- Email: [email protected]
Hydrogen-ion implantation in GaAs
Vacuum 63 (2001) 697}700
Hydrogen-ion implantation in GaAs G. Gawlik *, R. Ratajczak, A. Turos , J. Jagielski , S. Bedell, W.L. Lanford Institute of Electronic Materials Technology, ul. Wo& lczyn& ska 133, 01-919 Warsaw, Poland A. Soltan Institute of Nuclear Studies, 05-400 S! wierk/Otwock, Poland State University of New York, Albany, 12222 NY, USA
Abstract The results of the basic study on depth distribution of hydrogen atoms and corresponding damage pro"les produced by 50 keV H-ion implantation in (1 0 0) GaAs are reported. The in#uence of the H-ion dose and of the temperature of implantation and subsequent annealing were studied. The depth distribution of hydrogen was measured using the N(p, )C nuclear reaction, whereas the RBS/channeling was applied for defect analysis. Two temperature regions revealed: at temperatures below 903C independently of the hydrogen dose no blisters were found, whereas, at temperatures above 1203C blisters appear at doses exceeding the critical value. In the latter region hydrogen-defect complex formation was observed which are e!ective hydrogen traps and stabilize the radiation damage. With the increasing H-ion dose these complexes agglomerate into hydrogen gas bubbles. The increase of gas pressure in such bubbles upon subsequent thermal treatment can result in the splitting of a surface layer from the substrate. 2001 Elsevier Science Ltd. All rights reserved. Keywords: Direct bonding; Ion implantation; Blistering; Ion beam analysis
1. Introduction The performance of semiconductor micro- and optoelectronic devices can be greatly enhanced if one can freely integrate di!erent materials. An effective integration method has remained a challenge due to lattice constant mismatch between semiconductor systems. Heteroepitaxial growth of highly mismatched materials can result in layers containing large concentrations of defects which degrade or inhibit device operation. Because of this
* Corresponding author. Institute of Electronic Materials Technology, ul. WoH lczynH ska 133, 01-919 Warsaw, Poland. Fax: #48-3912-0764. E-mail address: gawlik}[email protected] (G. Gawlik).
limit of heteroepitaxial growth, a variety of methods of material joining techniques has been developed over the last decade [1]. One of the most elegant methods is direct wafer bonding technology. Bonding by Van der Waals forces occurs when two clean and smooth surfaces are placed in contact. While silicon is the most important material for microelectronic device fabrication, the majority of optoelectronic devices are based on the III}V semiconductor compounds. The role of GaAs in optoelectronic light emitting and high frequency devices is still incontestable. Wafer bonding technique opens new prospects for deposition of GaAs single crystal layers on almost all kinds of substrates. One of the most promising techniques for production of such material systems is `smart-cuta
0042-207X/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 2 - 2 0 7 X ( 0 1 ) 0 0 2 6 0 - 3
698
G. Gawlik et al. / Vacuum 63 (2001) 697}700
technology [2,3]. It is based on the fact that hydrogen ion implantation to large #uences leads to the formation of a buried layer of gas bubbles that eventually produces blistering upon thermal treatment. If hydrogen "lled bubbles form a uniform layer at a given depth, the uppermost part of a crystal can split o! preserving its crystalline structure. Such a layer can be bonded to a selected substrate and subjected to further processing [4]. `Smartcuta technology has been successfully applied in the silicon device technology. In the case of III}V semiconductor compounds it is still a subject of extended study and needs further development. This paper will focus on the study of the e!ects of hydrogen-ion implantation in GaAs single crystals at di!erent temperatures. Hydrogen depth pro"les, defect distributions and void formation were investigated. The elucidation of these fundamental phenomena is crucial for technological applications.
2. Experimental Hydrogen ion implantation in (1 0 0) GaAs single crystals was performed at ITME Warsaw at temperatures ranging from 453C to 1603C. Implantation energy was 50 keV and the doses ranged from 4;10 H/cm to 4;10 H/cm. Depth distribution of hydrogen was measured using the N(p, )C nuclear reaction at SUNY Albany, whereas, the 2 MeV He-ion channeling was applied for defect analysis at SINS Warsaw [5]. Surface morphology after hydrogen bombardment and thermal treatment leading to the surface layer exfoliation was studied by means of scanning electron microscopy (SEM) and optical microscopy. Surface roughness was determined by the Alphastep pro"lometer in terms of the roughness parameter Ra, i.e. the mean arithmetic value of the surface irregularities.
3. Results Surface blistering occurs if the ion dose exceeds a given critical value which, in turn depends on the target temperature during implantation. As shown in Fig. 1, no surface defects induced by ion im-
Fig. 1. Dose and temperature boundaries of blister formation in hydrogen implanted GaAs. The inset shows the roughness parameter dependence on the implanted hydrogen dose.
plantation at temperatures below 1003C were observed up to the highest used ion dose of 4;10 H/cm. Blistering and #aking occur upon ion implantation at temperatures exceeding 1003C at ion doses already less than 1;10 H/cm. The measured critical ion dose amounted to 9;10 H/cm at 1403C and decreased to 7;10 H/cm at 1603C. The increase of the ion dose over the critical value at a given temperature results in the increase of the amount of blisters and their dimensions. Variation of the surface roughness parameter, Ra, with the ion dose is shown in the inset in Fig. 1. One notes that blisters grow more rapidly at higher temperatures. Structural defect formation and transformation were studied by means of RBS/channeling. Fig. 2 shows random and aligned spectra for GaAs crystals implanted with H-ions under di!erent conditions. The increase of channeling yield in the vicinity of the channel range 550}650 re#ects the presence of radiation damage. From such spectra, defect contents and their depth distributions can be calculated [5]. Besides, the random spectrum and that of the unimplanted crystal (curve 1), Fig. 2 contains aligned spectra for the samples implanted at 453C (curves 2, 3, 6) and at 1403C (curves 4, 5). As can be seen, hydrogen implantation at 453C
G. Gawlik et al. / Vacuum 63 (2001) 697}700
699
Fig. 2. Random and aligned RBS spectra for GaAs single crystals implanted with various hydrogen doses at two di!erent temperatures.
Fig. 3. Hydrogen concentration pro"les for GaAs single crystals implanted with 6;10 H cm\ at di!erent temperatures as determined by the NRA analysis.
produces damage peaks that increase with the increase of the ion dose. In all cases, even for the heavily damaged crystal implanted with 2.3;10 H/cm only simple defects and/or small amorphous clusters were detected [6]. This is in agreement with the observation that no detectable blisters or other forms of surface deformation were observed after low temperature hydrogen implantation up to the #uence of 4.0;10 H/cm. Surprisingly, the damage peak for the samples implanted at 1403C was signi"cantly greater than that for the samples implanted with the same dose at a lower temperature (cf. curves 2 and 3, and curves 4 and 5 in Fig. 2). One could expect the opposite e!ect because of the increased defect mobility at elevated temperatures leading to the enhanced defect recombination probability. Results of the NRA analysis are shown in Fig. 3. The concentration pro"le of hydrogen implanted with a dose of 6;10 H/cm was obviously not in#uenced by the implantation temperature (1203C) and post implantation annealing up to 3003C. Additionally, similar results were obtained for implantation carried out at 2003C or post implantation annealing at 4003C. Taking into account very good depth resolution of the NRA (approx. 10 nm) only short-range rearrangement of
hydrogen atoms can occur at temperatures below 4003C.
4. Discussion and conclusions In the "rst part of the reported experiments, the conditions for blisters formation were investigated. Blistering is directly responsible for the formation of microcracks which in turn produce layer splitting. No blistering was observed at temperatures below 1003C even after implantation with a dose as large as 4;10 H/cm. In order to provoke blistering in such samples an annealing at temperatures exceeding 4003C is necessary [6]. Blistering is much more easily obtained upon hot implantation. There are two distinct temperature regions: at temperatures below 903C no blisters were found, whereas, at temperatures exceeding 1203C appreciably high blister density can be observed. In the second range the temperature at which blistering occurs depends on the implantation dose. The growth of blister dimensions with the increasing implantation temperature is re#ected in the rapid increase of surface roughness. The complementary investigations of defect behavior and hydrogen depth distributions gave
700
G. Gawlik et al. / Vacuum 63 (2001) 697}700
some insight into the mechanism of blister formation. The important conclusion from the NRA analysis is that the hydrogen concentration pro"le is altered neither upon hot implantation nor upon annealing at temperatures at least not exceeding 3003C. Thus, at these conditions, the distance of hydrogen atom migration most probably does not exceed a few nanometers. According to Rauhala and RaK isaK nen [7] the maximum damage distribution is located at the same depth as that of the hydrogen atom distribution. This was con"rmed by our RBS/channeling measurements. Moreover, the detailed study of the damage structure [8] revealed that after RT implantation only displaced atoms and/or small amorphous clusters are formed. The substantial increase of the damage peak after hydrogen implantation at temperatures above 1403C as compared to that measured for the sample implanted with the same dose at 453C is somewhat unusual. One could expect the opposite e!ect, i.e. a decrease of the damage peak due to greater defect mobility. It is worth pointing out that at this temperature only defects in the Ga sublattice are mobile in GaAs, whereas, defect mobility threshold for As sublattice lies above 3003C [9,10]. Thus, important defect transformations cannot be expected in the studied temperature range. Furthermore, the study of hydrogen di!usion in GaAs [11] clearly indicated that hydrogen mobility is largely suppressed by the presence of radiation damage. All these factors lead to the conclusion that the unusual increase of damage peaks in channeling spectra for the samples implanted at 1403C is due to the formation of hydrogen-defect complexes in which matrix atoms become displaced from their lattice sites. One notes that according to the data shown in Fig. 1 at the applied doses no blisters are formed. Since gas bubbles and voids have relatively low dechanneling cross-sections such defects cannot be the dominant kind of defect. Therefore, the
observed hydrogen-defect complexes of di!erent morphology containing multiple hydrogen atoms bounded to the displaced matrix atoms can be considered as precursors of the hydrogen bubble and eventually blister formation. They are e!ective traps for mobile hydrogen atoms and stabilize the radiation damage produced by ion implantation. The prerequisite of such complexes formation is high enough hydrogen concentration. Their formation is thermally activated: small complexes can only coalesce if the implantation temperature exceeds 1203C or after annealing at temperatures above 4003C.
Acknowledgements This work was partially supported by the IAEA Coordinated Research Project, RC No. 10035.
References [1] Mathine DL. IEEE J Sel Topics Quantum Electron 1997;3:952}9. [2] Bruel M. Nucl Instr and Meth B 1996;108:313}9. [3] Bruel M. Electron Lett 1995;31:1201}2. [4] Zhu Z-H, Ejeckam F, Qain Y, Ahang J, Zhang Z, Christenson G, Lo Y. IEEE J Sel Topics Quantum Electron 1997;3:927}36. [5] Ion Beam Analysis Handbook, Eds. [6] Jalgueier E, Aspar B, Pocas S, Michaud JF, Zussy M, Papon AM, Bruel M. Electron Lett 1998;34:408}9. [7] Rauhala E, RaK isaK nen J. Nucl Instr and Meth B 1994;94:245}50. [8] Stonert A, Turos A, Nowicki L, Breeger B. Nucl Instr and Meth B 2000;161}163:496}500. [9] Turos A, Stonert A, Breeger B, Wendler E, Wesch W, Fromknecht R. Nucl Instr and Meth B 1999;148:401}5. [10] Turos A, Stonert A, Breeger B, Wendler E, Wesch W. Nukleonika 1999;44:93}102. [11] RaK isaK nen J, Keinonen J, Karttunen V, Koponen I. J Appl Phys 1988;64:2334}36.
Hydrogen-ion implantation in GaAs G. Gawlik *, R. Ratajczak, A. Turos , J. Jagielski , S. Bedell, W.L. Lanford Institute of Electronic Materials Technology, ul. Wo& lczyn& ska 133, 01-919 Warsaw, Poland A. Soltan Institute of Nuclear Studies, 05-400 S! wierk/Otwock, Poland State University of New York, Albany, 12222 NY, USA
Abstract The results of the basic study on depth distribution of hydrogen atoms and corresponding damage pro"les produced by 50 keV H-ion implantation in (1 0 0) GaAs are reported. The in#uence of the H-ion dose and of the temperature of implantation and subsequent annealing were studied. The depth distribution of hydrogen was measured using the N(p, )C nuclear reaction, whereas the RBS/channeling was applied for defect analysis. Two temperature regions revealed: at temperatures below 903C independently of the hydrogen dose no blisters were found, whereas, at temperatures above 1203C blisters appear at doses exceeding the critical value. In the latter region hydrogen-defect complex formation was observed which are e!ective hydrogen traps and stabilize the radiation damage. With the increasing H-ion dose these complexes agglomerate into hydrogen gas bubbles. The increase of gas pressure in such bubbles upon subsequent thermal treatment can result in the splitting of a surface layer from the substrate. 2001 Elsevier Science Ltd. All rights reserved. Keywords: Direct bonding; Ion implantation; Blistering; Ion beam analysis
1. Introduction The performance of semiconductor micro- and optoelectronic devices can be greatly enhanced if one can freely integrate di!erent materials. An effective integration method has remained a challenge due to lattice constant mismatch between semiconductor systems. Heteroepitaxial growth of highly mismatched materials can result in layers containing large concentrations of defects which degrade or inhibit device operation. Because of this
* Corresponding author. Institute of Electronic Materials Technology, ul. WoH lczynH ska 133, 01-919 Warsaw, Poland. Fax: #48-3912-0764. E-mail address: gawlik}[email protected] (G. Gawlik).
limit of heteroepitaxial growth, a variety of methods of material joining techniques has been developed over the last decade [1]. One of the most elegant methods is direct wafer bonding technology. Bonding by Van der Waals forces occurs when two clean and smooth surfaces are placed in contact. While silicon is the most important material for microelectronic device fabrication, the majority of optoelectronic devices are based on the III}V semiconductor compounds. The role of GaAs in optoelectronic light emitting and high frequency devices is still incontestable. Wafer bonding technique opens new prospects for deposition of GaAs single crystal layers on almost all kinds of substrates. One of the most promising techniques for production of such material systems is `smart-cuta
0042-207X/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 2 - 2 0 7 X ( 0 1 ) 0 0 2 6 0 - 3
698
G. Gawlik et al. / Vacuum 63 (2001) 697}700
technology [2,3]. It is based on the fact that hydrogen ion implantation to large #uences leads to the formation of a buried layer of gas bubbles that eventually produces blistering upon thermal treatment. If hydrogen "lled bubbles form a uniform layer at a given depth, the uppermost part of a crystal can split o! preserving its crystalline structure. Such a layer can be bonded to a selected substrate and subjected to further processing [4]. `Smartcuta technology has been successfully applied in the silicon device technology. In the case of III}V semiconductor compounds it is still a subject of extended study and needs further development. This paper will focus on the study of the e!ects of hydrogen-ion implantation in GaAs single crystals at di!erent temperatures. Hydrogen depth pro"les, defect distributions and void formation were investigated. The elucidation of these fundamental phenomena is crucial for technological applications.
2. Experimental Hydrogen ion implantation in (1 0 0) GaAs single crystals was performed at ITME Warsaw at temperatures ranging from 453C to 1603C. Implantation energy was 50 keV and the doses ranged from 4;10 H/cm to 4;10 H/cm. Depth distribution of hydrogen was measured using the N(p, )C nuclear reaction at SUNY Albany, whereas, the 2 MeV He-ion channeling was applied for defect analysis at SINS Warsaw [5]. Surface morphology after hydrogen bombardment and thermal treatment leading to the surface layer exfoliation was studied by means of scanning electron microscopy (SEM) and optical microscopy. Surface roughness was determined by the Alphastep pro"lometer in terms of the roughness parameter Ra, i.e. the mean arithmetic value of the surface irregularities.
3. Results Surface blistering occurs if the ion dose exceeds a given critical value which, in turn depends on the target temperature during implantation. As shown in Fig. 1, no surface defects induced by ion im-
Fig. 1. Dose and temperature boundaries of blister formation in hydrogen implanted GaAs. The inset shows the roughness parameter dependence on the implanted hydrogen dose.
plantation at temperatures below 1003C were observed up to the highest used ion dose of 4;10 H/cm. Blistering and #aking occur upon ion implantation at temperatures exceeding 1003C at ion doses already less than 1;10 H/cm. The measured critical ion dose amounted to 9;10 H/cm at 1403C and decreased to 7;10 H/cm at 1603C. The increase of the ion dose over the critical value at a given temperature results in the increase of the amount of blisters and their dimensions. Variation of the surface roughness parameter, Ra, with the ion dose is shown in the inset in Fig. 1. One notes that blisters grow more rapidly at higher temperatures. Structural defect formation and transformation were studied by means of RBS/channeling. Fig. 2 shows random and aligned spectra for GaAs crystals implanted with H-ions under di!erent conditions. The increase of channeling yield in the vicinity of the channel range 550}650 re#ects the presence of radiation damage. From such spectra, defect contents and their depth distributions can be calculated [5]. Besides, the random spectrum and that of the unimplanted crystal (curve 1), Fig. 2 contains aligned spectra for the samples implanted at 453C (curves 2, 3, 6) and at 1403C (curves 4, 5). As can be seen, hydrogen implantation at 453C
G. Gawlik et al. / Vacuum 63 (2001) 697}700
699
Fig. 2. Random and aligned RBS spectra for GaAs single crystals implanted with various hydrogen doses at two di!erent temperatures.
Fig. 3. Hydrogen concentration pro"les for GaAs single crystals implanted with 6;10 H cm\ at di!erent temperatures as determined by the NRA analysis.
produces damage peaks that increase with the increase of the ion dose. In all cases, even for the heavily damaged crystal implanted with 2.3;10 H/cm only simple defects and/or small amorphous clusters were detected [6]. This is in agreement with the observation that no detectable blisters or other forms of surface deformation were observed after low temperature hydrogen implantation up to the #uence of 4.0;10 H/cm. Surprisingly, the damage peak for the samples implanted at 1403C was signi"cantly greater than that for the samples implanted with the same dose at a lower temperature (cf. curves 2 and 3, and curves 4 and 5 in Fig. 2). One could expect the opposite e!ect because of the increased defect mobility at elevated temperatures leading to the enhanced defect recombination probability. Results of the NRA analysis are shown in Fig. 3. The concentration pro"le of hydrogen implanted with a dose of 6;10 H/cm was obviously not in#uenced by the implantation temperature (1203C) and post implantation annealing up to 3003C. Additionally, similar results were obtained for implantation carried out at 2003C or post implantation annealing at 4003C. Taking into account very good depth resolution of the NRA (approx. 10 nm) only short-range rearrangement of
hydrogen atoms can occur at temperatures below 4003C.
4. Discussion and conclusions In the "rst part of the reported experiments, the conditions for blisters formation were investigated. Blistering is directly responsible for the formation of microcracks which in turn produce layer splitting. No blistering was observed at temperatures below 1003C even after implantation with a dose as large as 4;10 H/cm. In order to provoke blistering in such samples an annealing at temperatures exceeding 4003C is necessary [6]. Blistering is much more easily obtained upon hot implantation. There are two distinct temperature regions: at temperatures below 903C no blisters were found, whereas, at temperatures exceeding 1203C appreciably high blister density can be observed. In the second range the temperature at which blistering occurs depends on the implantation dose. The growth of blister dimensions with the increasing implantation temperature is re#ected in the rapid increase of surface roughness. The complementary investigations of defect behavior and hydrogen depth distributions gave
700
G. Gawlik et al. / Vacuum 63 (2001) 697}700
some insight into the mechanism of blister formation. The important conclusion from the NRA analysis is that the hydrogen concentration pro"le is altered neither upon hot implantation nor upon annealing at temperatures at least not exceeding 3003C. Thus, at these conditions, the distance of hydrogen atom migration most probably does not exceed a few nanometers. According to Rauhala and RaK isaK nen [7] the maximum damage distribution is located at the same depth as that of the hydrogen atom distribution. This was con"rmed by our RBS/channeling measurements. Moreover, the detailed study of the damage structure [8] revealed that after RT implantation only displaced atoms and/or small amorphous clusters are formed. The substantial increase of the damage peak after hydrogen implantation at temperatures above 1403C as compared to that measured for the sample implanted with the same dose at 453C is somewhat unusual. One could expect the opposite e!ect, i.e. a decrease of the damage peak due to greater defect mobility. It is worth pointing out that at this temperature only defects in the Ga sublattice are mobile in GaAs, whereas, defect mobility threshold for As sublattice lies above 3003C [9,10]. Thus, important defect transformations cannot be expected in the studied temperature range. Furthermore, the study of hydrogen di!usion in GaAs [11] clearly indicated that hydrogen mobility is largely suppressed by the presence of radiation damage. All these factors lead to the conclusion that the unusual increase of damage peaks in channeling spectra for the samples implanted at 1403C is due to the formation of hydrogen-defect complexes in which matrix atoms become displaced from their lattice sites. One notes that according to the data shown in Fig. 1 at the applied doses no blisters are formed. Since gas bubbles and voids have relatively low dechanneling cross-sections such defects cannot be the dominant kind of defect. Therefore, the
observed hydrogen-defect complexes of di!erent morphology containing multiple hydrogen atoms bounded to the displaced matrix atoms can be considered as precursors of the hydrogen bubble and eventually blister formation. They are e!ective traps for mobile hydrogen atoms and stabilize the radiation damage produced by ion implantation. The prerequisite of such complexes formation is high enough hydrogen concentration. Their formation is thermally activated: small complexes can only coalesce if the implantation temperature exceeds 1203C or after annealing at temperatures above 4003C.
Acknowledgements This work was partially supported by the IAEA Coordinated Research Project, RC No. 10035.
References [1] Mathine DL. IEEE J Sel Topics Quantum Electron 1997;3:952}9. [2] Bruel M. Nucl Instr and Meth B 1996;108:313}9. [3] Bruel M. Electron Lett 1995;31:1201}2. [4] Zhu Z-H, Ejeckam F, Qain Y, Ahang J, Zhang Z, Christenson G, Lo Y. IEEE J Sel Topics Quantum Electron 1997;3:927}36. [5] Ion Beam Analysis Handbook, Eds. [6] Jalgueier E, Aspar B, Pocas S, Michaud JF, Zussy M, Papon AM, Bruel M. Electron Lett 1998;34:408}9. [7] Rauhala E, RaK isaK nen J. Nucl Instr and Meth B 1994;94:245}50. [8] Stonert A, Turos A, Nowicki L, Breeger B. Nucl Instr and Meth B 2000;161}163:496}500. [9] Turos A, Stonert A, Breeger B, Wendler E, Wesch W, Fromknecht R. Nucl Instr and Meth B 1999;148:401}5. [10] Turos A, Stonert A, Breeger B, Wendler E, Wesch W. Nukleonika 1999;44:93}102. [11] RaK isaK nen J, Keinonen J, Karttunen V, Koponen I. J Appl Phys 1988;64:2334}36.