Download b - Center for High Technology Materials

Optimizing the Substrate Removal Technique for thermal management of 2µm VECSEL
grown on GaAs.
REU Student: Gregory Ballou
Graduate Mentor: Emma Renteria
Faculty Mentor: Dr. Ganesh Balakrishnan
1.A. Introduction:
Since 1960, the LASER which is an acronym for Light Amplification by Stimulated
Emissions of Radiation has been used for several applications. It has been used in military
applications as a rangefinder and for painting targets for smart bombs. Civilian uses include
anything from laser pointers to leveling devices to alignment of optical systems. These
applications are possible because of a laser’s properties of having highly monochromatic light
that is light of one wavelength and having light that is highly collimated meaning that the
photons are very close together while they propagate through the air or the medium they happen
to be in. It was Albert Einsteinin 1917 who first theorized about the process that makes lasers
possible called “Stimulated Emission”[1,2]. Before the laser was the maser which is acronym
for Microwave Amplification by Stimulated Emission of Radiation. This works on the same
principles as the laser but emits no light photons; it was mainly used to amplify radio signals.
The men who invented the maser, Charles Townes and Arthur Schawlow, were at Columbia
University [1,2]. In 1960, Theodore Maiman invented the first optical laser. It was called the
ruby laser because it used a ruby rod doped with chromium to emit light in a beam. However it
was pulsed and had few applications[1,2]. Ali Javan, in 1960, invented the first gas laser which
was the Helium-Neon which outputs visible light at 632.8 nm wavelength (red). These HeliumNeon lasers are widely used today. There are even Helium-Neon lasers that emit a green light at
around 532 nm wavelength. In 1962, a new kind of laser was invented by Robert Hall. This new
laser was a Semiconductor laser made of Gallium-Arsenide at General Electric Labs [1,2]. This
was followed closely by the first continuous wave solid state laser.
There are basically three types of lasers: gas lasers, solid state lasers, and semiconductor
lasers. Gas lasers have different uses that are determined by wavelength and power out. For
example, Helium-Neon lasers are usually low power but operate in the visible spectrum and
make good alignment lasers, CO2 lasers are higher power and operate at a wavelength of 10.6µ
in the IR range and they are good for cutting and welding metals, Argon lasers are good for
pumping other lasers dye and chemical lasers it can be used to optically excite other lasers; it
also makes good hologram slides. Solid –state lasers, such as ruby rod, and Nd: Yag operate in
the near IR wavelength. These lasers use a crystal doped with an ion which provides the
wavelength of the light that the laser operates at. The military has used ruby rod lasers in
rangefinder applications, they have also used Nd:Yag lasers as rangefinders, but they are most
commonly used in the medical field in surgery.The next types of lasers are the Semiconductor
lasers. There are several kinds of semiconductor lasers and they operate differently than solid
state lasers and gas lasers. Semiconductor lasers have an engineered band gap instead of
atraditional active medium.The photons that are released are spontaneous and most will combine
to be in phase, highly coherent, monochromatic light. Those photons that are not emitted are
absorbed by the diode as heat energy making the diode very hot which can lead to a breakdown
in the diodes structure. The internal heating can also interfere with the diodes ability to lase and
maintain a population inversion. Edge emitting diode lasers which are the first semiconductor
lasers that were used in mass quantities, also the first semiconductor lasers invented,are pumped
by electrical current which itself is a source of heat. Another type of semiconductor laser is the
Vertical External Cavity Surface Emitting Laser (VECSEL).
The lasers mentioned here all have one thing in common, that is that they are all very
inefficient when it comes to power use (which is power out over power in). The best that most
lasers can hope for is about10% - 30% especially in gas and solid state lasers. Semiconductor
lasers are about 50% efficient with some high power semiconductor lasers having 60% - 70%
efficiency. The reason that lasers are so inefficient is that the energy that is not emitted as light is
converted into heat. Therefore, elimination of this heat is crucial to improve laser efficiency. It
is especially true for semiconductor lasers where excessive heat can lead to structural break
down. For most lasers (gas and solid state) the problem of heat is solved by simply designing a
heat exchange system around the active medium using circulating water from the city or from
water cooling tower; some lasers come with or one can have ordered for it a system that has a
refrigerated water cooling system. The cooling system usually includes cooling the power
supply also since so much power is required to excite the lasers; this is normal practice because
most power supplies are made with semi-conductors and they create large amounts of heat while
in operation. The cost of, all of these solutions is very expensive and are all external solutions to
the problem of controlling heat in lasers. So it makes sense to engineer the laser so that the
creation of heat is at a minimum. The applications for VECSELs are many because of their
versatility in wavelength and power; plus they are compact in size and that makes them perfect
for science and commercial applications. Their photonic applications range from research to
military the following are just a few. VECSELs can be used in the fields RGB laser illuminators
for next generation TV’s, high accuracy gyroscopes for satellite navigation, radar, lidar, free
space communications, medical diagnostic and surgery, sensing and security, fiber optics
communications.
2.B Background:
1. Definition of a VECSEL.
A VECSEL is a type of semiconductor laser. Its name is an acronym for Vertical External
Cavity Surface Emitting Laser. Figure 1[3] shows its major components . The bottom mirror is a
highly reflective mirror usually a distributed Bragg reflector (DBR). A DBR is composed of
multiple layers of materials with different refractive index which allows up to 99.9% reflectivity.
The active region that can be composed of quantum well, quantum dashes, or quantum dots is
band gap engineered to absorb the pumped photons and emit photons at a desired wavelength
[4]. The external cavity is formed between the active region and the external mirror. The
external mirror, also call an output coupler, is partially reflective spherical mirror that defines the
laser transverse mode [4] and allows the extraction of a portion of the laser beam from the cavity.
Figure 1. Configuration of an optically pumped VECSEL.
Figure 2 shows the basic functions of the VECSEL components. High energy incident
pump photons are absorbed into pump absorbing region [4]. Carriers diffuse to the quantum
wells where electrons are excited and relaxed emitting photons with energy equal to the quantum
well band-gap energy [4]. The generated photons travel back and forth in the cavity. These
photons stimulate more electrons in high energy levels to drop to lower energy levels and
generate more photons with same wavelength and phase emitting coherent light as shown in
Figure 3 [5]. As it can be seen in Figure 3 in order to produce a considerable amount of photons
by stimulated emission, population inversion must be maintained hence the use of an external
pump laser [4-6]. Eventually, enough photons will be generated and they will escape the cavity
through the output coupler.
Figure 2. Operating principles of optically pumped VECSELs [4].
Figure 3. Stimulated emission processes [5].
2. Thermal Management
In a VECSELs, it seems that most of the heat comes from the pumping mechanism which is
the external laser that is exciting the active medium. Less than 100% of the energy pump into
the laser is converted to effective radiative power and hence the rest is converted to heat. This is
due to the difference in energy between the incident pump photons and emitter laser photons.
The pump photons have higher energy than the emitted photons and thus the energy difference is
dissipated as heat in the active medium. As the temperature increases in the active medium the
excited carriers star escaping the quantum wells until the laser gain is depleted and the power
output decreases. This effect is called thermal rollover and can be observed in Figure 4 [6].
Hence, thermal management is crucial for VECSEL power performance and it is usually
addressed by using heat spreader plates and heat sinks as will be explained next.
Figure 6. Output power from a 1 µm VECSEL with an unprocessed InGaAs/GaAs gain as
a function of incident pump power at two different temperature settings [6].
Several VECSELs designs had been developed to remove the undesired heat from the active
region. The ones most mention in literature [4, 6-8] are shown in Figure 7. The first one, Figure
7a, is the most simple design. The VECSEL is directly bonded to a heat sink exactly as it is
grown. This design is not too efficient due to the substrate thermal impedance. In addition to the
poor thermal conductivity of the substrates, (33 W m-1 K-1 for GaSb [7] and 45 W m-1 K-1 for
GaAs [8]), they are too thick, up to 500µm. The second design, Figure 7b, the substrate is
removed hence improving heat removal. However, this design can only be used when etch
chemistry permits a good selectivity between the substrate and an etch stop layer as it will be
described in the subsequent sections. For this design, The VECSEL structure is grown in
reverse order. The active medium is grown close to the substrate and the DBR is grown last. In
this way the heat is extracted through the DBR and even though DBR's thermal conductivity
could still be low, ( 11 W m-1 K-1 for Al0.5Ga0.5As DBR [6]), thermal diffusion length is much
less than with the presence of the substrate. In the third design, Figure 7c, heat is extracted
directly from the active region. The active region is bonded to heat spreader plate. Thus, the
substrate can be either removed or kept depending of its etch chemistry. However, the heat
spreader must be transparent such as transparent diamond, sapphire or silicon carbide.
Active-region
DBR
Figure 7. VECSEL structure designs for thermal management.
3. Substrate Removal techniques
There exist two types of etching mechanisms: physical etching which relies on physical
interaction of particles that erode the surface and wet/dry etching that relies on chemical
reactions to erode the surface [9]. Typically a combination of both methods is used to remove
substrates with thickness between 300µm and 500µm. First, the substrate is mechanically thin
down to about 100µmand then wet etching or dry etching is applied [7-8, 10-11]. However, dry
etching is highly associated with surface damaging since it uses bombardment of ions from
reactive gases to disassociate the substrate [9]. In view of the fact that a VECSEL surface must
be smooth, dry etching is undesirable leaving us with wet etching as the reminding option.
Wet etching is the process by which layers of material are removed by using liquid chemical
solutions. A wet etching process usually involves a three step reaction mechanism as shown on
Figure 8 [12]. First step, the reactants diffuse to the semiconductor surface. Second step, the
chemical reaction occurs at the surface by oxidation reduction. Third step, reaction products
diffuse from surface to dissolve into the solution. Therefore, the reaction chemistry between the
semiconductor material and the etchant solution must be compatible.
Figure 8. Reaction mechanism for a wet etching process.
Wet etching systems for GaAs had been studied extensively [11, 13-15]. A typically GaAs
system is shown in Figure 9. Usually, the etchant solution is composed of an oxidizing agent
such as hydrogen peroxide, H2O2, and a solvent agent such as ammonium hydroxide, NH4OH, or
citric acid, C6H8O7 [13]. These etchant solutions have a higher selectivity for GaAs material
than for AlGaAs which means GaAs is etched at a much faster rate than AlGaAs [13-15].
Therefore, AlGaAs acts as an etch stop layer. Once, the etchant solution reaches AlGaAs the
etch rate becomes so slow that the layer is conserved and it is an indication that the substrate has
been removed. In addition, AlGaAs layer can be etched in seconds with Hydrofluoric acid
which has aextreme selectivity for AlGaAs over GaAs [14]. Nonetheless, in this experiment the
system is GaSb/GaAs as shown is Figure 9b and the selectivity of etchant solutions for GaAs
over GaSb must be determined.
a
GaAs
Al0.95Ga0.05As
b
GaAs Substrate
GaSb (3 μm)
GaAs Substrate
Figure 9. Structure of etching systems for a) GaAs and b) GaSb/GaAs.
3.C. Experimental Procedure:
Previous experiments have shown that an etch stop layer is unnecessary for the substrate
removal of GaAs in GaSb/GaAs systems. Now, the goal of this study is to optimize a substrate
removal technique to obtain the smoothest surface possible. The GaSb layers used in this
experiment were grown by means of molecular beam epitaxy (MBE). A3μm thin GaSb film was
grown at a rate of 0.50μm/hr at a temperature of 510°C with a V to III ratio of Sb to Ga of 6.6
onGaAs substrate.
1. Preparing the sample
The sample were cleaved in squares of 1cm x 1cm for easy handling. Cleaving is performed
using a diamond scriber. The wafer is placed epitaxial side up on a clean room wipe sheet. With
a ruler on the straight side of the sample 1cm is measured and marked by scribing a small line.
Then the wafer is placed over a straighten out paper clip so that the scribe line aligns with the
paper clip and by applying equal pressure to both sides of the wafer with the ends of two
tweezers the sample fractures in a straight line. The same procedure is repeated until the
maximum possible 1cm2 chips are obtain from one wafer.
The next step is to chemically clean the sample to remove any impurities that may reside on
the wafer surface. Each sample was soaked in acetone, isopropanol, and methanol for 2 minutes
in each solution. Then the sample was rinsed with deionized (DI) water and dried with a
Nitrogen gun. Nomarski images of the clean surface were taken before the bonding procedure.
Microscope glass slides were cut into three pieces. A pinch of ApiezonW ® wax was placed
on top of the glass slide and then the slide was placed on a hot plate at a temperature of 150°C.
Once the wax was melted, the sample was placed on the wax with the GaSb layer facing down.
Slight pressure was applied on the sample to assure even contact, free of air bubbles, with the
wax. Then the sample was removed and allowed to cool.
2. Measuring etch rate and selectivity
Since there is no etch stop layer, it is essential to know with precision the rate at which the
etchant solution will etch GaAs as well as GaSb. To measure the etch rate on GaAs, the sample
bonded to glass was measured with a micrometer before and after etch at five spots as shown in
the diagram in Figure 10. Etch rate is equal to the thickness removed per unit time. The etch
time was set for forty minutes. The etch rate for each of the five spot in the sample was
calculated and averaged with the other to have a better estimate of the whole 1cm2 surface. To
measure the etch rate on GaSb, the sample bonded to glass with the GaSb layer facing up was
etched monitoring the time it took to etch the 3µm GaSb layer. The sample was measured and
you can see the measurements in microns. The top number is the before measurement the bottom
number is the after measurement. The bottom number is subtracted from the top number which
tells us how many microns were etched in that spot. After determine how many microns were
etched we take that number and calculate the etch rate; which is thickness / time. We do this for
each spot measured, add the results and divide by five, which gives us and overall average of
how much substrate was removed.
Figure 10. Diagram of measuring positions.
3. Removing the substrate.
The etchant solution of NH4OH: H2O2 at a volume ratio of 1: 33 was used. It was prepared
by mixing 500ml of H2O2 and 15ml of NH4OH. The sample was placed and secured on the jet
etcher sample holder as shown in Figure 11. The jet etcher is a homemade etcher that uses a
pump to circulate the etchant solution and to keep a constant flow falling on the sample. Once
the sample and the solution were ready, the pump was turned on and the valve open. The etch
process was monitored through all the etch time to assure a steady stream and therefore an even
etch. When the etch time was expired, the etch was stop by quenching the sample in DI water.
The surface was dried with nitrogen gun and its surface was analyzed under Nomarski
microscope.
Figure 11. Jet etcher setting.
4. C Results:
The substrate removal of GaAs was successfully achieved as it can be verified in the
XRD Etch analysis. Figure 12 shows the results from omega-two theta X-Ray diffraction study
of the sample before (red curve) and after (black curve) the etch process. It can be clraerly notice
that the substrate peak in the black curve, post etch, is completely gone leaving behind only the
GaSb layer peak. This is an indication that the NH4OH: H2O2 1:33 ratio has a higher selectivity
for GaAs than for GaSb.
Figure 12. Omega-two theta X-Ray diffraction study of the sample before (red curve) and
after (black curve) the etch process.
Nomarski microscope was used to analyze the surfaces of GaSb epi-layer before and after
the etch. Figure 13 shows the Nomarski images results which shows almost no difference in
surface quality. Therefore, further analysis need to be done to determine the roughness of the
surface.
a
b
Figure 13. Nomarski images of GaSb surfaces a) before etch, b) after etch.
References
1. Schulz, N., et al. "Optically pumped GaSb-based VECSEL emitting 0.6 W at 2.3 μm."
Photonics Technology Letters, IEEE 18.9 (2006): 1070-1072.
[1] Bellis, Mary. "History Of Lasers." About.com Inventors.N.p., n.d. Web. 23 Mar. 2013.
[2] "History of Lasers."History of Lasers.N.p., n.d. Web. 23 Mar. 2013.
[3] "Nano-Photonics Materials and Technologies for Multicolor High-Power Sources
(NATAL)." AG Bimberg: NATAL. N.p., 0.2 Sept. 2010. Web. 26 Mar. 2013.
[4] O.G. Okhotnikov, Semiconductor Disk Lasers: Physics and Technology, Weinheim,
Germany: Wiley-VCH Verlag GmbH & Co. KGaA, Apr. 2010.
[5] "Stimulated Emission." HyperPhysics. N.p., n.d. Web. 14 Apr. 2013.
[6] Tropper, A. C., et al. "Vertical-external-cavity semiconductor lasers." Journal of Physics D:
Applied Physics 37.9 (2004): R75.
[7] Perez, J. P., et al. "Technologies for thermal management of mid-IR Sb-based surface
emitting lasers." Semiconductor Science and Technology 25.4 (2010): 045021.
[8] Jacquemet, Mathieu, et al. "Single-frequency cw vertical external cavity surface emitting
semiconductor laser at 1003 nm and 501 nm by intracavity frequency doubling." Applied Physics
B 86.3 (2007): 503-510.
[9] Arokia, Nathan. "Etching and Chemo-Mechanical Polishing."N.p., 2005. Web. 14 Apr.
2013.
[10] Clayton, F., R. Westerman, and D. Johnson. "Characterization of a Manufacturable High
Rate GaAs Via Etch Process." GaAsMantech, Conference. 2002.
[11] Yeh, Hsi‐Jen J., and John S. Smith. "Integration of GaAs vertical‐cavity surface emitting
laser on Si by substrate removal." Applied physics letters 64.12 (1994): 1466-1468.
[12] Barron, Andrew R. "Formation of Silicon and Gallium Arsenide Wafers." Connexions.N.p.,
21 Sept. 2012. Web. 14 Apr. 2013.
[13] Novak, J., et al. "Wet chemical separation of low-temperature GaAs layers from their GaAs
substrates." Materials Science and Engineering: B 40.1 (1996): 58-62.
[14] Yablonovitch, Eli, et al. "Van der Waals bonding of GaAs epitaxial liftoff films onto
arbitrary substrates." Applied physics letters 56.24 (1990): 2419-2421.
[15] DeSalvo, Gregory C., Ron Kaspi, and Christopher A. Bozada. "Citric Acid Etching of
GaAs1− x Sb x, Al0. 5Ga0. 5Sb, and InAs for Heterostructure Device Fabrication." Journal of
The Electrochemical Society 141.12 (1994): 3526-3531.