Polycrystalline indium phosphide on silicon using a

Polycrystalline indium phosphide on silicon using a simple chemical route Wondwosen Metaferia, Pritesh Dagur, Carl Junesand, Chen Hu, and Sebastian Lourdudoss Citation: Journal of Applied Physics 113, 093504 (2013); doi: 10.1063/1.4794006 View online: http://dx.doi.org/10.1063/1.4794006 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/113/9?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Polycrystalline indium phosphide on silicon by indium assisted growth in hydride vapor phase epitaxy J. Appl. Phys. 116, 033519 (2014); 10.1063/1.4890718 Carrier recombination dynamics in Si doped InN thin films J. Appl. Phys. 110, 023703 (2011); 10.1063/1.3607271 Characterizations of InN films on Si(111) substrate grown by metal-organic chemical vapor deposition with a predeposited In layer and a two-step growth method J. Vac. Sci. Technol. A 25, 701 (2007); 10.1116/1.2740293 The role of the InGaAs surface in selective area epitaxy of quantum dots by indium segregation Appl. Phys. Lett. 84, 3031 (2004); 10.1063/1.1705731 Thermal strain in indium phosphide on silicon obtained by epitaxial lateral overgrowth J. Appl. Phys. 94, 2746 (2003); 10.1063/1.1593213

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 130.237.22.87 On: Tue, 02 Sep 2014 08:39:05

JOURNAL OF APPLIED PHYSICS 113, 093504 (2013)

Polycrystalline indium phosphide on silicon using a simple chemical route Wondwosen Metaferia, Pritesh Dagur, Carl Junesand, Chen Hu,a) and Sebastian Lourdudossb) Laboratory of Semiconductor Materials, School of Information and Communication Technology, KTH-Royal Institute of Technology, Electrum 229, 16440 Kista, Sweden

(Received 27 December 2012; accepted 14 February 2013; published online 4 March 2013) We describe a simple, aqueous and low thermal budget process for deposition of polycrystalline indium phosphide on silicon substrate. Using stoichiometric indium oxide films prepared from its spin-coated precursor on silicon as an intermediate step, we achieve stoichiometric indium phosphide films through phosphidisation. Both indium oxide and indium phosphide have been characterized for surface morphology, chemical composition, and crystallinity. The morphology and crystalline structure of the films have been explained in terms of the process steps involved in our deposition method. Incomplete phosphidisation of indium oxide to indium phosphide results in the restructuring of the partly unconverted oxide at the phosphidisation temperature. The optical properties of the indium phosphide films have been analyzed using micro photoluminescence and the results compared with those of a homoepitaxial layer and a theoretical model. The results indicate that good optical quality polycrystalline indium phosphide has been achieved. The Hall measurements indicate that the carrier mobilities of our samples are among the best available in the literature. Although this paper presents the results of indium phosphide deposition on silicon substrate, the method that we present is generic and can be used for deposition on any suitable substrate that is flexible and cheap which makes it attractive as a batch process for photovoltaic C 2013 American Institute of Physics. [http://dx.doi.org/10.1063/1.4794006] applications. V

I. INTRODUCTION

Several of the direct band gap III-V compound semiconductors such as gallium arsenide (GaAs), indium phosphide (InP), and gallium antimonide (GaSb) have high optical absorption coefficients and good values of minority carrier lifetimes and mobilities which make them excellent materials for making high efficiency solar cells.1,2 The two most potential III-V materials useful for single junction solar cells are GaAs and InP. The former one is already exploited more widely than the latter. Solar radiation with energy above that of the band gap is absorbed efficiently within a few micrometers of the surface because of their high absorption coefficients. Both materials have near optimum band gap values for single-junction conversion of the solar spectrum.2 InP solar cells are particularly attractive for space applications due to high radiation resistance compared to GaAs and Si,3,4 hence, high end-of-life efficiency. Theoretically, the conversion efficiency for homojunction solar cells is almost the highest for InP with respect to the other semiconductors.5 However, the cost of III-V solar cells in general is too high for terrestrial applications. This stems from the cost of raw materials, need for lattice matched substrate for epitaxial growth of single crystals and complex epitaxial growth processes.6 In comparison, polycrystalline InP offers the advantages of low cost, possibility of deposition on flexible substrates, e.g., metal7,8 and cheaper substrates, e.g., a)

Present address: INTEC, Gent University, Sint-Pietersnieuwstraat 41, 9000 Gent, Belgium. b) Author to whom correspondence should be addressed. Electronic mail: [email protected]. 0021-8979/2013/113(9)/093504/9/$30.00

glass.9,10 Various ways of depositing polycrystalline InP thin films such as metal organic chemical vapor deposition,7 flash spray pyrolysis,10 evaporation,11 pulsed laser deposition,12,13 RF magnetron sputtering,14 and co-evaporation technique15 are worth mentioning. However, most of the techniques are limited by parameters, such as requirements of high substrate temperature, multiple processing steps, stringent vacuum conditions, expensive precursors and, in some cases, problem of scalability for large scale production. In this paper, we report the results of a simple aqueous approach involving simple chemicals and relatively fewer process steps for preparation of polycrystalline InP films on Si substrate. The low thermal budget for the entire process, in addition, makes it particularly attractive for low cost, large area deposition of InP on Si. Deposition of In2O3 films on Si has been used as an intermediate step prior to InP conversion using a modified spin coating method adopted by Chang et al. for NiP deposition.16 Conversion of the deposited In2O3 to InP (phosphidisation) has been achieved using flowing dilute PH3 as the source precursor for phosphorus in a hydride vapor phase epitaxy (HVPE) reactor. The In2O3 and InP films thus prepared have been characterized for surface morphology, chemical composition, and crystallinity. The morphology and crystalline structure of the films have been explained in terms of the process steps involved in our deposition method. Incomplete phosphidisation of In2O3 to InP results in the restructuring of the unconverted In2O3 at the phosphidisation temperature. The optical properties of the InP films have been analyzed using room temperature and low temperature micro-photoluminescence (l-PL) and the results compared with those of a homoepitaxial layer and

113, 093504-1

C 2013 American Institute of Physics V


093504-2

Metaferia et al.

theoretical behavior. The results indicate that a high optical quality polycrystalline InP has been achieved. Hall measurements indicate that the carrier mobilities of our samples are among the best available in literature. Although this paper presents the results of InP deposition on Si substrate, the method that we present is generic and can be used for deposition on any suitable substrate that is flexible and cheap.

J. Appl. Phys. 113, 093504 (2013)

(LP-HVPE) reactor18 at 450 C and 20 mbar by using PH3 diluted in H2 (1:10) as the phosphorus source. Under these conditions of sufficiently low pressure and temperature, the number of collisions between the PH3 molecules is low and the decomposition of PH3 to P2 or P4 is unfavored.19 The phosphidisation process then should involve mostly PH3 molecules which can be written as: In2 O3 ðsÞ þ 2PH3 ðgÞ ! 2InPðsÞ þ 3H2 OðgÞ:

II. EXPERIMENTAL

To obtain polycrystalline InP, first In2O3 films on Si were prepared from indium acetate (In(CH3COO)3) solution and were later phosphidised to InP with dilute PH3 gas in the HVPE reactor. A. Preparation of In2O3 film

The spin-coating solution for preparing the In2O3 films was prepared from indium chloride (InCl3) as follows. A solution of InCl3.xH2O (0.25 M, x ¼ 3–4) in deionized water was treated with 28% ammonia solution. The In(OH)3 precipitate thus obtained was rinsed multiple times with deionized water and redissolved in acetic acid until a clear transparent solution was obtained. The pH of the solution was measured to be around 2. An aqueous solution of ammonium carboxymethyl cellulose (CMC, 1%) was added as binder to the prepared In3þ solution by 20 vol. %. The binder solution helped to increase the viscosity of the In3þ solution in order to uniformly coat the substrate during spin coating. To realize spin coating, Si(100) substrate was cleaned and attached to OPTIspin SST20 spin coater with vacuum suction. Adequate amount of spin coating solution was dropped on the substrate using a syringe. Spin coating was carried out at 2000 rpm for 30 s. After being air dried at 110 C for 15 min, the coated layer was heated to 400 C and kept at that temperature for 30 min in the ambient atmosphere. This heating process results in the formation of crystalline In2O3 film on Si substrate as crystallization temperature of In2O3 is known to be >280 C.17 The cooling was done in the furnace inside the clean room with a cooling rate of 5 degrees/ minute and this did not yield any cracks and we continued using this cooling procedure without any other specific measures taken for avoiding cracks. These steps were repeated several times to achieve films of increasing thicknesses. In this study, two samples, A and B, with 4 and 6 times spin coating cycles, respectively, were considered for further analysis and phosphidisation. No special surface treatment of silicon such as metal coating was undertaken to achieve good adhesion of the spin coated layer on silicon. The formation of the native SiO2 layer is practically impossible to avoid because of the nature of chemicals and solvents being used and a handling in air ambient during the spin coating process. So, it is more likely than not that the In2O3 layer is formed over an amorphous SiO2 layer which has resulted in good adhesion.

The In2O3 films were cleaned with isopropanol and acetone prior to loading into the HVPE reactor and the conversion of In2O3 to InP was done at different conversion times between 2.5 and 30 min. To study the effect of the phosphidisation time, sample A with In2O3 film resulting from 4 spin coating cycles was cleaved into three pieces, sample A1, A2, and A3 and were phosphidised for 2.5, 5, and 30 min, respectively. The details of spin coating and phosphidisation parameters for each sample are given in Table I. To study the quality of the In2O3 and InP films, a number of characterization techniques were employed. The surface morphology was examined by atomic force microscopy (AFM) and scanning electron microscopy (SEM). The chemical composition was analysed quantitatively using energy dispersive X-ray analysis (EDX). The crystallinity of the films was examined by X-ray diffraction (XRD) and the thickness determined by ellipsometer. The optical properties of the InP films were examined using micro-photoluminescence spectroscopy. Transmission electron microscopy was utilized to assess the interface between InP and Si. To carry out the Hall measurements, In2O3 film was formed on a semi-insulating InP:Fe substrate with 15 cycles as described above. This piece was cleaved into two. On one, Hall measurements were made to characterize the In2O3 film and on the other the same measurements were made after a 30-min phosphidisation as described above. Semiinsulating InP:Fe was used as the substrate only for the purpose of Hall measurements.

III. RESULTS AND DISCUSSION A. Characterization of In2O3 films

The average rate of deposition of In2O3 on Si substrate, as determined by ellipsometer, was 20 nm per spin coating cycle. The deposition rate was found to be consistent and reproducible for more than 6 cycles. AFM scans of the In2O3 films on samples A and B are shown in Fig. 1. The films were found to be extremely smooth with root mean square TABLE I. Summary of sample description, spin coating cycles for In2O3 formation and phosphidisation time. Sample ID A

B. Conversion of In2O3 to InP

Phosphidisation process, i.e., the conversion of In2O3 to InP was done in a low pressure hydride vapor phase epitaxy

(1)

B

A1 A2 A3

Number of spin-coating cycles for In2O3 formation

Phosphidisation time (min)

4

2.5 5 30 30

6


093504-3

Metaferia et al.

J. Appl. Phys. 113, 093504 (2013)

Fig. 2. Polycrystalline In2O3 with XRD peaks at 2h ¼ 30.6, 33.3, 48.06, 54.88, and 56.65 were observed corresponding to the crystal planes of (222), (321), (521), (600), and (611), respectively, from sample A and the same XRD peaks with stronger intensity and narrower full width at half maxima (FWHM) were observed from sample B. A similar trend of increasing peak intensity as a function of thickness and crystalline quality has been reported10,20 due to increased ability of ad-atoms to move towards stable sites in the lattice for films with higher thicknesses. It is observed from the deposition of CdTe films21 that as the thickness of the deposited film increases, the grain size increases and the FWHM decreases and the intensity of the XRD peaks become stronger. Recalling that the difference between the samples A and B is the difference in the thickness of the In2O3 layer, resulting from the number of spin coating cycles, the results shown by XRD in Fig. 2 clearly indicate that the structural quality improves with its thickness. The improvement of the crystalline quality in sample B may also be partly due to the additional annealing during the two more cycles that the sample was subjected to at 400 C during the formation of indium oxide. In fact, In2O3 film resulting from a two-cycle spin coated precursor was also investigated, but it did not exhibit clear XRD pattern which supports the above observation. Hence this two-cycle spin coated sample was not considered for further analysis. FIG. 1. 10 lm 10 lm AFM 3D surface plot of In2O3 films on (a) sample A and (b) sample B. The height scale is 6.5 nm.

B. Characterization of InP films

roughness (Rq) values of 0.5 and 0.4 nm for A and B, respectively, for AFM scans of 10 lm 10 lm. Scanning electron microscope (SEM) revealed films with no perceptible features. Chemical analysis using energy dispersive X-ray (EDX) yielded an In:O ratio of 1:1.53 (elemental composition of 37.3:57.2%), corresponding well with In2O3 composition. The In2O3 film on both samples is crystalline and the Xray diffraction (XRD) patterns from those films are shown in

The AFM scans of InP films from the samples A3 and B both resulting from the phosphidisation duration of 30 min are shown in Fig. 3. The Rq values were obtained from the 10 lm 10 lm scans. In general, the InP films had higher average roughness values compared with those of the corresponding In2O3 films. For example, the Rq values of InP resulting from the samples A3 and B were 142 and 77 nm, respectively; the corresponding values for In2O3 films were 0.5 and 0.4 nm, respectively. This change in roughness upon

FIG. 2. X-ray diffraction patterns for In2O3 films of samples A and B.


093504-4

Metaferia et al.

J. Appl. Phys. 113, 093504 (2013)

FIG. 3. 10 lm 10 lm AFM scans (a) and (c) and corresponding surface profiles (b) and (d) of InP films. (a) and (b): Sample A3; (c) and (d): Sample B. The surface profiles (b) and (d) correspond to the dotted lines on (a) and (c).

conversion to InP from In2O3 can be attributed to surface restructuring during the conversion. The phosphidisation process is an amalgamation of a number of simultaneous changes in the In2O3 films. The high mobility rendered to the ad-atoms due to the conversion temperature coupled with the chemical conversion occurring simultaneously can allow surface restructuring to a great extent, thereby inducing surface roughness. At the initial stage of conversion, the In2O3 film can agglomerate to form smaller islands as single grains or coalesced multiple grains. The size of these islands and extent of clustering depend on the thickness of the In2O3 film and the conversion temperature. The thinner In2O3 films give smaller islands and hence higher tendency of clustering due to faster grain migration. During the coalescence of the smaller islands (clustering), there is a strong driving force for roughing through surface atom diffusion and grain boundary motion.22 A close scrutiny of Figs. 3(a) and 3(c) indicate that InP converted from thinner In2O3 film (sample A3), shows continuous clustered islands of multiple grains that are bigger both in vertical and lateral dimensions than those islands arising from InP of sample B. The surface profiles shown in Figs. 3(b) and 3(d) were taken from two line sections (regions) to depict both the small and large clustered islands. The root mean square roughness or the interface width which follows the surface profile (the valleys, mountains and clustered islands) become smaller for increased film thickness. This tendency of decreased root mean square roughness due to the shrinkage of lateral and vertical dimension with increasing thickness was observed in indium tin oxide (ITO) films by Raoufi and Hosseinpanahi.23 The chemical composition, In:P ratio of the phosphidised In2O3 films determined by EDX analysis was found to

be very close to 1:1, within the errors of EDX measurements. The crystallinity of the InP films on samples A3 and B was examined using XRD patterns, depicted in Fig. 4. The observed InP diffraction peaks are (111), (220), (311), and (222) at 2h ¼ 26.62, 43.85, 51.96, and 54.9 , respectively, with (111) being the dominant one. The average size (D) of the crystallites of the InP films on the two samples was estimated by the use of the Debye-Scherrer formula24 given in the following equation: D¼

Kk ; b cos h

(2)

where, K is the shape factor (and has a typical value of 0.9), ˚ ), b is the peak k is the X-ray wavelength (Cu-Ka, 1.5418 A broadening (in radians), and h is the Bragg angle. For samples A3 and B, the corresponding broadening b is the same and it is 3.1 102 radians, corresponding to the strongest (111) diffraction peak which leads to the crystallite size of 48 nm. The dimensions of the grains, as observed from AFM, are in the range 50 nm to 150 nm. The lower range is not far from what is observed from AFM. The discrepancy with the higher values may be due to the fact that the peak broadening b in Debye-Scherrer formula is an approximation of crystallite size in the absence of other peak broadening effects. This also suggests that the single grain in AFM measurement may be constituted by various small crystallites formed at the initial stage of conversion as discussed before. It can be noted by comparing Figs. 2 and 4 that In2O3 (222) and (600) peaks are absent after conversion to InP but the In2O3 (321), (521) and (611) peaks still persist. The persistence of In2O3 peaks is an indication of incomplete


093504-5

Metaferia et al.

J. Appl. Phys. 113, 093504 (2013)

FIG. 4. X-ray diffraction patterns for InP films of samples A3 and B.

conversion/phosphidation of the In2O3 film. However, it is not clear if those In2O3 grains with (222) and (600) orientations are completely converted to InP or restructured to different crystalline orientations at the phosphidisation temperature. To investigate how these diffraction peaks behave with the different phosphidisation times, the phosphidisation duration was varied but by keeping all the other parameters constant. The result of this experiment is presented later. The crystalline nature of the films was confirmed by transmission electron microscopy (TEM) studies on sample A3, wherein different orientations of InP crystallites were observed. TEM micrograph, highlighting the interface and polycrystalline grains is shown in Fig. 5. The native SiO2 present at the interface is an inevitability due to the very nature of the processing involved, i.e., air ambient processing for In2O3 formation. The InP region is clearly seen with different crystallographic orientations. It can also be noted that the film is connected, i.e., no void seen at the interface between the InP and the native SiO2 layer. In order to study the effect of phosphidisation duration the conversion of In2O3 to InP, the samples A1, A2, and A3 arising from 4 cycles of spin coating were subjected to phosphidisation for 2.5, 5, and 30 min, respectively. Fig. 6 shows 5 lm 5 lm AFM scans of InP samples of A1, A2, and A3. It is observed that the surface roughness drastically increases as the conversion time increases: the surface roughness, in terms of Rq values was determined to be 41, 51, and 76 nm for samples A1, A2, and A3, respectively. The increased roughness of the film potentially arises from the increased phosphidisation duration of In2O3, to InP. An increased phosphidisation time implies a longer time for the ad-atoms to migrate and surface restructuring to take place, thereby leading to the formation of the planes with the least surface energy, InP(111),25 which is also supported by the predominance of XRD peak due to (111), see Fig. 7. It can also be

noted from the AFM image, Fig. 6(c), that the InP from a 30min conversion shows faceted grains/clustered islands. Fig. 7 shows the XRD mapping of the InP films on samples A1, A2, and A3 and Table II presents the FWHM of the InP(111) peak with the corresponding calculated grain sizes and lattice constant from these samples. It can be noted that as the phosphidisation time increases, the InP diffraction peaks get stronger suggesting an improvement of the crystalline nature of the film as well as the increase in the grain size of the InP film. However, no such clear trend was observed on the diffraction peaks from the unconverted In2O3 film due to the difference in phosphidisation time. To compare the relative composition of InP and unconverted In2O3, we calculate the ratio of the sum intensities of all the diffraction peaks of InP to those of In2O3 following the treatment by Moore.26 Note that we consider the intensity due to (hkl) orientation as the intensity count obtained from the measurement divided by the intensity of the (hkl) orientation of a standard sample taken from JCPDS-Joint Committee on

FIG. 5. Transmission electron micrograph of InP film on Si (sample A3) with visible grain boundaries and SiO2 interface.


093504-6

Metaferia et al.

J. Appl. Phys. 113, 093504 (2013)

FIG. 6. 5 lm 5 lm AFM scans of InP films of (a) sample A1, (b) sample A2, and (c) sample A3.

Powder Diffraction Standards file number 00-001-0929 for indium oxide and 00-010-0216 for InP. This ratio is shown in Fig. 8 as a function of phosphidisation time. The figure indicates that a continuous conversion of In2O3 to InP takes place within the experimented phosphidisation duration. However as we have no intermediate data between 5 min and 30 min, it is also difficult to ascertain if saturation is really achieved or not. To study the preferred orientation or the degree of preferential orientation in the unconverted In2O3 film, we investigated the variation of Rhkl, the ratio of the intensity of each diffraction peak due to a particular (hkl) orientation with respect to the sum of the intensities of all the peaks corresponding to all the orientations. Thus, Ihkl Rhkl ¼ X

;

(3)

Ihkl

all peaks

where Ihkl is the intensity due to (hkl) orientation27 divided by the intensity of the (hkl) orientation of a standard sample taken from JCPDS data. Fig. 9 summarizes Rhkl for the

orientations of (220), (600), (521), (611), and (321) of In2O3 with phosphidisation time. It is observed that the (222) and (600) diffraction peaks disappear after 2.5 min of phosphidisation. This indicates that those In2O3 grains with (222) and (600) orientations were either immediately converted to InP or restructured to In2O3 grains of other orientations due to the slightly higher phosphidisation temperature (450 C) than the oxidation temperature (400 C) used to obtain In2O3. It can also be noted that as the conversion time increases, the intensity ratios of (521) and (611) increase slightly after a small decrease while that of (321) decreases after a large increase which indicates the occurrence of restructuring of the In2O3. This restructuring is supposedly due to the simultaneous annealing of part of the In2O3 film unexposed to phosphine during phosphidisation. The InP film initially formed might have shielded this portion of the In2O3 film. Fig. 8 also indicates that the stability of In2O3 planes is in the order (321) > (611) > (521) > (600) > (222). For the InP films to be suitable for practical applications, e.g., polycrystalline solar cells, the optical characteristics of the films prepared are of paramount importance. The

FIG. 7. X-ray diffraction pattern for InP films of sample A1, A2, and A3.


093504-7

Metaferia et al.

J. Appl. Phys. 113, 093504 (2013)

TABLE II. XRD peak position, FWHM, crystallite size and lattice constant of InP corresponding to InP(111) of samples, A1, A2, and A3 phosphidised at different times. Sample Phosphidisation Peak position FWHM Crystallite Lattice ˚) ID time (min) (degree) (2h) (degree) size (nm) constant (A A1 A2 A3

2.5 5 30

26.58 26.53 26.53

0.23 0.18 0.17

40 50 53

5.80 5.81 5.81

photoluminescence of InP films prepared was measured at temperatures from 50 K to room temperature. The room temperature PL spectrum and the temperature variation of the band gap energy of the InP film from sample A3 are shown in Fig. 10. The PL spectrum is seen to peak at 1.35 eV, which corresponds to a wavelength of 919 nm and its FWHM is 28 nm. The PL spectrum shown in Fig. 10(a) is very similar to that of homoepitaxial InP in all respects, but the intensity as indicated by the normalised curves shown in the inset. The trend for the energy peak of the band–to-band emission as a function temperature is shown in Fig. 10(b) along with the theoretical prediction by Varshni,28 according to whom the band gap energy (Eg Þ and temperature are related as follows: Eg ðeVÞ ¼ Ego

aT 2 ; bþT

FIG. 9. The variation of degree of preferential orientation of the unconverted In2O3 film with phosphidisation time.

measurements were carried out on the prepared samples at room temperature. The mobility and the concentration of the In2O3 films were 15 cm2/Vs and 5 1019/cm3 and of InP, 110 cm2/Vs and 8 1017/cm3, respectively. Both the films were found to be of n-type. Our mobility and carrier

(4)

where Ego is the band gap energy of InP at 0 K (1.42 eV) and a and b are constants with the values, 4.9 104 eV K1 and 327 K, respectively. The experimentally measured results match very well with the above theoretical model. All these clearly indicate the good optical quality of the InP material that has been synthesized. C. Hall measurements

In order to determine the carrier mobility and carrier concentration of both In2O3 and the resulting InP, Hall

FIG. 8. The ratio of the sum of intensities of all diffraction peaks of In2O3 to those of InP.

FIG. 10. (a) Room temperature photoluminescence spectrum and comparison of the normalized spectrum with the homoepitaxial InP (inset); (b) temperature dependent band gap energy of InP film from sample A3 along with the theoretical prediction from Ref. 26.


093504-8

Metaferia et al.

concentration values of indium oxide are comparable to those of indium oxide deposited by spray pyrolysis,29 which are 10 cm2/Vs and 9 1019/cm3, respectively. For InP, the reported values are 36 cm2/Vs and 2 1016/cm3 for the film deposited by RF magnetron sputtering30 and in the order of 10 cm2/Vs and 1015–1017/cm3, for the film deposited by a chemical method,31 respectively. As noted, our mobility values are among the best available in the literature. It is not known what exactly the dopant source is. This can be silicon if it originates from the HVPE reactor during phosphidisation or something else coming from the chemicals used for the aqueous solutions. Since indium oxide is found to be heavily doped, as is the case with other investigators as well,29–31 the converted indium phosphide also tends to acquire a rather high doping although not as much as indium oxide. In this investigation, we did not attempt to study the variation of Hall mobility with grain size. The effect of grain size on mobility in polycrystalline InP was investigated by Roy et al. in Ref. 32. According to these authors, the carrier mobility varies from 29 to 691 cm2/Vs for the grain size of 15–2000 lm. It was shown that the mobility decreases as the grain size decreases due to the increase in grain boundary scattering. Our mobility value of 109 cm2/Vs for grains of nanometer size is extremely good. IV. CONCLUSIONS

We have demonstrated that InP films on Si(001) substrates can be deposited using a simple and less process intensive method via phosphidisation of indium oxide. Indium oxide in turn was prepared via oxidation of spin-coated indium acetate solution produced from indium chloride and other suitable reagents. EDX analysis of indium oxide and indium phosphide films confirms their expected stoichiometric compositions. It is established that the thickness of the intermediate indium oxide film can be controlled from the number of spin-coatings of indium acetate, which is 20 nm/ spin-coating cycle under our experimental conditions. AFM analysis indicates that the indium oxide film is extremely smooth with the root mean square roughness (Rq) value (611) > (521) > (600) > (222). From the PL studies, it is confirmed that the peak of the spectrum corresponds well with the band-to-band transition energy of InP, which is 1.35 eV at room temperature. Normalized spectra for the investigated InP sample and a homoepitaxial sample are similar. In addition, temperature dependent band gap energy from PL measurement is in agreement with theory. The carrier mobility and concentration obtained from the Hall measurements indicate the good quality of our In2O3 and InP films. The proposed deposition method in the current report compares well, if not better, with the contemporary reports for deposition of polycrystalline InP, but offers a lower thermal budget and less expensive approach. Preliminary investigation has shown promising results for deposition on glass and due to its inherent simplicity, the process, in principle, is scalable to large area substrates. Hence, the deposition method holds promise for cheap, bulk production of polycrystalline solar cells on Si and other suitable flexible and cheap substrates for photovoltaic applications. ACKNOWLEDGMENTS

The authors wish to thank Wubeshet Sahle and Hans Bergqvist for their help with TEM measurements and Andreas Fischer for X-ray diffraction measurements. Special thanks to Shagufta Naureen for her generous help with PL measurements and to Muhammet Toprak and Srinivasan Anand for their fruitful discussions. This work was partly realized within the framework of EURINDIA, an EU project within Erasmus Mundus External Cooperation Window, which provided a post-doctoral fellowship to one of the authors (Pritesh Dagur). 1

F. Dimroth, Phys. Status Solidi C 3, 373 (2006). R. W. Miles, K. M. Hynes, and I. Forbes, Prog. Cryst. Growth Charact. Mater. 51, 1 (2005). 3 M. Yamaguchi and K. Ando, J. Appl. Phys. 63, 5555 (1988). 4 C. J. Keavney, R. J. Walters, and P. J. Drevinsky, J. Appl. Phys. 73, 60 (1993). 5 M. K. Lee, D. S. Wuu, and R. H. Horng, Appl. Phys. Lett. 62, 1140 (1993). 6 M. W. Wanlass, T. J. Coutts, J. S. Ward, K. A. Emery, and G. S. Horner, J. Electron. Mater. 20, 1019 (1991). 7 M. Zheng, Z. Yu, T. JoonSeok, Y.-Z. Chen, R. Kapadia, K. Takei, S. Aloni, J. W. Ager, M. Wu, Y.-L. Chueh, and A. Javey, J. Appl. Phys. 111, 123112 (2012). 8 D. Kiriya, M. Zheng, R. Kapadia, J. Zhang, M. Hettick, Z. Yu, K. Takei, H.-H. Hank Wang, P. Lobaccaro, and A. Javey, J. Appl. Phys. 112, 123102 (2012). 9 P. Mitra, P. Gupta, S. Chaudhuri, and A. K. Pal, Mater. Chem. Phys. 38, 81 (1994). 10 € ¸, M. Bedir, R. Kayalı, and F. Aksoy, J. Mater. Sci.: Mater. M. Oztas Electron. 17, 841 (2006). 11 R. N. Gayen, S. N. Das, S. Dalui, R. Paul, R. Bhar, and A. K. Pal, Thin Solid Films 518, 3595 (2010). 12 A. Iribarren, R. Castro-Rodrıguez, L. Ponce-Cabrera, and J. L. Pe~ na, Thin Solid Films 510, 134 (2006). 13 M. Hafez, K. Elamrawi, and H. Elsayed-Ali, Appl. Surf. Sci. 233, 42 (2004). 14 G. H. Chandra, J. P. de la Cruz, and J. Ventura, Semicond. Sci. Technol. 26, 075017 (2011). 15 R. N. Gayen, S. Hussain, R. Bhar, and A. K. Pal, Vacuum 86, 1240 (2012). 2


093504-9 16

Metaferia et al.

W.-Y. Chung, G. Sakai, K. Shimanoe, N. Miura, D.-D. Lee, and N. Yamazoe, Sens. Actuators B 65, 312 (2000). 17 A. Rogozin, N. Shevchenko, M. Vinnichenko, F. Prokert, V. Cantelli, A. Kolitsch, and W. M€oller, Appl. Phys. Lett. 85, 212 (2004). 18 S. Lourdudoss and O. Kjebon, IEEE J. Sel. Top. Quantum Electron. 3, 749 (1997). 19 K. Gr€uter, M. Deschler, H. J€ urgensen, R. Beccard, and P. Balk, J. Cryst. Growth 94, 607 (1989). 20 Ziaul Raza Khan, M. Zulfequar, and Mohd. Shahid Khan, Chalcogenide Lett. 7, 431 (2010). 21 E. R. Shaaban, N. Afify, and A. El-Taher, J. Alloys Compd. 482, 400 (2009). 22 I. Petrov, P. B. Barna, L. Hultman, and J. E. Greene, J. Vac. Sci. Technol. A 21, S117 (2003). 23 D. Raoufi and F. Hosseinpanahi, J. Mod. Phys. 03, 645 (2012).

J. Appl. Phys. 113, 093504 (2013) 24

B. D. Cullity, Elements of X Ray Diffraction (Addison-Wesley Publishing Company, Inc., Boston MA, 1956), p. 262. 25 Y. Han and L. Lin, Solid-State Commun. 110, 403 (1999). 26 C. A. Moore, Clays Clay Miner. 16, 325 (1968). 27 L. Sun, J. He, H. Kong, F. Yue, P. Yang, and J. Chu, Sol. Energy Mater. Sol. Cells 95, 2907 (2011). 28 Y. P. Varshni, Physica 34, 149 (1967). 29 J. Joseph Prince, S. Ramamurthy, B. Subramanian, C. Sanjeeviraja, and M. Jayachandran, J. Cryst. Growth 240, 142 (2002). 30 G. Hema Chandra and J. P. de la Cruz, J. Cryst. Growth 354, 67 (2012). 31 T. Saitoh, S. Matsubara, and S. Minagawa, Jpn.J. Appl. Phys., Part 1 16, 807 (1977). 32 J. N. Roy, S. Basu, and D. N. Bose, J. Appl. Phys. 54, 847 (1983).
