Atomic Force Microscopy Studies on GaAs/In Bilayers Deposited on Si (100)

GaAs/In bilayers were prepared by RF Magnetron Sputtering in an Ar atmosphere on Si (100) substrates using high purity (95.95%) GaAs (100) and In targets. The growth temperatures were 300 ̊C and 580 ̊C for the high purity targets of In and GaAs, respectively. Three samples were prepared: the deposition time (td) for the GaAs layers was fixed to 30 minutes, while varied for the In layers from td = 10, 15, and 20 minutes. The morphological and optical studies of the samples were made by means of Amplitude Modulation Atomic Force Microscopy (AM-AFM). In order to analyze and correlate surface morphology and alloy composition properties, the as-prepared samples were cleaved along the [001] direction and subsequently studied by AM-AFM-micrographs. From topographic images, a statistical study of the roughness and grain size was made. Additionally, cross sectional AM-AFM-micrographs were performed for each sample, where the phase channel, which is sensitive to the material properties of the specimen, was of particular interest.


Introduction
Indium Gallium Arsenide (InGaAs) heterostructures, often regarded as GaAs and InAs binary alloys, are a III-V semiconductor compound largely used in the electronic and optoelectronic industry. The range of practical devices fabricated with this compound, grown on Si (100) substrates, due to the different band gaps related in the system (0.33 eV for InAs, 1.43 eV for GaAs and 1.12 eV for Si), varies from optical sensors, photodiodes, solar cells and many other kind of devices [1] [2].

Sample Preparation and Fabrication
The In/GaAs bilayers where grown on a (100) silicon substrate by Radio Frequency Magnetron Sputtering in an Argon (Ar) atmosphere. High purity (95.95%) GaAs (100) and In targets where used. Before deposition the Si (100) surfaces were firstly degreased with acetone and methanol. Secondly, they were cleaned with hydrofluoric acid at 2%, rinsed with deionized water, and dried by flowing nitrogen. Thirdly, the substrate was introduced into the sputtering chamber to start the vacuum process. When the base pressure of 1.2 × 10 −6 Torr was reached, the substrate temperature was raised up 650˚C to remove the oxide on surface. The substrate temperature was lowered and stabilized at 580˚C, and high purity Argon gas was introduced into the chamber until a working pressure of 1.2 × 10 −3 Torr, and the plasma is turning on. Next, the GaAs target source was powered on in order to deposit a GaAs buffer layer during 30 min, after then the GaAs shutter was closed. Subsequently, the substrate temperature was lowered to 300˚C and the Indium (In) source target was powered on, in order to deposit Indium during t d = 10 minutes (sample M1), 15 minutes (sample M2) and 20 minutes (sample M3), respectively. This process is then repeated once again in order to deposit two In/GaAs films. Finally, the structure was capped with a GaAs layer (t d = 20 min).

AFM Equipment Setup Description
The AM-AFM characterization where performed with the N8 NEOS SENTERRA system from Bruker, working at ambient conditions. This equipment combines an optics SENTERRA Raman spectrometer with the Nano's N8 NEOS Atomic Force Microscope, to allow the morphological, structural and chemical analysis of the same sample area. It is mounted on a very heavy table and everything is enclosed in a Faraday Box. The cantilever beam deflection sensor is based in a fiberoptical interferometric detection system. The cantilevers used were Point Probe Plus Non-Contact Long cantilever Reflex coated (PPP-NCLR) from Nanosensors, with resonance frequencies in vacuum of 190 Khz, and force constants around 48 N/m. The AFM machinery was set to work in AM-AFM contrast phase mode, with free oscillation amplitudes of 195 nm, set points of 50%, 65% and 80% and oscillation frequencies around 169 Khz. All the AFM images presented here shown the forward scanning direction. Free oscillation phase was set to 90˚, in a range going from 0˚ to 180˚ or from 0˚ to −180˚. The software's programs used to treat the AFM images were Gwyddion [11], WSxM of "Nanotec Electronica" [12] and Nano Scope Analysis [13] (the sample orientation with respect to the AFM cantilever is shown in Appendix). Figure 1 shows the optical images obtained with a 100× lent (upper part) and the corresponding 40 × 40 μm 2 AM-AFM topography images (lower part), for each of the three samples: In Figure 1 For these large scale AFM images, the corresponding statistical quantities were the following. For the sample M1 it was found a Maximum Height of 2.36 µm and the RMS roughness was 0.28 μm. For the sample M2, we found a Maximum Height of 3.18 µm and the RMS roughness was 0.38 μm. Finally, for the sample M3, we found a Maximum Height of 7.72 µm and the RMS roughness was 0.97 μm.

Analysis and Results
In order to obtain the grain discretization a 10 × 10 μm 2 zoom was obtained in this same area for each sample. The corresponding AFM topographic images are following shown in Figure 2.  Either the perpendicular features as well as the polygonal alike shapes, seems to be randomly localized. Note that the localization of these shapes is easily obtained from the grain discretized mask for the samples M2 and M3 but more difficult for the one representing the sample M1. Another interesting observation was obtained from the 3D representations, shown in Figures 2(g)-(i), for the samples M1, M2 and M3 respectively. Here, it is possible to note that the grain structures were presented in all the specimens; however a larger tilted planes with certain acute angles. This is more clearly appreciated for the samples M2 and M3, as following described.
The topography AFM images presented in Figure 3, shows a zoomed area of 5 × 5 μm 2 of the three samples. In this part of the specimens one of the larger grains found for each sample, was selected to be measured by a height profile line. These measurements showed these tilted planes (see Figure 3). Figure 3 shows now a series of height measurements for each topography image corresponding to each sample. In this area, the highest height was 1.84 μm, 3.49 μm and 5.4 μm, for the samples M1, M2 and M3, respectively. The image shown in Figure 3(a) is marked with black dotted lines some of the linear features found on the specimen. In this case, the highest height found is the product of an island, an agglomeration of grains. As the indium quantity increased, these agglomerations leaves to the formation of bigger grains giving rise to the bigger grains presented in Figure 3 shape, with a small planar section. On the other hand, the sample M3 seemed to have small terraces separated by certain heights.
In the following, the cross section experiments are presented.

Cross Section Imaging: Topography and Phase Micrographs
The following images shows the cross section morphology of the samples, see Figure 4. On top, we can see the optical images with the 100× lent. Below it is shown the corresponding topography AM-AFM image of the marked zone by the green rectangle area in each of the optical images. The optical images of Figure 4, shows the position of the In/GaAs/In/GaAs film with the red arrow, pointing toward the right side of the image, the [100] direction. The Si substrate is presented at the left side and the black color to the right is the free space after the specimen. Note that the 3 × 10 μm 2 AM-AFM topographic images, preserves the same orientation than that of the optical image. These AFM images with their height profiles are following compared to their corresponding phase images, presented with their corresponding Phase vs Distance Graphs, in Figure 5. A Code-V color representation for the topography and phase AM-AFM images was used. The presence of the In/GaAs/In/GaAs film is appreciated more clearly. The topographic channels are presented in the upper part, Figures 5(a)-(c) for the samples 1, 2 and 3, respectively. Here, the silicon substrate is exhibited with orange, hard-orange or red colors. The GaAs-In film is presented with yellow and/or with green color. Each image has a height profile line, which is shown below. In this particular locality, these height profiles show a film thickness of about 2.16 μm for the sample M1, 3.85 μm for the sample M2, and 6.15 μm for the sample M3. Note that these values are quite close to those observed by the height profiles of the big grains presented in Figure 3. Here the criteria followed to obtain this particular distance, was by selecting the first abrupt change in the phase channel that was followed by a different phase behavior. For the sample M1, Figure 5(a), the height profile starts with a linear tendency from the left (light orange color). Then the line profile presented a small increment until a maximal height value of 6 μm (orange color), that decreased until a height of 5.29 μm at about 7.52 μm of distance. This is what we considered to be the part of the Si substrate. After this the height value started to decrease to about 4.5 μm of height and remained there for a while (corresponding to the yellow contrast in Figure 5(a)). Then the height decreased linearly (green contrast) until a value close to 0.73 μm of height  Then the presence of some perturbations to 75˚ and 83˚ until a distance of 6.25 μm, followed by a small lobe to 85˚, appeared (yellow color). After this, the phase signal decreased to 76˚ at a distance of 7.5 μm, (green color), and then increments and decrements to 77˚ and 68˚ are continued (green, yellow, blue colors) until a clear fallen of the phase to 31˚ at 9.1 μm of distance (blue color). From here the phase went up to 62˚ and then to 77˚ at 9.68 μm of distance, where it remained until the end of the profile part, right side of the image (green color) corresponding to the free oscillation of the cantilever. For the sample M2, Figure 5(b), a similar feature is presented in the height profile measurement, but not with the same increment rate as it was presented in sample 1. Starting from the left side, a more straight linear tendency at about 17.82 μm is presented until a distance of about 4.06 μm (all the hard-orange/red color). This part is what we attributed to be the Si substrate. A small lobe is presented just before the height started to decrease. Then the profile is decreased (yellow contrast) until the profile goes to a value of about 13 μm of height and remained there for a while (green contrast). Then the height is decreased until a 0.44 μm of height is reached, at 7.91 μm of distance, where the signal remained until the end of the profile at the right part of the image (all the blue color). Here is where the cantilever is in the free space position. The phase presented values near 81˚, corresponding to the orange and red contrasts, left side of Figure 5(e), and then the phase increased gradually until a close value of 86˚, around a distance of 2.98 μm. Here it is presented a series of jumps near 85˚ until a fast fallen to 64.95˚ at 4.06 μm of distance. After this, increments to values close to 85˚ with intermittent jumps to 90˚ were presented (red, blue, green and orange zone) until a distance of 5.6 μm, where a noticeable decrease in the phase is observed. Here the phase presented different fallen and jumps to 75˚ and up to 85˚ (corresponding to the blue, green and yellow colors) and lasted until a distance of 7 μm. Then the phase presented jumps from 65˚ to 85˚ and 80˚. These phase oscillations were attenuated until a close value to 70.23˚ at 7.9 μm of distance, where the attenuated phase kept a constant value (blue contrast), presented at the right part of the image. This is the part where the cantilever is not any more interacting with the specimen cross section.
Finally, for the sample 3, starting from the left part of the image, the height profile presented a constant height value, at about 13.8 μm, which lasted to a distance of 2.13 μm. Then an increment to about 14.68 μm of height started at about 2.46 μm of distance (orange and red color). This is what we considered to be the Si substrate. The profile remained for a while around 16 μm and then decreased with a logarithmic alike behavior (yellow and green colors) for all over a distance of 5.49 μm and until a value near 1.75 μm of height, at 8.52 μm of distance. From here a constant value remained all over the right side of the image (blue contrast). Regarding the phase signal of the sample 3, the image started to the left part (blue contrast), corresponding to values near −85˚, Figure 5(f). At 1.41 μm of distance, a pick to −75˚ was observed, then returning to −85˚, followed by a decrement to around −90˚ at 2.11 μm of distance. Two picks are then presented, one to −7.7˚ at 2.46 μm and the other to −71˚ at 2.7 μm (green and blue colors). After that, values around −75˚ with intermittent fallen to −80˚ were presented until a distance of about 6 μm (green, blue, yellow colors). Then a sharp pick to −45˚ with a subsequent fallen to −80˚ followed by another pick to −50˚ and a consecutive fallen to −75˚ were presented. The phase went down to −85˚ with jumps to −70˚ and fallen phases to −80˚ (green, yellow, blue, hard-orange colors) until a constant value of around −63.61˚ at 8.61 μm of distance. At this phase value the signal was maintained until the end of the profile, at the right part of the image (hard-orange color).
To conclude with this section, the phase signal presented a different value in the GaAs/In/GaAs/In/GaAs film compared with the silicon substrate. It is important to note that the change in phase is presented in conjunction with a change in topography; however, even though the change in topography decreased with a logarithm alike behavior, the phase shift did not. Then after the existence of the cross section, were the tip was oscillating in the free space, the phase shift was not 90˚ as it was originally when far away from the surface.
Regarding the AFM study, from the watershed algorithm of the Gwyddion software, we could obtain the discretization of the grains and the corresponding mask, containing the grain boundaries for each of the three samples studied, Figures 2(d)-(f). From these masks, and with the use of the Maximum Height statistical quantities obtained from each of the AFM images, we could observe that the grain lateral dimensions and heights, as well as the film averaged thickness, were increased upon larger Indium deposition times. This observation is consistent with the Mean Grain Size and total gran volume quantities extracted from the grain statistics see Table 1. This suggests that there exists a growing tendency produced by the number of Indium layers deposited and the grain size. The AFM images presented in Figure 2, showed the following behavior. Figure 2(a) presented a variety of very small grains that in certain zones they formed agglomerations (islands), that exhibited higher heights and greater lateral dimensions. These agglomerations were not homogeneous around the 10 × 10 μm 2 surface. Instead, they apparently seem to be randomly arranged. By subsequent Indium deposition, theses agglomerations collapsed and formed bigger grains, leaving the new smaller grains to accumulate around the bigger ones, see Figure 2(b). These bigger grains started to present a large tilted plane towards the [011] direction, with acute angles that increased from 53˚, 58.5˚ and 61˚ while the Indium quantity increased. Also the plane in front of this tilted plane tended to form right angles from the [011] horizontal line and the top of the grains became flatter. This suggest a pure three dimensional growth with a high tendency of expansion towards the [100] direction. This three dimensional growth comes from the lattice parameters mismatches presented by the substrates and the alloys.
Regarding the statistical quantities and grain statistics, we obtained the Table 1 and relations. The increments of the In deposition time, starting from a total of 20 minutes for the sample M1, was scaled  Regarding the phases analysis, for the moment only qualitative information can be given, see Figure 6. As shown in Figure 1, for the sample 1, this particular profile line scan speed was made at 0.7 l/s, which is faster than those speeds of 0.2 l/s used for imaging the sample 2 and sample 3. However, while the topography signal started to decrease in a logarithmic form, the phase signal presented a linear tendency composed of jumps and falls. The topography effect observed on the Phase signal was, when the oscillating tip found an abrupt fall in the topography, a fallen for the sample 1 and 2 and a jump for the sample 3. This falls and jumps are indicated by the dotted green lines in the Figures 6(a)-(c). The overall behavior of the phase signal did not show any logarithmic or exponential tendency. For example in the sample 2, when the topography signal decreased after the small lobe, at around 4.5 μm of distance, the phase fallen form a while. Then it seems to be oscillating until a more stable signal was obtained. When the topography went down to 13 μm of height, the phase suffered also a fallen. Then it came back to around 70˚ and even thought the topography profile, started to decrease the phase signal presented values around 70˚, with intermittent falls and jumps. No exponential or logarithmic behavior was exhibited by the phase. This behavior is still more presented for the sample 3, Figure 6(c). Here, the topography profile clearly decreased logarithmically; while the phase seemed to presents a repetitive value around −75˚. Again when the topography signal presented a suddenly fall, at around 6.11 μm, the phase presented a big pick to around −50˚, then came back to values close the −75˚. Another abrupt fall in the topography signal at 6.90 μm was presented, and again, the phase presented a big jump to around −45˚, and then returned to the values between −80˚ and −70˚. Finally, regarding the corners shapes found in the sample 1, the hexagonal shapes found in the sample 2 and sample 3 as well as the grain three dimensional structures, the growth process of In-GaAs on top of clean Si (100) substrates probably leads to the formation structures with some order that runs in certain crystallographic orientations. A subsequent analysis by electron microscopy can give us new insides about these observations as well as a detailed study of the initial growth of GaAs on Si.

Conclusions
Three different samples containing In/GaAs films on Si (100) substrates with different Indium quantity were successfully fabricated by RF Magnetron Sputtering. The samples were successfully characterized by AM-AFM.
The AFM was implemented in order to obtain the morphology of the surface and of the inner structure features for each of the samples. For a total time of 20 minutes of In deposition, small grains were formed. Corner alike shapes were found, and some agglomerations of grains were presented. This initial formation presented a certain growth tendency towards the [111] direction and a tilted plane towards the [011] direction could be identified. For a total of 40 minutes, this agglomerations formed bigger grains, and for a total of 30 minutes of In deposition these grains presented larger tilted planes and a flatter structure on top.
Thanks to the statistical quantities obtained, we could observe that the samples reported Maximum Heights of 1.79, 3.77 and 5.44 µm, presenting RMS roughness of 0.237, 0.512 and 1.01 µm for the sample M1, M2 and M3, respectively. Thanks to the implementation of the watershed algorithm, these grains could be discretized. From where the grain population densities of 11.2 μm −2 , 1.71 μm −2 and 0.73 μm −2 where obtained. The Mean Grain Size, varied from 183 nm to 630 nm and to 820 nm, and the Total Grain Volume varied from 48 μm −3 to 137 μm −3 , and then to 227 μm −3 . These grains formed some shapes that could be identified. The corners shapes found in the sample 1, the hexagonal shapes found in the sample 2 and polygons found in the sample 3 as well as the grain three dimensional structures, leads to the supposition that the growth process of In-GaAs on top of clean Si (100) substrates probably give rise to the formation of structures with some order that runs in certain crystallographic orientations. A subsequent analysis by electron microscopy can give us new insides about these observations as well as a detailed study of the initial growth of GaAs on Si.
Finally, the qualitative analysis observed here, the cross section phase imaging presented different phases shift for each sample, comparing the Si (100) and the In/GaAs film. The more notable result was that while the topography profile decreased logarithmically, the tendency of the phase signal did not present this behavior. The sudden falls on the topography signal, produced a sharp pick in the phase signal. As an idea, it is clear that Silicon, Gallium Arsenide and Indium Arsenide have different mechanical properties, like different harness values. It would be very interesting to try to relay any mechanical properties of the alloys and substrates compositions with the phase signal evolution. But for the moment we are not able to relate anmechanical properties with the phase shifts.