Structural, Electrical and Optical Properties of

0 downloads 0 Views 2MB Size Report
May 15, 2018 - we studied the growth of InN films using a ZnO buffer layer on different ... bulk GaN, sapphire and quartz substrates have been studied in this ...
Journal of ELECTRONIC MATERIALS

https://doi.org/10.1007/s11664-018-6386-3 Ó 2018 The Minerals, Metals & Materials Society

Structural, Electrical and Optical Properties of Sputtered-Grown InN Films on ZnO Buffered Silicon, Bulk GaN, Quartz and Sapphire Substrates UMAR BASHIR,1,4 ZAINURIAH HASSAN,2 NASER M. AHMED,1 and NAVEED AFZAL3 1.—School of Physics, Universiti Sains Malaysia, 11800 Penang, Malaysia. 2.—Institute of Nano Optoelectronic Research and Technology (INOR), Universiti Sains Malaysia, 11800 Penang, Malaysia. 3.—Centre for Advanced Studies in Physics, GC University, Lahore, Pakistan. 4.—e-mail: [email protected]

Indium nitride (InN) films were grown on Si (111), bulk GaN, quartz and sapphire substrates by radio frequency magnetron sputtering. Prior to the film deposition, a zinc oxide (ZnO) buffer layer was deposited on all the substrates. The x-ray diffraction patterns of InN films on ZnO-buffered substrates indicated c-plane-oriented films whereas the Raman spectroscopy results indicated A1 (LO) and E2 (high) modes of InN on all the substrates. The crystalline quality of InN was found to be better on sapphire and quartz than on the other substrates. The surface roughness of InN was studied using an atomic force microscope. The results indicated higher surface roughness of the film on sapphire as compared to the others; however, roughness of the film was lower than 8 nm on all the substrates. The electrical properties indicated higher electron mobility of InN (20.20 cm2/Vs) on bulk GaN than on the other substrates. The optical band gap of InN film was more than 2 eV in all the cases and was attributed to high carrier concentration in the film. Key words: InN film, RF sputtering, ZnO buffer layer, different substrates, III–V nitrides

INTRODUCTION Indium nitride (InN) is an emerging and promising material in semiconductor industry with a large number of applications in the optoelectronics field.1 InN exhibits the smallest band gap (0.7 eV) in the III–V nitride family, high electron mobility and saturation velocity and small electron effective mass.2,3 InN exists in cubic zinc blende and hexagonal wurtzite phases. Among these, the hexagonal wurtzite structure is thermodynamically more stable. However, InN film growth is difficult due to its low dissociation temperature (500°C) and weak In–N bond energy (1.93 eV).4,5 The growth methods used to produce InN films include molecular beam

(Received September 21, 2017; accepted May 15, 2018)

epitaxy (MBE),6 metal organic chemical vapor deposition (MOCVD),7 radio frequency (RF) magnetron sputtering,8 high-pressure chemical vapor deposition (HPCVD)9 and plasma-assisted atomic layer epitaxy (PA-ALE).10 Apart from the above-mentioned difficulties, another difficulty faced by the InN-based devices is the lack of matched lattice substrate. To overcome this problem, several techniques have been used which include nitridation of the substrate surface, etching of substrate and induction of buffer layers.11–14 Since there is a lack of InN bulk substrates, a proper foreign substrate is needed that could ensure high-quality growth of InN films. The commonly used substrates for InN epitaxy with respective lattice mismatch (LM) are, silicon (LM 8%), sapphire (LM 25%), gallium nitride (LM 11%) and silicon carbide (LM 15%).3,15 Previous studies showed the use of different buffer layers to

Bashir, Hassan, Ahmed, and Afzal

accommodate the LM. For instance, AlN and GaN buffer layers were used for InN growth on the sapphire substrates using MBE growth technique.16,17 Laskar et al.18 used InN, AlN and GaN buffer layers on sapphire substrates grown by MOVPE. In our previous studies,19,20 we used a low-temperature InN buffer layer and Cu-ZnO compound buffer layer on silicon, GaN, sapphire and quartz substrates using RF magnetron sputtering. In all the cases, an improved quality of InN film was achieved after the application of these buffer layers. A ZnO buffer layer was also used by few researchers on sapphire and ZnO substrates using different growth techniques.1,15,21 Since InN has a hexagonal wurtzite structure (a ¼ 3:53 and ˚ ˚ c ¼ 5:70 A), ZnO (a ¼ 3:24 and c ¼ 5:20 A), which has a LM of 8.9%, was suggested as a suitable substrate/buffer layer for the InN film growth.22–24 In previous work, ZnO buffer layers were used on sapphire substrates for InN epitaxy. In this work, we studied the growth of InN films using a ZnO buffer layer on different substrates. The aim of this work is to compare the growth of InN on various ZnO-buffered substrates. The structural, optical and electrical properties of InN films on silicon, bulk GaN, sapphire and quartz substrates have been studied in this paper under identical conditions. EXPERIMENTAL WORK In this work, a ZnO buffer layer was deposited prior to the deposition of InN film on silicon, bulk GaN, sapphire and quartz substrates. The samples were first cut into 1 9 1 cm2 and then cleaned. Details of substrate cleaning is given in our previous work.19,20 After cleaning, the samples were loaded into the RF magnetron sputtering chamber (Model APX). A ZnO target was loaded inside the sputtering chamber for the deposition of ZnO buffer layer. The growth parameters for ZnO buffer layer are given in Table I. The growth of ZnO buffer layer was carried out in argon atmosphere, where a mass flow controller controlled and regulated the argon gas flow inside the chamber. After successful deposition of the ZnO buffer layer, the ZnO target was taken out and the indium target was loaded inside

the chamber. The fabrication of InN film on different ZnO-buffered substrates (Si, GaN, quartz and sapphire) was carried out in a mixture of argon and nitrogen ambient. The growth parameters for InN film are given in Table II. The deposited films were characterized by using various techniques. Structural analysis of the deposited films was carried out using PANanalytical X’pert x-ray diffractometer and Raman spectrometer (Jobin–Yvon HR 800 UV Ar+ laser). The surface roughness analysis was made using the atomic force microscope (AFM, model: Dimension Edge, Bruker). Hall effect measurements (model: Lakeshore Controller 601/DRC93CA) were carried out to determine the carrier concentration, mobility, resistivity and nature of the InN films. The optical band gap of InN films on different substrates was measured using an UV–Vis spectrophotometer. RESULTS AND DISCUSSIONS Raman Spectroscopy Analysis Figure 1 shows the Raman spectra of InN films grown on different ZnO-buffered substrates. These spectra were obtained with a Raman spectroscopy system by using a 514-nm Ar+ laser. The Raman peaks are observed at 490 cm1 and 586 cm1 that correspond to E2 (high) and A1 (LO) phonon modes of InN, respectively.25,26 The figure shows that the intensity of the A1 (LO) mode is higher than the intensity of the E2 (high) mode. The A1 (LO) mode ranges from 572–588 cm1 on quartz, 565–590 cm1 on sapphire, 572–585 cm1 on silicon and 581– 587 cm1 on bulk GaN substrates. The peak broadening of the A1 (LO) mode and low intensity of the E2 (high) mode are a result of weak crystal quality of InN films. The origin of peak broadening could be due to the coupling between free electron oscillations (plasmons) and LO phonons.27 The A1 (LO) mode is also quite sensitive to the lattice distortions. As a result, the non-stoichiometric defects adversely affect the long-range ionic ordering in the film. The relatively large intensity of the A1 (LO) mode and weak intensity of the E2 (high) mode could also be related to the small electron mobility of the

Table II. Growth deposition Table I. Growth parameters for ZnO buffer layer deposition Zno target Chamber pressure Argon flow Target to substrate distance RF power Deposition temperature Thickness

99.8% Purity 3 9 105 mbar 10 sccm 8 cm 100 W Room temperature 100 nm

parameters

Indium target Base pressure Argon flow Nitrogen flow Working pressure Target to substrate distance RF power Deposition temperature Deposition rate

for

InN

film

99.9% Purity 2.5 9 105 mbar 12 sccm 8 sccm 2.2 9 102 mbar 8 cm 60 W 250°C ˚ /s 3A

Structural, Electrical and Optical Properties of Sputtered-Grown InN Films on ZnO Buffered Silicon, Bulk GaN, Quartz and Sapphire Substrates

Fig. 2. XRD patterns of InN films grown on ZnO-buffered substrates. Fig. 1. Raman spectra of sputtered-grown InN films on ZnO-buffered substrates. (a) Silicon, (b) Bulk GaN, (c) Quartz, (d) Sapphire.

sputtered-grown InN films.26 The presence of A1 (LO) and E2 (high) modes of InN confirm the hexagonal nature of the films. The peaks at 567 cm1 and 739 cm1 correspond to bulk GaN substrate whereas the Si substrate peak appears at 520 cm1. X-ray Diffraction Analysis X-ray diffraction (XRD) studies were carried out for the structural analysis of InN films. Figure 2 shows the XRD patterns of InN films grown on ZnObuffered Si, bulk GaN, sapphire and quartz substrates. The XRD patterns show the growth of ZnO buffer layer and InN films with preferential orientation towards the c-plane on different substrates. The XRD patterns reveal InN peak positions at 31.236°, 31.228°, 31.236° and 31.169° on Si, quartz, bulk GaN and sapphire substrates, respectively. No indium oxide (In2O3), metallic In, metallic Zn and Zn3N2 peaks were detected, which is in accordance with the previous results.1,8,18,19,21 The hexagonal nature of InN films and ZnO buffer layers was confirmed from the XRD reference codes (00-0501239) and (00-036-1451), respectively. The comparison of XRD patterns of InN on different substrates shows that the films grown on sapphire in the presence of a ZnO buffer layer show larger intensity as compared to the films grown on the other ZnObuffered substrates. To evaluate the crystalline quality of InN films, various structural parameters were calculated. The values of full width at half maximum (FWHM; b) of InN films for the (002) plane are given in Table III. The structural parameters including lattice parameters c and a, strain (), crystallite size (D), bond length (l), positional parameter (u), dislocation density (d) and unit cell volume (V) of InN films along the (002) plane were calculated using the following equations.19,20,28–30

The lattice parameter (a) was calculated using the ratio c/a = 1.61 for bulk InN. The values obtained for these parameters on different substrates are given in Table III. c ¼ co ð þ 1Þ

ð1Þ

  bcosh  kk D ¼ 2sinh

ð2Þ



kk bcosh

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 2 ! u a3 1 t l¼  u c2 þ 2 3



ð3Þ

ð4Þ

a2 þ 0:25 3c2

ð5Þ

15bcosh 4aD

ð6Þ



V ¼ 0:866a2 c

ð7Þ

Table III shows there is a little variation in the crystallite size and strain of InN films grown on different substrates. The structural parameters of InN film grown on Si and bulk GaN substrates are similar whereas there is a slight variation in the structural parameters of InN on sapphire and quartz substrates. The crystallite size of InN film on sapphire and quartz is slightly higher as compared to the crystallite size of the film on the Si and GaN. This is followed by a lesser dislocations density in the films deposited on the sapphire and quartz substrates. These results show that the crystalline quality of InN films on sapphire and quartz are slightly better than those films grown on the bulk GaN and Si.

Bashir, Hassan, Ahmed, and Afzal

Table III. Structural parameters of InN films grown on ZnO-buffered substrates Substrate type Silicon Quartz Bulk GaN Sapphire

FWHM (o)

˚) c (A

˚) a (A

0.44 0.40 0.44 0.41

5.7200 5.7185 5.7200 5.7297

3.5491 3.5482 3.5491 3.5445

 3.515 3.253 3.515 5.212

9 9 9 9

107 107 107 107

D (nm)

l (nm)

21.12 23.23 21.12 22.65

0.3909141 0.3909040 0.3909141 0.3909797

d (nm22) 3.29 2.72 3.29 2.86

9 9 9 9

104 104 104 104

˚ 3) V (A 62.0890 62.0728 62.0890 62.1940

Fig. 3. AFM images of InN films (5 9 5 lm2) on different ZnO-buffered substrates: (a) Silicon, (b) Bulk GaN, (c) Quartz, and (d) Sapphire.

Surface Topography Analysis The surface topography of InN films grown on ZnO-buffered Si, bulk GaN, quartz and sapphire substrates are shown in Fig. 3. The AFM images were taken from a randomly selected 5 9 5-lm2 area of each sample. The route mean square (RMS) roughness of InN films was measured using AFM NanoScope software. The highest roughness of 8.0 nm is recorded for InN films on sapphire substrate whereas the lowest surface roughness of

2.6 nm is obtained on bulk GaN substrates. A roughness of 3.9 nm and 5.9 nm of InN films is found on Si and quartz substrates respectively. The variation in surface roughness can be attributed to the nature of different substrates because the ZnO buffer layer was deposited under identical conditions on all the substrates. In our previous studies,19 lesser values of surface roughness were reported for InN films grown on InN buffer layer. This means that the ZnO buffer layer has caused an increase in the

Structural, Electrical and Optical Properties of Sputtered-Grown InN Films on ZnO Buffered Silicon, Bulk GaN, Quartz and Sapphire Substrates

Fig. 4. (a) Variation in carrier mobility with change of substrate. (b) Variation in electrical resistivity with change of substrate. (c) Variation in carrier concentration with change of substrate.

surface roughness of InN films. The large value of surface roughness on the sapphire substrate may be due to the inhomogeneities in the film which can be seen from the large and non-uniform island growth as shown in Fig. 3. The island distribution on quartz substrate is more uniform and dense whereas the island size is also smaller as compared to the sapphire substrate. The same thing happens for Si (111) and for bulk GaN substrates where the islands are more dense, small and uniform. The large surface roughness may also be attributed to the large LM between InN film and the sapphire substrate. The 3D island growth is a result of Volmer– Weber growth process which takes place in a layerby-layer fashion and eventually results in 3D nanostructures after critical thickness (Stranski–Krastanov growth).31 The 3D islands are formed on substrate or buffer layer in order to reduce the

strain energy between substrate and film.32 To reduce the surface roughness, thermal annealing of InN films can be carried out; however, this may affect the crystalline quality of InN films due to more oxygen incorporation from the ZnO buffer layer and N2 atoms evaporating from the film. Hall Measurements The electrical properties of InN films such as electrical resistivity, carrier concentration and mobility were obtained through Hall effect measurements. To calculate these parameters, Al metal contacts were made on the samples using a thermal evaporator.33 The InN films were found to be n-type in nature on all the substrates. Figure 4a, b, and c shows the histograms of carrier mobility, resistivity and carrier concentration of InN films grown on

Bashir, Hassan, Ahmed, and Afzal

different substrates. The values of these parameters have been mentioned on the top of the histograms. The highest (20.20 cm2/Vs) and lowest (1.23 cm2/Vs) mobility values were recorded on bulk GaN and quartz substrates, respectively. The highest carrier concentration (2.64 9 1023/cm3) of InN was recorded for quartz and lowest (2.45 9 1022/cm3) for silicon substrates. The values of electrical resistivity of InN films on sapphire substrate were found to be lower as compared to the other substrates. The high carrier concentration in the film may be due to the incorporation of oxygen impurities inside the film. The sputtered-grown films usually show high carrier concentration as compared to films grown by other techniques.26 The product of carrier concentration and mobility is a reciprocal of resistivity33 which is in agreement with the obtained results. Here we can notice one more important thing in that the structural properties of InN are better on sapphire and quartz as compared to other substrates whereas these films show better electrical properties on bulk GaN and sapphire substrates than on the quartz and silicon. The films grown on bulk GaN substrate have higher mobility and lower carrier concentration as compared to the other substrates. The high resistivity shown by the film grown on silicon substrate indicates that the InN grown on Si might contain more defects relative to the other substrates. These defects can be impurities, grain boundaries and dislocations that restrict the mobility of carriers. Furthermore, the high carrier concentration and less carrier mobility in the films grown on quartz and silicon may be due to higher donor defects generated in the film during the growth process. Overall, films grown on the sapphire substrate have shown better structural quality, better electron mobility, lowest resistivity and smooth surface.

Fig. 5. Reflectance spectra of InN films grown on ZnO-buffered substrates.

Fig. 6. Band gap variations on different substrates.

Optical Band Gap Figure 5 shows the UV–Vis reflectance spectra of InN films grown on different ZnO-buffered substrates. These spectra were recorded at room temperature using a Cary 5000 series UV–Vis-nearinfrared (NIR) system in the wavelength range of 300–800 nm. Prominent Fabry–Perot interference fringes are observed due to multiple reflections from surfaces of films and the substrates. These fringes disappear in the wavelength region corresponding to absorption edge of the film.28,34 Therefore, the band gap of InN film was measured by finding the cutoff wavelength in the reflectance spectra. The fringes were observed to disappear at 515.8 nm for silicon, 544.9 nm for quartz, 543.3 nm for bulk GaN and 614.0 nm for sapphire. These correspond to the band gap values of 2.40 eV, 2.27 eV, 2.28 eV and 2.01 eV for silicon, quartz, bulk GaN and sapphire substrates, respectively. The band gap values are also shown in Fig. 6. The variations in band gap

values may be due to the variation in carrier concentration of InN films grown on different substrates. The large carrier concentration produces Burstein–Moss shift which widens the band gap.35 This is mostly true for the sputtered-grown InN films that exhibit high carrier concentration as compared to MBE- and MOCVD-grown films.35–37 This means that the carrier concentration in InN films has a strong effect on optical absorption edge and hence on the band gap of these films. On the other hand, it is observed that there is a small variation in the band gap values of InN on quartz and bulk GaN whereas in the case of quartz, the carrier concentration is higher than that of the bulk GaN. This discrepancy may be due to stoichiometric-related defects in the films that affect its band gap. The results of the present work are in agreement with the previous studies.38–43

Structural, Electrical and Optical Properties of Sputtered-Grown InN Films on ZnO Buffered Silicon, Bulk GaN, Quartz and Sapphire Substrates

CONCLUSIONS InN films were deposited on ZnO-buffered silicon, bulk GaN, quartz and sapphire substrates. The Raman spectroscopy results confirmed the hexagonal nature of films whereas the XRD results showed that the films were highly oriented along the c-plane on all substrates. Films deposited on sapphire showed good crystal quality whereas the films deposited on bulk GaN exhibited good electrical properties. The higher RMS surface roughness was obtained for the films grown on sapphire; however, the roughness of InN on bulk GaN substrate was found to be the lowest among other substrates. The optical band gap of InN on sapphire was found to be lowest as compared to its value on other substrates. The variations in optical band gap of InN were associated to variations in the carrier concentration. ACKNOWLEDGEMENTS Support from an RU Top-Down Grant (1001/CINOR/870019) through the Universiti Sains Malaysia is gratefully acknowledged. REFERENCES 1. C.-H. Shih, I. Lo, W.-Y. Pang, and C.-H. Hiseh, J. Phys. Chem. Solids 71, 1664 (2010). 2. J. Wu, W. Walukiewicz, W. Shan, K. Yu, J. Ager III, and S. Li, J. Appl. Phys. 94, 4457 (2003). 3. A.G. Bhuiyan, A. Hashimoto, and A. Yamamoto, J. Appl. Phys. 94, 2779 (2003). 4. C. Gallinat, G. Koblmu¨ller, J. Brown, and J. Speck, J. Appl. Phys. 102, 064907 (2007). 5. H. Xiao, X. Wang, J. Wang, N. Zhang, H. Liu, and Y. Zeng, J. Cryst. Growth 276, 401 (2005). 6. H. Lu, W.J. Schaff, L.F. Eastman, and C. Stutz, Appl. Phys. Lett. 82, 1736 (2003). 7. B. Zhang, H. Song, J. Wang, C. Jia, J. Liu, and X. Xu, J. Cryst. Growth 319, 114 (2011). 8. S. Inoue, T. Namazu, T. Suda, and K. Koterazawa, Vacuum 74, 443 (2004). 9. N. Dietz, M. Alevli, V. Woods, M. Strassburg, H. Kang, and I. Ferguson, Phys. Status Solidi B (b) 242, 2985 (2005). 10. N. Nepal, N.A. Mahadik, L.O. Nyakiti, S.B. Qadri, M.J. Mehl, and J.K. Hite, Cryst. Growth Des. 13, 1485 (2013). 11. T. Tsuchiya, H. Yamano, O. Miki, A. Wakahara, and A. Yoshida, Jpn. J. Appl. Phys. 38, 1884 (1999). 12. Y. Cho, O. Brandt, M. Korytov, M. Albrecht, V.M. Kaganer, and M. Ramsteiner, Appl. Phys. Lett. 100, 152105 (2012). 13. B. Liu, T. Kitajima, D. Chen, and S.R. Leone, J. Vac. Sci. Technol. A 23, 304 (2005).

14. H. Lu, W.J. Schaff, J. Hwang, H. Wu, G. Koley, and L.F. Eastman, Appl. Phys. Lett. 79, 1489 (2001). 15. S.-Y. Kuo, F.-I. Lai, W.-C. Chen, W.-T. Lin, C.-N. Hsiao, and H.-I. Lin, Diam. Relat. Mater. 20, 1188 (2011). 16. O. Laboutin and R. Welser, Appl. Phys. Lett. 92, 223103 (2008). 17. S.-Y. Kuo, W.-C. Chen, C. Kei, and C. Hsiao, Semicond. Sci. Technol. 23, 055013 (2008). 18. M.R. Laskar, T. Ganguli, A. Kadir, N. Hatui, A. Rahman, and A. Shah, J. Cryst. Growth 315, 233 (2011). 19. U. Bashir, Z. Hassan, and N.M. Ahmed, J. Mater. Sci. Mater. Electron. 28, 9228 (2017). 20. U. Bashir, Z. Hassan, N.M. Ahmed, A. Oglat, and A.S. Yusof, Mater. Sci. Semcond. Proc. 71, 166 (2017). 21. T. Ohgaki, N. Ohashi, H. Haneda, and A. Yasumori, J. Cryst. Growth 292, 33 (2006). 22. I. Lo, W. Wang, M. Gau, J. Tsai, S. Tsay, and J. Chiang, Appl. Phys. Lett. 88, 082108 (2006). 23. S. Ohuchi and T. Takizawa, J. Electron. Mater. 34, 424 (2005). 24. R. Zhang, P. Zhang, T. Kang, H. Fan, X. Liu, and S. Yang, Appl. Phys. Lett. 91, 162104 (2007). 25. X. Pu, J. Chen, W. Shen, H. Ogawa, and Q. Guo, J. Appl. Phys. 98, 033527 (2005). 26. M. Amirhoseiny, S. Ng, and Z. Hassan, Mater. Sci. Semcond. Proc. 35, 216 (2015). 27. B. Barick and S. Dhar, J. Cryst. Growth 416, 154 (2015). 28. N. Afzal, M. Devarajan, and K. Ibrahim, Mater. Sci. Semcond. Proc. 51, 8 (2016). 29. P. Shokeen, A. Jain, A. Kapoor, and V. Gupta, Plasmonics 11, 669 (2016). 30. S. Singhal, J. Kaur, T. Namgyal, and R. Sharma, Physica B 407, 1223 (2012). 31. E. Bauer and J.H. van der Merwe, Phys. Rev. B 33, 3657 (1986). 32. J. Tersoff and F. Legoues, Phys. Rev. Lett. 72, 3570 (1994). 33. N. Afzal, M. Devarajan, and K. Ibrahim, J Alloys Compd. 652, 407 (2015). 34. Q. Guo, T. Tanaka, M. Nishio, and H. Ogawa, Jpn. J. Appl. Phys. 47, 612 (2008). 35. K.S.A. Butcher and T. Tansley, Superlattice Microst. 38, 1 (2005). 36. H. He, Y. Cao, R. Fu, H. Wang, J. Huang, and C. Huang, J. Mater. Sci. Mater. Electron. 21, 676 (2010). 37. H. He, Y. Cao, R. Fu, W. Guo, Z. Huang, and M. Wang, Appl. Surf. Sci. 256, 1812 (2010). 38. H. Hovel and J. Cuomo, Appl. Phys. Lett. 20, 71 (1972). 39. T. Tansley and C. Foley, J. Appl. Phys. 59, 3241 (1986). 40. T. Inushima, V. Mamutin, V. Vekshin, S. Ivanov, T. Sakon, and M. Motokawa, J. Cryst. Growth 227, 481 (2001). 41. K. Butcher, H. Hirshy, R.M. Perks, M. Wintrebert-Fouquet, and P. Chen, Phys. Status Solidi A 203, 66 (2006). 42. K. Scott, A. Butcher, M. Wintrebert-Fouquet, P.P.T. Chen, K.E. Prince, and H. Timmers, Phys. Status Solidi (c) 2, 2263 (2005). 43. O. Briot, B. Maleyre, S. Ruffenach, B. Gil, C. Pinquier, F. Demangeot, and J. Frandon, J. Cryst. Growth 269, 22 (2004).