Single pixel isolation of Type II inAs/GaSb strained layer superlattice infrared detectors by chemical etching

Single pixel isolation of Type II InAs/GaSb strained layer superlattice infrared detectors by chemical etching

REU Student: Travis Nelson

Primary Mentor: Elena Plis

Faculty Mentor: Sanjay Krishna

Introduction

The drive to develop high performance infrared (IR) detectors has been generated primarily by its applications in the military as well as the medical industry including human body detection for surveillance and non-evasive medical imaging and diagnostics. Type II InAs/GaSb strained layer superlattics (SLS), first proposed as an IR detecting material in the 1980’s [14], [15], is now thought as the most promising avenue to further advance IR sensing technology. Although there are other materials, such as mercury cadmium telluride (MCT) or bulk indium antimonide (InSb), which can be used for this same purpose, the InAs/GaSb SLS is theoretically more versatile and efficient than its competitors. Improved properties include; suppressed auger recombination rates due to special separations in the SLS, stronger absorption of normal incident light without the need for gratings used in MCT leading to higher quantum efficiency, a larger electron effective mass than MCT reducing tunneling currents and a more uniform structure [16]. While InAs/GaSb has many expected benefits, it currently has not reached its fullest potential. Part of the reason for the underperformance of the InAs/GaSb SLS detector arises from impurities created when etching the detector material. Complications arise due to the complex chemical composition of the superlattice. A fundamental understanding of SLS structure as well as the mechanisms behind etching is necessary in order to create experiments and improve SLS IR detection.

Background


Fig. 1. Subtraction image obtained with dual-color InAs=GaSb SL camera. [1]
All objects give off IR radiation, and much of this radiation is in the 3-30µm range that Type II InAs/GaSb SLS’s can detect. Visible light is the most familiar region in the electromagnetic spectrum because it is the wavelength that our eyes can detect. IR radiation, felt as heat can be useful to identify because everything with a temperature above 0 K emits IR radiation. The Infrared region is broken up into different regions, the regions the SLS can detect are mid wavelength infrared (MWIR), long wavelength infrared (LWIR) and very long wave infrared (VLWIR). Higher temperatures give off wavelengths in the MWIR region while the cooler temperatures the VLWIR region. Different gasses give off wavelengths of different regions so dual band, or multiple wavelength detectors can distinguish between their signatures. The human body’s temperature gives off radiation best absorbed by LWIR detectors. Because SLS detectors can be modified to detect each of these regions and as such is extremely versatile. Not only do objects emit IR radiation at different wavelengths but also at different intensities. These characteristics can be used to detect differences in objects with similar chemical compositions especially when they have little or no distinguishable features in the visible light region. The most obvious use of IR detection is as night vision, in the LWIR spectrum, because IR wavelengths are given off by objects at all times, including in the dark. Other important applications of IR detectors are monitoring and recording global temperature profiles, LWIR, missile detection and tracking MWIR, and detection of gas leaks, dual band. For example, exhaled carbon dioxide, while indistinguishable from air, can be detected through a dual-color IR camera as seen in Figure 1. It has also long been known that heat is associated with the health of the human body, LWIR; any deviation from standard temperature can be cause for alarm. The different temperatures can be detected by IR sensors. Cancer also gives off different heat signals and can be detected. All of these applications can be used for early detection of illness and complications as well as non evasive diagnostics. The IR detectors also have a useful tool in astronomy to characterize astronomical features, VLWIR. Planets and other bodies can be characterized through their IR signal such as temperature and composition.


Fig. 2. Zincblende structure
In order to detect IR light a material must respond selectively to specific wavelengths which make up the IR region of the electromagnetic spectrum. This response is an electric current or voltage created by the incident photons. A photon transfers its energy into the valence electrons in the material, and the electrons move up in energy into the conduction band. The difference in energy between the valence and conduction bands is known as the bandgap, and the bandgap energy directly corresponds to the wavelength of light to be detected. The MWIR and LWIR light regions have a wavelength of 3-5µm 8-12µm respectively and are defined by their atmospheric absorption, or the absorption due to water vapor. The absorption for these wavelengths is very small. GaSb and InAs are both III-V semiconductors. Both of these have a crystalline lattice structure known as zincblende as depicted in Figure 2. While both of these semiconductors have the same lattice structure, the bond lengths differ between the two. When the two lattice structures are grown epitaxially, or one on top of the other, the bonds that form do not perfectly matchup. This causes strain in the bonds, a result of the lattice mismatch. The layers of each material are very thin, consisting of several monolayers of GaSb and InAs and switching between the two materials until a desired number of alterations are formed as seen in Figure 3. The name of the material strained layer superlattice is derived from the properties of the total compound. These properties are the strain between the alternating layers and the complex multi layered crystal lattice structure. The thickness of individual monolayers determines the bandgap of the material which determines the wavelength that will excite a valence electron to the conduction band. The SLS structures are typically grown using a process known as molecular beam epitaxy (MBE). This is capable of growing monolayers of each structure accurately with abrupt interfaces.

Fig. 3. Schematic of SLSL IR absorber formed by alternating layers of InAs and GaSb

This leads to the next aspect of the characteristic of the SLS material, which has to deal with the relationship of the bandgap of InAs to GaSb. This bandgap orientation is known as a Type II bandgap. The orientation of our InAs/GaSb superlattice is characterized by two materials that have no overlap of their bandgaps. In our case the valence band of GaSb is higher than the conduction band in InAs which would logically mean that both the conduction and valence band in GaSb is higher than the conduction and valence band in InAs. There are three different known band gap alignments, ranging from type I to type III [9]. The types are characterized by the position of the valance and conduction bands in relation to one another. To review, the valence band is the highest energy band in which electrons resides and the conduction band is the band with higher energy that is the next level up from the valence band. All of the band structures are illustrated in Fig. 4. In a type I superlattice, one of the materials, material A, has a bandgap that resides completely inside the other material, material B. So material A has a valence band with a higher energy than material B’s valence band, and a conduction band with lower energy than material B’s conduction band. Type II’s structure was described earlier and is the orientation of the GaSb/InAs SLS. Type III is similar to type two but differs in that there is some overlap of the two materials bandgaps. While material A has a conduction band higher than material B’s conduction band and a valance band with greater energy than material B’s valance band, in Type III SLS the valence band of material A is not at a higher energy state than material B’s conduction band. As mentioned earlier, the IR light that we are most interested in has wavelengths of 3-12 µm. The individual layers of the superlattice are smaller than this wavelength. This results in a property known as quantum confinement. The importance of this property is that when the material is smaller than the de Broglie wavelength, or the wavelength of photons, the material exhibits different properties than the bulk material. The reason very small materials exhibit differing properties is because of quantum effects. The holes and electrons can tunnel and interact from one material to the other. The combination of quantum confinement due to the thickness of the layers and the Type II band alignment leads to the special detector characteristics of InAs/GaSb SLS. These features allow this material to cater its bandgap to a specific wavelength of IR radiation. This SLS can change its effective bandgap from 3-30 µm which means that both MWIR and LWIR devices can be made out of this material.

Fig. 4 Schematics of various bandalignments [9]

In order to create an imaging device, detector materials must have a specific architecture and receptors that are able to distinguish between different regions of photons. A detector is first grown using MBE. The layers of GaSb and InAs are grown on top of a GaSb substrate wafer. The simplest structure first has a p doping, or region with extra holes, then the SLS structure, followed by an n doped region, or an electron rich region. This structure is referred to as a pin diode, p for the p doped region i for the SLS region and n for the electron rich region. The detector region is only several microns thick. While growth is a very important step in how the detector will perform, many of the inefficiencies arise in the fabrication of the device. Fabrication is necessary in order to isolate regions in the semiconductor for individual detection and metallization. These additions to the Superlattice are necessary to create an image that we can interpret. Pixel isolation is a very important fabrication step. During pixel isolation, regions in the semiconductor are separated from one another. By separating the semiconductor into regions known as pixels, individual signals can be read from each pixel when IR radiation is entering that region. In this way, a picture can be formed by one of two isolation techniques. The first method uses a large region of pixels, each with individual signals. The second system makes use of a single pixel by scanning the pixel across an area and compiling the data received from each point in the scan. These two different ways of obtaining an image also have differences in fabrication. The first process requires many small pixels to be in the same region as each other. This type of detector is known as a Focal Planar Array or FPA. For our FPA detectors a 320 x 256 approximately 20 x 20 µm each is used. On the other hand single, much larger pixels, anywhere from 100 µm to 400 µm, are used for the scanning pattern and are mainly used for research purposes. On the other hand, FPA’s are used in practical applications. In order to isolate these pixels a process known as etching is performed.

Fig. 5. Wafer region with pixel isolation top and side views

There are two types of device definition, wet chemical etching and dry plasma etching. Figure 5 shows the orientation and a rough, enlarged representation of what a semiconductor piece looks like before and after etching. The views are from a top down looking at the wafer, and what the wafer would look like if a pixel were to be looked at from the side after first being cut down the middle. Wet chemical etchants chemically react with the semiconductor, changing the exposed solid to a fluid that detaches and is transported into the solution. The wet plasma etch works by bombarding the surface with ions. Because only a small part of the surface needs to be etched, before etching we need to protect part of the surface. This is done by depositing a layer of a material of photochemical reactive material, photoresist (PR), on the whole semiconductor surface. Next a mask is placed on top of the material. After the mask is in place the sample can be exposed to light, the light only reaches the PR that needs to be removed. This process is known as lithography. There are two basic types of PR used in lithography, positive, which was used for this experiment, and negative. When light hits positive PR, the material becomes soluble in a specific solution while the PR that is not exposed is insoluble. The opposite is true for negative PR. The exposed region is insoluble while the unexposed region is soluble in solution. Positive photoresist is the more commonly used photoresist because it is easier to define small regions while negative photoresist is used because of its better adhesion in on certain materials. Usually only a small area of the photoresist, which means a simpler mask can be patterned for positive photoresist. This is because the mask is composed of a region blocking out light deposited on top of a translucent material. The blocking material must then be removed; it is easier to remove a small portion of this blocking material than a large portion. This means if only 10% of the PR needs to be exposed only 10% of the blocking material needs to be removed for the positive photoresist while 90% would need to be removed for the negative photoresist. The final result is a material protected in specific areas by PR. A final image of a complete, simple detector material with PR can be seen in figure 6. In this image the substrate is first covered with PR and then this PR is removed as described above. Etching can now take place after the PR has been deposited and unwanted PR removed.

Fig. 6. Depiction of deposition and selective removal of positive photoresist

Etching is the process of systematically removing unwanted material. While this is specific to wet chemical etching, an understanding of plasma dry etching is also important. For this reason these two processes will next be discussed along with benefits and drawbacks of both processes.

1. 
   Plasma-assisted (dry) etch
   

Fig. 7 Wet and dry chemical etching used to isolate pixels
For plasma etch, the wafer is first placed in a special chamber. Plasma ions are then directed towards the surface of the wafer. The plasma etch takes place at a specific etch rate and the PR is chosen so that the plasma ions cannot penetrate though it to the underlying semiconductor. The etching is stopped once a specified layer is exposed. The time of etch is determined by the etch rate of the plasma through the material.

2. 
   Chemical (wet) etch
   

For wet chemical etching the wafer is placed in a solution for an amount of time that is determined by its etch rate. After the desired depth has been achieved, the PR is removed by placing the wafer in a dissolving solution to prepare for the next stage of processing. Figure 7 shows a basic representation of these two processes. While this image shows vertical sidewalls that end abruptly at the N-contact layer, or the layer at which the user determines that etching should stop, this is not the case for real etches. In reality, etches have several problems that must be resolved in order to obtain better performance from the detector.

There are two main properties of etching, selectivity and isotropy. Selectivity is determined by the etch rate of the plasma or chemical solution and arises when one material in the superlattice structure has an etch rate much larger than another. Selectivity can be both a wanted and unwanted property. In Figure 8, selectivity is shown as favorable because etching of the N-contact is unfavorable. If the etch rate is much faster for the p-contact and SLS absorber than for the N-contact, then overetching, or etching into the N-contact region, is much less likely to occur. However selectivity can also be an undesirable property, especially in the case of SLS. This is due to the fact that the SLS is composed of two different materials with varying chemical properties. These materials have separate etch rates which can cause non-uniformity in the material hindering its performance. The other property that is important when characterizing an etching material is its isotropy. Isotropic means in all directions equally. The opposite of this is anisotropic which is directionally dependent. When dealing with pixel isolation an anisotropic etch is desirable. This is because if the etchant etches in all directions, as would occur in an isotropic etch, then the PR is undercut. This creates a problem because while the undercut is usually only a few microns thick, the area that is to be defined is usually on this order of magnitude as well. This undercut can make the pixels completely unusable. The diagram below shows the possible combinations of these two properties. In figure 8, the top left image is the most desirable while the bottom left image is the least.

Fig. 8. Selectivity and isotropy and pixel sidewalls

Next we will focus on the wet chemical etching process in more detail. In particular, we will review how a wet chemical etch differs in performance from a plasma etch, what figures of merit are used to characterize the etch and the process of wet etching. Plasma etching, while able to define small pixels, is also very rough on the sidewall surface and leaves etch residue as well as other surface defects that result in surface leakage current. A wet chemical etch is not as harsh and the surface leakage currents are much lower. Because the wet etch is carried out by chemical reaction, the solution etches in all directions. This etching rate may not be in all directions equally because of the crystal lattice structure, but the process still results in a large undercut. This undercut is acceptable for a single pixel detector where the individual pixels can be around 400 µm. However, when dealing with focal planar arrays where the pixel size is much smaller this undercut compromises the pixels. Due to this reason wet etching is rarely used for the fabrication of FPA’s. The dry plasma etching mechanism consists of the ions being directionally propelled directionally propelled and creates a much more anisotropic etch than the wet etching does. While the plasma etch’s ease of use and preferable side wall angle usually means that it is the etchant of choice, the rough side walls it produces is cause for enough problems that other avenues such as wet etching are being explored to solve this problem. One solution that has been proposed is to first plasma etch and then use a short chemical clean up etch after the plasma etch [2].

While wet chemical etching has not been able to define these small features, there are many unexplored avenues. While the basic properties of wet etching are understood, due to the complex nature of etching arising from the many etching solutions and semiconductor properties a complete picture of the reaction process has not been developed for most solutions. For this reason, much of the methodology for experimentation must rely on overlying principles and test results to determine how a new etching solution will perform.

Fig. 9. Acids used in wet chemical etching of GaSb/InAs SLS

An etching solution consists of a diluent, an oxidant and an acid. The diluent acts as a medium in which the reaction can take place. The most common diluent is water or H₂0. No other diluents were commonly used. The next component is the oxidant. The oxidant removes electrons from the semiconductor. While the oxidant reacts, it does not change the solid into a fluid that can be transported away from the surface of the material. For this process an acid is used. The acid reacts with the oxidized surface creating either a gas or a liquid that is no longer a part of the crystal lattice structure of the semiconductor. Now the newly formed product is free to move in solution. All the sources studied in this research used water as the diluents, the oxidizers and acids varied. Oxidizers used were hydrogen peroxide, bromine and nitric acid. The acid is, the most varied between solutions. The main difficulty in choosing an acid arises from the etch rate differences of InAs and GaSb. All of the acid studied showed some selectivity towards one of these components. To resolve this issue, two acids can be used in a final solution. The acids used can be seen in Figure 9 along with their chemical composition and selectivity. While all of them are selective, some of them are much less selective than others. However even the least selective of the components etch one material fast enough that a non-uniform surface can be expected. These components all contribute to the rate of reaction.

Fig. 10. Steps of wet chemical etch

The processes of the reactions are illustrated in Figure 10 and characterize the etching process. There are two main factors determining the speed of the reaction; the diffusion of the material and the reaction rate of the material. There is a property that describes a material’s tendency to move from an area of high concentration to an area of lower concentration. This property is known as diffusion. A common example of diffusion is the dispersion of food die when a drop is placed in a glass of water. Diffusion is the rate or how fast the material is transferred from one point to another. Diffusion plays an important role in the etching rate. The reaction rate also plays an important role in the etching of the SLS. In chemical reactions a molecules chemical composition is altered. In the etching process this change is characterized by the oxidizer and the acid changing the solid semiconductor to a fluid. This process contains multiple steps that can take place quickly or over a long period of time. The time for the complete change of chemical composition from the beginning product to the final product is known as the reaction rate. The limiting factor of a reaction can either be the diffusion rate or the reaction rate. If the reaction is diffusion rate limited, the reaction is taking place faster than the etchants can reach the surface. In reaction rate limited processes, the reaction rate does not take place fast enough to react with all of the etching material before more can reach the surface. Figure 10 shows the steps to the reaction and the limiting factor in each step. The first step that must take place is the transport of the etchant to the surface of the material. In this step, the oxidizer and acids must move through the diluent and reach the surface of the semiconductor. This step is diffusion limited. In the second step the etching chemicals react with the semiconductor. These reactions may also take place on the PR. The PR, however, etches much slower than the semiconductor material, so this reaction is neglected. After the reaction takes place the new products must be transported away from the materials surface to make room for fresh etchant material. This step is also diffusion limited. These processes are important because knowing which step is the limiting factor will help determine which variables should be changed, such as more stirring, to create more favorable reaction conditions.

The main variables are the temperature of the solution, agitation and the composition of each component of the solution. The temperature of the solution affects both the transport and the reaction rate. A higher temperature means more energy and more movement of molecules. This results a larger diffusion and reaction rate. However, the rates don’t increase at the same ammount, so raisng or lowering the temperature can change which step is rate limiting. Agitation moves the molecules of the etchant around in the material to aid in diffusion and helps to switch the rate determineing step from diffusion rate limited to reaction rate limited. The individual concentration can also have a major effect on the rate and mode of etching. A high concentration of diluent can both lower the reaction rate and transport rate. Also, if more than one acid is used, the concentration of each acid can determine the uniformity of the sidewalls. Overall, by changing any one of these variables the end result of the etch can vary considerably. As such it is necessary to not only understand how changing these variables may affect your solution, but also to test various paramaters in order to arrive at an optimized final solution. An optimized solution will be defined as a solution, with the same main components, that through the changing of variables maintains the most consistent and sought after results.

Research Objective

The purpose of the conducted research was to develop a chemical etch solution for InAs/GaSb SLS that would improve upon established etching procedure. These improvements include anisotropic side walls, low selectivity between GaSb/InAs, a smooth glass like surface and consistant predictable etch rates. Various journal articles were used as reference in order to develop the final procedure used in this. From this data a novel soultion composition was formed, optimized, and characterized.

Methodology

To first develop a new solution, the acids and oxidizers used in other etching solutions are documented and analyzed. Of these results, the most notable and closest to the solution studied in this paper was proposed by the 2009 paper by Chaghi et al. The solution they devoloped contained orthophosphoric acid, citric acid, hydrogen perozide and water. The composition of the soluiton of the final optimized solution was not provided. Another important soluiton, a solution currently used at the University of New Mexico by Prof. Zhaobing Tian, is composed of Tartaric acid, orthophosphoric acid, hydrogen peroxide and water in a ratio of 1:1:1:2. The research also shows that GaSb etching acids are much more prevalent than InAs etching acids. Tartaric acid and Citric acid are the only acids documented as InAs selective. For this reason, I utilized the more commonly and accessible Citric acid as the InAs etching acid in the solution to be tested. I also noted that Citric acid and Tartaric acid were similar not only in atomic structure (they both gain their acidity from carboxylic acids) but they are also common organic acids known for their contribution to the sour taste in many foods. Fig. 11. Oxalic acid and its corresponding stabilized resonance structure after losing a proton

Based on this information I performed experiments on another common organic carboxylic acid, oxalic acid, the chemical structure shown in Figure 11. The compound contains two carboxylic acid groups, and has a similar structure to citric and tartaric acid. However, oxalic acid is smaller and does not contribute to the sour taste in foods. There are also other natural acids that could be used in a chemical etch, such as lactic acid. While both the solutions noted earlier used orhtophosphoric acid as the Primary GaSb etching agent, there are many more that can etch GaSb. For my experiment I chose Hydrochloric acid (HCl). While HCl is selective towards GaSb it is noted to have a smooth surface by M.N. Kutty et al. in the Journal of Electronic Materials. It is also recorded as having a low selectivity of GaSb over InAs of 1.59 by Dier in Semiconductor Science and Technology. I chose to use water as my diluents and hydrogen peroxide as my oxidizer because they are common and readily available. these last two chemicals form a complete solution containing citric acid, hydrochloric acid, hydrogen peroxide and water. Further research into acids and oxidizers and diluents may be done based on these same principles. After the acids etching rates have been characterized for GaSb and InAs, more solutions can be tested to gain a further understanding of the topic.

Description of Experiments/Materials

The next section details the experimental components and procedures. The Semiconductor material architecture is described in Figure 12. The SLS regions consist of alternating layers of GaSb/InAs. This is the structure of the wafer that was etched by the etching solutions. Further experiments could etch a surface consisting of only GaSb or InAs to determine their individual etch rates. Also thicker layers of GaSb and InAs could be used to show the differences in etch rates.

Fig. 12. SLS structure

The superlattice layers in the wafer above were too small to gain any usable information using a scanning electron microscope (SEM). All processing took place in a class 100 clean room. The first step is to add PR and perform lithography on the wafer. Since wet chemical etching is not effective on small features, a PR mask defining single pixel detectors ranging from 25 to 400 µm was used. Once the PR was placed on the semiconductor wafer, and unwanted regions removed after exposure to UV radiation, the etching solution was prepared. All etching took place on an acid bench. Water was always measured before adding the acid or hydrogen peroxide. Citric acid becomes saturated in water when in a ratio of 1g to 1ml, this solution should be made prior to experimentation. For most of the experiments the wafer was placed in the solution for 4 min at a room temperature of 20 ºC. Three different solution groups were tested. The first was a solution with HCl, H₂O₂ and H₂O in a ratio of 100:1:100. This solution composition remained constant while the temperature was changed. Temperatures at which the etching took place were 20 ºC, 30 ºC, 40 ºC. The second solution of HCl:H₂O₂:H₂O:Citric Acid had varying compositions. The last two solutions were used to test oxalic acid. The compositions were H₂O₂:Oxalic Acid:H₂O in a ratio of 1:1:10 and HCl:H2O2:Oxalic Acid:H2O in a ratio of 1:1:1:10. All solutions were manually stirred to help facilitate diffusion of the chemicals. Original composition was determined based on the best results of the previous experiment. Concentration ratios were tested and analyzed based on surface roughness.

Results

After etching, the wafer was imaged under an optical microscope with varying magnifications. A profilomiter was used to measure the etch depth. Multiple depth readings were taken across the length of the wafer showing the uniformity of the surface. SEM images were taken of important samples to gain further insight into the surface quality produced by etching. The tests were run in 5 cycles; all data is reported in Table 1. The first test performed was on solutions 1 thorough 3, the HCl solution with a small amount of hydrogen peroxide. The etch rates for this situation were small in comparison to the other solutions. The etching rate, as expected, increased as temperature increased. The next set of experiments, solutions 4 through 7 were used to try and discern what effect each element had on the surface of the etch. The control was a solution with equal parts of all components. The best results were found in solution 5, where the etching rates were lower and produced less holes in the substrate. All off the solutions except the 5^th solution etched fairly deep into the substrate, passing the SLS structure. The solutions had many holes which may be due to HCl revealing defects in the substrate surface. Also, the color of each of the first seven solutions except solution number 5 turned some shade of yellow. This is possibly due to a reaction with HCl and the semiconductors resulting in a yellowish material in a solution that started out transparent.

Table 1. Data for wet chemical etch

After discovering that a dilute solution showed the best results, the next three solutions were all molded after this. Solutions 8- 10 were all dilute and like the previous experiment attempted to reveal the role each chemical played by changing the composition and the ratio between each solution. An even more dilute solution was prepared, as well as a solution with the same ratio of hydrogen peroxide, water and citric acid, but half as much hydrochloric acid. Diluting the solution had little effect other than to lower the etching rate, lowering the HCl percentage showed some improvement. The solution that showed the best results was accomplished when lowering the ratio of hydrogen peroxide. This resulted in, while not the most uniform etches, one of the best and by far the smoothest surface. To make sure that the solution was actually performing the best and not only a function of where, in the SLS structure the etch reached, a wafer was etched for a little less than twice the amount of time as the previous test. This again resulted in a very uniform and clean surface. The last tests were performed using Oxalic acid at the same temperature and time as before. The surface produced was extremely rough, so much so that the profilomiter couldn’t measure the etch depth for the final solution. While these results were not the most favorable the etch depth weren’t very deep into the substrate. More testing on this acid with shorter etching times, or more dilute composition, may end in better results. The reasons for the rougher etch of Oxalic acid, as opposed to the citric acid and tartaric acid solutions, may be due to its higher acidity and more toxic nature of the molecule. After this solution was optimized, solution number 10 was used to etch more relevant architecture. At the time of this article, the wafers were still being processes and, because of the differing architecture results, cannot be directly compared.

Fig. 13. Etching rates for different solutions

Figure 13 is a graphical representation of the etch rates of the different solutions. The graph can be used for future experiments to target a specific etch depth. The etch rate was calculated by averaging the etch depth over several areas of the wafer and then dividing by the total time of the etch.

A previously optimized sample of tartaric acid, phosphoric acid, hydrogen peroxide and water was compared for quality and used as a reference for surface quality. An SEM image of the wafer after the etch is displayed in Figure 14. These two solutions should not be directly compared because they were etching using different SLS architechtures. However the clean surface of the tartaric acid etch can be used as a guideline for the characteristics of a good etching solution.


Fig. 14. SEM image




Fig. 15 Individual Pixel Images

An image of each of the pixels in Figure 15, shows the contrast between the etches and how the surface quality improves as the samples composition is changed. An interesting feature to note is the double profile on the first three images, and that is accentuated as the temperature increases. This is because HCl etches GaSb faster than InAs. looking back at the device architecture, Figure 12, the first 50 nm of the structure is composed of GaSb. The depth of this first etch stops abruptly at 50nm. This feature also helps to reiterate how a wet chemical etch undercuts the surface of the PR. The bottom layer more closely represents the area the PR covers. Also SEM images of solution 1 shown in Figure 16 shows the undercut. The quality of solution 10 can also been seen and compared with previous wafers created by the solution. The solution defects, most notably seen as holes and waves in the surface, are most likely caused by defects in the substrate. Reavealing these defects is an unwanted trait of the etching solution. The oxalic treated wafers were very dark, perhaps hinting at the roughness of its surface. The defects in the sidewalls of the surface especially in soluiton 13, are also shown in these images.

The SEM images in Figure 16, show in more detail the sidewall surfaces of the wafer. Also a few profilometer profiles of key solutions are provided. The number of the solution can be seen in the top left corner. The later solutions surfaces are at a more uniform depth.







Fig. 16.The SEM image of solution 5 shows in more detail some of the holes created by the etch . The profilometer profile is looking at the sample from a side view while the SEM images are takes from an angle.

Conclusions

By understanding the properties of a SLS structure and the chemicals used to etch the SLS material, a solution to dark current in SLS detectors may be found. While the solutions proposed in this essay didn’t show signs of solving the under etching problem plagued by chemical wet etching of FPA, a detailed explanation of how this problem was described. Further research in this area may focus on other acids similar to Tartaric and Citric acid, as well as testing the samples figures of merit such as their dark currents. The most favorable solution found in this study was solution 10. Without more data and sample sets, no definite conclusions can be drawn. However this may be a direction for the future and spark interest to further optimize chemical etches to improve the performance of IR detectors.

References

[1] Rehm, R.; Walther, M.; Schmitz, J.; Fleissner, J.; Ziegler, J.; Cabanski, W.; Breiter, R.; , "Dual-colour thermal imaging with InAs/GaSb superlattices in mid-wavelength infrared spectral range," Electronics Letters , vol.42, no.10, pp. 577- 578, 11 May 2006

[2] M.N. Kutty, E. Plis,  A. Khoshakhlahg,  S. Myers, N. Gautam, S. Smolev,  Y. D. Sharma, R. L. Dawson, S. Krishna , S.J. Lee, and S.K. Noh,   “Study of Surface Treatments on InAs/GaSb Superlattice LWIR Detectors”, Journal of Electronic Materials. 39, 2203 (2010)

[3] L. Reijnen, R. Brunton, I.R. Grant, GaSb single-crystal growth by vertical gradient freeze, Journal of Crystal Growth, Volume 275, Issues 1–2, 15 February 2005

[4] Chaghi R, Cervera C, Ait-Kaci H, Grech P, Rodriguez J B and Christol P 2009 Wet etching and chemical polishing of InAs/GaSb superlattice photodiodes Semicond. Sci. Technol. 24 065010

[5] Dier O, Lin C, Grau M and Amann M 2004 Selective and non-selective wet-chemical etchants for GaSb-based materials Semicond. Sci. Technol. 19 1250–3

[6] Reverchon J, Carras M, Marre G, Renard C, Berger V, Vinter B and Marcadet X. “Design and fabrication of infrareddetectorsbased on lattice-matched InAs_0.91Sb_0.09 on GaSb” 2004 Physica E 20 519–22

[7] Campbell, S. A. Fabrication engineering at the micro and nanoscale. 3. New York: Oxford University Press, USA, 2008.

[8] Elena A. Plis, Maya Narayanan Kutty, and Sanjay Krishna, “Passivation techniques for InAs/GaSb  strained layer superlattice detectors”,  Laser Photonics Rev., 1–15 (2012)

[9] E. Plis, J.B. Rodriguez, S. Krishna, 6.06 - InAs/(In)GaSb Type II Strained Layer Superlattice Detectors, In: Editors-in-Chief:  Pallab Bhattacharya, Roberto Fornari, and Hiroshi Kamimura, Editor(s)-in-Chief, Comprehensive Semiconductor Science and Technology, Elsevier, Amsterdam, 2011, Pages 229-264, ISBN 9780444531537, 10.1016/B978-0-44-453153-7.00017-1. (http://www.sciencedirect.com/science/article/pii/B9780444531537000171)

[10] J. Nguyen, A. Soibel, D. Z.Y. Ting, C. J. Hill, M. C. Lee,and S. D. Gunapala, Appl Phys. Lett. 97, 051108 (2009)

[11] Kim HS, Plis E, Rodriguez JB, et al. (2008) Mid-IR focal plane array based on type-II InAs/GaSb strain layer superlattice detector with nbn design. Applied Physics Letters 91: 263504-263506.

[12] M.N. Kutty, E. Plis, A. Khoshakhlagh, S. Myers, N. Gautam, S. Smolev, Y.D.

Sharma, R.L. Dawson, S.J. Lee, S.K. Noh, S. Krishna, J. Electron. Mater. 39 (2010)

2203–2209

[13] Williams, Ralph E. Gallium Arsenide Precessing Techniques. Dedham, MA: The Artech House Microwave Library, 1984. Print.

[14]Tsu R, Esaki L, and Sai-Halasz BA (1977) A new semiconductor superlattice. Applied

Physics Letters 30: 651–653.

[15]Smith DL and Mailhiot C (1990) Theory of semiconductor superlattice electronic structure.

Reviews of Modern Physics 62: 173–234.

[16]P Martyniuk, J Wr´obel, E Plis, P Madejczyk, A Kowalewski, WGawron, S Krishna and A

Rogalski, “Performance modeling of MWIR InAs/GaSb/B–Al0.2Ga0.8Sb type-II superlattice nBn detector”, Semicond. Sci. Technol. 27 (2012)