laser is to be optimized) is set by applying a cor ¬ responding electrical DC voltage between the upper metal con- tact 60 and the lower metal contact 70. Further, a small si ¬ nusoidal AC voltage is applied to the contacts.

The frequency of the sinusoidal signal (AC voltage) is swept in a given frequency range, and the small signal response of the radiation R that leaves the laser 10 at the output inter ¬ face 100 is measured. The measured small signal response is fitted to the following transmission function H(f): and the parameters f R , γ , A and f P , are determined, wherein f R describes the relaxation resonance frequency, γ the damp- ing parameter, A an offset and f P the 3 dB frequency of the laser's parasitics.

Then, the design parameter H(f R ) is calculated by calculating the value of the transmission function H(f) at the relaxation resonance frequency f R based on the parameters that were de ¬ termined as explained above.

If the design parameter H(f R ) is in the range between -2.5 and -3.2 dB, the thickness of the additional layer 70 is deemed optimal.

If the design parameter H(f R ) is outside the range between - 2.5 and -3.2 dB, the thickness of the additional layer 70 is deemed suboptimal and will be increased by depositing more layer material. Then, the above steps can be repeated again and again until the design parameter H(f R ) is inside the range between -2.5 and -3.2 dB . Since the mirror reflectivity is periodic in optical thick- nesses of it is possible to find a suitable coating thickness to decrease or increase the reflectivity. If the reflectivity has to be increased beyond this interval, entire DBR mirror pairs can be deposited. A required coating thick- ness is produced by repeated deposition followed by subse ¬ quent small-signal measurements, such that the top mirror has the desired optical loss. Than H(R) corresponds to about -3 dB and an optimal " R) shape is obtained. This process of adjusting the cavity photon lifetime is non ¬ destructive and does not rely upon a complex etching process as described above with reference to Figure 1. The coating process is reliable, fast, and can be performed on a volume production basis.

Experiments - Examples:

Figure 7 shows the simulated ratio of bandwidth and relaxa ¬ tion resonance frequency as a function of log for one VCSEL with two different mirror reflectivities (round/square symbols: larger/smaller reflectivity, respectively) for increasing current. The dashed and full straight lines connect- ing the symbols present the variation of f^^/f . for decreas ¬ ing current (left to right) for smaller (dashed) and a larger (full) values of the damping parameter γ, respectively. De ¬ creasing damping corresponds to a shift along the faintly dashed curves. The two vertical lines display the window for maximum bandwidth.

In summary, optimizing the seemingly contradictory properties of maximum bandwidth and energy efficiency can be achieved by simultaneously optimizing the interplay of all laser diode parameters, without any exception. The method presented above optimizes for the first time the energy-efficiency of verti ¬ cal-cavity surface-emitting lasers (VCSELs) , while also achieving large bandwidths . Embodiments of the method allows:

- A small-signal modulation bandwidth increase of more than 10%.

- Energy reduction for data transmission at a given bit rate of up to 50%.

- An increase in the laser diode output power by more than 100%.

The embodiments are based on the discovery of a laser di ¬ ode optimization parameter, that leads to the definition of a general performance optimization procedure. This pro- cedure eliminates the need to perform time-consuming laser testing and analysis of similar batches of the laser di ¬ odes .

- The optimization process is destruction-free, requiring no material removal, such that the lasers are ready to be used in their as-processed, original state.

- The optimization technology is fast, requires minimal han ¬ dling, and is low cost.

- The optimization by the deposition of a thin surface layer proceeds with nanometer-scale precision, and can be easily controlled by ellipsometry .

References

[1] P. Wolf, P. Moser, G. Larisch, and D. H. Bimberg, "High- Speed and Temperature-Stable oxide-confined 980-nm

VCSELs for Optical Interconnects, " IEEE Journal of Selected Topics in Quantum Electronics, vol. 19, pp. 1- 7, 2013.

[2] A. Mutig, J. A. Lott, S. A. Blokhin, P. Moser, P. Wolf, W. Hofmann, et al . , "Highly temperature-stable modulation characteristics of multioxide-aperture high ¬ speed 980 nm vertical cavity surface emitting lasers, " Applied Physics Letters, vol. 97, pp. 151101-1-3, 2010.

[3] D. Ellafi, V. Iakovlev, A. Sirbu, G. Suruceanu, Z.

Mickovic, A. Caliman, et al . , "Control of cavity lifetime of 1.5-Dm wafer-fused VCSELs by digital mirror trimming," Optics Express, vol. 22, p. 32180, 2014.

[4] P. Westbergh, J. S. Gustavsson, B. Kogel, A. Haglund, and A. Larsson, "Impact of Photon Lifetime on High-Speed VCSEL Performance, " IEEE Journal of Selected Topics in

Quantum Electronics, vol. 17, pp. 1603-1613, Nov-Dec 2011. [5] M. Kohler and A. Wiegand, Etching in microsystem technology. Weinheim [etc.]: Wiley-VCH, 1999.

[6] C. A. B. G. C. DeSalvo, J. L. Ebel, D. C. Look, J. P.

Barrette, C. L. A. Cerny, R. W. Dettmer, J. K.

Gillespie, C. K. Havasy, T. J. Jenkins, K. Nakano, C. I.

Pettiford, T. K. Quach, J. S. Sewell and G. D. Via, "Wet

Chemical Digital Etching of GaAs at Room Temperature, "

J. Electrochem. Soc, vol. 143, 1996.

[7] A. Larsson, "Advances in VCSELs for Communication and

Sensing, " IEEE Journal of Selected Topics in Quantum

Electronics, vol. 17, pp. 1552-1567, 2011.

[8] L. A. Coldren and S. W. Corzine, Diode lasers and photonic integrated circuits. New York, NY, USA: Wiley,

1995.

[9] G. Larisch, P. Moser, J. A. Lott, and D. Bimberg, "Impact of Photon Lifetime on the Temperature Stability of 50 Gb/s 980 nm VCSELs," IEEE Photonics Technology Letters, vol. 28, pp. 2327-2330, 2016.

[10] G. Larisch, P. Moser, J. A. Lott, and D. Bimberg, "Correlation of photon lifetime and maximum bit rate for 55 Gbit/s energy-efficient 980 nm VCSELs," in 2016 IEEE Optical Interconnects Conference (01), 2016, pp. 16-17.