Figure 2 shows the X-ray diffraction patterns of the layers deposited on silicon (S0-S2) and glass (V0-V2) substrates, respectively. For comparison, in Figure 2 we show the GaAs (blue) and InAs (red) x-ray PDF from the data base. The 2θ position of all the diffraction peaks of the InGaAs alloys is shifted to lower values in 2θ regarding the characteristic values for GaAs. This is a good indicative on the formation of the ternary semiconductor InGaAs and it depends on the In concentration. The crystallites sizes (D) of thin films were estimated using the Scherer formula:

D = k λ β c o s θ (1)

Figure. 2

where k is a constant = 0.9, β is the full width at half maximum (FWHM) and λ is the wavelength of X-ray used. The mean crystallite size was calculated using the Scherrer equation to the (111) diffraction peak. The mean crystallite sizes of the layers were of 20 nm approximately. The results are summarized in Table 1.

Table 1

Layer	RF (W)	2θ (°)	L.P. (Å)	FWHM	C. S. (nm)
Glass substrate
GaAs- V0	0	27.3	5.65	0.405	20.18
InGaAs-V1	6	27.2	5.68	0.449	18.19
InGaAs-V2	8	27.1	5.69	0.381	21.44
Silicon substrate
GaAs-S0	0	27.3	5.65	0.375	21.79
InGaAs-S1	6	27.0	5.70	0.440	18.56
InGaAs-S2	8	26.9	5.73	0.781	10.45

L.P. lattice parameter, C.S. Crystallite Size

As a first approximation, we assume the Vegard's law, commonly used on determining the composition of ternary and quaternary semiconductor III-V ^[20], and used it for determining the Indium concentration at our samples, as a function of the Indium sputtering power (Table 2).

Table2. Atomic

Layer	As(%) EDS	Ga(%) EDS	In(%) EDS	In(%) XRD	In(%) Raman
Glass substrate
V0	52.58	47.42	−	−	−
V1	51.65	40.61	7.73	7.31	8.10
V2	50.11	35.43	14.46	9.75	13.00
V2(P1)	26.77	28.27	44.95	−	−
V2(P2)	48.26	41.41	10.33	−	−
Silicon substrate
S0	51.54	48.46	-	-	-
S1	54.11	32.71	13.19	12.19	11.70
S2	55.13	28.86	16.01	19.51	17.40
S2(P1)	23.51	18.78	57.51	−	−
S2(P2)	46.03	41.06	12.91	−	−

In all layers, a preferential growth along the direction <111> was observed. However, for the S2 film, additional peaks (marked with * on Figure 2) associated to the crystallographic planes (111) and (200) of the tetragonal Indium phase appear, due to segregation of In during the film deposition. We will discuss this point in the morphological characterization and SIMS sections.

In the first order Raman spectra shown in Figure 3 for GaAs layers (S0 and V0), characteristic modes were observed, while Raman TO and LO frequencies of InAs-like and GaAs-like vibrations appear in the InGaAs layers. A fitting was performed in all spectra using Lorentzian profiles, in order to determine the position of each phonon frequency. The vibrational frequency for TO InAs-like for the layer deposited at 6 W is shifted Δω ≈ 9.9 cm^-1 with respect to the typical value of the TO mode (221 cm^-1 ) for InAs . Similarly, a shift of Δω ≈ 8.3 cm^-1 was observed for the layers deposited at 8 W. These shifts in the Raman frequencies of the In_xGa_1-xAs alloy is a combined result of the varying concentration and stress, due to lattice mismatch at the layer-substrate interphase ^[21]. The Raman shifts confirm the formation of the InGaAs ternary semiconductor. By considering the linear relation between the optical phonon frequency and the composition (x) according to ref. ^[22], it is estimated that the In concentrations (x) is 0.08 for sample V1 and 0.13 for sample V2, respectively.

Figure. 3

The layers reported in this work were also characterized morphologically by scanning electron microscopy (FE-SEM) and atomic force microscopy (AFM). The FE-SEM cross-sectional images show a columnar growth mode with a thickness about 650 nm for layer S0 and 540 nm for layer S1 (Figure 4a) and c), respectively). For the S2 layer, a well-defined growth mode is not observed (Figure 4e)), possibly due to indium excess during growth, in good agreement with the XRD results. The estimated thickness from the SEM image was around 600 nm. Figure 5 show the FE- SEM micrographs taken on a plane view with a magnification of 100.000X. In the GaAs layers S0) and V0) a smooth surface is observed, while in the S1 sample, particles (≈ 150 nm diameter) with different geometries can be seen. In the S2 sample, change on the surface morphology is observed attributed to indium segregation. An increase in both, particle size and roughness due to the influence of the In sputtering power, was confirmed by the values of roughness and particle size obtained from the AFM images (not shown here).

Figure. 4

Figure 5

The morphology of the V1 and V2 films (Figure 5d) and f)) are similar to the S1and S2 layers. The sections marked with a red circle (P1) in Figure 5 denote In rich regions (due to In segregation), while the regions marked with a blue circle (P2) denote regions without segregation.

The EDS spectra, showed signals associated to the Kα-lines of In, Ga and As, respectively. For the V1 and V2 layers, smaller concentrations of In were determined, likely due to the lower thermal conductivity coefficient of glass with respect to the Si (100) substrates, affecting the surface mobility of each layer during growth for a nominal temperature of 580 °C. In order to determine the distribution of elements on the surface, a bi-dimensional analysis was performed on each layer. The S1, S2 samples showed a gradual increase of the In concentration (%In) when sputtering power increased, as shown at table 2.

A detailed EDS analysis was performed on all layers. For S1 and V1 layers the composition was measured at four different points without finding significant variations in the atomic percentages. The S2 and V2 layers showed composition variation at different regions as summarized in Table 2. The points indicated by P1 correspond to segregation regions, while points P2 correspond to the bulk of the material. Figure 5 schematically indicates the different points in each of the layers. Although, regions with different compositions are evident due to bulk segregation of In in the S2 and V2 samples, the results do not allow discarding the occurrence of other InGaAs phases with different composition to a lesser concentration. The difference in the In composition as determined by different methods (table 2), is attributed to the nature of each of the characterization techniques used. In the Raman case, the shifts due to stress were omitted while for X-ray diffraction, a single peak associated with the (111) plane was considered discarding other phases present in the alloy.

From the results of secondary ion mass spectroscopy SIMS (Figure 4), it was observed that for sample S1, Ga, In and As are almost uniformly distributed along most of the depth. However, a little change in intensity at the layer-substrate interface is observed, attributed to imperfections and decoupling surfaces. On sample S2 there is an equally uniform distribution throughout the thickness, but a change of the (In, Ga) concentration is also observed at the surface, confirming the segregation of In there. It was also determined that the layers have low levels of contaminants such as carbon and oxygen, but their quantification was not possible.

The transmission spectra of the layers deposited on glass V0-V2 were taken in a range of 300-2000 nm From these spectra, the bandgap of the ternary alloys were determined. From the transmittance spectrum the absorption coefficient (α) was calculated using the Lambert's law:

α = 1 d ln ⁡ 1 T (2)

where T is the transmittance and d is the layer thickness. It is well known that the InGaAs is a direct band gap semiconductor, thus the absorption coefficient (α) can be described as a function of the incident photon energy by the relation,

α h v 2 ∝ h v - E g (3)

it is associated to the inter-band electronic transition between the valence band and the conduction band by extrapolating the curve (αhv)² vs hv when (αhv)²→0. Direct band gaps of 1.44, 1.19 and 1.00 eV were obtained for samples V0, V1 and V2, respectively (Figure 6.). These results agree well with the expected values, intermediate between the band gaps of InAs (0.36 eV) and GaAs (1.43 eV), depending upon the composition ^[20,23]. The dependence of InGaAs band gap as a function of the sputtering power for In is evident. It is also observed that there is a band gap shift (E_g) towards lower energies, when increasing the sputtering power for In deposition up to 8 W. This behavior is due to the incorporation of substitutional In atoms into the GaAs lattice.

Figure 6

The thicknesses of the layers were determined from the interference oscillations present in the transmission spectra. The values obtained were 615, 520.1 and 548.5 nm for V0, V1 and V2 samples, respectively, which agree well with those observed in the SEM images ^[24-25].