Nuclear Instruments and Methods in Physics Research B 216 (2004) 407–413 www.elsevier.com/locate/nimb Direct ion beam synthesis of II–VI nanocrystals U.V. Desnica a,*, M. Buljan a, I.D. Desnica-Frankovic a, P. Dubcek a, S. Bernstorff b, M. Ivanda a, H. Zorc a a Department of Physics, Rudger Boskovic Institute, Bijenicka 54, 10000 Zagreb, Croatia b Sincrotrone Trieste, SS 14 km163,5, 34012 Basovizza, Italy Abstract We have studied the direct synthesis of nanoparticles formed by dual implantation of large and equal doses of Cd + S, Zn + Te, Cd + Te or Pb + Te ions into SiO2 substrate. Grazing incidence small angle X-ray scattering (GISAXS), transmittance measurements and Raman spectroscopy were used to investigate implanted composites. The 2D GISAXS patterns suggest the synthesis of nanoparticles already during ion implantation, performed either at 300 or at 77 K, while annealing at higher T causes an increase of the fraction and the average size of synthesized nanoparticles. After high-T annealing both optical methods detected nanocrystals of compound semiconductors CdS, ZnTe or CdTe through the appearance of the respective first optical gaps, Eg , in transmittance measurements and characteristic LO peaks in Raman spectra. It is proposed that at high ion doses a fraction of implanted atoms synthesize already during implantation into amorphous aggregates of compound semiconductor, which transform into crystalline nanoparticles after annealing.
2003 Elsevier B.V. All rights reserved. PACS: 61.10.Eq; 81.07.)b; 61.72.V; 61.46.+w; 81.20.)n; 81.05.Ys Keywords: Nanocrystals; Implantation; GISAXS; Raman; Transmittance; CdS; CdTe; ZnTe; PbTe; Quantum dots 1. Introduction There is an intense interest and strong research activity going on to develop technology for the efficient and well reproducible synthesis of nanoparticles (NPs) [1–4]. Control over the particle size enables strong modifications of electronic, optical and other properties of NPs, offering a number of potential applications in * Corresponding author. Tel.: +385-1-4561-173; fax: +385-14680-114. E-mail address: desnica@rudjer.irb.hr (U.V. Desnica). semiconductor and other industries [1]. The main interest in wide band-gap semiconductors NPs comes out from the pronounced effects of quantum confinement – strongly dependent on the particle size – which make possible the tunability of the band-gap and of the powerful visible photoluminescence. Sequential implantation of equal doses of constituent atoms offers a unique way to produce compound-semiconductor NPs [1,3–6]. In this paper the synthesis of NPs of a number of binary semiconductors by high dose ion implantation was investigated by means of grazing incidence small angle X-ray scattering (GISAXS), transmittance measurements and Raman spectroscopy. 0168-583X/$ - see front matter
2003 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2003.11.068 408 U.V. Desnica et al. / Nucl. Instr. and Meth. in Phys. Res. B 216 (2004) 407–413 2. Experimental details Samples were produced by implanting equal doses of metal ions (Cd, Zn or Pb) and chalcogen ions (S or Te) into amorphous SiO2 (fused silica, Corning 7940). All ions were singly charged and ion rates were close to 25 lA/cm2 . Ion doses were 4 · 1016 cm
2 of each type of ions in all samples, except for CdS samples, where doses were 1017 cm
2 (Table 1). Samples nos. 1–4 were implanted at room temperature (RT) and samples (5–10) at liquid nitrogen temperature (LN2). Half of the samples (even numbers in Table 1) were subsequently annealed at high annealing temperature (Ta ¼ 1073 or 1273 K) in the flow of Ar + 4%H2 for 1 h (except sample no. 4, which underwent rapid thermal annealing for 16 min. Ion energies were selected using SRIM calculations to create similar, overlapping concentration profiles of metal and chalcogen ions. Doses, ion energies, temperatures of implantation, Ti , and temperatures of annealing, Ta , are given in Table 1 for all samples. SRIMcalculated peak-volume concentrations of each of implanted atoms were in range 6–8 · 1021 cm
3 at depth of 60–70 nm (exemptions were samples 1 and 2, for which both implant doses and maximum-depth positions were approximately twice as large). For sample #1 the overlapping of depth profiles of Cd and S atoms was experimentally confirmed using Rutherford back scattering, with concentrationÕs maxima being around 6.3 · 1021 cm
3 , at depth of 130 nm. GISAXS experiments were carried out with the X-ray wavelength, k ¼ 0:154 nm, at the Austrian SAXS beamline at Elettra, Sincrotrone Trieste, Italy [7]. The GISAXS patterns were recorded with Table 1 List of all samples, including their number, name, ion doses and energies for each ion, implantation temperatures, Ti , and annealing temperatures, Ta , if any, as well as calculated values of the average diameter, R, and the interparticle distance, L, of nanocrystals, as obtained from the LMA analysis Sample name Doses (ions cm
2 ) Implantation energy (keV) Ti K
1 Ta K
1 R (nm) ±DR (nm) 1 CdS1 Cd(1E17) S(1E17) Cd(320) S(115) RT – 1.85 0.03 7.95 0.1 2 CdS2 Cd(1E17) S(1E17) Cd(320) S(115) RT 1073 2.07 0.03 7.2 0.09 3 ZnTe3 Zn(4E16) Te(4E16) Zn(115) Te(190) RT – 2.41 0.01 8.8 0.03 4 ZnTe4 Zn(4E16) Te(4E16) Zn(115) Te(190) RT 1073 2.55 0.03 9.12 0.05 5 ZnTe5 Zn(4E16) Te(4E16) Zn(110) Te(160) LN2 – 2.49 0.02 9.33 0.05 6 ZnTe6 Zn(4E16) Te(4E16) Zn(110) Te(160) LN2 1273 3.03 0.02 13.2 0.06 7 CdTe7 Cd(4E16) Te(4E16) Cd(155) Te(160) LN2 – 2.2 0.02 8.57 0.05 8 CdTe8 Cd(4E16) Te(4E16) Cd(155) Te(160) LN2 1273 3.42 0.02 11.3 0.06 9 PbTe9 Pb(4E16) Te(4E16) Pb(210) Te(160) LN2 – 2.05 0.011 7.5 0.02 10 PbTe10 Pb(4E16) Te(4E16) Pb(210) Te(160) LN2 1273 3.54 – Sample no. L (nm) ±DL (nm) 13.2 – DR and DL denote estimated errors in R and L, respectively. RT denotes room temperature and LN2 the temperature of liquid nitrogen (77 K). U.V. Desnica et al. / Nucl. Instr. and Meth. in Phys. Res. B 216 (2004) 407–413 1024 · 1024 pixels 2D CCD-detector, placed in the plane perpendicular to the specular plane, at the distance of 200 cm from the sample. Spectra where first corrected for background intensity and detector response, and then for refraction and absorption effects [8]. Transmittance measurements were done using the Perkin Elmer UV/VIS spectrometer lambda 25, while Raman spectra were taken using the triple spectrometer (DILOR Z-24), excitation line of 2.57 eV from an Ar-ion laser. 409 3. Results 3.1. GISAXS results Fig. 1 shows 2D GISAXS patterns for investigated samples. Left panes of Fig. 1 (odd numbers) represent patterns from as-implanted samples, while spectra presented on the right side refer to samples annealed at high temperatures (even numbers). The 2D patterns represent maps of the scattering intensities in reciprocal space, q (wave Fig. 1. Two-dimensional GISAXS patterns of double-species ion implanted SiO2 . Numbers 1; 2; . . . ; 10 refer to samples from Table 1. U.V. Desnica et al. / Nucl. Instr. and Meth. in Phys. Res. B 216 (2004) 407–413 vector q ¼ ð4p=kÞ sin b, 2b being the scattering angle). An inserted beam stopper blocks strong surface signals in/close to the specular plane (reflected beam, Yoneda, etc.), permitting better sensitivity for the weak diffuse scattering at larger q. The incidence angle of the X-ray beam, ai , used during measurements presented in Fig. 1 was ai ¼ ac þ 0:05, where ac denotes the critical angle for total external reflection. For this angle of incidence the penetration depth corresponds roughly to, or close to the maximal volume density of the implanted atoms. All GISAXS patterns in Fig. 1 comprise a highly intense region close to the center (remaining part of surface contributions, which are not relevant for further discussion) and pronounced half rings. GISAXS patterns of non-implanted SiO2 (not shown) is substantially different, comprising only the reflected signal in the specular plain. The presented rings are circular and generally similar in shape in all directions. In all annealed samples (right) rings were of relatively stronger intensity and of smaller radius than in respective non-annealed samples. Comparative 1D scattering intensity profiles, IðqÞ, taken along the polar angle ¼ 70 from some of the 2D spectra in Fig. 1, are shown in Fig. 2. The annealing at high temperature causes a shift of the intensity maximum toward lower q as well as somewhat steeper decrease of IðqÞ on the right side of the peak maximum. Spectra were analyzed using either simple particle scattering model (Guinier approximation) or local mono-disperse approximation (LMA) [9]. Numerical results obtained from LMA for the average aggregate radius, R, and for the average distance among NPs, L, are shown in Table 1. Values achieved from Guinier approximation were quite similar. GISAXS results suggest that a considerable fraction of NPs has been synthesized already during implantation. 3.2. Optical characterization Results of transmittance (TR) measurements on samples from Fig. 2 are shown in Fig. 3. In all annealed samples there is a distinct ÔkinkÕ in the TR curves at energies corresponding to the first energy gap, Eg , of the respective semiconductor. 2 1 Cd+S ions implanted 10 6 5 Intensity (a.u.) 410 1 8 Zn+Te ions implanted 7 0.1 Cd+Te ions implanted 0.4 0.6 0.8 1.0 1.2 q (nm-1) 1.4 1.6 Fig. 2. 1D intensity profiles from the 2D spectra shown in Fig. 1, taken along the detector polar angle / ¼ 70. Numbers 1; 2; . . . ; 8 refer to samples from Table 1. Result suggests that in all three annealed samples (CdS2, ZnTe6 and CdTe8), some of the implanted ions synthesized into compound semiconductors. Changes in the TR are better perceived in the first derivative of TR (Fig. 3(b)). The maximum of the peak position corresponds to the Eg of the respective semiconductor nanocrystals [10,11]. There is a shift of the peak maxima towards higher energies (Ôblue shiftÕ) in comparison to bulk values for respective Eg , reflecting quantum confinement of isolated semiconductor nanocrystals formed in an insulating, amorphous matrix. However, for non-annealed samples, analogous changes in TR (and its first derivative) are completely absent. Examples of Raman data are shown in Fig. 4 for the same set of samples. Presented Raman spectra are obtained after the subtraction of the wide background signal originating from the amorphous SiO2 . Distinct lines around 170 cm
1 for sample CdTe8, close to 205 cm
1 for ZnTe6 and around 295 cm
1 for CdS2 were found, respectively. These signals can be confidently U.V. Desnica et al. / Nucl. Instr. and Meth. in Phys. Res. B 216 (2004) 407–413 (a) 90 Transmitance, % 1 7 2 5 8 6 80 70 (b) x1/3 1st derivative 0.1 conductors, respectively. Hence, Raman results confirm the synthesis of implanted constituent ions into semiconductors after high-temperature annealing. Again, like with the (non)appearance of characteristic Eg in transmittance results, there is no positive sign of well-defined inter-atomic bonds, characteristic for respective crystalline semiconductors (no LO peaks) in spectra of asimplanted samples CdS1, ZnTe5 and CdTe7. 2 ZnTe bulk 1 CdTebulk CdSbulk 0.0 8 7 6 5 500 600 700 800 900 Wavelength (nm) Fig. 3. Optical transmission spectra (a), and their first derivative (b), of three types of as-implanted samples (dashed lines) and implanted + annealed samples (full lines): 1 ¼ CdS1, 2 ¼ CdS2, 5 ¼ ZnTe5, 6 ¼ ZnTe6, 7 ¼ CdTe7, 8 ¼ CdTe8. Ion doses and energies are given in Table 1. The transmittance of SiO2 is also shown (dotted line in (a)). The bulk values of gap energy Eg for CdS, CdTe and ZnTe are indicated in (b). ZnTe6 8000 CdS2 CdTe8 Counts (a.u.) 411 6000 x1/4 Air 4000 lines 2000 0 140 160 180 200 220 240 260 280 300 320 340 360 Raman shift (cm-1) Fig. 4. Raman spectra of samples 1 ¼ CdS1, 2 ¼ CdS2, 5 ¼ ZnTe5, 6 ¼ ZnTe6, 7 ¼ CdTe7, 8 ¼ CdTe8, using the 2.41 eV excitation line. assigned to longitudinal optical (LO) peaks, expected for crystalline CdTe, ZnTe and CdS semi- 4. Discussion GISAXS has been recently successfully applied to study nanoparticles embedded in matrices with low electron density, including NPs obtained by ion implantation [6,12]. Studied NPs consisted of either single metal atoms, (Au, Ag, Fe, Pt, Co; . . .) [9,13,14] or metal alloy (Cu–Ni) NPs [12] or binary semiconductors (CdS, ZnTe, ZnS) [4–6,15,16]. In implanted samples, in principle, nanoparticles can be observed by GISAXS when atoms, implanted uniformly into the substrate, become de-mixed and aggregated into nanoparticles. Spontaneous clustering is usually induced and controlled by thermal treatment. If nanoparticles are spatially correlated, a characteristic ÔhaloÕ (ring-like shape) is observed in 2D GISAXS pattern [9,13–16], which is circular if NPs are distributed isotropically. Generally, the appearance of a ÔringÕ in GISAXS pattern is considered as a sure sign of successful synthesis of nanoparticles in the nm size-range [13–17]. The size of NPs, as well as their shape, were in numerous cases compared with sizes/shape determined from the cross-sectional TEM analysis, and satisfactory agreements between the two methods were, generally, observed [9,13–17], confirming that the GISAXS signal indeed corresponds to synthesized nanoparticles and reflects their properties. Hence, the presented GISAXS results (Figs. 1 and 2) reveal that some NPs are synthesized in all studied samples: not only in annealed ones but also in as-implanted samples, even when the implantation was done at LN2 temperature (samples 5–10). Annealed samples: Both optical methods confirmed that after high-T annealing (at least part) of 412 U.V. Desnica et al. / Nucl. Instr. and Meth. in Phys. Res. B 216 (2004) 407–413 the implanted Cd + S, Zn + Te or Cd + Te ions synthesize into nanocrystals of the respective binary semiconductor – CdS, ZnTe or CdTe. This finding agrees well with a recent study of Cd + S implanted samples, in which, after annealing at 1273 K, for a range of ion doses CdS nanocrystals were detected by several methods [3]. For these CdS NPs the GISAXS-determined size and size distribution of NPs [16] agreed very well with sizes/ size distribution determined from TEM [3] as well as with sizes determined from the blue shift of Eg in TR [3]. As-implanted samples: The intensity of the NPsrelated GISAXS signals, at least for samples implanted with Cd + S or Zn + Te (Fig. 2), is similar for as-implanted and annealed ones. Consequently, the quantity of clustered material should be comparable in both cases. For these same samples the LMA analysis gave the percentages of synthesized implanted ions of about 40% and 60% before and after annealing, respectively. Complete lack of characteristic signal in TR (particularly in its first derivative, Fig. 3(b)), and particularly complete lack of the LO peak in Raman signals (Fig. 4), clearly show that nanoclusters in as-implanted samples do not consist of crystalline semiconductor phase. A question arises what is the chemical nature of NPs observed by GISAXS in as-implanted samples. One possibility is that GISAXS-observed NPs are single-species, metal clusters. This hypothesis implies that at some Ta these metal NPs would dissolve while the other kind of nanoparticles (compound semiconductor CdS, ZnTe, CdTe) would form. However, in a recent detailed comparative studies of a series of samples implanted with the same high dose of Cd and S ions (1017 cm
2 each), but annealed at different temperatures in the Ta ¼ RT to ¼ 1173 K range, a set of very smoothly changing GISAXS spectra was obtained [6]. Only small, gradual increase of average particle size was observed with each Ta step. There was not any discontinuity or other change that would suggest that one sort of nanoparticles (i.e. Cd clusters) is transforming into another (CdS NPs). The other possibility is that the high-dose implantation, at RT or lower temperatures, creates small-size clusters of compound semiconductor material but in the amorphous phase. Such amorphous aggregates would be equally well observable by GISAXS but features characteristic for the crystalline phase of semiconductor would be absent in TR and Raman spectra. Obtaining amorphous aggregates with high-dose, low-T implantation would be consistent with the findings of some other methods of NPs synthesis, where small amorphous aggregates were obtained in low-T non-equilibrium deposition processes, and the crystallization into NPs was achieved only after subsequent thermal processing [18,19]. Very little is known about NPs directly produced by implantation, since practically all published data refer to NPs obtained only after high-T annealing. Hence, further characterization of directly synthesized NPs in as-implanted samples is needed. 5. Conclusions Direct synthesis of nanoparticles formed by dual implantation of large and equal doses of Cd + S, Zn + Te, Cd + Te or Pb + Te ions into SiO2 substrate was studied. Grazing incidence small angle X-ray scattering (GISAXS), Transmittance measurements and Raman spectroscopy were used to investigate implanted composites. The 2D GISAXS patterns suggest the synthesis of nanoparticles already during ion implantation either at RT or at LN2 temperature, while annealing at higher Ta caused an increase of the fraction and the average size of synthesized nanoparticles. In all cases an isotropically distributed 3D ensemble of these nanoparticles in the amorphous SiO2 was established. After highT annealing both optical methods detected nanocrystals of compound semiconductors CdS, ZnTe, or CdTe through the appearance of the respective first optical gaps, Eg , in transmittance measurements and characteristic LO peaks in Raman spectra. The chemical nature of nanoparticles observed by GISAXS in as-implanted samples remains to be determined. A hypothesis is put forward that at high ion doses a fraction of implanted atoms synthesize already during implantation into amorphous aggregates of U.V. Desnica et al. / Nucl. Instr. and Meth. in Phys. Res. B 216 (2004) 407–413 compound semiconductor, which transform into crystalline nanoparticles after annealing. Acknowledgements We thank C.W. White, A. Meldrum and H. Karl for providing implanted samples. This work was supported by the Ministry of Science and Technology, Republic of Croatia. References [1] A. Meldrum, R.F. Haglund Jr., L.A. Boatner, C.W. White, Adv. Mater. 13 (2001) 1431. [2] J.D. Budai, C.W. White, S.P. Wiothrow, M.F. Chisholm, J. Zhu, R.A. Zuhr, Nature 390 (1997) 384. [3] C.W White, A. Meldrum, J.D. Budai, S.P. Withrow, E. Sonder, R.A. Zuhr, D.M. Hembree Jr., M. Wu, D.O. Henderson, Nucl. Instr. and Meth. B 148 (1999) 991. [4] H. Karl, I. Grosshans, W. Attenberger, M. Schmid, B. Stritzker, Nucl. Instr. and Meth. B 178 (2001) 126. [5] U.V. Desnica, I.D. Desnica-Frankovic, O. Gamulin, C.W. White, E. Sonder, R.A. Zuhr, J. Non-Cryst. Solids 299–302 (2002) 1100. [6] U.V. Desnica, P. Dubcek, I.D. Desnica-Frankovic, M. Buljan, S. Bernstorff, S.W. White, J. Appl. Cryst. 36 (2003) 443–446. 413 [7] H. Amenitsch, S. Bernstorff, M. Kriechbaum, D. Lombardo, H. Mio, M. Rappolt, P. Laggner, J. Appl. Cryst. 30 (1997) 872. [8] M. Buljan, K. Salamon, P. Dubcek, S. Bernstorff, I.D. Desnica-Frankovic, O. Milat, U.V. Desnica, Vacuum 71 (2003) 65. [9] D. Babonneau, I.R. Videnovic, M.G. Garnier, P. Oelhafen, Phys. Rev. B 63 (2001) 195401. [10] B.G. Potter Jr., J.H. Simmons, J. Appl. Phys. 68 (1990) 1218. [11] M. Sotelo-Lerma, M.A. Quevedo-Lopez, R.A. OrozcoTeranm, R. Ramirez-Bon, F.J. Espinoza-Beltran, J. Phys. Chem. Solids 59 (1988) 145. [12] E. Cattaruzza, F. dÕAcapito, F. Gonella, A. Longo, A. Martorana, G. Mattei, C. Maurizio, D. Thiaudiere, J. Appl. Cryst. 33 (2000) 740. [13] D. Babonneau, A. Naudon, D. Thaudiere, S. Lequien, J. Appl. Cryst. 32 (1999) 226. [14] A. Naudon, D. Thiaudiere, J. Appl. Cryst. 30 (1997) 822. [15] I.D. Desnica-Frankovic, U.V. Desnica, P. Dubcek, M. Buljan, S. Bernstorff, H. Karl, I. Großhans, B. Stritzker, J. Appl. Cryst. 36 (2003) 439. [16] U.V. Desnica, P. Dubcek, I.D. Desnica-Frankovic, M. Buljan, K. Salamon, O. Milat, S. Bernstorff, C.W. White, Nucl. Instr. and Meth. B 200 (2003) 191. [17] J.R. Levine, J.B. Cohen, Y.W. Chung, P. Georgopoulos, J. Appl. Cryst. 22 (1989) 528. [18] R. Ramirez-Bon, F.J. Espinoza-Beltran, H. Arizpe-Chavez, O. Zelaya-Angel, F. Sanchez-Sinencio, J. Appl. Phys. 77 (1995) 5461. [19] Ch. Ye, G. Meng, Y. Wang, Z. Jiang, L. Zhang, J. Phys. Chem. B 106 (2002) 10338.