Electrical properties of boron nitride thin films grown by neutralized nitrogen ion assisted vapor deposition Ming Lu, A. Bousetta,a) and A. Bensaoula Space Vacuum Epitaxy Center, University of Houston, Houston, Texas 77204-5507
K. Waters and J. A. Schultz IONWERKS, 2472 Bolsover Suite 255, Houston, Texas 77005
~Received 29 September 1995; accepted for publication 20 November 1995! Boron nitride ~BN! thin films ~containing mixed cBN/hBN phase! have been deposited on Si~100! substrates using neutralized nitrogen beam and electron beam evaporation of boron. All as-deposited BN films were p type with a room-temperature carrier concentration in the range of 531016 to 131017 cm23 . The Mg-doped BN films showed carrier concentrations in the range of 1.2 31018 cm23 to 5.231018 cm23 when the Mg cell temperature was varied from 250 to 500 °C. The films were analyzed for both majority elements ~B and N! and dopant/impurity ~Si, Mg, Fe, etc.! incorporation using secondary ion mass spectroscopy and mass spectroscopy of recoiled ions ~MRSI!. MRSI is shown to be superior for dopant characterization of boron nitride thin films. © 1996 American Institute of Physics. @S0003-6951~96!01605-1#
Boron nitride single crystals, with zinc blende structure ~cBN!, were first synthesized from hexagonal BN ~hBN! by Wentorf in 1957 using high-pressure and high-temperature process in the presence of a suitable catalyst.1,2 More recently, several groups3–5 have been able to synthesize BN thin films with various amounts of the metastable cubic phase. The increased interest in this material is due to the fact that the cubic phase boron nitride ~cBN! has many properties similar to that of diamond: wide band gap ~.6 eV!, high thermal conductivity, and good transmittance over a large spectral range from UV to visible.6 – 8 In addition and unlike diamonds, cBN can be doped both p and n type.9,10 Envisioned applications for this material in a thin film form include its use as hard and protective coatings for cutting tools and optical instruments, as well as its use in the fabrication of UV optical detectors and emitters and hightemperature electronic devices.11 To our knowledge, the only demonstrated cBN device is a p-n junction diode made from bulk materials produced under high pressure.10 In this letter we report the growth of mixed phase-BN thin films on Si~100! using a neutralized nitrogen beam and electron beam evaporation of boron and its successful doping with Mg to achieve electrically active p-type BN thin films. A schematic diagram of the experimental apparatus is shown in Fig. 1. The neutralizer atomic beam source ~NABS!5 was fitted onto a 3 cm diam Kaufman-type ion source; 1 in. diam Si~100! n-type wafers were degreased using standard solvents and etched for 1 min in a buffered HF solution. The substrates were thermally cleaned in situ at 850 °C for 15 min. The substrate temperature was maintained in the range of 400–500 °C during the film deposition. During growth, the samples were exposed to either a pure nitrogen beam or a fixed nitrogen/argon controlled by setting the individual Ar and N2 flow controllers. A constant presa!
Author to whom correspondence should be addressed. Electronic mail: [email protected]
Appl. Phys. Lett. 68 (5), 29 January 1996
sure of 331024 Torr was maintained for all experiments in the ion source chamber. The base pressure in the growth chamber was 431028 Torr which rose to 1026 Torr during deposition. Boron ~99.5% purity! was evaporated from an electron beam evaporator at a rate maintained at about 0.2 Å/s using a quartz crystal monitor. For the doping studies, Mg ~5N purity! was evaporated using a standard effusion cell. The total thickness and stoichiometry ~B/N ratio! were measured using electron probe microanalysis ~EPMA!. The Mg incorporation and metallic impurities in the BN films were measured by secondary ion mass spectroscopy ~SIMS! and the results were compared to those obtained using mass spectroscopy of recoiled ions ~MSRI!.12 The electrically active dopant concentration levels in the film were measured by standard four-point probe resistivity and Hall techniques. For the purpose of Hall and resistivity measurements indium pellets were mechanically pressed onto the sample surface. The sample was then annealed also in a Ne atmosphere at 500 °C for 30 min. The annealing process was critical in obtaining contacts with good adhesion and adequate ohmic properties. Fourier-transform infrared ~FTIR! measurements on the BN films show a large spread in their cubic content ~from 0% to 30%!. The stoichiometry ~B/N ratios ranges from 3 to nearly 1! also varies dramatically with both the growth tem-
FIG. 1. Schematic diagram of the experimental growth chamber. The neutralizer atomic beam source is shown fitted to a Kaufman-type ion source.
© 1996 American Institute of Physics
TABLE I. EPMA B/N ratio, resistivity, carrier concentrations, and cubic phase content of as-grown samples and Mg-doped samples at different Mg cell temperatures. All samples were grown under similar conditions, 75/25 N2 /Ar gas mixture and 450 °C substrate temperature, except as noted. a! grown with 50/50 N2 /Ar gas mixture, b! grown with a substrate temperature of 500 °C, and c! substrate temperature of 400 °C.
T Mg ~°C!
r (31022 V cm)
Mg concentration (31018 cm23 )
Mobility (cm2 /v s)
% cubic ~FTIR!
Not used Not used 250 °C 300 °C 350 °C 400 °C 450 °C 450 °C 450 °C 500 °C
BN127a BN228b BN242 BN243 BN244 BN233 BN234 BN235c BN236d BN237
NA 1.11 1.59 1.31 1.24 2.31 2.17 1.42 NA 1.55
78.8 8.27 10.5 0.893 5.71 12.8 6.69 13.2 5.29 6.93
0.0535 0.427 1.25 26.8 2.09 1.89 4.18 1.63 4.46 5.16
148.0 177.0 47.4 26.2 52.4 35.2 22.3 29.1 26.5 17.5
0% 16% 28% 21% 0% 0% 33% 0% 15% 15%
50/50 N2 /Ar gas mixture and 500 °C substrate temperature. 75/25 N2 /Ar gas mixture and 400 °C substrate temperature. c 50/50 N2 /Ar gas mixture and 450 °C substrate temperature. d 75/25 N2 /Ar gas mixture and 500 °C substrate temperature. b
perature, the ion beam energy, and the Ar/N2 flow mixture. As reported in Table I, the resistivity and Hall measurements show that the residual doping of the as-grown BN films is p type with resistivity ranging from 0.08 to 1 V cm, carrier concentrations between 531016 and 131017 cm23 with corresponding mobilities of 177 to 48 cm2 /V s. B/N ratio, and the cubic phase content are also reported in Table I. This intrinsic p-type behavior is consistent with results reported in the literature13 and is possibly attributed to trace metal impurities inadvertently added to the film during deposition or to B antisites as a result of neutral bombardment.14 The Mg doping results are summarized in Fig. 2 from which the activation energy for Mg in BN thin films is found to be 0.3 eV. The activation energy is rather low but within the range published by Tanigushi ~0.25 to 1 eV! ~Ref. 15! and is also consistent with our difficulty in achieving low level doping for this material. The corresponding resistivities
FIG. 2. Hall measurements of Mg dopant levels in BN films as a function of inverse Mg temperature. The activation energy from these results is found from the slope to be 0.3 eV. The data points are taken from Table I and the open circle points indicates the 50/50 N2 /Ar gas mixture film. Appl. Phys. Lett., Vol. 68, No. 5, 29 January 1996
in these films, their Hall mobilities and carrier concentrations are also reported in Table I. In order to verify the incorporation of Mg dopants in the BN films, the latter were analyzed using SIMS sputter profile and MSRI analysis. In the latter technique a pulsed keV ion beam is impinged at grazing incidence onto a surface and the surface atoms which are directly recoiled and ionized during binary collisions with primary ions are subsequently mass analyzed by time of flight. A significant advantage of MSRI spectra are that they contain only atomic information and are, therefore, unambiguously interpreted. This occurs because only recoils having a few keV of energy are accepted by a time-of-flight ~TOF! analyzer which thus eliminates slow molecular ions. More detailed information on the MSRI analysis technique has been published elsewhere.12 The SIMS results are presented in Fig. 3 and the MSRI spectra from the same BN thin films used in the SIMS analysis are shown in Fig. 4. In SIMS profiles, Mg is identified both in a highly doped BN film (2.031018 cm23 ) and somewhat in an as-grown BN film at a much higher level.
FIG. 3. SIMS results showing the presence of contaminants from stainless steel from ~a! a 2.031018 cm23 Mg-doped BN film and ~b! an undoped BN film. Notice the apparent Mg level in the undoped film which is seemingly higher than in the doped sample. MSRI has confirmed the presence of Mg in the doped film and its absence in the undoped. Therefore, the SIMS signal at mass 24 must be due to BN1 molecular interference. Lu et al.
FIG. 4. MSRI spectra from the ~a! 2.031018 cm23 Mg-doped BN film and ~b! an undoped BN film. MSRI shows the presence of Fe, Ni, and CR in the doped sample. For the undoped sample we have MSRI analysis showing no Mg. In ~b! we have, however, intentionally allowed secondary ions into the analyzer during MSRI analysis of the undoped sample in order to illustrate the complexity of the SIMS analysis. The spikes are not noise but are rather from molecular ions at each mass unit.
However, MSRI analysis shown in Fig. 4 has verified the presence of Mg in the doped film and its absence in the undoped film. Therefore, the SIMS signal at mass 24 in the undoped film is not related to Mg but rather to a BN1 molecular interference. The results show the effectiveness of MSRI in analyzing the dopant presence in the film. In SIMS analysis, the concentration of Ni, Cu, Fe, and Cr contaminations have been assigned by resorting to elemental standard in implanted Si3 N4 since no such data base exists for BN films. The mass 56 intensity in the SIMS spectrum has been shown by isotopic analysis to, at least par1 tially, come from Si1 2 which is isobaric with Fe . This interference can easily be seen at the Si interface where the apparent Fe signal increases as Si1 2 ions are formed during sputtering through the BN film into the Si substrate. The true Fe levels are undoubtedly much lower than those presented in Fig. 3 because of this interference. The numerous compound fragments and molecular species that are generated during ion bombardment of boron nitride ~and perhaps CN and AlN as well! makes the identification of dopant elements by SIMS analysis rather difficult. In the MSRI spectra, impurities such as Al, Si, Fe, and Cr are also seen. The Fe and Cr may originate from the stainless-steel structure of the neutralizer ~NABS! assembly. Also some of these impurities probably come from the boron source itself since the pellets uses are only 99.5% pure. The remaining 0.5% impurities comprise in part 0.008% Al, 0.07% Fe, 0.028% Mn, and 0.08% Si. Analysis of films grown after prolonged use of the B source show much less impurities due to the boil off of the higher vapor elements
Appl. Phys. Lett., Vol. 68, No. 5, 29 January 1996
from the crucible during each successive deposition run. Figure 4~b! shows a MSRI spectrum from the as-grown BN sample which was intentionally biased to collect SIMS ions simultaneously with the directly recoiled ions. Although we have verified with MSRI that Mg was not in the undoped film, we show the combined MSRI/SIMS spectra in Fig. 4~b! to illustrate the complexity of the SIMS analysis for such materials. Although there is no Mg in the undoped sample, the molecular ions from mass 20 on up are isobarically interfering with Na, Mg, and Al analyses. We have reproducibly doped BN thin films ~containing mixed hBN/cBN phases! with Mg to a high carrier concentration level of 631018 cm23 (2.031019 cm23 in one sample!. SIMS analyses of the films was shown to be of only limited use because of the numerous molecular interferences 1 (BN1 /Mg1 and Si1 2 /Fe ). MSRI analysis, however, demonstrated unequivocally that Mg is incorporated in the BN films. Although a complicated electrical behavior in the BN films might be expected because of multiple phases and the presence of several trace of impurities, the present work, nevertheless, holds hope for improvements in these materials and offers new tools for their characterization to be superior to SIMS for detecting dopants even when performed ex situ. This work is supported at the University of Houston by NASA, Grant No. NAGW977 and State of Texas ARP Grant No. 00365224. The work at Ionwerks was in part supported by the Army Research Office through SBIR Contract No. DAAH04-93-C-0009 and the National Science Foundation through Grant No. III-9361718. R. H. Wentorf, Jr., J. Chem. Phys. 26, 956 ~1957!. R. H. Wentorf, Jr., J. Chem. Phys. 34, 809 ~1961!. 3 D. J. Kester and R. Messier, J. Appl. Phys. 72, 506 ~1992!. 4 T. Wada and N. Yamashta, J. Vac. Sci. Technol. A 10, 515 ~1992!. 5 M. Lu, A. Bousetta, R. Sukach, A. Bensaoula, K. Waters, K. Eipers-Smith, and J. A. Schultz, Appl. Phys. Lett. 64, 1516 ~1994!. 6 J. N. Plendl and P. J. Gielisse, Phys. Rev. B 125, 828 ~1972!. 7 L. Vel, G. Demazeau, and J. Etourneau, Mater. Sci. Eng. B 10, 149 ~1991!. 8 S. P. Arya and A. D’Amico, Thin Solid Films 157, 267 ~1988!. 9 R. H. Wentorf, Jr., J. Chem. Phys. 36, 1990 ~1962!. 10 O. Mishima, K. Era, J. Tanaka, and S. Yamoka, Appl. Phys. Lett. 53, 962 ~1988!. 11 R. F. Davis, Proc. IEEE 79, 702 ~1991!. 12 K. Waters, A. Bensaoula, A. Schultz, K. Eipers-Smith, and A. Freundlich, J. Cryst. Growth 127, 972 ~1993!. 13 Y. Bar-Yam, T. Lei, T. D. Moustakas, D. C. Allan, and M. P. Tejer, Mater. Res. Soc. Symp. Proc. 242, 335 ~1992!. 14 N. Badi, A. Bousetta, M. Lu, and A. Bensaoula, in Applied Diamond Conference Proceedings, Third International Conference, 1995 ~NIST, Gaithersburg, MD, 1995!, p. 849. 15 T. Taniguchi, J. Tanaka, O. Mishima, T. Osawa, and S. Yamaoka, Appl. Phys. Lett. 62, 576 ~1993!. 1 2
Lu et al.