ARTICLES

Shallow donors with high n-type electrical conductivity in homoepitaxial deuterated boron-doped diamond layers ZÉPHIRIN TEUKAM1, JACQUES CHEVALLIER∗1, CÉCILE SAGUY2, RAFI KALISH2, DOMINIQUE BALLUTAUD1, MICHEL BARBÉ1, FRANÇOIS JOMARD1, ANNIE TROMSON-CARLI1, CATHERINE CYTERMANN2, JAMES E. BUTLER3, MATHIEU BERNARD4, CÉLINE BARON4 AND ALAIN DENEUVILLE4 1

Laboratoire de Physique des Solides et de Cristallogénèse,UMR CNRS 8635,1 place A.Briand,92195 Meudon Cedex,France Physics Department and Solid State Institute,Technion,Haifa 32000,Israel 3 Naval Research Laboratory,Code 6174,Washington,DC 20375,USA 4 Laboratoire d’Etudes des Propriétés Electroniques des Solides,CNRS B.P.166,38042 Grenoble Cedex 09,France *e-mail: [email protected] 2

Published online: 22 June 2003; doi:10.1038/nmat929

Diamond is a unique semiconductor for the fabrication of electronic and opto-electronic devices because of its exceptional physical and chemical properties. However, a serious obstacle to the realization of diamond-based devices is the lack of n-type diamond with satisfactory electrical properties.

Here

we

show

that

high-conductivity

n-type diamond can be achieved by deuteration of particularly selected homo-epitaxially grown (100) borondoped diamond layers. Deuterium diffusion through the entire boron-doped layer leads to the passivation of the boron acceptors and to the conversion from highly p-type to n-type conductivity due to the formation of shallow donors with ionization energy of 0.23 eV. Electrical conductivities as high as 2 Ω–1 cm–1 with electron mobilities of the order of a few hundred cm2 V–1 s–1 are measured at 300 K for samples with electron concentrations of several 1016 cm–3. The formation and break-up of deuterium-related complexes, due to some excess deuterium in the deuterated layer, seem to be responsible for the reversible p- to n-type conversion. To the best of our knowledge, this is the first time such an effect has been observed in an elemental semiconductor.

D

iamond is especially attractive as a semiconductor because of its unique combination of good electrical, optical, thermal and chemical properties1. A major challenge in the fabrication of diamond-based electronics and opto-electronics devices is the achievement of high-conductivity n-type diamond. Up to now, the best n-type diamond has been obtained for homoepitaxial phosphorus-doped layers, but with electrical properties that are still unsatisfactory for most device applications. The conductivities at room temperature for such phosphorus-doped samples are usually lower than 10–4 Ω–1 cm–1, and the electron mobilities2 at 300 K are below 250 cm2 V–1 s–1. These low values are due to the high ionization energy (0.6 eV) of phosphorus in diamond, and may also be due to the presence of growth-related defects in the (111) epilayers. Here, we show that high-conductivity n-type diamond can be achieved by deuteration of homo-epitaxial (100) boron-doped diamond films. The formation of deuterium-related complexes seems to be responsible for the p- to n-type conversion.These complexes break-up under thermal annealing until the samples return to the original boron-related p-type conductivity. Homoepitaxial layers of boron-doped diamond,0.5 µm thick,were grown at 900–950°C in a conventional 1.5kW Astex HPMM microwave plasma chemical vapour deposition (MPCVD) system. The substrates were 3.5 × 3.5 × 1.5 mm (100)-type Ib high-pressure high-temperature synthetic diamonds. The gas mixture was composed of 0.44% CH4 and 99.56% H2 at a pressure of 15torr,and the total flow-rate was 906s.c.c.m. The growth rate was 0.17 µm h–1. Boron doping was achieved by the addition of diborane to the gas mixture. These boron-doped samples were then exposed to a microwave deuterium plasma at 550 °C for 8 h at a deuterium pressure of 10 torr. The deuterated samples were subjected to annealing under vacuum at 520–750 °C. The samples were structurally and, following each step, chemically, electrically and optically characterized as described below. Electron-channelling patterns (ECP) and scanning electron microscopy (SEM) images were performed on a JEOL 840 SEM to check the structural quality of our homoepitaxial layers. The boron nature materials | VOL 2 | JULY 2003 | www.nature.com/naturematerials

482 © 2003 Nature Publishing Group

ARTICLES a Deuterium (after D plasma 550 °C, 8 h)

1020

Concentration (cm–3)

1019

1018

Boron

D plasma + 750 °C, 1 h

1017

1016 Sample 1 1015 0.0

0.2

0.4 Depth (µm)

0.6

0.8

b Figure 2 Secondary-ion mass spectrometry (SIMS) analysis of sample 1.The boron and deuterium concentration profiles are obtained after 8 h of deuteration at 550 °C,and after deuteration followed by thermal annealing at 750 °C for 1 h.

Figure 1 Representative electron-channelling pattern (ECP) and scanning electron microscopy (SEM) image of the samples investigated in this work. a,The ECP exhibits the first- and second-order lines of the backscattered electron diffraction.The analysed volume has an area of 100 µm diameter and a depth of about 100 nm.Using the angular scale and the shift of the pole with respect to the figure centre,one deduces that this (100) epilayer has a misorientation of about 1°.b,SEM image showing the surface morphology of the homoepitaxial layers.The striations are likely due to (h11) microfaceting.

and deuterium concentrations were measured by secondary-ion mass spectrometry (SIMS) analysis using a CAMECA IMS4f instrument. Electrical conductivity and Hall-effect measurements were performed in the van der Pauw configuration at both the Laboratoire de Physique des Solides et de Cristallogénèse at room temperature and at the Technion Institute from 20 °C to 600 °C in a magnetic field of 0.8 T. The electrical contacts consisted of Ag paint dots deposited into four corners of the sample. The surface of the samples was oxidized for one hour in a boiling solution of perchloric/sulphuric/fuming nitric acid (1:3:4) before the electrical measurements to remove the deuterium-related surface conductivity of virgin and deuterated samples. Infraredtransmission experiments at 10 K were performed with a BOMEM DA8 Fourier-transform interferometer to investigate the presence of neutral boron acceptors. In addition, Raman spectroscopy (Renishaw system 1000 microRaman and LABRAM INFINITY Dilor Raman equipment) at room temperature was used to verify the absence of graphite, and X-ray photoelectron spectroscopy (XPS; VG Scientific 220iXL) to investigate the chemical nature

of the near-surface region of the diamond epilayers. The results are presented for two samples (samples 1 and 2) with respective boron concentrations of 1–4 × 1019 cm–3 and 7 × 1018 cm–3. Figure 1a presents a typical ECP of the diamond films investigated in this work. This clearly indicates that the films are (100)-oriented homoepitaxially grown. In Fig. 1b, the SEM image shows a homoepitaxial surface morphology of our diamond films. Figure 2 presents the boron and deuterium concentration profiles in sample 1 after deuteration and following high-temperature (750 °C, 1 h) annealing. As can be seen, the deuterium plasma conditions were sufficient to diffuse deuterium through the entire 0.45-µm-thick epilayer. As expected for deuterium diffusion temperatures equal to or below 550 °C, the deuterium profile closely follows the boron concentration3–6. This shows that boron–deuterium complex formation has occurred throughout the whole thickness of the sample. Similar results were obtained for sample 2. The interaction between boron and deuterium (or hydrogen) gives rise to a strong passivation effect of the boron acceptors, noticeable by the disappearance of the boron acceptor states located 0.37 eV above the valence band maximum of diamond3,7,8. This boron passivation effect is directly observable in Fig.3,which shows the infrared transmittance of the deuterated sample 2 by the disappearance of the absorption bands, related with the two electronic transitions of neutral boron acceptors at 2,450 cm–1 and 2,800 cm–1, corresponding to the transitions from the fundamental state to the first excited state and to the second excited state of neutral boron respectively. Hall effect and conductivity measurements were performed in the as-grown state, the as-deuterated state and after deuteration followed by a low-temperature annealing state, and a higher temperature annealing state of these samples. Table 1 summarizes the results obtained at room temperature; also listed are data for a sample (sample 3), which did not convert to n-type despite prolonged deuteration. The reasons for this absence of conversion are discussed below. After 8 hours of deuteration, both samples 1 and 2 exhibit a conversion from p-type to n-type conductivity. Just after deuteration, the electron concentration of sample 1 is 2 × 1019 cm–3 and its conductivity is 6.4 Ω–1 cm–1 at room temperature. After 520 °C, 0.5 h annealing, this sample shows an electron concentration of 7 × 1016 cm–3 with a mobility of 180 cm2 V–1 s–1, and a conductivity of 2 Ω–1 cm–1.

nature materials | VOL 2 | JULY 2003 | www.nature.com/naturematerials

483 © 2003 Nature Publishing Group

ARTICLES Table 1 Results of SIMS analysis of boron, and Hall effect and conductivity measurements at 300 K for samples 1, 2 and 3 in different states. The free-carrier concentration, carrier mobility and conductivity type of these samples are given in the as-grown state and as-deuterated states (550 °C, 8 h) and after deuteration followed by thermal annealings (sample 1: 520 °C, 0.5 h and then 650 °C, 1 h; sample 2: 600 °C, 0.5 h). Boron concentration

As-grown state

After deuteration 550 °C, 8 h, no annealing

After deuteration plus 520 °C annealing

After deuteration plus 650 °C or 600 °C annealing

Sample 1

1–4 × 1019 cm–3

p = 2.5 × 1019 cm–3 µp = 2.5 cm2 V–1 s–1 E < 0.09 eV

n = 2 × 1019 cm–3 µn = 2 cm2 V–1 s–1

n = 7 × 1016 cm–3 µn = 180 cm2 V–1 s–1 E = 0.24 eV

650 °C annealing, 1 h p = 1 × 1018 cm–3 µp = 5 cm2 V–1 s–1 E = 0.09 eV

Sample 2

7 × 1018 cm–3

p = 1 × 1016 cm–3 µp = 140 cm2 V–1 s–1 E = 0.24 eV

n = 6 × 1015 cm–3 µn=15 cm2 V–1 s–1 E = 0.22 eV

Sample 3

1.5–2.5 × 1019 cm–3

p = 5 × 1016 cm–3 µp = 40 cm2 V–1 s–1 E = 0.21 eV

p = 9 × 1015 cm–3 µp = 130 cm2 V–1 s–1 E = 0.22 eV

These conductivities are more than four orders of magnitude higher than the conductivities of phosphorus-doped n-type diamond2. Similar results, though less ‘dramatic’, were obtained for sample 2 (see Table 1). Figure 4a,b presents the temperature dependence of the free-carrier concentration and the carrier mobility after deuteration for sample 1. The evolution of the electrical properties of the sample following deuteration and subsequent annealings at increasing temperatures are marked with increasing numbers. The following two points should be noted: (1) The electrical properties change dramatically following short annealings at temperatures as low as 520 °C. Directly after deuteration, the material is n-type with a room-temperature carrier concentration (nRT) of 2 × 1019 cm–3 (point 2 in Fig. 4a) and a mobility (µn) of 2 cm2 V–1 s–1. After a first 520 °C thermal treatment for 0.5 hour, nRT has decreased to 7 × 1016 cm–3 (point 3 in Fig. 4a) and µn increased to 180 cm2 V–1 s–1 (Table 1 and Fig. 4b).After a second anneal at 520 °C for another 0.5 hour,nRT reduced further to 1 × 1016 cm–3 (point 4 in Fig. 4a) and µn reached 430 cm2 V–1 s–1.This evolution of the electron mobility is consistent with the well-known increase of mobility with decreasing carrier concentration. Between 300 K and 500 K, the electron mobility has ∼T–3 temperature dependence.This power law is close to the law T–2.5 found above 400 K for electron mobilities in natural diamond and attributed to intervalley phonon scattering9. (2) After annealing at 600 °C for 15 min, sample 1 has lost its clear n-type nature, and no definite type can be deduced from the Hall measurements.Following further annealing at temperatures at or above 600 °C, both samples exhibited unambiguous p-type conductivity (Table 1).As the annealing temperature of sample 1 was increased from 650 °C (line 5 in Fig. 4a) to 750 °C (line 6 in Fig. 4a), the hole concentration increased while their activation energy decreased from 0.14 to 0.09 eV (slopes of curves 5 and 6 in Fig.4a).Moreover,a very high carrier concentration was measured at 600 °C (7 × 1019 cm–3 with no saturation reached). Hence, the acceptor concentration is at least 1 × 1020 cm–3, which yields an activation energy10 of ∼0.09 eV. After annealing at 750 °C for 1 hour,most of the deuterium has diffused out, as seen in Fig. 2 for sample 1. It should be noted that the hole concentration, measured at high temperature, sometimes exceeds the boron concentration obtained by SIMS. The reason for this is that the particular boron-doped

600 °C annealing, 0.5 h p = 2 × 1015 cm–3 µp = 185 cm2 V–1 s–1 E = 0.26 eV

diamonds used for this work were grown on very thick (1.5 mm) type Ib substrates and homoepitaxial boron-doped growth has also occurred on the sample sides. Therefore, the resistivity and Hall-effect results, which provide information on the entire measured sample, exhibit the results of conductivities (both n- and p-type) along the boron-doped top layer together with that due to the four boron-doped sides, whereas the SIMS gives only a local result. Indeed, SIMS measurements performed on one side of sample 1 showed a boron concentration as high as 1 × 1020 cm–3 extending to a depth of 600 nm. This effect somewhat overestimates the free-carrier concentrations in the top layer, and introduces some uncertainty into the absolute mobility values. Nevertheless, comparative electrical results for the same sample following different treatments should not be grossly affected by the above ‘edge-effects’. After annealing sample 2 at 600 °C for 0.5 h,the sample returns to ptype conductivity. The hole concentration is smaller and the hole mobility is higher than in the as-grown state (Table 1). This means that, although sample 2 is already p-type, a significant fraction of the boron acceptors remains passivated after this annealing step at 600°C as shown in Fig. 3. The first conclusion of these results is that shallow donors govern the electrical properties of these deuterated boron-doped diamond homoepitaxial layers. Their ionization energy of 0.23 eV is remarkably low compared with the 0.6 eV ionization energy of phosphorus in diamond. The ionization energy of these shallow donors is almost the value predicted by the effective mass approximation11 (0.22 eV); from the high-temperature region of curves 3, 5 and 6 in Fig. 4a, we see that their concentration is close to the acceptor concentration. From the above experimental data,it is clear that in samples 1 and 2, following deuteration, all boron acceptors have been passivated by deuterium, and that new donor centres located 0.23 eV below the conduction-band edge appear. The break-up of deuterium-related complexes related to n-type conductivity starts at an annealing temperature below 520 °C. Complete dissociation of these donor complexes results in a highly compensated material, for which no clear conclusion type can be determined: this is the ‘transition state’ between the n-type conductivity of deuterium-related complexes and the pure p-type boron-related conductivity. The break-up of boron–deuterium pairs related to boron-acceptor passivation sets in above 550 °C,and the conversion from n- to p-type is clear above 600 °C. Out-diffusion of nature materials | VOL 2 | JULY 2003 | www.nature.com/naturematerials

484 © 2003 Nature Publishing Group

ARTICLES a

4,000

1020

Sample 2 2 1

3,000

As-grown

2,500

n or p type concentration (cm–3)

Absorption coefficient (cm–1)

3,500

Deuterated

2,000

D + 600 °C, 0.5 h

1,500 1,000 D + 750 °C, 2 h

5 3 1016

2,400

2,600 Wavenumber (cm

2,800

1014

3,000

4 Sample 1 Open symbols are for p-type Full symbols are for n-type

500 0 2,200

6

1018

–1)

0.001

0.002

0.003

0.004

0.005

1/T (1/K)

b Figure 3 Infrared absorption spectra of sample 2 at 10 K in different states. Before deuteration (as-grown sample); after 8 h of deuteration at 550 °C; after deuteration followed by a first annealing at 600 °C for 0.5 h, and a second annealing at 750 °C for 2 h.The ordinate of each spectrum has been displaced for clarity.

Sample 1 Open symbols are for p-type Full symbols are for n-type

400

D + 520 °C, 0.5 h D + 520 °C, 1 h D + 650 °C, 1 h D + 750 °C, 1 h

deuterium4 starts at 700 °C. This is the first time, to the best of our knowledge,that conduction type in an elemental semiconductor can be reversibly switched between p- to n-type by the formation and break-up of deuterium-related defect complexes. To determine the origin of the shallow donors and to eliminate possible mistakes, we performed various experiments as described below. Confocal Raman spectra taken with excitation at 488 nm, and conventional Raman spectra, taken with 634 nm illumination to enhance the signal of the near-surface ‘graphitic’ phases, show no evidence for their existence before or after deuteration. The XPS signal before and after deuteration is similar to those of the best graphite-free diamond samples. Hence, we can rule out the possibility of graphite formation to be responsible for the observed n-type conductivity. Extensive SIMS analysis was performed on sample 1, in search of impurities that can give rise to n-type conductivity. These have shown that the phosphorus and sulphur concentrations are below 1017 cm–3, whereas the nitrogen and oxygen concentration in the epilayer were below the SIMS detection limit of 5 × 1018 cm–3.So there are no common impurities that can give rise to n-type conductivity in concentrations above 2×1019 cm–3 in this sample.Moreover,the as-grown epitaxial layers do not contain concentrations of compensating donor impurities or defects above 2 × 1019 cm–3, because it is known that, in such a compensated boron-doped diamond, the deuterium concentration profile does not follow the boron concentration5.However,native defects acting as deuterium deep traps are often present in boron-doped epilayers12.This is clearly noticeable in sample 3,which did not exhibit the p- to n-type conversion following deuteration.Before deuteration of this sample, SIMS experiments show a strong incorporation of hydrogen on the first 0.25 µm of the boron-doped layer from the interface with the substrate. This reveals the existence of defects in the growing films. For this sample, prolonged deuteration did not result in p- to n-type conversion, despite its very large deuterium uptake measured by SIMS (over five times more deuterium concentration than boron).Clearly,the deuterium was preferentially captured in this sample on defects rather than on the boron dopants.The hydrogen incorporation observed in the as-grown sample 3 is totally absent in the as-grown samples 1 and 2. From the above considerations, we conclude that some deuteriumrelated complex is likely to be the origin of the shallow donors in deuterated boron-doped epilayers of diamond.

Mobility (cm2 V–1 s–1)

300

200

100

0 200

400

600

800

Temperature (K)

Figure 4 Hall effect and conductivity measurements for sample 1. a, Carrier concentration as a function of the reciprocal temperature in the following states: 1, as-grown; 2, as-deuterated; 3, deuterated and annealed at 520 °C for 0.5 h, and 4, for 1 h; 5, annealed at 650 °C for 1 h; 6, annealed at 750 °C for 1 h. b, Mobility as a function of temperature after deuteration and then annealing at different temperatures.

According to theorists, isolated hydrogen acts as a deep donor in p-type diamond7.Its energy level was located by Goss et al.8 at 3eV below the conduction band minimum Ec of diamond, whereas Saada et al.13 predict its position at Ec–2.2 eV. The same holds true for deuterium, because the position of the energy level does not depend on the mass. Consequently, isolated deuterium impurities are not likely to be the shallow donors in diamond. At this point, we examine two possible origins for the shallow donors. First, we consider the creation of some specific deuteriumdefect complexes that may give rise to a donor level, in addition to the boron–deuterium complex that passivates the p-type boron-related conductivity. These defects can, in principle, be present in the as-grown sample or can be induced by the plasma deuteration. However, whatever the origin of these defects is, their concentration has to be unrealistically high to account for the observed donor concentrations, unless they are related to the boron dopants. Furthermore, in this mechanism the shallow donor concentration

nature materials | VOL 2 | JULY 2003 | www.nature.com/naturematerials

485 © 2003 Nature Publishing Group

ARTICLES would be rather independent of the boron doping level contrary to our experimental results. Hence, an alternative and more plausible explanation for the appearance of donor states in boron-containing deuterated diamond assumes the formation of some particular (boron–deuterium) related complexes. This finds support in the observation that the higher the boron concentration the higher the free-electron concentration. It is possible that a fraction of the excess deuterium in the samples that exhibit n-type is bonded to existing boron–deuterium pairs, thus giving rise to (boron–multideuterium) complexes such as (B,D2), which would have a shallow donor level. Taking into account the uncertainties in the absolute values of the measured deuterium and boron concentrations in Fig. 2, it is a reasonable assumption that the excess deuterium is related to the appearance of the n-type features of the sample. The existence of dopant–multihydrogen complexes has been demonstrated in silicon14. (B,H2) complexes, with one acceptor boron and two donor hydrogen atoms, have potentially a donor character. It has been theoretically shown that the interaction between the acceptor and the donors may give rise to less-deep electronic states than for isolated donors15. Detailed computations are needed to investigate the stability of the various possible boron–multihydrogen complexes in diamond, and to predict the electronic levels in the gap they give rise to. The two possibilities described above are being investigated experimentally and theoretically, to clarify the parameters controlling the origin of the shallow deuterium-related donors in diamond. In particular, it will be important to carefully investigate the role of the structural quality of the boron-doped samples that do exhibit the p-type to n-type conversion, and the role that boron plays in the donorformation mechanism.

Received 10 March 2003; accepted 28 May 2003; published 22 June 2003. References 1. Chow, T. P. & Tyagi, R. Wide bandgap compound semiconductors for superior high-voltage unipolar power devices. IEEE Trans. Electron. Dev. 41, 1481–1483 (1994). 2. Koizumi, S., Teraji, T. & Kanda, H. Phosphorus-doped chemical vapor deposition of diamond. Diam. Relat. Mater. 9, 935–940 (2000). 3. Chevallier, J. et al. Hydrogen-boron interactions in p-type diamond. Phys. Rev. B 58, 7966–7969 (1998). 4. Ballutaud, D. et al. Diffusion and thermal stability of hydrogen in homoepitaxial CVD diamond films. Diam. Relat. Mater. 9, 1171–1174 (2000). 5. Chevallier, J. et al. Hydrogen-acceptor interactions in diamond. Diam. Relat. Mater. 10, 399–404 (2001). 6. Uzan-Saguy, C. et al. Hydrogen diffusion in B-ion-implanted and B-doped homo-epitaxial diamond: passivation of defects vs passivation of boron acceptors. Diam. Relat. Mater. 10, 453–458 (2001). 7. Mehandru, S. P. & Anderson, A. B. The migration of interstitial H in diamond and its pairing with substitutional B and N. J. Mater. Res. 9, 383–395 (1994). 8. Goss, J. P. et al. Theory of hydrogen in diamond. Phys. Rev. B 65, 115207 (2002). 9. Nazaré, M. H. & Neves, A. J. Properties, Growth and Applications of Diamond (Emis Data Review Series 26, INSPEC, London, 2001). 10. Lagrange, J. P., Deneuville, A. & Gheeraert, E. Activation energy in low compensated homoepitaxial boron-doped diamond films. Diam. Relat. Mater.7, 1390–1393 (1998). 11. Gheeraert, E., Koizumi, S., Teraji, T., Kanda, H. & Nesladek, M. Electronic states of phosphorus in diamond. Diam. Relat. Mater. 9, 948–951 (2000). 12. Teukam, Z. et al. Trap limited diffusion of hydrogen in boron-doped diamond. Diam. Relat. Mater. 12, 647–651 (2003). 13. Saada, D., Adler, J. & Kalish, R. Lowest-energy site for hydrogen in diamond. Phys. Rev. B 61, 10711–10715 (2000). 14. Liang, Z. N., Haas, C. & Niesen, L. Multiple trapping of hydrogen in antimony-doped silicon. Phys. Rev. Lett. 72, 1846–1849 (1994). 15. Katayama-Yoshida, H., Nishimatsu, T., Yamamoto, T. & Orita, N. Codoping method for the fabrication of low-resistivity wide band-gap semiconductors in p-type GaN, p-type AlN and n-type diamond: prediction versus experiments. J. Phys. Condens. Matter 13, 8901–8914 (2001). Correspondence and requests for materials should be addressed to J.C.

Competing financial interests The authors declare that they have no competing financial interests.

nature materials | VOL 2 | JULY 2003 | www.nature.com/naturematerials

486 © 2003 Nature Publishing Group