INTRODUCTION
Ag-III-VI2 based ternary and quaternary chalcopyrite semiconductors have been widely investigated because of their potential application in photovoltaics and nonlinear optics. For example, AgGaTe2 (band gap 1.16 eV) thin-film based solar cells with efficiency of 4.8% at laboratory scale have been demonstrated (Mandal et al., Reference Mandal, Smirnov, Roy and Burger2003). Efficiency of 7.3% has been reported (Yamada et al., Reference Yamada, Hoshino and Nakada2006) for a tandem solar cell with an AgInGaSe2 absorber layer. AgGaSe2, another compound of this family, exhibits a large nonlinear optical coefficient and high transmission in the IR region (Xue et al., Reference Xue, Betzler and Hesse2000). In fact, AgGaSe2 crystals for second harmonic generation in the IR region are commercially available (EKSPLA, 2011). When the occupancy of cations in the ideal chalcopyrite lattice is disturbed, a series of compounds with general formula AIB2n+1III C3n+2VI, where n = 0, 1, 2, have been reported to stabilize (Zhang et al., Reference Zhang, Wei and Zunger1997). These compounds have been studied with equal interest, ever since the experimental observation of CuIn3Se5 as a segregated phase over CuInSe2 film surface (Schmid et al., Reference Schmid, Ruckh, Granwald and Schock1993). While a number of Cu based compounds have been investigated, Ag based systems have attracted much less attention. Reported work on Ag-based vacancy compounds, mainly concern with Ag-III5-VI8. The exception being AgIn3Te5, on which preliminary work was done by O’Kane and Mason (Reference O’Kane and Mason1964), and in fact, its structure has not been resolved yet. Since, the structure of AgInTe2 (I 42d) and AgIn5Te8 (P 42m) have been understood fairly well (Diaz et al., Reference Diaz, De Chalbaud, Sagredo, Tinco and Pineda2000; Sanchez et al., Reference Sanchez, Melendez, Castro, Hernandez, Hernandez and Durante Rincon2005), it is of interest to investigate AgIn3Te5, which is located in between the above two on the Ag2Te-In2Te3 tie line in the Ag-In-Te ternary phase diagram (Chiang et al., Reference Chiang, O’Kane and Mason1967). The present investigation aims at determining the structure of melt-quenched AgIn3Te5 unambiguously through proton induced X-ray emission (PIXE), X-ray diffraction (XRD) and selected area electron diffraction studies.
The isostructural nature of Cu and Ag based compounds in the I-III-VI2 family implies that the structure of AgIn3Te5 may be related to that of CuIn3VI5. Therefore, it is not unreasonable to consider the structures suggested for CuIn3VI5 compounds based on Rietveld and electron diffraction analysis. Thiogallate, stannite, and P-type tetragonal structures have been proposed for Cu-In-Se, a widely studied system with compositional ratio close to 1:3:5. Tseng and Wert (Reference Tseng and Wert1989) have suggested thiogallate structure (space group: I 4) based on electron diffraction data of commercially grown sample with composition CuIn3Se5.33. A similar conclusion has been arrived at by Bodnar et al. (Reference Bodnar, Vaipolin, Rud and Rud2006) from an XRD study of single crystal CuIn3Se5. Hanada et al. (Reference Hanada, Yamana, Nakamura, Nittono and Wada1997) have concluded that CuIn3Se5 crystallizes with stannite structure (space group: I 42m) based on XRD and electron diffraction analysis of CuIn3Se5 powder obtained from ingots grown by air cooling the melt. Rietveld refinement of crystal structure (Paszkowicz et al., Reference Paszkowicz, Lewandowska and Bacewicz2004) using XRD data of CuIn3Se5 powder, obtained from single crystal CuIn3Se5, has led to similar conclusions as well. P-type tetragonal structure with P 42c space group has also been suggested by several authors. The studies include Rietveld refinement of crystal structures using: (i) XRD data of CuIn3Se5 single crystal quenched from 1023 K into water (Honle et al., Reference Honle, Kühn and Boehnke1988) and (ii) XRD data of CuIn2Se3.5 and CuIn3Se5 powders obtained from furnace cooled ingots (Merino et al., Reference Merino, Mahanty, Leon, Diaz, Rueda and Martin de Vidales2000) as well as electron diffraction study of single crystal CuIn3Se5 grown from vertical gradient freeze method (Tham et al., Reference Tham, Su, Neumann, Schubert-Bischoff, Beliharz and Benz2000). On the other hand, for the relatively less studied CuIn3Te5, XRD analysis (Marin et al., Reference Marin, Delgado, Wasim, Rincon, Sanchez Perez, Mora, Bocaranda and Henao2000) of CuIn3Te5 powder obtained from ingots grown by vertical Bridgman method has yielded a tetragonal chalcopyrite related structure, while Rietveld refinement of crystal structure using XRD data of CuIn3Te5 powder taken from ingots quenched from 1358 to 473 K has yielded P-type tetragonal structure with P 42c space group (Diaz et al., Reference Diaz, Bisson, Agullo-Rueda, Lefdil and Rueda2005).
It may be understood from the above observations that the structure of CuIn3Se5 and CuIn3Te5 depends on changes in composition as well as on the thermal cycle followed for sample synthesis. It may also be inferred that the structure analysis of I-In3-VI5 compounds, in general, is not straight forward and an exhaustive study is required before any conclusion can be drawn. Accordingly, while refining the structure of AgIn3Te5, we have considered all space groups in I- and P-type tetragonal systems, which possess 4 symmetry and also preserve the anion sublattice arrangement of the chalcopyrite structure. We have also corroborated the Rietveld refinement results using selected area electron diffraction measurements.
EXPERIMENTAL
Synthesis of bulk AgIn3Te5
Polycrystalline AgIn3Te5 was synthesized by melt-quench technique, which involved fusion of constituent elements and subsequent quenching in water. 5N purity Ag, In, and Te pieces with compositional ratio 1:3:5 and total mass of about ∼5 g were taken in a cleaned quartz ampoule, which was then evacuated to a pressure of ∼5 × 10−6 mbar and sealed. The sealed ampoule was heated in a vertical tubular furnace, the temperature of which was controlled to an accuracy of ±2 K. The furnace temperature was initially increased to 523 K at a rate of 4 K per min and was maintained at this temperature for 3 h in order to allow the exothermic reaction between In and Te to take place at a controlled rate. Heating was continued at the same rate up to 1273 K, which is higher than the melting points of the constituent elements. The charge in the ampoule was allowed to react in this molten state for 60 h with intermittent shaking in order to improve the homogeneity. Finally, the ampoule with the charge in the molten state was quenched in water. The ingot thus obtained was polycrystalline and was black in color with shiny surface.
Characterization of AgIn3Te5
PIXE measurement
PIXE measurements were performed using tandem pelletron accelerator facility available at the Centre for Ion Beam Analysis, National University of Singapore, Singapore. The sample for PIXE measurements was in the form of pellet (thickness ∼1 mm and diameter 8 mm) made out of AgIn3Te5 powder. The sample was irradiated with collimated proton beam of energy 2 MeV, and the spectra were recorded with a maximum beam current of ∼6 nA. The characteristic X-rays emitted from the sample were detected by a liquid N2 cooled Si(Li) detector having a resolution of 170 eV at 5.9 keV. The detector was kept at an angle of 170° with respect to the incident beam and the characteristic X-rays were collected through a 50-μm thick Mylar window. The spectra were recorded using a Canberra series S-100 multichannel analyzer and were analyzed using the software “GUPIX” (Campbell et al., Reference Campbell, Hopman, Maxwell and Nejedly2000) in order to determine the elemental concentrations. A digital filter in the software was used to remove the background in the experimental spectra. GUPIX uses a nonlinear least square procedure to fit the experimental data and converts the intensities of peaks into concentration of the respective constituent elements. The main experimental parameters, which are needed for the analysis, are incident ion energy, experimental geometry, detector details, filter details, the instrumental constant H, and the collected charge.
XRD measurement
Powder X-ray diffraction measurements were made at 298 K using a PANalytical (model-X’Pert PRO, equipped with X’Celerator detector) XRD unit with Cu- Kα radiation source in the 2θ range 10 to 130° with step size of 0.0170° 2θ and integration time of ∼40 s. The X-ray generator was operated at 40 kV and 30 mA, and the wavelengths of Kα1 and Kα2 lines were 1.54060 and 1.54443 Å, respectively. Theta compensating slit was fixed at 1°, and K β lines present in the diffracted beam were filtered using a nickel filter. Prior to the measurements, the XRD unit was calibrated with silicon (external) standard. Specimen (mortar and pestle ground powder) for XRD measurements was taken in an aluminum well and was gently pressed (without affecting its random orientation) to present a smooth flat surface. Rietveld refinement was carried out using the software GSAS (Larson and von Dreele, Reference Larson and Von Dreele2004) supported by EXPGUI graphical user interface, which allows simultaneous refinement of two wavelengths (Kα1 and Kα2 with intensity ratio 0.5, in the present case). Shifted Chebyschev polynomial was used to describe the background and a Pseudo-Voigt profile function was used to model the peak shapes. The quality of fit was described by the discrepancy factors (Young, Reference Young1993) R p, R wp, R exp, R F, and χ2, the goodness of fit (GOF).
Electron diffraction measurement
Selected area electron diffraction measurements were made at different zone axis using a Philips CM-12 transmission electron microscope with an accelerating potential of 120 kV and a beam current of a few μA. The diameter of the probing electron beam was about 150 nm. For TEM measurements, fine particles of AgIn3Te5 were collected on carbon coated copper grids from a solution consisting of AgIn3Te5 powder dispersed in acetone/methanol.
RESULTS AND DISCUSSION
PIXE analysis
A typical 2 MeV PIXE spectrum of bulk AgIn3Te5 is shown in Figure 1. The inset to Figure 1 shows, on an
Figure 1. Typical PIXE spectrum of bulk AgIn3Te5. The circles represent experimental data and the continuous line denotes the calculated curve. The inset shows the spectrum in the range 2.7 to 4.25 keV, on an expanded scale.
expanded scale, the region between 2.7 and 4.3 eV. The characteristic X-ray lines from the constituent elements are marked in the spectrum for convenience. The instrument constant and energy calibration parameters were obtained using a standard sample (quartz), which emitted both K and L X-ray lines in the energy range 2 to 30 keV. Assuming bulk AgIn3Te5 as a matrix consisting of Ag, In, and Te with unknown ratio, a thick target PIXE analysis was carried out. The weights (in percentage) of Ag, In, and Te determined from the analysis are 9.76, 31.18, and 59.05, respectively, which show that the synthesized material is near stoichiometric.
Rietveld refinement
Structure refinement based on I 4, I 42m, and P 42c space groups
Powder diffraction data (symbols) of AgIn3Te5 is shown in Figure 2. Rietveld refinement was initiated assuming thiogallate structure with I 4 space group. In this structure, the atomic position 2a at (0, 0, 0) is shared by both Ag and In atoms, whereas the positions 2b at (0, 0, 1/2), 2c at (0, 1/2, 1/4) and 2d at (0, 1/2, 3/4) are occupied by In atoms. Te atoms occupy the site 8g at (x, y, z), where x, y, and z
Figure 2. Experimental XRD pattern and the results of Rietveld refinement for AgIn3Te5. The inset shows the data in the range 85° to 130°, on an expanded scale. The crosses and the overlying solid line represent the experimental data and the calculated profile respectively. The short vertical markers indicate possible Bragg reflections whereas the bottom curve denotes the difference between the experimental and calculated data.
are free parameters. The assigned occupancies were (0.8Ag + 0.2In) at 2a, 1 at 2b, 1 at 2c, 0.2 at 2d, and 1 at 8g. It is to be noted here that all Rietveld refinements have been carried out assuming the ideal compositional ratio (1:3:5) as the composition of AgIn3Te5 is found to be near stoichiometric. The refinement converged satisfactorily to R wp = 7.53% (GOF: 2.031). The calculated pattern, however, exhibited an additional reflection 002, the structure factor of which is 1.6fAg, where fAg is Ag atomic scattering factor. Since the 002 reflection was not observed in the experimental pattern, refinement was performed with redistributed In occupancy (0.1 at 2a and 0.3 at 2d) that reduces the 002 reflection structure factor to 1.6fAg – 0.4fIn, where, fIn is the atomic scattering factor for In. The refinement converged with a marginal reduction in R wp value (7.38%, GOF: 1.954), but the anion positional parameters had to be kept fixed during the refinement.
Refinement was continued, assuming stannite structure with I 42m space group. In the unit cell, Ag atoms occupy 2a at (0, 0, 0), In atoms occupy 2b at (0, 0, 1/2) and 4d at (0, 1/2, 1/4), while Te atoms occupy 8i at (x, y, z). The assigned site occupancies were 0.8 at 2a, 0.4 at 2b, 1 at 4d, and 1 at 8i. The refinement converged to R wp= 6.38% (GOF: 1.463), which is lower than that obtained for the refinement done assuming I 4 space group. However, the calculated pattern showed a low-intensity additional reflection 002, similar to the previous refinement. In order to verify if reduction in the intensity of 002 reflection would improve R wp, refinement was performed by reducing the structure factor from 1.6fAg – 3.2fIn to 1.6fAg – 2.6fIn by redistributing In occupancy (0.55 at 2b, 0.925 at 4d) and leaving Ag and Te occupancies and composition unaltered. However, the refinement converged to higher value of R wp (6.62%, GOF: 1.575), suggesting that the quality of refinement decreases.
Refinement was continued further assuming P-type tetragonal structure with P 42c space group. In the unit cell, Ag atoms occupy 2e at (0, 0, 0), In atoms occupy 2b at (1/2, 0, 1/4), 2d at (0, 1/2, 1/4) and 2f at (1/2, 1/2, 0), and Te atoms occupy 8n at (x, y, z). Vacancies are assumed to be present at the Ag site (2e) and the In site (2b). The refinement converged satisfactorily with R wp = 6.09% (GOF: 1.328), which is the lowest among the R wp values obtained so far. The optimized occupancies are 0.8 at 2e, 0.4 at 2b, 1 at 2f, 1 at 2d, and 1 at 8n. The results are presented in Figure 2 along with the experimental pattern. Unlike in the previous two cases, the calculated pattern did not contain any additional peaks. Further, all peaks observed in the experimental XRD pattern were found to obey the general reflection conditions for the space group P 42c given in the International Tables of X-ray Crystallography (Henry and Lonsdale, 1965). The unit-cell parameters obtained are a = 6.2443(8) and c = 12.5058(4) Å, with a tetragonal distortion (η = c/2a) ∼ 1.001. The refined structural parameters are given in Table I.
The experimental and calculated powder diffraction data for bulk AgIn3Te5 synthesized by melt-quench method is listed in Table II. For the sake of brevity, only the reflections with relative intensity more than 0.3% are considered while forming Table II. The reflection with the highest intensity, 112, is observed at 2θ = 24.556° and is mainly due
TABLE I. Refined structure parameters for AgIn3Te5 synthesized by melt-quench method. Structure: P-type tetragonal, space group: P 42c, unit-cell parameters: a = 6.2443(8) and c = 12.5058(4) Å, unit-cell formula weight = 1744.525, and density = 5.491 gm/cm3. The discrepancy factors are R p= 4.75%, R wp= 6.09%, R exp= 5.28%, and R F2 = 7.22%. The goodness of fit, χ2 = 1.328.
to anion sublattice as in AgInTe2. Reflections such as 202, 114, 212, 214, 306, 334, and 426 are not allowed as per the general reflection conditions for chalcopyrite structure. They are unique to off-stoichiometric or defect chalcopyrite structure. Reflections with extinction condition h + k + l = 2n + 1 are characteristic of P-type tetragonal class while those with extinction condition h + k + l = 2n are allowed for both P- and I-type tetragonal systems. Specifically, reflections which obey the conditions, l = 4n (with odd h, k-indices) or l = 4n + 2 (with even, h, k-indices) are expected for defect I-type structure. For P-type defect structure, reflections with condition l = 4n+2 (with odd/even h, k-indices) are expected in addition. Thus, for instance 212, the highest intensity reflection among these additional ones is expected for P 42c alone. The above observations clearly favor P 42c space group over I 4 and I 42m space groups. However, choice of P 42c space group without further consideration is not possible at this stage since the experimental powder XRD data did not contain all reflections expected for P 42c space group. The 002 reflection for instance, although allowed in P 42c space group, was absent in the experimental XRD pattern. The reason for this may be that either these characteristic reflections are too weak to contribute or the structure itself is different. In order to eliminate the latter possibility, it is essential to consider the space groups in P- and I-type tetragonal systems with 4 symmetry, in general.
Discussion on other space groups in P- and I-type tetragonal systems with 4 symmetry
The remaining space groups with 4 symmetry in the P-type tetragonal system are P 4, P 42m, P 421c, P 421m, P 4m2, P 4c2, P 4n2, and P 4b2 and those in the I-type tetragonal system are I 4m2 and I 4c2 (Henry and Lonsdale, 1965). However, not all of them need to be considered for the structure refinement of the title compound. This is because, although several structures have been proposed for I-III3-VI5 defect compounds, common to all of them is that they are tetragonal with the anion sublattice remaining same as that in a chalcopyrite unit cell. This requirement restricts the available space groups to P 4, P 42m, P 421m, and P 421c. The rest of them (P 4m2, P 4c2, P 4n2, and P 4b2 space groups in P-type tetragonal system and I 4m2 and I 4c2 space groups in I-type tetragonal system) can be safely excluded from the analysis as they do not preserve the anion sublattice.
Structure refinement performed assuming P 4 space group with Ag on 1a at (0, 0, 0) and on 1b at (0, 0, 1/2), In on 1c at (1/2, 1/2, 0), 1d at (1/2, 1/2, 1/2), 2g at (0, 1/2, 1/4) and 2g at (0, 1/2, 3/4), and Te on 4h at (x, y, z) and 4h at (x, y, z+1/2) was found to converge only with fixed anion positional parameters. The minimum R wp obtained was 7.92% (GOF: 2.249, site occupancy factors: 0.9 at 1a, 0.7 at 1b, 0.4 at 1c and 1d, 1 at 2g and 1 at 4h). In P 42m space group, Ag occupies 1a and 1c, respectively, at (0, 0, 0) and (0, 0, 1/2), In occupies 1b at (1/2, 1/2, 1/2), 1d at (1/2, 1/2, 0), and 4m at (0, 1/2, 1/4), and Te occupies 4n at (x, x, z) and 4n at (x, x, z+1/2). The minimum possible R wp obtained from structure refinement was 8.21% (GOF: 2.420, site occupancy factors: 0.6 at 1a, 1 at 1c, 0.23 at 1b, 0.85 at 1d, 0.93 at 4m, and 1at 4n). In P 421m space group, the atomic positions in the unit cell are Ag on 2a at (0, 0, 0), In on 2b at (0, 0, 1/2), 2c at (0, 1/2, 1/4), and 2c at (0, 1/2, 3/4), and Te on 4e at (x, 1/2+x, z) and 4e at (x, 1/2+x, z+1/2). The refinement converged to
TABLE II. Powder XRD data for AgIn3Te5 synthesized by melt-quench method. The space group is P 42c. Reflections with relative peak heights <0.3% in the experimental data are not included in the table. Values of 2θobs, d obs and I obs are obtained from the experimental data using the software X’PERT HIGHSCORE PLUS, while values of 2θcal, d cal and I cal are calculated from refined unit-cell parameters and Miller indices (hkl).
R wp= 7.65% (GOF: 2.098, site occupancy factors: 0.8 at 2a, 1 at 2b, 0.4 for z = 1/4 and 1 for z = 3/4 at 2c, and 1 at 4e).
In all the above refinements, the calculated patterns exhibited low-intensity hhl reflections 002, 003/111, and 113, regardless of Ag and In site occupancies. In addition, the refinements based on these space groups yielded higher R wp values compared to that obtained for P 42c space group. Finally, structure refinement assuming P 421c space group was carried out with Ag on 2a at (0, 0, 0), In on 2b at (0, 0, 1/2) and 4d at (0, 1/2, 1/4), and Te on 8e at (x, y, z). The refinement converged only with fixed anion positional parameters, and the minimum R wp obtained was 8.94% (GOF: 2.868, site occupancy factors: 0.8 at 2a, 0.5 at 2b, 0.95 at 4d, and 1 at 8e), which is higher than that obtained for P 42c. Further, the calculated pattern was always found to contain a 002 peak with finite intensity, regardless of the occupancies at In sites.
Discussion on hhl reflections
In all refinements except that based on P 42c space group, the 002 reflection with observable intensity was present in the calculated pattern. In addition, the refinements with P 4, P 42m and P 421m space groups, the 003/111 and 113 reflections appeared in the calculated pattern. The 002 reflection is characteristic of defect phases (extinction condition l = 4n + 2 with even h, k indices) and is allowed in the space groups P 42c and I 42m under the general reflection condition hhl: l = 2n, while in I 4 space group, it is allowed under the condition h+k+l = 2n. The reflections 003/111 and 113 are allowed in addition to 002 in P 4, P 42m, and P 421m space groups under the condition hhl: l = odd or even. In order to verify if the intensities of these reflections are really negligible in the experimental data, high-resolution scans (step size: 0.002° and acquisition time 60 s) were taken. Figures 3a and 3b are examples of high-resolution scans taken around 14° and 21°, respectively. The large circles in Figures 3a and 3b indicate the regions where the 002 and 003/111 reflections are expected. Absence of any feature indicates that these reflections, if present, have negligibly small relative intensities. The reason for the absence of the 002 reflection may be that its structure factor in P 42c space group (1.6fAg – 0.8fIn) has the least value among the structure factors in the space groups considered and therefore its intensity can be annulled easily, while maintaining stoichiometry. The absence of 003/111 reflections in addition to 002 indicates that the space group P 4 or P 42m or P 421m is not suitable. Figure 3c shows powder diffraction data around the region in which the 114 peak appears. Deconvolution of the data clearly reveals the nearby 212 reflection, which is disallowed for the I-type tetragonal system. Although the 212 peak is close to the 114 peak, it could be resolved well in electron diffraction patterns, a discussion of which will appear in the latter sections. Thus, it may be concluded from the Rietveld analysis that among all space groups considered, P 42c is the most likely space group for near stoichiometric AgIn3Te5 synthesized by melt-quench technique.
Structure of AgIn3Te5
The choice of P 42c space group ensures that the anion sublattice remains undisturbed as that in chalcopyrite structure. The cationic sublattice is occupied by Ag and In atoms, and vacancies at four different Wyckoff positions (Table I) with no evidence of mixed occupancies of cations. The vacancies may be distributed statistically or arranged in a particular order. Further, the rearrangement of Ag and In atoms and the introduction of vacancies in cationic sublattice lead to the displacement of Te atoms (with 3 degrees of freedom) from their ideal position 1/4, 1/4, 1/8. Moreover, the refined Ag-Te bond length (2.72 Å) in AgIn3Te5 unit cell was found to be smaller compared to that (2.76 Å) in AgInTe2 (Xue et al., Reference Xue, Betzler and Hesse2000), which may be attributed to the presence of Ag vacancies. Similar observations were made on CuIn3Te5 and CuIn3Se5 compounds by Rincon et al. (Reference Rincon, Wasim, Marin, Hernandez, Delgado and Galibert2000) and Paszkowicz et al., (Reference Paszkowicz, Lewandowska and Bacewicz2004), respectively. Furthermore, the different bond lengths obtained between In and Te atoms can be explained on the basis of displacement and electronegativity of Te atoms, and the occupancies of In atoms.
Figure 4 shows the unit cell of AgIn3Te5, which was constructed with the refined structural parameters as guide. When the site occupancies are normalized to Te atom occupancy, the chemical formula of the compound can be written as Ag0.8In2.4Te4 with two formula units in a single unit cell. The dotted lines represent the unit-cell boundary while the bonding between the atoms is shown by thick solid lines. It
Figure 3. High resolution (step size: 0.002°, acquisition time 60 s) XRD data of AgIn3Te5 recorded in different 2θ ranges. The large circle in panel (a) marks the region where 002 reflection is expected, while that in panel (b) depicts the region where 003 reflection is expected. The presence of two closely spaced reflections (114 and 212) can be seen clearly in panel (c). In panel (c), the gray lines show the individual deconvolutions while the dark line shows the overall sum.
can be seen from Figure 4 that the unit cell has five different types of tetrahedral structures, viz., (□ + Ag)-Te4, (□ + In1)-Te4, In2-Te4, In3-Te4, and (□ + AgIn3)-Te, where □ denotes vacancy. The bond angles in the tetrahedra varied from 103.76° to 116.77°. However, the average bond angle in each of the tetrahedron was found to be close to the ideal value of 109.47°.
Electron diffraction analysis
Selected area electron diffraction patterns of AgIn3Te5 particles recorded along two different zone axes are shown in Figures 5a and 5b. The zone axis for the pattern shown in Figure 5a is [331] and that for the pattern shown in Figure 5b is [221]. A well-ordered array of spots observed in the patterns indicates the crystalline quality of the particle under investigation. The patterns were indexed following the usual procedure. The pattern indexed with the zone axis [3331] [Figure 5a] shows 103, 123, 220 and 323 reflections while that indexed with the zone axis [221] [Figure 5b] exhibits 220, 204, 024, 102, and 122 reflections. The spots corresponding to the reflections 102 and 122 are weak in intensity [Figure 5b] when compared to the others and may be attributed to their small structure factors. The observed reflections may be categorized as follows. The reflections 220, 220, 204, and 024 are typical of zinc blende lattice (extinction condition h+k+l = 2n and even l-indices). The reflections 103, 123, and 323 are specifically from chalcopyrite lattice (h + k + l = 2n and odd l-indices). The reflections 102 and 122 are unique to P-type tetragonal structure with P 42c space group. Thus, while the zone axis [ 331] limits
Figure 4. (Color online) Unit cell of AgIn3Te5 constructed using refined cell parameters as guide. Site occupancy factors of Ag, In1, In2, In3, and Te atoms are 0.8, 0.4, 1, 1, and 1, respectively. The dotted lines denote the cell boundary.
Figure 5. Electron diffraction patterns recorded on fine AgIn3Te5 particles. The zone axes are (a) [331] and (b) [221].
the observable reflections [Figure 5a] to some of those from zinc blende and chalcopyrite lattices, the zone axis [221] allows some of the reflections, which are unique to P-type tetragonal structure to be observed as well. The presence of 122 reflection (denoted 212 in the XRD pattern) corroborates well the conclusions drawn from Rietveld analysis on the existence of P 42c. It is worth to note here that while the 122 reflection could be identified uniquely through selected area electron diffraction pattern, it was close to the 114 reflection in the XRD pattern. The angular separation between 114 and 212 reflections in XRD pattern is determined by the tetragonal distortion. In the absence of tetragonal distortion they will be degenerate, but can still be resolved in electron diffraction pattern. Thus, from the combined analysis of PIXE, XRD, and electron diffraction, we can safely conclude that melt-quenched AgIn3Te5 with near stoichiometric composition crystallizes with P 42c space group.
CONCLUSION
The structure of AgIn3Te5, which is synthesized by melt-quench method, has been analyzed using a combined study of PIXE, XRD, and electron diffraction. Composition analysis by PIXE has confirmed the near stoichiometric nature. Rietveld refinement of crystal structure has been done by considering the space groups P 4, P 42m, P 42c, P 421m, P 421c, I 4, and I 42m, which preserve the anion sublattice arrangement of the chalcopyrite unit cell. The results have shown that near stoichiometric AgIn3Te5 synthesized by melt-quench method crystalizes with P-type tetragonal structure (space group: P 42c), the existence of which has also been confirmed by selected area electron diffraction patterns.
ACKNOWLEDGMENTS
The authors wish to thank the reviewers for their constructive views, technical comments, and useful suggestions. One of the authors (C. Rangasami) gratefully acknowledges the All India Council for Technical Education (AICTE), Government of India for a Research Fellowship under the QIP scheme.
