Damage recovery in ZnO by post-implantation annealing

Nuclear Instruments and Methods in Physics Research B 268 (2010) 1842–1846

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb

Damage recovery in ZnO by post-implantation annealing A. Audren a,*, A. Hallén b, M.K. Linnarsson b, G. Possnert a a b

Ion Physics, Ångström Laboratory, Dept. of Engineering Sciences, P.O. Box 534, Uppsala University, SE 751 21 Uppsala, Sweden Royal Inst. of Technology, School of Communication and Information Technology, Microelectronics and Applied Physics, Electrum 229, SE 16440 Kista, Sweden

a r t i c l e

i n f o

Article history: Available online 25 February 2010 Keywords: ZnO Ion implantation RBS/channeling SIMS Radiation damage Dopant diffusion

a b s t r a c t ZnO bulk samples were implanted with 200 keV-Co ions at room temperature with two fluences, 1  1016 and 8  1016 cm 2, and then annealed in air for 30 min at different temperatures up to 900 °C. After the implantation and each annealing step, the samples were analyzed by Rutherford backscattering spectrometry (RBS) in random and channeling directions to follow the evolution of the disorder profile. The RBS spectra reveal that disorder is created during implantation in proportion to the Co fluence. The thermal treatments induce a disorder recovery, which is however, not complete after annealing at 900 °C, where about 15% of the damage remains. To study the Co profile evolution during annealing, the samples were, in addition to RBS, characterized by secondary ion mass spectrometry (SIMS). The results show that Co diffusion starts at 800 °C, but also that a very different behavior is seen for Co concentrations below and above the solubility limit. Ó 2010 Elsevier B.V. All rights reserved.

1. Introduction Diluted Magnetic Semiconductors (DMS) are semiconducting materials in which transition metals (TM), or appropriate rare earth ions are substituted onto cation sites which couple with the free charge carriers to yield ferromagnetism via indirect interaction [1]. This ferromagnetic property has triggered a great interest for the DMS due to their application potential in spintronics [2,3]. One of the major problems for spintronic devices is the Curie temperature which must be above room temperature (RT) to have practical applications. The mean-field Zener model, proposed by Dietl et al. [4], predicts that wide band gap semiconductors, as ZnO, doped with TM ions exhibit ordering temperatures above 300 K, provided there is a sufficient charge carrier density [5]. Several experiments have shown promising results, however, the results differ from one experiment to another depending for instance on the TM doped ZnO synthesis technique. Some of them have shown ferromagnetic properties with a Curie temperature above RT [6], whereas other experiments exhibit spin-glass [7], or paramagnetic behavior [8]. These studies reveal also that the ferromagnetism in DMS requires a high carrier concentration as well as a high TM dopant concentration substituting the cations [4]. Moreover, the crystalline quality and the residual defects seem to have an impact on the ferromagnetic properties [9]. Another problem is the presence of TM precipitates and/or the formation of secondary phases. In many cases, the ferromagnetism observed is due to the presence of TM precipitates which present superpara* Corresponding author. E-mail address: [email protected] (A. Audren). 0168-583X/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2010.02.032

magnetic properties, rather than from carrier mediated magnetic dopants [10,11]. One method to introduce TM dopants into bulk ZnO crystals is the ion implantation technique which provides several advantages, namely reproducibility, precise control of the fluence, use of an isotopically pure beam and the possibility to overcome the solubility limit of atoms in a material [5]. As the ferromagnetic strength depends on the TM concentration in the DMS, this ion implantation method seems interesting for the study of ferromagnetic properties in DMS. However, this method also presents several drawbacks. The major one is the generation of structural defects in the lattice which are detrimental not only for a proper inclusion of implanted TM ions on substitutional sites [9], but also for compensating the free carriers. A possibility to prevent the disorder creation in the crystal during ion implantation is to increase the implantation temperature. However, this is not the case in ZnO crystal; an implantation temperature increase to 623 K induces a significant suppression of point defects at the surface, but is not able to decrease the disorder rate in the bulk [5]. In this material, a post-implantation annealing at high temperature is necessary to completely recover the lattice structure [12]. A drawback of post-implantation annealing at high temperature is the possible diffusion of the implanted ions and the formation of nano-clusters. The aim of this work is to study the disorder recovery and possible diffusion of the implanted species during thermal annealing. To do that, ZnO samples were implanted with Co atoms and subsequently annealed at different temperatures. Two Co fluences were chosen in order to observe how the Co concentration affects the ZnO disordering and the Co diffusion. The disorder was studied with Rutherford backscattering spectrometry in the channeling

1843

A. Audren et al. / Nuclear Instruments and Methods in Physics Research B 268 (2010) 1842–1846

Commercial ZnO bulk single-crystalline samples with a wurtzite structure (ZnOrdic AB) were implanted with 200 keV-Co ions from the implanter of the Ion Technology Centre at Uppsala University. The projected range is around Rp = 90 nm according to the SRIM code. Two fluences (1  1016 and 8  1016 cm 2) were used in order to introduce a maximum atomic concentration of 1% and 9%, respectively (SRIM code [13]). These doses are intended to be below and above the solid solubility limit of Co in ZnO, which is around 6.5 at.% [14]. During implantation, the samples were tilted 7° relative to the incident ion beam to minimize channeling. As the increase of the implantation temperature seems to promote the formation of Co clusters and does not reduce the damage production [5], the implantations were performed at RT. The samples were subsequently annealed up to 900 °C, for 30 min, in air in order to avoid a loss of oxygen [15]. No thermal treatments were preformed above 900 °C to prevent the implanted layer decomposition [16]. After the implantation and after each annealing step, the samples were analyzed by Rutherford backscattering spectrometry in random (RBS) and channeling (RBS-C) direction using the tandem accelerator at Uppsala University. The spectra were collected with a 2 MeV-He+ beam at a backscattering angle of 170°. To collect the RBS-C spectra, the samples were mounted on a goniometer, which allows aligning of the He beam with the (0 0 0 1)-axis. The spectra were analyzed with the SIMNRA program [17]. The vmin is the ratio of the minimum backscattering yield at channeling condition to that of a random beam. It indicates the degree of disorder in the sample and in our virgin samples, vmin was around 2%, which indicates a very high crystalline order. Analysis of depth distribution of cobalt was made by secondary ion mass spectrometry (SIMS) utilizing a Cameca ims 4f instrument. A primary sputter beam of 8.2 keV 32(O2)+ ions was applied and secondary ions of 59Co+ were detected. In addition, a primary beam of 4.5 keV Cs+ ions has been used and secondary ions of 59 Co133Cs+ were detected. For the depth profiling, the primary ions were rastered over an area of 200  200 lm2 and secondary ions were detected from the central part of this area (60 lm in diameter). 3. Results and discussion The RBS spectrum recorded in random direction after implantation of a Co fluence of 1  1016 cm 2 at RT is shown in Fig. 1a (full line). The SRIM code gives a maximum atomic concentration of 1.1 at.%, which is below the Co solubility limit of 6.5 at.% in ZnO [14]. The Zn and Co surfaces are indicated by the arrows in the RBS spectrum. According to SIMNRA, the peak corresponding to the implanted Co atoms should be centered at 1490 keV. In this spectrum, this Co peak cannot be seen, which means that the fluence is too low to be detected by RBS. To study the damage in the Zn sublattice, RBS spectra were recorded in channeling direction along the (0 0 0 1) axis on the as-implanted sample and after annealings at 500, 600 and 800 °C (see Fig. 1a). The RBS-C spectrum recorded on the as-implanted sample exhibits a large peak around 1480 keV, which is due to the superposition of the displaced Zn atoms and the incorporated Co atoms. But in this sample, the Co fluence is low and the Co atoms contribution can be neglected, as it is indicated by the random spectra. For this dose, we consider the wide peak at 1480 keV to represent mainly the damage in

300

Normalized yield

2. Experimental

a 16

as-impl. random

1x10

cm

-2

Co Zn chann. spectra

200

as-impl. 500 °C 600 °C 800 °C virgin

100

0 1200

1300

1400

1500

1600

1700

1800

Energy (keV)

b 300

Normalized yield

direction (RBS-C). To analyze the Co diffusion, we used the RBS and SIMS techniques. The results show a disorder recovery, but also Co diffusion during annealing at temperatures above 800 °C.

as-impl. random

16

8x10

Co

200

100

1300

1400

1500

-2

chann. spectra Zn

0 1200

cm

1600

as-impl. 500 °C 700 °C 800 °C 900 °C virgin

1700

1800

Energy (keV) Fig. 1. RBS spectra recorded in random (full line) and channeling direction (other symbols) on ZnO single-crystals implanted at RT with 200 keV-Co ions at 1  1016 cm 2 (a) and 8  1016 cm 2 (b) and subsequently annealed in air at different temperatures for 30 min.

the Zn sublattice. The maximum of this damage peak does not reach the level of the random spectrum, which means that the sample is only partially disordered. This absence of amorphization is in accordance with previous results observed after implantation of heavier ions, or at higher temperature [5,15,18]. After annealing at 500 and 600 °C, no changes were observed concerning the damage peak on the RBS-C spectra. After annealing at 800 °C, however, the RBS-C spectrum exhibits a huge decrease of the damage peak. That means that the disorder recovery in this sample starts between 600 and 800 °C. The analysis of the RBS-C spectra allows the determination of the disorder amount and the conversion of the energy scale into depth in the material. The results of this analysis (Fig. 2a) show that the disorder maximum is centered at 100 nm as predicted by SRIM. Before annealing, the maximum disorder was around 45%. No decrease of the disorder rate was observed after annealing at 500 and 600 °C, but after annealing at 800 °C, the disorder decreases to 17%. In order to increase the Co concentration in ZnO, another sample has been implanted with a higher fluence (8  1016 cm 2) at RT. In this sample, the maximum Co concentration (9 at.%) overcomes the Co solubility limit in ZnO (6.5 at.%). Fig. 1b presents the RBS spectra recorded in random and channeling directions for this sample. The RBS spectrum recorded in random direction in the as-implanted sample exhibits a peak at 1490 keV, which is related to the Co atoms. This Co peak will be discussed later, but it indicates that in this sample the Co implanted fluence is too high to be neglected in relation to the damage. As a consequence, the broad peak at 1480 keV on the RBS spectra recorded in channeling direction is due to the superposition of two contributions: the damage in the Zn sublattice and the implanted Co atoms occupying

A. Audren et al. / Nuclear Instruments and Methods in Physics Research B 268 (2010) 1842–1846

interstitial positions. However, before annealing, this peak does not reach the level of the random spectrum, revealing that, even after the high Co fluence implantation, the sample is not totally disordered. This result is in agreement with previous studies [5,15,18]. To determine the amount of disorder in this sample, the Co atoms contribution must be subtracted before analysis of RBS-C spectrum, but this is very difficult since the relative amount of interstitial and substitutional Co is not known. Furthermore, the contribution to the total RBS-C spectra from Co will be overestimated since the yield from Zn atoms is higher than from Co due to its larger atomic number. In the as-implanted sample, the disorder is high and the Co contribution to the backscattering yield on the channeled and the random spectra are almost the same. The Co contribution on the RBS-C sample can then be estimated from the random spectrum. The thermal annealing induces disorder recovery and may also result in incorporation of interstitial Co at substitutional sites, in which case the Co contribution to the RBSC spectra will be less than for the random spectra. Despite of these difficulties, Fig. 2b presents the final results of the spectra analysis where the Co atoms contribution has been obtained from the random spectrum and subtracted from the channeled spectra. In the as-implanted sample, the maximum disorder level is around 80%, which is much higher than in the low dose implanted sample (45%). The thermal treatments induce a disorder recovery and the disorder level decreases to 50% and 25% after annealing at 500 and 700 °C, respectively (Fig. 2b). Similar damage extraction from the channeled spectra recorded after annealing at 800 and 900 °C is not meaningful because a large part of the Co atoms will now be at substitutional sites (this will be discussed further in relation to Fig. 4). Nevertheless, it can be seen on RBS spectra

a

1.0 16

1x10

cm

-2

Disorder rate

0.8

as-impl. 500 °C 600 °C 800 °C

0.6 0.4 0.2 0.0 0

50

100

150

200

250

1e+21 -3 Co concentration (cm )

1844

1x1016 cm-2 1e+20 Annealing: 1e+19 as-impl. & 600 °C 1e+18

800 °C 900 °C

1e+17 0

100

200

300

400

500

600

Depth (nm) Fig. 3. Co concentration measured by SIMS in ZnO single-crystals implanted at RT with 200 keV-Co ions at 1  1016 cm 2 and subsequently annealed in air at different temperatures for 30 min.

(Fig. 1b) that the disorder recovery continues during annealing at 800 °C and the disorder level seems to be very low. However, the RBS-C spectrum does not reach the level of the spectrum obtained from the virgin sample, which means that the disorder recovery is not complete. A subsequent annealing at 900 °C (not shown) induces only a slight disorder recovery. In the low dose implanted sample, the Co profile cannot be seen on the RBS spectrum. In that case, to study the Co profile, SIMS analyses were performed on a second sample implanted and annealed under the same conditions. Results from the SIMS analysis with oxygen primary ions are displayed in Fig. 3. Furthermore, cesium primary ions have been employed and 59Co133Cs+-complexes has been detected to investigate influence on ionization yield due to possible changes in the matrix during heat treatment (not shown). Both kind of SIMS measurements indicate that before annealing, the Co profile exhibits one peak located around 90 nm, which is in good accordance with the Rp predicted by the SRIM simulation. No changes were observed after the thermal treatments up to 600 °C, but for annealings at 800 and 900 °C a Co concentration increase at the surface and a slight shift of the Co concentration maximum can be seen. In the high dose implanted sample, the Co dose was high enough to be detected by RBS and the spectra recorded in random direction exhibit a peak around 1490 keV, which is related to the Co atoms (Fig. 4). This peak is stable during annealing till 700 °C and begins to decrease after annealing at 800 and 900 °C. The SIM-

300

Depth (nm)

Normalized yield

8x1016 cm-2 0.8

Disorder rate

16

250

b 1.0 not ann. 500 °C 700 °C

0.6 0.4

8x10

250 200

150

150

HCo random

0 1200

100 1300 50

100

150

200

250

300

as-impl. 700 °C 800 °C 900 °C

Ann. 800°C chann.

0.2

0

-2

200

50

0.0

cm

1300

HCo+dis.

1400

1400

1500

1600

1500

1600

Energy (keV)

Depth (nm) Fig. 2. Depth distribution of damage accumulated in Zn sublattice of ZnO singlecrystals implanted at RT with 200 keV-Co ions at 1  1016 cm 2 (a) and 8  1016 cm 2 (b) and subsequently annealed in air at different temperatures for 30 min.

Fig. 4. RBS spectra recorded in random direction in ZnO single-crystals implanted at RT with 200 keV-Co ions at 8  1016 cm 2 and subsequently annealed in air at different temperatures for 30 min. A comparison between RBS spectra recorded in random and channeling direction on the 800 °C annealed sample is given in the inset.

1845

A. Audren et al. / Nuclear Instruments and Methods in Physics Research B 268 (2010) 1842–1846

Co amount (%)

12.5 as-implanted annealed at 700°C (7.7 1016 Co.cm-2)

10.0 7.5

annealed at 800°C 16 -2 (7.2 10 Co.cm )

5.0 2.5 0.0 0

50

100

150

200

250

300

Depth (nm) Fig. 5. Depth distribution of Co atoms in ZnO single-crystals implanted at RT with 200 keV-Co ions at 8  1016 cm 2 and subsequently annealed in air at different temperatures for 30 min.

1e+23 -3 Co concentration (cm )

NRA program allows the analysis of these RBS spectra and gives the Co profile as a function of depth in the sample. The result of this analysis is reported in Fig. 5. The Co profile after implantation and thermal annealings up to 700 °C presents a maximum at 35 nm with a Co concentration of 11%. The integration of this profile gives a Co dose around 8  1016 cm 2. During annealing at 800 °C, the Co profile broadens and the profile maximum is centered at  60 nm with a concentration of 7%. The total Co dose decrease to  7  1016 cm 2, which reveals a small out-diffusion of Co at 800 °C and this Co loss seems to continue at 900 °C according to the RBS spectrum in Fig. 4. As SIMS is a more sensitive technique, it provides complementary information concerning the Co profile behavior during annealing. A SIMS measurements on this sample were performed after annealing at 700 °C (Fig. 6). As the SIMS may not be linear when the Co atomic concentration overcomes 1%, the spectrum recorded on this sample was calibrated by taking into account that the profile integration is equal to the dose of 8  1016 cm 2. Two peaks around 45 and 75 nm separated by a plateau, or even a region with reduced concentration, are observed. In both Figs. 4 and 5, the Co concentration maximum appears closer to the surface than in the low dose implanted sample. A possible explanation to this effect is surface erosion by sputtering occurring during the high dose implantation. A simulation was performed with the SRIM code, which gives a sputtering of 30 nm under these experimental conditions. The surface binding energy used in this simulation was the ZnO sublimation energy found in the literature [19]. For the high dose sample it is also possible that the surface decomposition and evaporation during annealing further decreases the distance between the surface and Co peak. In addition, it should be noted that changes in stoichiometry and swelling of the material due to the high dose makes analysis difficult and the results, especially in Fig. 5, includes some errors. In both samples, the SIMS measurements and the RBS spectra indicate that the Co diffusion starts between 600 and 800 °C and continue at 900 °C. It is interesting to notice that for the high dose implanted sample annealed at 800 °C, the height of the peak in the RBS-C spectrum is higher than in the random spectrum (see the inset in Fig. 4). This indicates that a part of the Co atoms are incorporated at substitutional sites and the same phenomenon is also observed after annealing at 900 °C. Similar incorporation has been observed by photoluminescence following thermal annealing under high vacuum [20]. In the low dose implanted sample, the SIMS measurements reveal a Co concentration increase at the surface, which indicates that a Co diffusion occurs toward the surface. This

8x1016 cm-2

1e+22

Annealing at 700 °C

1e+21 1e+20 1e+19 1e+18 0

100

200

300

400

Depth (nm) Fig. 6. Co concentration measured by SIMS in ZnO single-crystals implanted at RT with 200 keV-Co ions at 8  1016 cm 2 and annealed in air at 700 °C for 30 min.

phenomenon is also observed by SIMS on the high dose implanted sample after annealing at 700 °C. But in this spectrum, the two peaks around 45 and 75 nm separated by a lower concentration region indicate that the Co diffusion cannot be explained by one simple diffusion mechanism. This particular shape of the SIMS profile may be an indication of a formation of a new compound during annealing. 4. Conclusions The ZnO samples were implanted with 200 keV-Co ions at RT with two fluences (1  1016 and 8  1016 cm 2) to study the effect on the ZnO damage formation and annealing. The samples were annealed at different temperatures up to 900 °C for 30 min in air and the damage evolution was recorded by RBS channeling. In addition, the Co depth profile was analyzed, both with RBS and SIMS. The RBS-C spectra have shown a disorder recovery, which starts between 600 and 800 °C in the low dose implanted sample and already between 300 and 500 °C in the high dose implanted sample. SIMS measurements reveal Co diffusion toward the surface starting at 800 °C and also, in the high dose sample, SIMS indicates the formation of secondary phases. The RBS spectra in the high dose implanted sample confirms the diffusion of Co at 800 and 900 °C. Due to the difficulty to separate the Co and the disorder contribution using RBS, it will be interesting to also analyze the samples by elastic recoil detection analysis. References [1] J.K. Furdyna, J. Appl. Phys. 64 (1988) R29. [2] S.A. Wolf, D.D. Awschalom, R.A. Buhrman, J.M. Daughton, S. von Molnar, M.L. Roukes, A.Y. Chtchelkanova, D.M. Treger, Science 294 (2001) 1488. [3] K. Sato, H. Katayama-Yoshida, Semicond. Sci. Technol. 17 (2002) 367. [4] T. Dietl, H. Ohno, F. Matsukura, J. Cibert, D. Ferrand, Science 287 (2000) 1019. [5] S. Zhou, K. Potzger, G. Talut, H. Reuther, J. von Borany, R. Grötzschel, W. Skorupa, M. Helm, J. Fassbender, N. Volbers, M. Lorenz, T. Herrmannsdörfer, J. Appl. Phys. 103 (2008) 023902. [6] H.J. Lee, S.Y. Jeong, C.R. Cho, C.H. Park, Appl. Phys. Lett. 81 (2002) 4020. [7] J.H. Kim, H. Kim, Y.E. Ihm, W.K. Choo, J. Appl. Phys. 92 (2002) 6066. [8] G. Lawes, A.S. Risbud, A.P. Ramirez, R. Seshadri, Phys. Rev. B 71 (2005) 045201. [9] Y.W. Heo, M.P. Ivill, K. Ip, D.P. Norton, S.J. Pearton, J.G. Kelly, R. Rairigh, A.F. Hebard, T. Steiner, Appl. Phys. Lett. 84 (2004) 2292. [10] D.P. Norton, M.E. Overberg, S.J. Pearton, K. Pruessner, J.D. Budai, L.A. Boatner, M.F. Chisholm, J.S. Lee, Z.G. Khim, Y.D. Park, R.G. Wilson, Appl. Phys. Lett. 83 (2003) 5488. [11] K. Potzger, S. Zhou, H. Reuther, A. Mücklich, F. Eichhorn, N. Schell, W. Skorupa, M. Helm, J. Fassbender, T. Herrmannsdörfer, T.P. Papageorgiou, Appl. Phys. Lett. 88 (2006) 052508. [12] S.O. Kucheyev, J.S. Williams, S.J. Pearton, Mater. Sci. Eng. R 33 (2001) 51. [13] . [14] C.H. Bates, W.B. White, R. Roy, J. Inorg. Nucl. Chem. 28 (1966). [15] S.O. Kucheyev, J.S. Williams, C. Jagadish, J. Zou, Cheryl Evans, A.J. Nelson, A.V. Hamza, Phys. Rev. B 67 (2003) 094115.

1846

A. Audren et al. / Nuclear Instruments and Methods in Physics Research B 268 (2010) 1842–1846

[16] V.A. Coleman, H.H. Tan, C. Jagadish, S.O. Kucheyev, J. Zou, Appl. Phys. Lett. 87 (2005) 231912. [17] . [18] S.O. Kucheyev, J.S. Williams, C. Jagadish, Vacuum 73 (2004) 93.

[19] A.V. Makarov, S.G. Zbezhneva, V.V. Kovalenko, M.N. Rumyantseva, Inorg. Mater. 39 (2003) 705–709. [20] V. Ney, S. Ye, T. Kammermeier, A. Ney, H. Zhou, J. Fallert, H. Kalt, F.-Y. Lo, A. Melnikov, A.D. Wieck, J. Appl. Phys. 104 (2008) 083904.