Tailoring the Band Gap in the ZnS/ZnSe System: Solid Solutions by a Mechanically Induced Self-Sustaining Reaction

Inorganic Chemistry 58 (2019) 2565-2575 1 Tailoring the band gap in the ZnS-ZnSe system: Solid solutions by mechanically induced self-sustaining reaction Miguel A. Avilés†, José M. Córdoba†‡, María J. Sayagués†, and Francisco J. Gotor*† † Instituto de Ciencia de Materiales de Sevilla (US-CSIC), 41092 Sevilla, Spain ‡ Departamento de Química Inorgánica, Universidad de Sevilla, 41012 Sevilla, Spain ABSTRACT: The complete ZnSxSe1-x solid solution was successfully obtained by the mechanochemical process denoted as mechanically-induced self-sustaining reaction (MSR). An excellent control of the chemical stoichiometry of the solid solution was possible by adjusting the atomic ratio of the starting Zn/S/Se elemental mixture subjected to milling. A mixture of both wurtzite-2H (hexagonal) and zinc blende (cubic) structures was always obtained, although for a similar milling time the proportion of the zinc blende structure increased with the Se content in the solid solution. However, wurtzite was the major phase for S-rich compositions when milling was stopped just after ignition. It was demonstrated that milling induces the wurtzite-to-zinc blende phase transition. The 8H hexagonal polytype was also observed in samples subjected to long milling times. The variation of the lattice parameters for both structures with the x value in the solid solution presented an excellent linearity, confirming the validity of the Vegard’s law. However, the variation of the band gap energy (Eg) with x was not perfectly linear and a small bowing parameter of 0.34 was obtained. It was possible to tune the Eg value between those of the end members of the solid solution in a continuous manner by adjusting the stoichiometry of the solid solution. The morphology and the crystalline domain size can be also controlled by adjusting, in this case, the post-ignition milling time of the mechanochemical process. KEYWORDS: Mechanochemistry, Combustion process, Semiconductor, Optical properties INTRODUCTION Zinc sulfide (ZnS) and zinc selenide (ZnSe) are IIVI semiconductor materials with wide direct band gap that attract much attention due to a great variety of electronic and optoelectronic applications, including light emitting diodes, electroluminescence, lasers, flat panel displays, infrared windows and sensors, among others1-2. ZnS presents two stable polymorphs at atmospheric pressure, the cubic zinc blende or sphalerite (F-43m) and the hexagonal wurtzite-2H (P63mc), being the sphalerite phase thermodynamically stable below 1293 K3. A cubic rock salt structure (Fm-3m) has also been reported, but this phase can only be obtained under high pressure conditions3,4. Although, most synthesis methods yield ZnS with the sphalerite structure, the synthesis of ZnS with wurtzite structure using mild- or low-temperature processes has also been reported5–7. It has been shown that the formation of a nanometric microstructure and the control of the particle morphology can favor the formation of the wurtzite structure8,9. ZnSe presents only one thermodynamically stable phase under normal pressure, the stilleite with zinc blende structure. Although the predicted temperature for the zinc blende-to-wurtzite phase transition in ZnSe is above its melting point10, examples in which the wurtzite structure was obtained using similar methods as for ZnS can also be found in the literature11–13. This is related to a small energy difference between the two structures14. A high-pressure cubic structure has also been reported for ZnSe15. The band gap energy (Eg) of ZnS is 3.6 and 3.8 eV for the zinc blende and wurtzite structures, respectively, and 2.58 eV for ZnSe with zinc blende structure16. Since ZnS and ZnSe have identical crystal structures and small lattice parameter mismatch, they can form a solid solution (ZnSxSe1-x) over the whole compositional range. This solid solution provides the possibility of tuning Eg and facilitate optical and elec- 2 tronic engineering by controlling the chemical composition17. It has been shown in ZnSxSe1-x with zinc blende or wurtzite structure that Eg varies gradually with the chemical composition18–31. Although lattice parameter in ZnSxSe1-x nearly follows the Vegard’s law18,19,21,24,28, it has been frequently observed that the dependence of Eg with the stoichiometry of the solid solution is not strictly linear19-23,25-31. A quadratic function expresses this dependence more precisely with a bowing parameter, b, in the 0.35-0.68 eV range19–23,26– 29. However, few authors have shown a linear relationship with the composition24,31. The ZnSxSe1-x solid solution has been generally synthesized by different thermal evaporation processes21–24,26–30. In some cases, the authors state that the solid solution was obtained with the zinc blende structure and in other cases with the wurtzite structure; however, a careful observation of results shown, especially the X-ray diffraction (XRD) diagrams, suggests that in many cases a mixture of both structures was in fact obtained24,26,27,29,30. Mechanochemical synthesis procedures are generally performed in high-energy ball-milling equipment, in which solid state reactions are induced by applying mechanical energy to a reactant mixture32. It has been reported that ZnS and ZnSe can be obtained by milling their respective Zn-S and Zn-Se elemental mixtures33–45, although other mechanochemical processes based on displacement reactions have also been informed46–48. The reactions of formation of ZnS and ZnSe are highly exothermic49 and fulfil the empirical Merzhanov's criterion for combustion processes (adiabatic temperature (Tad) > 1800 K). If the required energy to ignite a self-propagating reaction is provided by mechanical energy, not by heat as in the self-propagating high-temperature synthesis (SHS) process, the method is referred to as a mechanically induced self-sustaining reaction (MSR). In MSR, ignition occurs after a critical milling time called the ignition time (tig), which depends on the reactant system and the milling conditions50–52. The occurrence of an MSR process during milling Zn-S or Zn-Se mixtures has been reported in several works33–35,43,44. However, in other studies, the MSR effect has not been outlined36–42,45, although it cannot be excluded as the products were generally inspected after long milling times, after ignition was likely to have occurred. The formed ZnS or ZnSe is mainly obtained with the zinc blende structure, although the wurtzite phase and even a mixture of both phases has also been reported. The different milling times employed in the different studies, which extend from a few minutes to tens of hours, can be at the origin of these contradictory results concerning the resulted structure, since it has also been shown the occurrence of phase transformation phenomena driven by highenergy ball-milling. Unfortunately, there are also contradictory results in this issue. Whereas some authors have shown that milling induces the wurtzite-to-zinc blende phase transition in ZnS and ZnSe41,53, other authors have shown the opposite effect, that milling promotes the zinc blende-to- wurtzite phase transformation in ZnS54. In this work, the synthesis by MSR of the complete solid solution in the ZnS-ZnSe system was performed. MSR is a dry, time saving and waste-free method, which could be interesting for a subsequent semiconductor technology. The main goal of this study was to present an easy and adequate synthesis method that allows the tailoring of Eg by controlling the stoichiometry of the ZnSxSe1-x series. Moreover, we clarified the contradictory results found in the bibliography on which phases, the zinc blende or wurtzite, are obtained from mechanochemical processes. The quantification of both phases as a function of the chemical composition and the milling time after ignition was studied, always taking into account the possible changes in Eg. EXPERIMENTAL SECTION Zinc powder (99% pure, < 325 mesh, Strem Chemicals), sulfur powder (99.5% pure, < 325 mesh, Alfa Aesar), selenium powder (99.5% pure, < 325 mesh, Alfa Aesar) and high-purity argon gas (H2O and O2 < 3 ppm, Linde) were used in this work for the synthesis of ZnSxSe1-x solid solution by an MSR process using a modified planetary ball mill (Micro Mill Pulverisette 7, Fritsch). All of the milling experiments were conducted under an Ar atmosphere (6 bar) using 3 g of Zn/S/Se elemental mixtures with different atomic ratios according to the ZnSxSe1-x stoichiometry intended. Six AISI 420C stainless steel balls (d = 15 mm and m = 13.7 g), a 60 mL tempered steel vial (inner diameter = 45 mm) and a rotational speed of the supporting disk of 400 rpm were employed. Detection during milling of tig in MSR process was possible since the vial with a special lid was continuously connected to the gas cylinder by a rotating union (model 1005-163-038, Deublin) and a semi-rigid polyamide tube (Legris), and the Ar pressure was monitored by a pressure transducer (AKS, Danfoss) connected to a paperless recorder (Ecograph T RSG35, Endress + Hauser). At ignition, the heat released by the exothermic reaction of the ZnSxSe1-x formation provoked an instantaneous increase in the total pressure of the system that resulted in the appearance of a peak in the pressure-time record, from which tig was obtained. 3 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 50 60 70 80 ZnS tig(min) x tig(x) = x tigZnS + (1-x) tigZnSe - (59 ± 5) x (1-x) ZnSe Figure 1. Ignition time (tig) of the MSR process involving the milling of different Zn/S/Se elemental mixtures to obtain the ZnSxSe1-x solid solution. The x value corresponds to the [S/(S+Se)] atomic ratio in the starting mixture. XRD patterns of milled powders were obtained on a X'Pert Pro MPD diffractometer (PANalytical) equipped with a θ/θ goniometer, a graphite-diffracted beam monochromator and a solid-state detector (X'Cellerator). The diffraction patterns were acquired using Cu Kα radiation over a 2θ-range of 10° to 140° with a step size of 0.033° and a counting time of 800 s/step. Silicon powder (Standard Reference Material 640c, NIST) was used for calibration of the diffraction line positions. The quantification of phases detected in the XRD patterns was performed by means of the Rietveld method using the FULLPROF program55. A Thompson-Cox-Hastings pseudo-Voigt function convoluted with an axial divergence asymmetry function was chosen to generate the peak shapes. The following parameters were refined: background points, zero point, scale factor, pseudo-Voigt parameters of the peak shape and cell parameters. The following reference diffraction patterns from the PDF-4 + database of the International Centre for Diffraction Data (ICDD) were used as the initial model: ZnS (F-43m, a = b = c = 5.4060 Å; 05-0566), ZnSe (F-43m, a = b = c = 5.6688 Å; 37-1463), ZnS (P63mc, a = b = 3.8210, c =6.2573 Å; 36-1450) and ZnSe (P63mc, a = b = 3.9960, c = 6.5500 Å; 15-0105). Powder samples after milling were dispersed in ethanol, and droplets of the suspension were deposited onto a carbon coated copper grid for microscopy analysis. The scanning electron microscopy (SEM) images were obtained on a Hitachi S-4800 SEM-Field Emission Gun microscope. The transmission electron microscopy (TEM) and high resolution (HRTEM) images, electron diffraction patterns (EDP) and energy dispersive X-ray (EDX) spectra were taken on a 200 kV JEOL-2100-PLUS microscope equipped with a LaB6 filament (point resolution = 0.25 nm). The HR micrograph analysis, lattice spacing, Fast Fourier Transform (FFT) and phase interpretation were done with the Gatan Digital Micrograph software (Gatan Inc.) and the Java version of the Electron Microscope Software (JEM). Raman spectroscopy was performed using a dispersive Horiba Jobin Yvon HR800 confocal Raman Microscope (HORIBA) equipped with a chargecoupled device (CCD) detector at a laser excitation wavelength of 532 nm (green laser). The spectral resolution was 4 cm−1. The laser beam was focused on the powder samples with a confocal objective of 100×. The diffuse reflectance UV–Vis spectroscopy (DRS) of the powder samples was performed using a Cary 300 spectrophotometer from 200 to 800 nm. Eg value was calculated from the corresponding Kubelka–Munk functions, F(R∞), which are proportional to the absorption of radiation, by plotting (F(R∞) × hν)1/2 against hν. RESULTS AND DISCUSSION Effects of the ZnSxSe1-x stoichiometry on the crystal structure and the optical properties. Seven different Zn/S/Se powder mixtures with stoichiometric atomic ratios according to the ZnSxSe1- x solid solution with x = 1, 0.8, 0.6, 0.5, 0.4, 0.2 and 0 were submitted to milling using the experimental conditions described in the experimental section. For all mixtures, ignition was detected in the timepressure records (a sharp peak appeared) and the corresponding tig values determined are shown in Figure 1. For this set of samples, milling was prolonged 60 min after ignition. Figure 1 shows as tig was gradually shortened as the Se content in the reactant mixture increased. The tig variation with the x value, which corresponds to the [S/(S+Se)] atomic ratio in the starting Zn/S/Se mixture, was not linear, but a quadratic function defines more precisely this dependence. Note that the standard enthalpy of formation for ZnS (−205 kJ/mol) is greater than for ZnSe (−170.3 kJ/mol)56, however, in this case, it follows that the more exothermic reactant mixture, the longer tig. Although one would expect the opposite, more exothermic reactions ignite before, in metal-chalcogen systems, it has been shown that tig correlates better with the hardness of the reactants; it is shorter for softer metals and harder chalcogens57. 4 20 30 40 50 60 70 wurtzite ZnSxSe1-x zinc blende ZnSxSe1-x unreacted Zn ZnS ZnS0.8Se0.2 ZnS0.6Se0.4 ZnS0.5Se0.5 ZnS0.4Se0.6 2q (º) ZnSe ZnS0.2Se0.8 a) 25 26 27 28 29 30 31 32 ZnS ZnS0.8Se0.2 ZnS0.6Se0.4 ZnS05Se0.5 ZnS0.4Se0.6 2q (º) ZnSe ZnS0.2Se0.8 (100)W(002)W (111)ZB (101)W (100)W (002)W (111)ZB (101)W b) Figure 2. (a) XRD patterns of the different Zn/S/Se mixtures after the MSR process (milling time = tig + 60 min) that show the formation of the ZnSxSe1-x solid solution. (b) XRD patterns of the same samples in the 25º–32º 2q region where the (100), (002) and (101) reflections of the wurtzite phase and the (111) reflection of the zinc blende phase are observed. Figure 2a presents the XRD patterns of the products after the MSR process that confirmed the formation of the ZnSxSe1-x solid solution in all cases. The presence of a small amount of unreacted Zn was also observed. However, no unreacted S and/or Se were noticed, which may be the result of the capacity of high-energy ball milling to induce S and Se amorphization58,59. Figure 2 shows that a mixture of both the zinc blende and wurtzite structures was always obtained after the MSR process and that the proportion of the zinc blende structure increased with the Se content. The existence of both phases was evidenced even in samples with low or no Se content by the relative intensity of the central XRD peak in the triplet appearing in the 25-32º 2θ range (Figure 2b). It is the result of the overlapping of the (111) peak of the zinc blende structure, which is the most intense, with the (002) peak of the wurtzite structure. Note that the most intense XRD peak of the wurtzite structure corresponds to the (100) family planes. Figure 2a also shows a clear shift of the XRD peaks of the zinc blende and wurtzite phases to lower angle values as the Se content increased in the starting mixture, which correlates well to an increasingly richer Se stoichiometry in the ZnSxSe1-x solid solution, leading to an increase in the cell parameters. This is observed more clearly in Figure 2b, where the 25º–32º 2q region is enlarged. The broadening and symmetry of the XRD peaks, together with the continuous peak shifting, rule out the possibility of phase separation and suggest the formation of a homogeneous solid solution. To quantify all these different aspects, the Rietveld refinement of the XRD patterns were performed (Figure S1), from which the percentage of different phases and the lattice parameters of the solid solution were determined. The quantification of the zinc blende and wurtzite phases as determined by the Rietveld method is presented in Figure 3 and confirms the presence of both structures in all samples. Note that even the ZnS sample, with a characteristic XRD pattern similar to others found in the literature, from which the formation of ZnS with wurtzite structure is frequently established, has a 50 wt.% content of zinc blende. This amount increases progressively with the Se content in the solid solution. Remember that according 5 to the Zn-Se phase diagram10,60, the zinc blende structure is the only thermodynamically stable phase up to the melting point of ZnSe. Nevertheless, also for the ZnSe sample, a small amount of the wurtzite phase was observed (11.3 wt.%). The amount of unreacted Zn in samples was in the 2.4-3.1 wt.% range. 0.0 0.2 0.4 0.6 0.8 1.0 10 15 20 25 30 35 40 45 50 ZnS x wt.% wurtzite ZnSe 50 55 60 65 70 75 80 85 wt.% zinc blende Figure 3. Quantification of the zinc blende and wurtzite phases present in the ZnSxSe1-x solid solution obtained by MSR by Rietveld refinement of the XRD patterns in Figure 2. 0.0 0.2 0.4 0.6 0.8 1.0 155 160 165 170 175 180 185 190 ZnS x volume zinc blende structure (Å 3) ZnSe 76 78 80 82 84 86 88 90 92 volume wurtzite structure (Å 3) Figure 4. Unit cell volume of the zinc blende and wurtzite structures in the ZnSxSe1-x solid solution obtained by MSR. Lattice parameters for the zinc blende and wurtzite structures in the different samples were also refined and the results are shown in Table 1. The parameters for both structures in ZnS and ZnSe samples were in good agreement with those of the reference diffraction patterns of the ICDD database (see the experimental section). As expected, an increasing of the cell parameters with the Se content in the solid solution was observed. Figure 4 presents the variation of the unit cell volume of both structures for the different ZnSxSe1-x samples. The good linearity observed in both cases with the x value, i.e., the [S/(S+Se)] atomic ratio in the starting Zn/S/Se mixture, confirms the validity of the Vegard’s law in the ZnS-ZnSe system, as well as that the composition of the ZnSxSe1-x solid solution correlates well with the Zn/S/Se atomic ratio in the starting mixture. Figure 4 also points out that for all samples the stoichiometry of the solid solution was the same for both structures formed. TABLE 1. Lattice parameters of the wurtzite and zinc blende structures present in the ZnSxSe1-x solid solution obtained by MSR. Sample Lattice parameters Wurtzite Zinc blende a (Å) c (Å) a (Å) ZnS 3.8151 6.2577 5.4044 ZnS0.8Se0.2 3.8521 6.3166 5.4513 ZnS0.6Se0.4 3.8884 6.3752 5.5074 ZnS0.5Se0.5 3.9090 6.4081 5.5343 ZnS0.4Se0.6 3.9222 6.5008 5.5548 ZnS0.2Se0.8 3.9615 6.5394 5.6083 ZnSe 4.0002 6.5636 5.6638 Figure 5 exhibits representative microstructural results for selected samples, in particular, for ZnS (Fig. 5a), ZnS0.5Se0.5 (Fig. 5b) and ZnSe (Fig. 5c). In the SEM micrographs, it can be seen that the three samples are formed by agglomerated particles, which size ranged between 100 nm and 1µm. The larger particles present platelets shape and the smaller are rounded, that in many cases are deposited on the platelet surface. TEM images show the same type of morphology and the EDPs give information about the crystallinity of the samples. The ring-EDP obtained for the three samples (inset in the TEM images) was the evidence of a nanocrystalline character and, therefore, the particles observed by SEM are in fact constituted by nanocrystalline domains. The (h, k, l) planes corresponding to the rings (white marked) were indexed and belong to both crystal symmetry, the zinc blende (cubic, space group F-43m) and the wurtzite (hexagonal 2H, space group P63mc). This information is in good agreement with the XRD results. The nanocrystalline character of the three samples can also be confirmed in the HRTEM micrographs, in which small misoriented nanodomains are observed. Several nanodomains along the [0 0 1] zone axis of the wurtzite structure are white marked in the HRTEM images and the corresponding FFT and simulated EDP are included as insets. The chemical analysis of the samples was carried out by EDX in both, the SEM and the 6 Figure 5. Microstructural characterization (SEM, TEM, EDP, HRTEM and EDX) of (a) ZnS, (b) ZnS0.5Se0.5 and (c) ZnSe samples milled for 60 min after ignition. Small nanodomains oriented along the [0 0 1]2H zone axis are white marked and the corresponding FFT and simulated EDP are inset in the HRTEM images. [0 0 1] Hex-2H [0 0 1] Hex-2H S -At% = 51±3 Zn -At% = 49±3 Se -At% = 53±3 Zn -At% = 47±3 [0 0 1] Hex-2H S -At% = 25±3 Se -At% = 26±3 Zn -At% = 49±2 500 nm a) ZnS 500 nm b) ZnS0.5Se0.5 500 nm c) ZnSe (3 3 1)C (1 1 1)C (0 0 2)H (2 2 0)C (1 1 0)H (3 1 1)C (1 1 2)H (4 2 2)C (3 3 1)C (1 1 1)C (0 0 2)H (2 2 0)C (1 1 0)H (3 1 1)C (1 1 2)H (4 2 2)C (3 3 1)C (1 1 1)C (0 0 2)H (2 2 0)C (1 1 0)H (3 1 1)C (1 1 2)H SEM SEM SEM TEM TEM TEM EDX HRTEM HRTEM HRTEM EDX EDX ED ED ED 7 TEM microscopes, finding similar results. The average values shown in the EDX spectra presented for the three samples in Figure 5 were in agreement with the expected composition and the atomic ratio in the starting Zn/S/Se mixtures. 200 250 300 350 400 ZnSe-TO ZnSe-LO normalized intensity Raman shift (cm-1) ZnS ZnS0.8Se0.2 ZnS0.6Se0.4 ZnS05Se0.5 ZnS0.4Se0.6 ZnSe ZnS0.2Se0.8 ZnS-LO Figure 6. Raman spectra of the different ZnSxSe1-x samples obtained by MSR. Figure 6 displays the Raman spectra of the different samples, in which the composition-dependent Raman scattering is clearly observed. The Raman spectrum of the ZnS sample is dominated in the 200- 400 cm-1 region by a main peak located at ~ 346 cm-1 that corresponds to longitudinal optical (LO) modes in both wurtzite and zinc blende structures61. The ZnSe spectrum in Figure 6 is characterized by longitudinal optical (LO) and transverse optical (TO) modes at ~ 244.5 and 200 cm-1, respectively62. For the ZnSxSe1-x samples containing S and Se, the Raman spectra present both ZnS- and ZnSe-like optical modes. Moreover, a shift in the optical phonon frequencies was observed depending on the stoichiometry of the solid solution. As the S content decreases (or the Se content increases), the ZnS-like LO mode reduces continuously its frequency, while the ZnSelike LO mode shifts to higher frequencies. However, the ZnSe-like TO mode shifts to lower frequencies, which causes both ZnSe-like modes to increasingly overlap as the S content increases in the solid solution. Note that the relative intensity of the ZnS- and ZnSe-like modes correlates well with the stoichiometry of the solid solution. Similar results have been found in the bibliography for ZnSxSe1-x obtained by evaporation methods23,27,29 and confirms that the MSR process allows obtaining homogeneous ZnSxSe1-x solid solutions. 300 400 500 600 700 800 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 normalized absorbance wavelength, l (nm) ZnS ZnS0.8Se0.2 ZnS0.6Se0.4 ZnS0.5Se0.5 ZnS0.4Se0.6 ZnS0.2Se0.8 ZnSe a) 0.0 0.2 0.4 0.6 0.8 1.0 2.6 2.8 3.0 3.2 3.4 3.6 Eg(x) = x EgZnS + (1-x) EgZnSe - 0.34 x (1-x) ZnS ZnSe band gap, Eg (eV) x b) Figure 7. (a) DRS spectra and (b) variation of Eg as a function of the stoichiometry of the ZnSxSe1-x solid solution obtained by MSR. Figure 7a shows the DRS spectra of the different ZnSxSe1-x samples in the 250-800 nm range. It is clearly visible that the absorption edge presents a continuous redshift (towards longer wavelengths) with increase Se content, which implies a decrease in Eg as observed in Figure 7b. The Eg values for ZnS and ZnSe samples coincide with the experimental and 8 20 30 40 50 60 70 wurtzite ZnS0.5Se0.5 zinc blende ZnS0.5Se0.5 unreacted Zn 2q (º) tig + 0 min tig + 10 min tig + 30 min tig + 60 min tig + 120 min tig + 480 min tig + 240 min a) 25 26 27 28 29 30 31 32 (111)ZB 2q (º) tig + 0 min tig + 10 min tig + 30 min tig + 60 min tig + 120 min tig + 480 min tig + 240 min (100)W(002)W(101)W b) Figure 8. (a) XRD patterns of the Zn/S/Se mixture corresponding to ZnS0.5Se0.5 subjected at increasing milling times after ignition. (b) XRD patterns of the same samples in the 25º–32º 2q region where the (100), (002) and (101) reflections of the wurtzite phase and the (111) reflection of the zinc blende phase are observed. theoretical values for bulk ZnS and ZnSe16. Figure 7b shows as it is possible to tune Eg in the ZnS-ZnSe system by adjusting the stoichiometry of the ZnSxSe1-x solid solution that it is accurately controlled by the Zn/S/Se atomic ratio in the starting mixture submitted to milling. The evolution of Eg with the composition value x in the solid solution was not strictly line- ar, as frequently observed in the bibliography, and a bowing parameter b of 0.34 eV was obtained. This low value suggests the absence of important compositional fluctuations, as well as short-range order in the solid solution25, confirming again a homogeneous distribution of S and Se in the crystal structure. Effects of the post-ignition milling time on the crystal structure and the optical properties. In this second set of milling experiments, the influence of the milling time on the structure and microstructure of the ZnSxSe1-x solid solution, and how the optical properties were modified, was studied. The Zn/S/Se powder mixture with atomic ratios according to the ZnS0.5Se0.5 stoichiometry was chosen as a model sample. Figure 8 shows the XRD patterns for these samples subjected to increasing milling times after ignition (post-ignition milling time). The first experiment corresponds exactly to a milling time equivalent to tig, i.e., the milling process was stopped exactly at ignition with no further milling. The others experiments correspond to samples, for which the milling was prolonged after ignition for 10, 30, 60, 120, 240 and 480 min, respectively. Figure 8 clearly shows how the ZnS0.5Se0.5 solid solution formed immediately after ignition was obtained practically only with the wurtzite structure, as illustrated in Figure 9, where the results of the quantification by Rietveld (Figure S2) are presented. It is well known that extremely high temperatures are reached locally at ignition, which could favor the formation of the hightemperature structure. As heat losses between the milling media and the environment are important in milling processes, a quench-like phenomenon could have occurred and the wurtzite structure was stabilized at room temperature once the milling was stopped. By extending the post-ignition milling time, the wurtzite structure was continuously transformed into the zinc blende structure and the full transformation 9 TABLE 2. Lattice parameters, crystalline domain size (D) and lattice strain (e) of the wurtzite and zinc blende structures present in the ZnS0.5Se0.5 sample obtained by MSR as a function of the post-ignition milling time. Post-ignition milling time (min) Wurtzite Zinc blende a (Å) c (Å) D (nm) e x 10-3(%) a (Å) D (nm) e x 10-3(%) 0 3.9079 6.4048 55.3 0.29 5.5212 58.3 0.30 10 3.9089 6.4061 46.1 0.37 5.5285 37.9 0.41 30 3.9045 6.4043 30.0 0.52 5.5291 40.0 0.42 60 3.9090 6.4081 16.8 0.89 5.5343 38.5 0.43 120 3.9011 6.4220 17.3 0.86 5.5335 17.9 0.84 240 3.8790 6.4430 7.6 1.94 5.5306 11.5 1.29 480 3.8822 6.4448 6.9 2.27 5.5382 8.9 1.65 was achieved practically after 480 min (Figures 8 and 9). As the stoichiometry was always the same in this set of experiments, no significant XRD peak shift was detected. Only the phase transformation was observed, without modification of the chemical composition of the solid solution. This fact was confirmed by the constancy of the lattice parameters determined for both structures with milling time (Table 2). Note how the XRD peak broadening was increased with milling time as a result of the expected decrease and increase in particle size and structural defects, respectively (Table 2). Also note that a similar trend was observed for all samples in the ZnSxSe1-x solid solution. Although at tig the amount of wurtzite phase formed was dependent on the x value, if the milling was prolonged, the wurtzite phase was continuously transformed into zinc blende. 0100 200 300 400 500 0 20 40 60 80 100 milling time after ignition (min) wt.% wurtzite 0 20 40 60 80 100 wt.% zinc blende Figure 9. Quantification of the zinc blende and wurtzite phases present in the ZnS0.5Se0.5 sample subjected at increasing post-ignition milling time by Rietveld refinement of the XRD patterns in Figure 8. Figure 10 exhibits representative SEM/TEM results for the samples stopped at ignition (ZnS0.5Se0.5-0) and milled for 480 min after ignition (ZnS0.5Se0.5-480). The difference between both samples can be clearly seen in the SEM micrographs. Well-defined hexagonal prisms were formed in the sample ZnS0.5Se0.5-0 with sizes ranging between 0.2-1 µm in the base and 0.5-2 µm in height. Spherical particles (0.2-1 µm of diameter) are also observed, and they are formed in turn by smaller particles, as can be seen in the enlargement that is presented in the inset below to the right. Note that this sample was not subjected to milling after ignition and the morphology observed was the result of the self-sustaining reaction. On the other hand, the sample ZnS0.5Se0.5-480, presents the typical appearance of powder samples obtained by mechanical milling, formed by an agglomeration of small rounded particles. The differences between samples were also observed in the TEM images and EDP. The ZnS0.5Se0.5- 480 sample shows ring-EDPs with diffuse scattering (corresponding (h, k, l) are marked), which implies that the sample was nanocrystalline. However, in the ZnS0.5Se0.5-0 sample some microcrystallites were detected with characteristic spot-EDPs. A representative one is presented in the up-left corner, which is oriented along the [621]2H axis. Ring-EDPs were as well found, but formed in this case by many diffraction spots (indexed in the EDP), in agreement with larger crystalline domains. The HRTEM micrograph show a large crystalline domain oriented along the [0 0 1]2H zone axis of the hexagonal structure and the corresponding FFT and simulated EDP are inset (Fig.10a). Taking into account that these two samples are the ends of the experiment analyzing the influence of the post-ignition milling time, it is worth to compare them with the sample ZnS0.5Se0.5 after 60 min of