- Email: [email protected]
Evidence of hydrogen–boron interactions in diamond from deuterium diffusion and infrared spectroscopy experiments
Diamond and Related Materials 8 (1999) 278–282
Evidence of hydrogen–boron interactions in diamond from deuterium diffusion and infrared spectroscopy experiments J. Chevallier a,*, A. Lusson a, B. Theys a, A. Deneuville b, E. Gheeraert b a Laboratoire de Physique des Solides de Bellevue, CNRS, 1 place A. Briand, 92195 Meudon, Cedex, France b Laboratoire d’Etudes des Proprie´te´s Electroniques des Solides, CNRS, BP 166, 38042 Grenoble, France Received 25 July 1998; accepted 17 October 1998
Abstract We report on the first experimental evidence of hydrogen–boron interactions in boron-doped diamond. Deuterium diffusion studies in homoepitaxial B-doped diamond films reveal that hydrogen diffusion is limited by the B concentration and is characterized by a low effective diffusion activation energy. Infrared spectroscopy experiments show that boron acceptor electronic transitions disappear under hydrogenation. These results are consistent with hydrogen ionization and diffusion of fairly mobile H+ which form pairs with B−. © 1999 Elsevier Science S.A. All rights reserved. Keywords: Diamond; Doping; Hydrogen; Diffusion; Infrared absorption
1. Introduction Hydrogen in conventional semiconductors has been extensively studied in the last decade because this impurity may be non-intentionally incorporated in the crystal lattice during epitaxial or bulk growth and during a number of steps used for semiconductor device processing [1–4]. The attachment of hydrogen to the crystal imperfections causes important modifications of the electronic properties. For example, hydrogen makes active impurities which are normally inactive: this is the case for C and Si in germanium and for C in silicon. (C,H ) and (Si,H ) are shallow acceptors in germanium [2], and (C,H ) related complexes act as deep centers in silicon [5]. Hydrogen modifies the electronic properties of deep levels by passivating them or by introducing new deep centers: this is the case for silicon doped with transition elements where H–transition metal complexes introduce additional deep levels in the band gap [6 ]. Furthermore, a great deal of attention has also been given to the hydrogen–dopant interactions in silicon, III–V and II–VI compounds, because the hydrogen– acceptor and hydrogen–donor complex formations are at the origin of the neutralization of the dopants [7]. The understanding of the behavior of hydrogen in * Corresponding author. Fax: +33 145 0758 41; e-mail: [email protected]
diamond is very important, because the deposition of both polycrystalline and monocrystalline diamond (instead of graphite) films has to be achieved within a very active hydrogen plasma. One consequence is the presence of up to several atomic per cent of hydrogen at the grain boundaries of polycrystalline diamond films [8]. It has also been recognized that the exposure, at 300 K, of undoped or p-type doped diamond to hydrogen plasma induces a superficial (20 nm) highly p-type conductive layer which is associated with the presence of a hydrogen accumulation layer at the surface [9,10]. However, until now, there has been no experimental study of the hydrogen diffusion in p-type diamond. In this work, we present the first results on hydrogen diffusion in boron-doped diamond. We demonstrate the existence of hydrogen–boron interactions, with the hydrogen being trapped by boron acceptors. This understanding is important for a proper control of the conductivity of p-type diamond films.
2. Experimental We order avoid aries.
0925-9635/99/$ – see front matter © 1999 Elsevier Science S.A. All rights reserved. PII: S0 9 2 5- 9 6 3 5 ( 9 8 ) 0 0 32 6 - 4
have used homoepitaxial films of diamond in to control the boron incorporation level and to parasitic trapping of hydrogen at grain boundSeveral films with various boron contents from
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
2×1018 to 3×1020 cm−3 and a low compensation ratio [11,12] have been deuterated or hydrogenated at various temperatures from 235 to 490 °C. From their SIMS deuterium profiles, we have studied the hydrogen diffusion process. From infrared transmission experiments, we have investigated the influence of hydrogenation on the boron acceptor electronic transitions. Monocrystalline boron-doped diamond epitaxial layers were grown on (100) type Ib synthetic diamond substrates by microwave plasma assisted chemical vapor deposition (MPCVD) at T=820 °C with 4% methane in hydrogen and at a total pressure of 30 torr. The thickness of the epilayers was 3 mm. The boron doping was achieved by adding diborane into the gas phase with [B]/[C ] ratios in the gas phase of 25, 200 and 800 ppm. The deposition process is stopped, first by cutting off the methane and diborane flows, then by switching off the plasma 10 min later, and finally by cooling the sample down to room temperature under a hydrogen gas flow. The epitaxial layers are treated at 80 °C during 30 min in boiling H SO saturated by 2 4 CrO which is known to remove the effect of hydrogen 3 from the near surface region of the film [10,13]. Under such preparation conditions, the compensation ratio of these films is rather low (<10%), as measured by conductivity experiments up to 1000 K [12]. These layers exhibit the infrared absorption bands characteristic of the electronic transitions from the ground to the excited states of almost unionized boron acceptors at 300 K [11]. Then, they are exposed for 2 h to a radio frequency deuterium or hydrogen plasma at the above mentioned temperatures. The depth diffusion profiles of deuterium and the absolute boron concentration in these epilayers were measured using a CAMECA IMS 4f secondary ion mass spectrometer (SIMS ). Quantifications of deuterium and boron were carried out using a homoepitaxial undoped MPCVD diamond standard where the two elements were co-implanted. For [B]/[C ] ratio in the gas phase of 25, 200 and 800 ppm, the boron concentrations found in the diamond epilayers are, respectively, 2×1018, 5×1019 and 3×1020 cm−3. The infrared absorption experiments on the as-grown and the hydrogenated boron-doped diamond films were measured at 10 K with a BOMEM DA 8 Fourier transform interferometer. The resolution of 0.5 cm−1 was well below the half-widths of the absorption bands.
279
Fig. 1. Evolution of the deuterium profile in homoepitaxial borondoped diamond as a function of the boron concentration.
boron concentration, and finally a sharp decrease of the deuterium concentration. The plateau extent increases as the boron concentration decreases. The dependences of the plateau deuterium solubility and of the plateau extent L on the boron concentration provide strong D evidence for boron-limited diffusion of hydrogen [14] and for (B,H ) pair formation in diamond. Fig. 2 presents the temperature dependence of the deuterium diffusion profiles for the boron concentration
3. Results and discussion Fig. 1 shows the deuterium diffusion profiles in different boron-doped diamond epilayers exposed together to a deuterium r.f. plasma (T=290 °C, t =2 h). The diffusion profile is composed of a surface 0 deuterium accumulation region, followed by a plateau where the deuterium concentration is very close to the
Fig. 2. Evolution of the deuterium diffusion profile in homoepitaxial boron-doped diamond as a function of diffusion temperature. The uniform boron concentration is also shown.
280
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
of 5×1019 cm−3. The hydrogen plateau concentration remains very close to the boron concentration for diffusion temperatures as high as 480 °C. The effective diffusion coefficient L2 /t increases from D 0 8×10−15 cm2/s at 235 °C to 1×10−13 cm2/s at 480 °C. Plotting L2 /t as a function of the reciprocal diffusion D 0 temperature yields an activation energy of 0.35 eV for the deuterium diffusion. Qualitatively, the above results are similar to those for deuterium diffusion in boron-doped silicon [14,15]. In silicon, the features of (B,H ) pair formation during hydrogen diffusion disappear for diffusion temperatures above 250 °C as a result of the thermal dissociation of (B,H ) pairs [16 ]. However, in diamond, we observe that the trapping of hydrogen on boron acceptors remains at diffusion temperatures as high as 480 °C, which means that (B,H ) pairs are more stable in diamond than in silicon. Theoretical studies on diamond have concluded that H+ would exhibit a low migration energy E of m 0.1 eV [17] (instead of 0.4–0.5 eV in silicon [14,18,19]), and that B–H complexes would have a binding energy E of 0.3 eV [17] (instead of 0.6–0.8 eV in silicon b [14,19]). Assuming that the (B,H ) pairs dissociate into B− and H+ as in silicon, and that the frequency factors for the B–H pair thermal dissociation are of the same order of magnitude in silicon and in diamond, our above experimental results seem to contradict the calculations, giving a lower dissociation energy of these pairs in diamond [E +E (H+)=0.4 eV ] compared with silicon b m (1.0–1.3 eV ) [14,19]. Until now, only one hydrogen diffusion coefficient of 2.4×10−13 cm2/s at 860 °C has been reported by Popovici et al. [20], with no report of the activation energy. This diffusion coefficient was measured after diffusion of several very different chemical species in type IIa natural diamond: there was a first step of N, Li and O diffusion at 860 °C before co-diffusion of B and H. The nearly complementary error function (erfc) type hydrogen diffusion profile was in the 1019 cm−3 range, while those of N, B and O were in the 1020 cm−3 range. For Li, it was in the 1017 cm−3 range. The existence of vacancies, vacancy complexes, microvoids and dislocations in IIa natural diamond crystals was invoked to justify the close values (2.4 to 4.4×10−13 cm2/s) of the reported diffusion coefficients of H, N and O in this sample. In Figs. 1 and 2, we observe, near the surface, more usual deuterium diffusion profiles with low diffusion coefficients and high surface concentrations. These profiles are of the same type as those existing after film deposition [10] or room temperature plasma deuteration (hydrogenation) of CrO /H SO treated samples [9]. 3 2 4 While this accumulation layer induces a strong superficial p-type conductivity [9,10], knowledge of the physicochemical characteristics of this layer remains very poor. For a concentration of 5×1019 cm−3 boron
acceptors, its depth increases from about 50 to 100 nm as the diffusion temperature increases from 235 to 480 °C ( Fig. 2). Because the hydrogen concentration in this superficial layer is higher than that of boron, diffusion of single hydrogen cannot take place as H+, and theoretical works [17,18] predict much higher migration energy (1.9 eV and 2.5 eV, respectively, for H0 and H−), leading anyway to much lower diffusion coefficients for hydrogen as found here. It would be interesting to know whether this layer contains point defects with hydrogen atoms or extended defects like platelets, as found at these very high hydrogen concentrations in silicon [21]. For diamond, a theoretical work predicts a higher stability for H* complexes (with one hydrogen at the 2 bond center site and the other one at the antibonding site of a C–C bond [22–24]) than for H molecules in 2 tetrahedral sites [18]. On the contrary, in silicon, H 2 molecules are slightly more stable than H* complexes, 2 but both species have been detected [18,25,26 ]. In order to investigate the influence of hydrogenation on the boron acceptor electronic properties in diamond, we have studied the modifications under hydrogen diffusion of the infrared absorption spectrum related to the electronic transitions from the ground to the excited states of boron acceptors. Fig. 3 presents the infrared transmission spectra of a diamond film doped with boron at a concentration of 1.5×1018 cm−3 before and
Fig. 3. Infrared transmission spectrum of a boron-doped diamond film before and after hydrogen plasma (t=14 h, T=490 °C ). Before plasma, we clearly see the two absorption peaks at 2450 and 2820 cm−1 attributed to the electronic transitions between the ground state and the excited states of boron acceptors. The spectrum of an undoped diamond film is shown for comparison.
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
after hydrogen plasma exposure. For comparison, it also shows the spectrum of an undoped diamond film. Before hydrogen plasma, we clearly observe two absorption peaks at 2450 and 2820 cm−1, respectively, attributed to the ground statefirst excited state electronic transition and to the ground statesecond excited state transition. After hydrogen diffusion for 14 h at 490 °C, these absorption peaks almost vanish, which means that the concentration of neutral boron acceptors significantly decreases under hydrogenation. The residual absorption after hydrogen diffusion is due to the remaining neutral boron acceptors in the hydrogenated region and/or an insufficient hydrogen diffusion depth. In terms of (B,H ) pair formation mechanism, we expect some difference between silicon and diamond since, in silicon, all boron acceptors are ionized while, in diamond, most of the boron acceptors are neutral because of their high ionization energy (0.38 eV ). For example, in diamond containing 1018 cm−3 boron acceptors and having a compensation ratio of 10%, 84% of boron acceptors are neutral at a temperature of 320 °C. This difference will have some consequences for the (B,H ) pair formation mechanism. Because there is experimental evidence that H+ is a diffusing species in p-type silicon [27], the (B,H ) pair formation is considered as a result of the compensation of free holes followed by the Coulombic attraction between H+ and B−. A theoretical work has predicted that hydrogen introduces a mid-gap H+/0 donor level in the band gap of diamond [17], therefore above the fundamental level of boron. As a result, neutral hydrogen atoms entering boron-doped diamond will provide their electron: (i) to compensate the small concentration of free holes coming from the ionization of boron acceptors, knowing that only a small fraction of the boron concentration is ionized in diamond; and (ii) to ionize all the remaining neutral boron acceptors. This second effect is specific to the high ionization energy of boron acceptors in diamond. Then, in the plateau region where the hydrogen concentration is equal to the boron concentration, diamond behaves as a compensated semiconductor, similar to silicon. The ionization of all boron acceptors in B− and of all hydrogen in H+ is followed by (B,H ) pair formation at the considered diffusion temperatures. The ionization of boron acceptors explains the decrease of the absorption related to boron electronic transitions presented in Fig. 3. Another explanation of this decrease is the possible removal of the boron acceptor electronic states, which could result from the formation of (B,H ) pairs.
4. Conclusion We have provided direct evidence of deuterium diffusion limited by boron concentration in p-type diamond
281
as a result of trapping of deuterium by boron acceptors. The effective diffusion coefficient of deuterium varies from 8×10−15 cm2/s at 235 °C to 1×10−13 cm2/s at 480 °C for a boron concentration of 5×1019 cm−3. Due to a mid-gap donor behavior of hydrogen in diamond (theoretical work), boron acceptors become negatively charged under hydrogen incorporation. Hydrogen diffuses as H+ and forms pairs with charged boron acceptors. Additionally, there is a superficial deuterium accumulation layer with the usual diffusion profile and temperature behavior at least up to 480 °C, but whose physicochemical nature remains unknown. To our knowledge, this is the first experimental work showing direct evidence of hydrogen–boron acceptor interactions in diamond.
Acknowledgements We are very grateful to O. Kaitasov (CSNSM, Orsay) for the preparation of implanted diamond standards. We also thank E. Bustarret (LEPES) for critical reading of the manuscript and to F. Pruvost (LEPES) for the growth of some of the samples used in this work.
References [1] C.H. Seager, in: J.I. Pankove, N.M. Johnson (Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 17–33. [2] E.E. Haller, in: J.I. Pankove, N.M. Johnson (Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 351–380. [3] J. Chevallier, B. Clerjaud, B. Pajot, in: J.I. Pankove, N.M. Johnson ( Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 447–510. [4] S.J. Pearton, in: S.J. Pearton ( Ed.), Hydrogen in Compound Semiconductors (Trans. Tech. Publications), Mater. Sci. Forum 113 (1994). [5] A.L. Endro¨s, W. Kru¨hler, F. Koch, J. Appl. Phys. 72 (1992) 2264. ¨ . Sveinbjo¨rnsson, O. Engstro¨m, Phys. Rev. B 52 (1995) 4884. [6 ] E.O [7] J. Chevallier, Defect Diffus. Forum 131–132 (1996) 9. [8] J. Ku¨ppers, Surf. Sci. Rep. 22 (1995) 249. [9] M.I. Landstrass, K.V. Ravi, Appl. Phys. Lett. 55 (1989) 1391. [10] K. Hayashi, S. Yamanaka, H. Watanabe, T. Sekiguchi, H. Okushi, K. Kajimura, J. Appl. Phys. 81 (1997) 744. [11] E. Gheeraert, A. Deneuville, J. Mambou, P. Ribot, in: C.A.J. Ammerlaan, B. Pajot ( Eds.), Shallow-level Centers in Semiconductors, Amsterdam, July 1996, World Scientific, Singapore, 1997, p. 159. [12] J.P. Lagrange, A. Deneuville, E. Gheeraert, Diamond Relat. Mater., in press. [13] S.A. Grot, G.Sl. Gildenblat, C.W. Hatfield, C.R. Wronski, A.R. Badzian, T. Badzian, R. Messier, IEEE Electron Device Lett. 11 (1990) 100. [14] C.P. Herrero, M. Stutzmann, A. Breitschwerdt, P.V. Santos, Phys. Rev. B 41 (1990) 1054. [15] N.M. Johnson, M.D. Moyer, Appl. Phys. Lett. 46 (1985) 787. [16 ] J.C. Mikkelsen Jr., Appl. Phys. Lett. 46 (1985) 882.
282
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
[17] S.P. Mehandru, A.B. Anderson, J. Mater. Res. 9 (1994) 383. [18] S.K. Estreicher, M.A. Roberson, Dj.M. Maric, Phys. Rev. B 50 (1994) 17018. [19] C. Herring, N.M. Johnson, in: J.I. Pankove, N.M. Johnson ( Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 225–350. [20] G. Popovici, R.G. Wilson, T. Sung, M.A. Prelas, S. Khasawinah, J. Appl. Phys. 77 (1995) 5103. [21] N.M. Johnson, in: J.I. Pankove, N.M. Johnson (Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 113–137.
[22] P. Briddon, R. Jones, G.M.S. Lister, J. Phys. C 21 (1988) (1988) L1207. [23] P. Dea`k, L.C. Snyder, J.W. Corbett, Phys. Rev. B 37 (1988) 6887. [24] K.J. Chang, D.J. Chadi, Phys. Rev. Lett. 62 (1989) 937. [25] K. Murakami, N. Fukata, S. Sasaki, K. Ishioka, M. Kitajima, S. Fujimura, J. Kikuchi, H. Haneda, Phys. Rev. Lett. 77 (1996) 3161. ¨ berg, Phys. [26 ] J.D. Holbech, B. Bech-Nielsen, R. Jones, P. Sitch, S. O Rev. Lett. 71 (1993) 875. [27] A.J. Tavendale, A.A. Williams, D. Alexiev, S.J. Pearton, Mater. Res. Soc. Symp. Proc. 59 (1986) 469.
Evidence of hydrogen–boron interactions in diamond from deuterium diffusion and infrared spectroscopy experiments J. Chevallier a,*, A. Lusson a, B. Theys a, A. Deneuville b, E. Gheeraert b a Laboratoire de Physique des Solides de Bellevue, CNRS, 1 place A. Briand, 92195 Meudon, Cedex, France b Laboratoire d’Etudes des Proprie´te´s Electroniques des Solides, CNRS, BP 166, 38042 Grenoble, France Received 25 July 1998; accepted 17 October 1998
Abstract We report on the first experimental evidence of hydrogen–boron interactions in boron-doped diamond. Deuterium diffusion studies in homoepitaxial B-doped diamond films reveal that hydrogen diffusion is limited by the B concentration and is characterized by a low effective diffusion activation energy. Infrared spectroscopy experiments show that boron acceptor electronic transitions disappear under hydrogenation. These results are consistent with hydrogen ionization and diffusion of fairly mobile H+ which form pairs with B−. © 1999 Elsevier Science S.A. All rights reserved. Keywords: Diamond; Doping; Hydrogen; Diffusion; Infrared absorption
1. Introduction Hydrogen in conventional semiconductors has been extensively studied in the last decade because this impurity may be non-intentionally incorporated in the crystal lattice during epitaxial or bulk growth and during a number of steps used for semiconductor device processing [1–4]. The attachment of hydrogen to the crystal imperfections causes important modifications of the electronic properties. For example, hydrogen makes active impurities which are normally inactive: this is the case for C and Si in germanium and for C in silicon. (C,H ) and (Si,H ) are shallow acceptors in germanium [2], and (C,H ) related complexes act as deep centers in silicon [5]. Hydrogen modifies the electronic properties of deep levels by passivating them or by introducing new deep centers: this is the case for silicon doped with transition elements where H–transition metal complexes introduce additional deep levels in the band gap [6 ]. Furthermore, a great deal of attention has also been given to the hydrogen–dopant interactions in silicon, III–V and II–VI compounds, because the hydrogen– acceptor and hydrogen–donor complex formations are at the origin of the neutralization of the dopants [7]. The understanding of the behavior of hydrogen in * Corresponding author. Fax: +33 145 0758 41; e-mail: [email protected]
diamond is very important, because the deposition of both polycrystalline and monocrystalline diamond (instead of graphite) films has to be achieved within a very active hydrogen plasma. One consequence is the presence of up to several atomic per cent of hydrogen at the grain boundaries of polycrystalline diamond films [8]. It has also been recognized that the exposure, at 300 K, of undoped or p-type doped diamond to hydrogen plasma induces a superficial (20 nm) highly p-type conductive layer which is associated with the presence of a hydrogen accumulation layer at the surface [9,10]. However, until now, there has been no experimental study of the hydrogen diffusion in p-type diamond. In this work, we present the first results on hydrogen diffusion in boron-doped diamond. We demonstrate the existence of hydrogen–boron interactions, with the hydrogen being trapped by boron acceptors. This understanding is important for a proper control of the conductivity of p-type diamond films.
2. Experimental We order avoid aries.
0925-9635/99/$ – see front matter © 1999 Elsevier Science S.A. All rights reserved. PII: S0 9 2 5- 9 6 3 5 ( 9 8 ) 0 0 32 6 - 4
have used homoepitaxial films of diamond in to control the boron incorporation level and to parasitic trapping of hydrogen at grain boundSeveral films with various boron contents from
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
2×1018 to 3×1020 cm−3 and a low compensation ratio [11,12] have been deuterated or hydrogenated at various temperatures from 235 to 490 °C. From their SIMS deuterium profiles, we have studied the hydrogen diffusion process. From infrared transmission experiments, we have investigated the influence of hydrogenation on the boron acceptor electronic transitions. Monocrystalline boron-doped diamond epitaxial layers were grown on (100) type Ib synthetic diamond substrates by microwave plasma assisted chemical vapor deposition (MPCVD) at T=820 °C with 4% methane in hydrogen and at a total pressure of 30 torr. The thickness of the epilayers was 3 mm. The boron doping was achieved by adding diborane into the gas phase with [B]/[C ] ratios in the gas phase of 25, 200 and 800 ppm. The deposition process is stopped, first by cutting off the methane and diborane flows, then by switching off the plasma 10 min later, and finally by cooling the sample down to room temperature under a hydrogen gas flow. The epitaxial layers are treated at 80 °C during 30 min in boiling H SO saturated by 2 4 CrO which is known to remove the effect of hydrogen 3 from the near surface region of the film [10,13]. Under such preparation conditions, the compensation ratio of these films is rather low (<10%), as measured by conductivity experiments up to 1000 K [12]. These layers exhibit the infrared absorption bands characteristic of the electronic transitions from the ground to the excited states of almost unionized boron acceptors at 300 K [11]. Then, they are exposed for 2 h to a radio frequency deuterium or hydrogen plasma at the above mentioned temperatures. The depth diffusion profiles of deuterium and the absolute boron concentration in these epilayers were measured using a CAMECA IMS 4f secondary ion mass spectrometer (SIMS ). Quantifications of deuterium and boron were carried out using a homoepitaxial undoped MPCVD diamond standard where the two elements were co-implanted. For [B]/[C ] ratio in the gas phase of 25, 200 and 800 ppm, the boron concentrations found in the diamond epilayers are, respectively, 2×1018, 5×1019 and 3×1020 cm−3. The infrared absorption experiments on the as-grown and the hydrogenated boron-doped diamond films were measured at 10 K with a BOMEM DA 8 Fourier transform interferometer. The resolution of 0.5 cm−1 was well below the half-widths of the absorption bands.
279
Fig. 1. Evolution of the deuterium profile in homoepitaxial borondoped diamond as a function of the boron concentration.
boron concentration, and finally a sharp decrease of the deuterium concentration. The plateau extent increases as the boron concentration decreases. The dependences of the plateau deuterium solubility and of the plateau extent L on the boron concentration provide strong D evidence for boron-limited diffusion of hydrogen [14] and for (B,H ) pair formation in diamond. Fig. 2 presents the temperature dependence of the deuterium diffusion profiles for the boron concentration
3. Results and discussion Fig. 1 shows the deuterium diffusion profiles in different boron-doped diamond epilayers exposed together to a deuterium r.f. plasma (T=290 °C, t =2 h). The diffusion profile is composed of a surface 0 deuterium accumulation region, followed by a plateau where the deuterium concentration is very close to the
Fig. 2. Evolution of the deuterium diffusion profile in homoepitaxial boron-doped diamond as a function of diffusion temperature. The uniform boron concentration is also shown.
280
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
of 5×1019 cm−3. The hydrogen plateau concentration remains very close to the boron concentration for diffusion temperatures as high as 480 °C. The effective diffusion coefficient L2 /t increases from D 0 8×10−15 cm2/s at 235 °C to 1×10−13 cm2/s at 480 °C. Plotting L2 /t as a function of the reciprocal diffusion D 0 temperature yields an activation energy of 0.35 eV for the deuterium diffusion. Qualitatively, the above results are similar to those for deuterium diffusion in boron-doped silicon [14,15]. In silicon, the features of (B,H ) pair formation during hydrogen diffusion disappear for diffusion temperatures above 250 °C as a result of the thermal dissociation of (B,H ) pairs [16 ]. However, in diamond, we observe that the trapping of hydrogen on boron acceptors remains at diffusion temperatures as high as 480 °C, which means that (B,H ) pairs are more stable in diamond than in silicon. Theoretical studies on diamond have concluded that H+ would exhibit a low migration energy E of m 0.1 eV [17] (instead of 0.4–0.5 eV in silicon [14,18,19]), and that B–H complexes would have a binding energy E of 0.3 eV [17] (instead of 0.6–0.8 eV in silicon b [14,19]). Assuming that the (B,H ) pairs dissociate into B− and H+ as in silicon, and that the frequency factors for the B–H pair thermal dissociation are of the same order of magnitude in silicon and in diamond, our above experimental results seem to contradict the calculations, giving a lower dissociation energy of these pairs in diamond [E +E (H+)=0.4 eV ] compared with silicon b m (1.0–1.3 eV ) [14,19]. Until now, only one hydrogen diffusion coefficient of 2.4×10−13 cm2/s at 860 °C has been reported by Popovici et al. [20], with no report of the activation energy. This diffusion coefficient was measured after diffusion of several very different chemical species in type IIa natural diamond: there was a first step of N, Li and O diffusion at 860 °C before co-diffusion of B and H. The nearly complementary error function (erfc) type hydrogen diffusion profile was in the 1019 cm−3 range, while those of N, B and O were in the 1020 cm−3 range. For Li, it was in the 1017 cm−3 range. The existence of vacancies, vacancy complexes, microvoids and dislocations in IIa natural diamond crystals was invoked to justify the close values (2.4 to 4.4×10−13 cm2/s) of the reported diffusion coefficients of H, N and O in this sample. In Figs. 1 and 2, we observe, near the surface, more usual deuterium diffusion profiles with low diffusion coefficients and high surface concentrations. These profiles are of the same type as those existing after film deposition [10] or room temperature plasma deuteration (hydrogenation) of CrO /H SO treated samples [9]. 3 2 4 While this accumulation layer induces a strong superficial p-type conductivity [9,10], knowledge of the physicochemical characteristics of this layer remains very poor. For a concentration of 5×1019 cm−3 boron
acceptors, its depth increases from about 50 to 100 nm as the diffusion temperature increases from 235 to 480 °C ( Fig. 2). Because the hydrogen concentration in this superficial layer is higher than that of boron, diffusion of single hydrogen cannot take place as H+, and theoretical works [17,18] predict much higher migration energy (1.9 eV and 2.5 eV, respectively, for H0 and H−), leading anyway to much lower diffusion coefficients for hydrogen as found here. It would be interesting to know whether this layer contains point defects with hydrogen atoms or extended defects like platelets, as found at these very high hydrogen concentrations in silicon [21]. For diamond, a theoretical work predicts a higher stability for H* complexes (with one hydrogen at the 2 bond center site and the other one at the antibonding site of a C–C bond [22–24]) than for H molecules in 2 tetrahedral sites [18]. On the contrary, in silicon, H 2 molecules are slightly more stable than H* complexes, 2 but both species have been detected [18,25,26 ]. In order to investigate the influence of hydrogenation on the boron acceptor electronic properties in diamond, we have studied the modifications under hydrogen diffusion of the infrared absorption spectrum related to the electronic transitions from the ground to the excited states of boron acceptors. Fig. 3 presents the infrared transmission spectra of a diamond film doped with boron at a concentration of 1.5×1018 cm−3 before and
Fig. 3. Infrared transmission spectrum of a boron-doped diamond film before and after hydrogen plasma (t=14 h, T=490 °C ). Before plasma, we clearly see the two absorption peaks at 2450 and 2820 cm−1 attributed to the electronic transitions between the ground state and the excited states of boron acceptors. The spectrum of an undoped diamond film is shown for comparison.
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
after hydrogen plasma exposure. For comparison, it also shows the spectrum of an undoped diamond film. Before hydrogen plasma, we clearly observe two absorption peaks at 2450 and 2820 cm−1, respectively, attributed to the ground statefirst excited state electronic transition and to the ground statesecond excited state transition. After hydrogen diffusion for 14 h at 490 °C, these absorption peaks almost vanish, which means that the concentration of neutral boron acceptors significantly decreases under hydrogenation. The residual absorption after hydrogen diffusion is due to the remaining neutral boron acceptors in the hydrogenated region and/or an insufficient hydrogen diffusion depth. In terms of (B,H ) pair formation mechanism, we expect some difference between silicon and diamond since, in silicon, all boron acceptors are ionized while, in diamond, most of the boron acceptors are neutral because of their high ionization energy (0.38 eV ). For example, in diamond containing 1018 cm−3 boron acceptors and having a compensation ratio of 10%, 84% of boron acceptors are neutral at a temperature of 320 °C. This difference will have some consequences for the (B,H ) pair formation mechanism. Because there is experimental evidence that H+ is a diffusing species in p-type silicon [27], the (B,H ) pair formation is considered as a result of the compensation of free holes followed by the Coulombic attraction between H+ and B−. A theoretical work has predicted that hydrogen introduces a mid-gap H+/0 donor level in the band gap of diamond [17], therefore above the fundamental level of boron. As a result, neutral hydrogen atoms entering boron-doped diamond will provide their electron: (i) to compensate the small concentration of free holes coming from the ionization of boron acceptors, knowing that only a small fraction of the boron concentration is ionized in diamond; and (ii) to ionize all the remaining neutral boron acceptors. This second effect is specific to the high ionization energy of boron acceptors in diamond. Then, in the plateau region where the hydrogen concentration is equal to the boron concentration, diamond behaves as a compensated semiconductor, similar to silicon. The ionization of all boron acceptors in B− and of all hydrogen in H+ is followed by (B,H ) pair formation at the considered diffusion temperatures. The ionization of boron acceptors explains the decrease of the absorption related to boron electronic transitions presented in Fig. 3. Another explanation of this decrease is the possible removal of the boron acceptor electronic states, which could result from the formation of (B,H ) pairs.
4. Conclusion We have provided direct evidence of deuterium diffusion limited by boron concentration in p-type diamond
281
as a result of trapping of deuterium by boron acceptors. The effective diffusion coefficient of deuterium varies from 8×10−15 cm2/s at 235 °C to 1×10−13 cm2/s at 480 °C for a boron concentration of 5×1019 cm−3. Due to a mid-gap donor behavior of hydrogen in diamond (theoretical work), boron acceptors become negatively charged under hydrogen incorporation. Hydrogen diffuses as H+ and forms pairs with charged boron acceptors. Additionally, there is a superficial deuterium accumulation layer with the usual diffusion profile and temperature behavior at least up to 480 °C, but whose physicochemical nature remains unknown. To our knowledge, this is the first experimental work showing direct evidence of hydrogen–boron acceptor interactions in diamond.
Acknowledgements We are very grateful to O. Kaitasov (CSNSM, Orsay) for the preparation of implanted diamond standards. We also thank E. Bustarret (LEPES) for critical reading of the manuscript and to F. Pruvost (LEPES) for the growth of some of the samples used in this work.
References [1] C.H. Seager, in: J.I. Pankove, N.M. Johnson (Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 17–33. [2] E.E. Haller, in: J.I. Pankove, N.M. Johnson (Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 351–380. [3] J. Chevallier, B. Clerjaud, B. Pajot, in: J.I. Pankove, N.M. Johnson ( Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 447–510. [4] S.J. Pearton, in: S.J. Pearton ( Ed.), Hydrogen in Compound Semiconductors (Trans. Tech. Publications), Mater. Sci. Forum 113 (1994). [5] A.L. Endro¨s, W. Kru¨hler, F. Koch, J. Appl. Phys. 72 (1992) 2264. ¨ . Sveinbjo¨rnsson, O. Engstro¨m, Phys. Rev. B 52 (1995) 4884. [6 ] E.O [7] J. Chevallier, Defect Diffus. Forum 131–132 (1996) 9. [8] J. Ku¨ppers, Surf. Sci. Rep. 22 (1995) 249. [9] M.I. Landstrass, K.V. Ravi, Appl. Phys. Lett. 55 (1989) 1391. [10] K. Hayashi, S. Yamanaka, H. Watanabe, T. Sekiguchi, H. Okushi, K. Kajimura, J. Appl. Phys. 81 (1997) 744. [11] E. Gheeraert, A. Deneuville, J. Mambou, P. Ribot, in: C.A.J. Ammerlaan, B. Pajot ( Eds.), Shallow-level Centers in Semiconductors, Amsterdam, July 1996, World Scientific, Singapore, 1997, p. 159. [12] J.P. Lagrange, A. Deneuville, E. Gheeraert, Diamond Relat. Mater., in press. [13] S.A. Grot, G.Sl. Gildenblat, C.W. Hatfield, C.R. Wronski, A.R. Badzian, T. Badzian, R. Messier, IEEE Electron Device Lett. 11 (1990) 100. [14] C.P. Herrero, M. Stutzmann, A. Breitschwerdt, P.V. Santos, Phys. Rev. B 41 (1990) 1054. [15] N.M. Johnson, M.D. Moyer, Appl. Phys. Lett. 46 (1985) 787. [16 ] J.C. Mikkelsen Jr., Appl. Phys. Lett. 46 (1985) 882.
282
J. Chevallier et al. / Diamond and Related Materials 8 (1999) 278–282
[17] S.P. Mehandru, A.B. Anderson, J. Mater. Res. 9 (1994) 383. [18] S.K. Estreicher, M.A. Roberson, Dj.M. Maric, Phys. Rev. B 50 (1994) 17018. [19] C. Herring, N.M. Johnson, in: J.I. Pankove, N.M. Johnson ( Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 225–350. [20] G. Popovici, R.G. Wilson, T. Sung, M.A. Prelas, S. Khasawinah, J. Appl. Phys. 77 (1995) 5103. [21] N.M. Johnson, in: J.I. Pankove, N.M. Johnson (Eds.), Hydrogen in Semiconductors, Semiconductors and Semimetals, vol. 34, Academic Press, San Diego, 1991, pp. 113–137.
[22] P. Briddon, R. Jones, G.M.S. Lister, J. Phys. C 21 (1988) (1988) L1207. [23] P. Dea`k, L.C. Snyder, J.W. Corbett, Phys. Rev. B 37 (1988) 6887. [24] K.J. Chang, D.J. Chadi, Phys. Rev. Lett. 62 (1989) 937. [25] K. Murakami, N. Fukata, S. Sasaki, K. Ishioka, M. Kitajima, S. Fujimura, J. Kikuchi, H. Haneda, Phys. Rev. Lett. 77 (1996) 3161. ¨ berg, Phys. [26 ] J.D. Holbech, B. Bech-Nielsen, R. Jones, P. Sitch, S. O Rev. Lett. 71 (1993) 875. [27] A.J. Tavendale, A.A. Williams, D. Alexiev, S.J. Pearton, Mater. Res. Soc. Symp. Proc. 59 (1986) 469.