Open Access
Add to BibSonomyAbstract
Polycrystalline ZnS taken from an industrial chemical vapor deposition (CVD) process was thermally treated with and without hot isostatic pressing (HIP). Both annealed materials show an increased light transmittance in the visible range, a recrystallized and reorganized microstructure as well as an absence of the Zn-H bonds observed in CVD-ZnS. However, the increased light transmission essential for most optical applications is only observed in the HIP-ZnS. This is caused by a significantly reduced porosity of the material. The majority of pores initially form at grain boundaries during thermal treatment and the application of simultaneous high pressure reduces their number to a minimum. X-ray diffraction (XRD), scanning electron microscopy (SEM) including electron back scatter diffraction (EBSD) as well spectroscopic measurements and optical microscopy are used for analysis. It is concluded, that the density of crystal faults in CVD-ZnS is significantly reduced by the HIP-step.
© 2014 Optical Society of America
1. Introduction
Most materials used in optical or photonic technologies are either glasses or single crystals. The preparation of the latter in the most cases requires high technical expenditures making these materials comparably expensive. On the other hand, not all the desired properties can be achieved using glasses. In principle, these problems can be overcome by using polycrystalline materials, although light scattering is still a problem for photonic applications. So far, only few polycrystalline materials show a sufficient transparency to be used in optical components.
The cubic modification of ZnS has a zincblende structure with a large band gap of 3.72 eV [1] and “a nearly free electron band structure in a crystal” [2]. It has a high transparency in a wavelength range from 380 nm to 10.5 µm [3]. Industrial polycrystalline ZnS is generally produced by a CVD-process which enables the preparation of bulk materials with a thickness of 30 mm and more. This process differs notably from those used to obtain thin CVD-ZnS films. Bulk CVD-ZnS is grown from H2S-gas and zinc vapor and some hundred kilograms of product are produced in large reaction chambers with a volume of more than 1 m3 within a time of more than one week. In order to obtain optical grade ZnS, the produced material must subsequently be treated by a thermal process carried out under high temperature and elevated isostatic pressure (HIP). This treatment leads to a change in the microstructure and to a notable increase in the transparency, especially in the visible range [3–6]. Experiments on CVD-ZnS annealed without elevated pressure have also been reported [5–7], although the resulting material is of insufficient quality for optical applications. Here pores may play an important role [7]. It was shown that the Zn-H bonds detected in CVD-ZnS [3–6] are neither detected in annealed CVD-ZnS [5] nor in HIP-ZnS [3–6]. Annealing CVD-ZnS at elevated temperature in a vacuum leads to the evaporation of the material [3].
The recent EBSD-analysis of the microstructure of untreated CVD-ZnS [8,9] showed well pronounced textures at different stages of deposition [8] and two layers of vastly different microstructure [8] in which many grains are heavily twinned [3–6,8,9]. In the bulk of the material, the grains are fairly small, show a 001-texture and internal stresses are observed [8]. The hexagonal wurtzite modification of ZnS was also shown to occur adjacent to the substrate and the immediate growth front, most probably due to the dynamic conditions during process initiation and termination [9].
As it is well known that the key to improving the properties of polycrystalline materials often lies in the microstructure, the focus of this article will be systematically analyzing and comparing the microstructure of the three ZnS types discussed for industrial production. We describe changes of the grain structure and texture as well as in the optical transmission of the respective bulk materials.
2. Experimental setup
Polycrystalline ZnS was produced via CVD from zinc vapor and H2S-gas at about 650 °C to 700 °C in an industrial production process designed to produce ZnS layers with a thickness up to 35 mm with graphite as the substrate material (Vitron GmbH, Jena, Germany). Post production treatment was performed by hot isostatic pressing at about 1000 °C and about 100 MPa for more than 24 h and annealing CVD-ZnS for the same time at 850 °C in an Ar-atmosphere of normal pressure. The HIP-parameters are preset by the industrial process and cannot be changed easily. Because the sublimation of ZnS increases with increasing temperatures, but does not play any role during the HIP-step due to the high pressure, it was decided to use 850°C during the annealing treatment at atmospheric pressure as a practicable compromise.
The samples were studied by X-ray diffraction (XRD) using a SIEMENS D5000 diffractometer and CuKα radiation. Scanning electron microscopy (SEM) and electron backscatter diffraction (EBSD) were performed using a Jeol JSM-7001F equipped with a TSL Digiview 3 EBSD-camera. Surface charging in the SEM sample was avoided by contacting the samples with Ag-paste and applying a thin layer of carbon at a pressure of about 10−3 Pa.
EBSD-scans were captured and evaluated using the programs TSL OIM Data Collection 5.31 and TSL OIM Analysis 5.31. The EBSD-scans were performed using a voltage of 20 kV. For the calculation of pole figures (PF) only reliably indexed data points with a confidence index (CI) ≥ 0.1 were considered [10]. All PFs presented in this article were calculated using the stereographic projection.
Ultraviolet-visible range-near Infrared spectra (UV-VIS-NIR) were recorded using a Shimadzu 3101 PC spectrometer. Midwave infrared spectra (MIR) were obtained by a Shimadzu IR-Affinity-1 FTIR spectrometer. Optical microscopy was performed using an Axio Imager Z1M LSM5-Pascal (Carl Zeiss AG, Oberkochen, Germany).
3. Results and discussion
Three XRD-patterns obtained from ZnS samples cut parallel to the substrate plane in the bulk of the material are presented in Fig. 1.The graphs (a), (b) and (c) present the respective patterns of CVD-ZnS, annealed ZnS and HIP-ZnS. Graph (d) presents a pattern of statistically oriented zincblende for comparison. All patterns were normalized to the 111-peak. The peak ratio is given by dividing the intensity of the peak in the normalized pattern by the intensity of the normalized theoretical intensity of the same peak. The low structure factor of the {002}-lattice planes leads to lower intensities in the theoretical pattern (d). The peak ratios in pattern (a) are 4.92 for the 001-peak, 0.33 for the 022-peak and 2.83 for the 113-peak, i.e. the 001-peak clearly dominates but the 113 is enlarged as well, indicating a 001-texture parallel to the former substrate plane in CVD-ZnS.
Fig. 1 XRD-patterns obtained from compact ZnS samples cut parallel to the substrate plane: (a) CVD-ZnS, (b) annealed ZnS and (c) HIP-ZnS. The intensities for statistically oriented zincblende are plotted as graph (d) (JCPDS No. 005-0566) for comparison.
In the case of the specific grain structure described for CVD-ZnS [8], it is possible to explain the elevated intensity of the 113-peak as part of the 001-texture: there are three non-equivalent 001-planes and twelve 113-planes in the fcc unit cell. Four 113-planes exist at an angle of only 25.2° for every 001-plane. Taking into account the grain fragmentation described in CVD-ZnS [8], a continuous orientation change of up to 28° within 12 µm of a single grain may occur. As the information depth of XRD extends for tens of µm into the sample, this enables parts of a single grain to contribute to the XRD-pattern with one 001-plane as well as with a 113-plane. This does not apply to the other lattice planes contributing to the XRD-data because their angle to the 001-plane is much larger.
The XRD-patterns (b) and (c) are very different: the 002-peak is no longer discernible and the 111-peak has the largest intensity. Hence, XRD indicates the 001-texture of CVD-ZnS transformed to a 111-texture during annealing and the HIP-step in agreement with the literature [3–6].
Figures 2 , Fig. 3, and Fig. 4 feature cut planes parallel and perpendicular to the graphite substrate through the bulk of the respective materials. They show the combined inverse pole figure (IPF) and image quality (IQ) maps of performed EBSD-scans.
Fig. 2 IPF + IQ-maps of EBSD-scans performed on CVD-ZnS samples cut parallel and perpendicular to the substrate plane.
Fig. 3 IPF + IQ-maps of EBSD-scans performed on annealed ZnS samples cut parallel and perpendicular to the substrate plane.
Fig. 4 IPF + IQ-maps of EBSD-scans performed on HIP-ZnS samples cut parallel and perpendicular to the substrate plane.
Figure 2 shows results of a CVD-ZnS sample at a distance of approximately 27 mm from the substrate. The grains in the cut plane parallel to the substrate do not show a systematic anisotropy and may reach ca. 20 µm in diameter. The grains in the cut plane perpendicular to the substrate appear systematically elongated parallel to the growth direction and extend more than 25 µm in length in the presented cut plane. Previous studies have reported that many of these grains show lamellar twins like those marked by the arrows in Fig. 2 [3,8,9]. The black data points and areas did not allow reliable EBSD-pattern indexing. For single points this may be due to superimposed patterns at grain boundaries and twin boundaries while areas often represent pores, contamination particles on the surface or places where material was torn out during the polishing process. Hence statements towards porosity based only on EBSD-maps are not reasonable.
Comparable results of annealed ZnS are presented in Fig. 3 and show that the microstructure has changed significantly: large crystals occur, some of them with a size of more than 300 µm. The grains do not show an obvious anisotropy in either of the cut planes. Twinning is observed, although the lamellar twin systems of CVD-ZnS are no longer detected. Instead, the twins in annealed-ZnS are very large (tens of µm thick) and do not intersect the whole grain as has been shown to occur for the nm-thick twins in CVD-ZnS [3,8,9]. Unreliably indexed areas occur more frequently than single points in comparison to Fig. 2.
The corresponding results of HIP-ZnS are presented in Fig. 4. While the microstructure is comparable to that of the annealed ZnS featured in Fig. 3, unreliably indexed areas are rarely observed in these optical grade samples.
Additionally, all four possible twin orientations of a grain lattice can often be identified within one grain in annealed ZnS and HIP-ZnS. This is illustrated in detail in Fig. 5, where in (a) the IPF + IQ-map of a detailed EBSD-scan performed on a HIP-ZnS sample cut parallel to the substrate plane is presented. The PFs 1-5 of the orientations at the locations 1-5 are presented in (b) to illustrate the orientation of the main grain 1 and the respective twins 2-5. Hence all four possible orientations of Σ3-twinning are observed within one grain.
Fig. 5 (a) IPF + IQ-map of a detailed EBSD-scan performed on a HIP-ZnS sample cut parallel to the substrate plane. (b) The image shows the main orientation of a ZnS-grain in 1: purple and the four possible twin orientations (in 2: red, 3: yellow, 4: blue and 5: violet) within one grain.
Pole Figures (PFs) of textures calculated from the entire EBSD-scans featured in Figs. 2-4 (only parts of the scans were presented in the figures to enhance clarity) are presented in Fig. 6.The PFs of CVD-ZnS show a well pronounced texture describing that a 001-plane of the grains is preferably oriented parallel to the substrate plane. For an ideal 001-orientation, the 001-PF should have a maximum in the centre and a ring at 90° from the center pole, whereas the 111-pole figure should have a ring at 54.7°. The 111-PFs of annealed ZnS and HIP-ZnS show a high probability in the centre and a corresponding ring with an angular distance of approximately 70° due to a statistical rotation of the grain orientations around the aligned <111>-direction. The 001-PFs are also presented and show a ring at about 55° in contrast to the 001-PF of CVD-ZnS.
Fig. 6 001- and 111-PFs of textures calculated from EBSD-scans performed on the respective materials. The schematic 001- and 111-PFs of the HIP-ZnS texture are presented to ease interpretation.
The PFs of textures calculated from the EBSD-scans performed on cut planes of HIP-ZnS cut perpendicular to the substrate are also presented in Fig. 6 and confirm the texture information acquired from the cut planes parallel to the substrate. The schematic 111-pole figures for such a texture show maxima at the top and bottom and two semicircles at an angular distance of 70.5° to the nearest maximum. The number of grains in the scans is not high enough to result in an ideal ring in the calculated PFs. The textures of annealed and HIP-ZnS describe a preferred orientation of the fcc zincblende unit cells with a <111>-direction parallel to the initial growth direction of CVD-ZnS, where a <001>-direction is comparably aligned. XRD-results, indicating the described texture change, have previously been published [3,5,6].
Spectroscopic measurements were performed in order to analyze the effect of annealing and the HIP-step on the optical properties of the material. Figure 7 presents FTIR transmission spectra of the three materials which all show sharp absorption edges at about 900 cm−1 but differ significantly at larger wavenumbers. The absorption edges at 900 cm−1, a wavelength of about 11 µm, are probably due to the Zn-S stretching vibrations [11]. The absorption peak at 1,667 cm−1 occurs due to Zn-H bonds [3] and is neither observed for annealed ZnS nor HIP-ZnS. Besides the absorption due to this specific vibration, the background of the transmission decreases steadily with higher wavenumbers from a maximum of 75% at 1,000 cm−1 to 66% at 3,333 cm−1. These bonds are formed as lattice defects during the CVD-process because H2S is used as one reactant in the CVD-process. They disappear from the material at elevated temperatures in agreement with the literature [5,6]. The transmission of annealed ZnS decreases steadily from 71% at 1,000 cm−1 to 51% at 3,333 cm−1. HIP-ZnS shows an almost wavenumber independent transmission of more than 71% from 3,333 to 1,000 cm−1.
The UV-VIS transmission spectra of the samples described above are shown Fig. 8. The spectra of annealed ZnS and HIP-ZnS show an UV cut off at approximately 340 nm. The detailed inset highlights that CVD-ZnS already loses its transmittance at 380 nm, i.e. before the material exposed to higher temperatures. The transmission of HIP-ZnS is generally much higher, especially in the visible range where it decreases steadily from 71% at 785 nm to 50% at 385 nm. The transmission of CVD-ZnS and annealed ZnS decreases continuously with decreasing wavelengths. However, the spectra show that annealing CVD-ZnS increases the transmission from 380 nm to 740 nm but decreases the transmittance above 740 nm.
The results presented above show that the high transmittance of HIP-ZnS compared to CVD-ZnS and annealed ZnS is not a result of the recrystallization alone, nor is it mainly caused by Zn-H bonds. The optical micrographs presented in Fig. 9 allow an integral impression of the porosity in the respective materials. Figure 9(a) shows the optical micrograph of the CVD-material with some distributed pores which could not be localized to grains or grain boundaries, because these are too small to contrast using optical microscopy. Annealed ZnS shows more and larger pores in Fig. 9(b). Figure 9(c) was obtained from an incomplete HIP-process to illustrate that the pores in the material initially form and accumulate at the grain boundaries before they leave the microstructure. The pores appear to be elongated along the grain boundary direction. The grain structure of the material cannot be discerned after the complete HIP-step and the number of pores is reduced dramatically. Some pores may be found in a detailed analysis of the optical grade HIP-ZnS featured Fig. 9(d).
Fig. 9 Optical transmission micrographs of (a) CVD-ZnS and (b) annealed ZnS (50 h at 850 °C in an Ar atmosphere). Micrographs of HIP-ZnS are presented after (c) an incomplete and (d) a complete HIP-procedure. Pores appear black.
The spectra of the HIP-ZnS presented here are in agreement with the spectra presented in [3], McCloy et al., where it is stated that pores were not observed and should not occur because the spectra show no indication of Rayleigh scattering. Hence it may be concluded, that the pores are too few to significantly affect the optical properties of the material. Otherwise the results presented here are in agreement with the literature regarding CVD- and HIP-ZnS [4] as well as annealed ZnS [5,6].
The formation of pores may result from the recrystallization during which the healing of dislocations and lattice mismatches, e.g. in the form of small angle grain boundaries, should lead to the formation and agglomeration of free volume (e.g. pores) at grain boundaries. This supports the still unproven concept described in [12], Humphreys et al., which suggests that “…vacancies, swept up by a moving grain boundary, will increase the free volume of the grain boundary…” meaning they can agglomerate. This is supported by the fact, that the pores in annealed ZnS or ZnS from the incomplete HIP-step are larger than those in CVD-ZnS.
The transparency of the material increases notably during HIPping. The transmission spectra of HIP-ZnS show a high transmittance of more than 71% in the wavelength range from 785 nm to 10 µm. This runs parallel to a strong coarsening of the microstructure, a change in the texture and the disappearance of Zn-H bonds. In a cubic lattice, the refractive index of the crystal does not depend on its orientation and hence coarsened microstructures do not necessarily give rise to light scattering in polycrystalline materials. Hence, a coarsening of the microstructure in the HIP-ZnS is not in contradiction with lower scattering losses. The reason of high scattering in the CVD-ZnS should hence not be the size of the ZnS grains, but of regions which show refractive indices different from ZnS. Here pores and frequently occurring twin boundaries which may function as optical phase boundaries are the most likely contributors.
4. Summary
Industrially produced polycrystalline CVD-ZnS was thermally treated with and without isostatic pressure. This recrystallization resulted in a notable coarsening of the grain structure and a change in the grain orientation as well as the texture of the material. While a 001-texture was observed perpendicular to the surface in CVD-ZnS, a 111-texture was observed perpendicular to the surface in annealed ZnS and HIP-ZnS. The lamellar twinning of CVD-ZnS disappears during recrystallization and all possible Σ3-twins occur in many grains. Neither the grain fragmentation nor the Zn-H bonds detected in CVD-ZnS are observed after recrystallization. However, only HIP-ZnS shows the notable increase of the light transmission necessary for the application of the material as the elevated pressure during annealing reduces the porosity of the material to an insignificant amount. The density of crystal faults in CVD-ZnS is significantly reduced by the HIP-step.
Acknowledgment
This work was funded by the Bundesministerium für Wirtschaft, BMWI (ZIM program KF2519702FK1).
References and links
1. T. K. Tran, W. Park, W. Tong, M. M. Kyi, B. K. Wagner, and C. J. Summers, “Photoluminescence Properties of ZnS Epilayers,” J. Appl. Phys. 81(6), 2803–2809 (1997). [CrossRef]
2. P. Y. Yu and M. Cardona, Fundamentals of Semiconductors: Physics and Materials Properties (Springer, 2005).
3. J. S. McCloy, R. Korenstein, and B. Zelinski, “Effects of temperature, pressure, and metal promoter on the recrystallized structure and optical transmission of chemical vapor deposited zinc sulfide,” J. Am. Soc. 92(8), 1725–1731 (2009).
4. A. F. Shchurov, E. M. Gavrishchuk, V. B. Ikonnikov, E. V. Yashina, A. N. Sysoev, and D. N. Shevarenkov, “Effect of hot isostatic pressing on the elastic and optical properties of polycrystalline CVD ZnS,” Inorg. Mater. 40(4), 336–339 (2004). [CrossRef]
5. P. Ramavath, P. Biswas, R. S. Kumar, V. Mahendar, G. V. N. Rao, U. S. Hareesh, R. Johnson, and N. E. Prasad, “Effect of sphalerite to wurtzite crystallographic transformation on microstructure, optical and mechanical properties of zinc sulphide ceramics,” Ceram. Int. 37(3), 1039–1046 (2011). [CrossRef]
6. P. Biswas, R. S. Kumar, P. Ramavath, V. Mahendar, G. V. N. Rao, U. S. Hareesh, and R. Johnson, “Effect of post-CVD thermal treatments on crystallographic orientation, microstructure, mechanical and optical properties of ZnS,” Ceramics J. Alloy. Compd. 496(1–2), 273–277 (2010). [CrossRef]
7. V. I. Bredikhin, E. M. Gavrishchuk, V. B. Ikonnikov, E. V. Karaksina, L. A. Ketkova, S. P. Kuznetsov, and O. A. Mal’shakova, “Optical losses in polycrystalline CVD ZnS,” Inorg. Mater. 45(3), 235–241 (2009). [CrossRef]
8. T. Zscheckel, W. Wisniewski, and C. Rüssel, “Microstructure and texture of polycrystalline CVD-ZnS analyzed via EBSD,” Adv. Funct. Mater. 22(23), 4969–4974 (2012). [CrossRef]
9. T. Zscheckel, W. Wisniewski, A. Gebhardt, and C. Rüssel, “Mechanisms counteracting the growth of large grains in industrial zns grown by chemical vapor deposition,” ACS Appl. Mater. Interfaces 6(1), 394–400 (2014). [CrossRef] [PubMed]
10. D. P. Field, “Recent advances in the application of orientation imaging,” Ultramicroscopy 67(1–4), 1–9 (1997). [CrossRef]
11. C. A. Klein and R. N. Donadio, “Infrared active phonons ins cubic Zinc sulfide,” J. Appl. Phys. 51(1), 797–800 (1980). [CrossRef]
12. F. J. Humphreys and M. Hatherly, Recrystallization and Related Annealing Phenomena (Elsevier, 2004).
