- Email: [email protected]
Characterisation of electron irradiated boron-doped diamond
Diamond and Related Materials 11 (2002) 681–685
Characterisation of electron irradiated boron-doped diamond S.J. Charlesa,*, J.W. Steedsa, D.J.F. Evansb, S. Lawsonb, J.E. Butlerc a
Department of Physics, University of Bristol, Bristol, UK b DTC Research Centre, Bristol, UK c Naval Research Laboratory, Washington, WA, USA
Abstract Various boron-doped polycrystalline chemical vapour deposition (CVD) and single crystal high pressure–high temperature (HPHT) diamonds of different doping levels have been investigated by transmission electron microscopy (TEM) electron irradiation and subsequent study by low-temperature photoluminescence spectroscopy in a Renishaw microspectrometer. Several new optical centres have been discovered that have only been seen in irradiated B-doped diamonds that are evidently closely related, with zero phonon lines at 635.7, 651.2 and 666.0 nm. UV laser irradiation causes inter-centre conversion. In the presence of the 666.0-nm centre, the 635.7-nm centre grows enormously at the expense of the 666.0-nm centre at temperatures above 100 K and the initial relative intensities of these centres determine the details of the temperature dependence. Other unrelated centres are also normally obtained with zero phonon lines at 506.0, 648.0 and 732 nm. The 648.0-nm centre evidently has an associated local mode. Polycrystalline CVD samples of B-doped diamond were examined at DTC Maidenhead to measure the free and boron-bound excitons of the samples and surface maps of the intensity were created. These quantitative results show that the boron level varies from grain to grain by up to 10 times presumably as a result of facet orientation on each crystallite. This detail corresponds exactly with the contrast shown under a scanning electron microscope (SEM) under low beam currents; however, after a long exposure to the electron beam the contrast in the SEM changes to show much larger structures that are believed to be related to grain clusters with a common growth direction (while the exciton map remains constant). 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Diamond defects; Boron doping; Photoluminescence; Vacancy complexes
1. Introduction Boron is a well-established dopant, that controls the (p-type) electrical conductivity of diamond. Its use has enabled the production of a variety of electronic devices, sensors and electrodes. The uptake of the boron during growth has been found to rely on surface facet orientation and on the presence of oxygen in the growth chamber. Others indicate that there may be B depletion at grain boundaries or a grain size or boundary dependence on the uptake, but our previous work does not support these hypotheses w1x. The study of point defects is facilitated by their introduction via irradiation, usually done with electrons. *Corresponding author. H.H. Wills Physics Laboratory, Tyndall Avenue, Bristol BS8 1TL, UK. Tel.: q44-117-928-8750; fax: q44117-925-5624. E-mail address: [email protected] (S.J. Charles).
When irradiating with electrons, simple point defects are formed, mainly interstitials and vacancies. Borondoped diamonds have an abundance of holes and could be a candidate to find the elusive positive vacancy (Vq) in diamond. To study these defects, microscopic photoluminescence (PL) is carried out, which enables local studies of the spectra, and hence boron levels in different regions of the diamond. The polycrystalline chemical vapour deposition (CVD) samples covered a wide range of B concentrations and were typically 100–200 mm in thickness and had occasional large grains of up to 50 mm in diameter, while the high pressure–high temperature (HPHT) samples had boron levels of approximately 1018 cm3. The growth directions of the CVD samples were either N110M or N112M. All samples had been mechanically polished to give smooth surfaces. Electron irradiation of the samples was performed in an ion-free electron beam in a transmission electron
0925-9635/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 9 6 3 5 Ž 0 1 . 0 0 5 5 2 - 0
682
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
microscope (TEM). The usual procedure used here was to irradiate the bulk samples with a 200-mm (or less) diameter area at 200 keV to electron doses of between 1018 –1021 e y cmy2 with dose rates of approximately 1016 –1018 e y cmy2 sy1. After irradiation the samples were transferred to a Renishaw micro-Raman spectrometer fitted with an Oxford Instruments Microstat that allowed microscopic PL examination down to 6.4 K with a spatial and depth resolution of approximately 4 mm. The PL was excited using the 488-nm line of an Arq laser (although a few experiments were carried out using the 325-nm line of a HeCd laser). CL in the visible region was carried out on a JEOL 6400 scanning electron microscope (SEM) using exciting electrons of 20 keV at currents of 10y6 amperes and photographs were taken of the results using an Olympus camera. Also CL in the UV region (in the region of the free and boron-bound excitons) was carried out at DTC Maidenhead. Exciton maps of the CVD samples were obtained and these results were used to get quantitative values for the boron level of each grain facet and hence show its distribution throughout the samples. Secondary electron images on the surface of the CVD diamond samples were taken on a modified Coates and Welter field emission gun SEM operating at low voltage (1 keV). 2. Experimental results 2.1. Electron irradiation experiments Expanding on previous work done in this group on electron irradiated B-doped diamond w1x, further studies have been carried out on polycrystalline CVD diamond and now also on HPHT diamond. These results show numerous centres that have previously only been reported by some of these authors and only in B-doped diamonds. However, as a result of the use of the homogeneous HPHT diamonds, systematic and internally consistent results have been obtained, which would have been almost impossible on CVD samples. The zero phonon lines (ZPLs) seen include 635.7 nm, 651.2 nm, 666.1 nm, 506.0 nm, 648.0 nm and 732 nm, the GR1 (neutral vacancy) luminescence centre is also seen in Fig. 1. Some of the centres seen have interesting properties and these will be discussed in the sections below. 2.1.1. Luminescence: 635.7 nm and 666.1 nm One or other of these two centres occur in all irradiated B-doped diamonds and generally both occur together w2x. They seem to have a close relationship with each other. It is usual to find that, initially, the 635.7-nm centre is quite weak and the 666.1-nm centre is relatively stronger (as in Fig. 1). Investigating the
Fig. 1. Spectra showing the 600–690-nm region of a boron-doped CVD diamond sample at 7 K. Note in particular the 635.7-nm and 666.1-nm centres.
intensity of these centres with temperature shows an increase in the intensity of the 635.7-nm centre and a decrease in that of the 666.1-nm centre. At 220 K the 635.7-nm centre reaches its maximum, as we have shown previously w3x. It is also noticeable that the 666.1-nm centre reduces in intensity as the 635.7-nm ZPL increases, and this relationship can be shown, as in Fig. 2. Laser power dependence results were also taken on the irradiated areas and it is possible to see a difference with power of these centres and of the GR1. This particular work was done on an irradiated HPHT sample, to eliminate the possibility that the changes are caused by slight changes of beam position in homogeneous material. As can be seen in Fig. 2, the intensity, relative to the Raman intensity, of the 635.7-nm and 666.1-nm centres increases with decreasing power, but the relative intensity of the GR1 decreases. The change in intensity has no hysteresis and returning to 100% or 1% power restores the original spectrum. This phenomenon is unique to B-doped diamond and there is generally no power dependence of the GR1 centre in undoped or nitrogen doped specimens. Looking at the spectra at 1% power (0.04 mW) the effect is enhanced even more dramatically and it is clear that at the lower laser powers the GR1 is greatly reduced relative to the other two centres. As well as the power and temperature dependence, there is also an occasional positional dependence shown between the 666.1-nm ZPL and the GR1. One particular series of scans across a sample, at 7 K, with 10-mm steps between each spectrum, shows that at one point in the sample the intensity of the GR1 is reduced, while at the same point, the intensity of the ZPL at 666.1 nm has increased. This is attributed to a local change, perhaps of the Fermi level, within the diamond lattice. One CVD sample was found to have only GR1 in the irradiated area when excited with 488-nm laser light, and this sample was subsequently exposed to 325-nm (near UV) excitation and then re-examined with 488-
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
683
Fig. 2. The temperature dependence of the 635.7-nm, 666.1-nm and GR1 centres showing how they change with applied laser power. 100% power is equivalent to 4 mW on the sample. Also note that the 100% power on the GR1 is only applicable at above 100 K, as below that the detector on the Renishaw spectrometer is saturated.
nm excitation. After the UV treatment the irradiated area showed 635.7-nm luminescence, where there had previously been none and a lack of GR1 where it had previously been detected, but only in the area that had been exposed to the ultraviolet laser. This centre does not decay over short time scales, and persists for months afterwards. A local mode has also been discovered, associated with the 635.7-nm centre, it occurs at 187meV from the ZPL neither the energy of this nor that of ZPL depends on the change of the boron isotope to 10 B.
phonon replica of the 635.7-nm centre and also displays unusual temperature dependence. Unlike the 635.7-nm centre, however, this centre has a maximum intensity at 150 K. It has not yet been seen in HPHT boron-doped diamonds, and only occurs infrequently in CVD Bdoped diamonds. The unusual temperature dependence can only be investigated in the absence of the 635.7-nm and the 666.1-nm centres due to the overlap that otherwise occurs.
2.1.2. Other luminescence centres Although the forgoing has concentrated on the two centres that have been studied most in this work, there are also several other centres that occur in electronirradiated boron-doped diamond. The 648.0-nm centre occurs in all of these diamonds, as does the 506.0-nm centre. These two have not been studied in detail yet, but they do not show any unusual temperature dependence. Also a 732-nm centre has been encountered quite recently w1x. This has an irregular shape and seems to only occur in localised regions of a sample. In one sample it was seen only in a 20-mm-wide strip across a 200-mm irradiated area. The final centre of that we have studied has a ZPL at 651.2 nm. It occurs just at the point of the first
Maps of free and bound exciton luminescence intensity, taken on B-doped polycrystalline CVD diamond samples, show clear inhomogeneities between different crystal facets. If spot spectra are taken at different points of the sample, in areas with different luminescence intensities it is possible to see different ratios of the free to bound exciton peaks. These ratios can be used to determine the boron levels within each region of the sample w4x. If these maps are correlated with visible cathodoluminescence topography images then it is possible to see that the different colours in these images correspond to the different boron levels, shown on the exciton maps (Fig. 3). In fact the dark grey areas correspond to approximately 1 ppm or less of boron, while the mostly light grey areas relate to levels of 7
2.2. Cathodoluminescence spectroscopy
684
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
contrast only reverts after the sample has been brought to atmosphere and then returned to the SEM. 3. Discussion of results As has already been reported, electron irradiation, followed by microscopic PL in CVD B-doped diamond produces several unique centres in the visible spectral region. These centres have now also been seen in Bdoped HPHT diamond, leading to the conclusion that it is the boron involvement in the samples that produces these centres. Furthermore, not only are the 635.7-nm centre and the 666.1-nm centre related to each other (as demonstrated by their temperature dependence), they also maintain a very close relationship with GR1. This relationship can be explained if it is assumed that these two centres represent a positive vacancy (Vq). The laser power dependencies with the 488-nm laser indicate that at higher powers an electron is being removed from the valance band and put onto the Vq, converting it into the neutral vacancy. When the illumination is removed the V0 captures a hole and converts back into the Vq. This process is facilitated by the fact that Bdoped diamond is p-type and has an abundance of holes
Fig. 3. A free and bound exciton intensity map, shown with a correlated area in visible CL. The different greyscale areas correlate with different B-doping levels within the diamond lattice.
ppm or above. The mid-grey areas give values somewhere in the middle. These values for the boron doping and the size of the grains also correspond accurately with the SIMS results that we have shown previously w1 x . 2.3. SEM contrast of B-doped CVD diamonds Studying the surface of the diamond in a low voltage (1 keV) SEM shows grain contrasts that also corresponds with the boron level and grain size seen in the CL maps above, see Fig. 4. However, if the voltage is increased to above 5 keV then the contrast can be seen to change to that corresponding to larger area grains (approx. 100 mm in size) of approximately the size of the grain clusters seen in the TEM (Fig. 4). In this work the beam current was approximately 0.5 nA. The contrast can also be changed with extended exposure to higher beam currents, of the order of mA. Additionally, even after reducing the beam voltage to 1 keV (after changing the contrast at 5 keV) the contrast remains in its changed form with large grain clusters evident. This
Fig. 4. (a) SEM image of CVD B-doped diamond, studied at 1 keV and 0.5 nA beam current; (b) SEM image of CVD B-doped diamond studied at 5 keV and 0.5 nA beam.
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
in the lattice. The UV result supports this conclusion. Thu UV photons have sufficient energy to ionise the V0 centre by removal of an electron which is subsequently trapped, leaving the centre in a Vq state. The lack of electrons in the conduction band prevents its back conversion to V0. This explains why the positive vacancy persists, even after the UV excitation has been removed. An explanation of the positional variation of the GR1 and 666.1-nm centre can be attributed to a greater density of boron acceptors at this position in the lattice. More acceptors at this point will pin the Fermi level lower in the band gap at this point and, by Boltzmann theory, make the population of V0 lower while increasing the population of Vq. This information combined with the fact that the local mode of the 635.7nm centre is boron isotope independent indicates that we may indeed have identified the positive vacancy in diamond. From the CL exciton maps it can be seen that the levels of boron within each crystallite can vary by an order of magnitude, on a scale of 5 mm or less. This inhomogeneity can also be closely linked to the colours shown in the visible CL topographs and the secondary electron contrast in the SEM. The higher levels of boron give brighter contrast in the SEM and more yellow colours in the CL topographs. This gives the opportunity of a simple method of quickly gauging the doping levels from crystallite to crystallite in a CVD sample. The SEM contrast, as has been mentioned, can correlate exactly to the distribution of the boron within the sample. However, it is possible to lose this contrast with prolonged exposure to an electron beam. The only explanation possible for this is that the electron beam is somehow changing the character of the diamond surface, and in fact this fits a reported theory of the surface conductivity in diamond w5x. The diamond surface, when exposed to air attracts impurities, such as water, to it. These then form a layer on the surface of the diamond which, when combined with band bending at the surface of the sample, produces enhanced electron emission. As the electron beam probes the surface this adsorbed water
685
layer is being removed and eventually the crystallite contrast goes, to replaced with channelling related contrast from the much larger ‘grain clusters’ which have a precise, common growth direction. The original contrast does not return while the sample is left in the vacuum, but does after it has been to atmosphere. Again this shows that a surface impurity layer is likely the cause of this phenomenon, and is related to the surface conductivity of the diamond. 4. Conclusions We have found the first evidence of a likely candidate for the positive vacancy in diamond, as can be seen by its clear relationship to the GR1 centre. Another important point is that now it is possible to measure the variation in boron doping levels from grain to grain in diamonds doped to the 1018 –1019 cmy3, where previously this was thought not possible. This level can also be correlated with simple ‘quick-look’ methods to get an idea of boron levels. Finally we have shown that there is a distinction between crystal facets and grain clusters in the SEM, which is evident from the behaviour of the diamond surface under an electron beam. References w1x J.W. Steeds, A. Gilmore, S. Charles, P. Heard, B. Howarth, J.E. Butler, Use of novel methods for the investigation of the boron distribution in CVD diamond, Acta Mater. 47 (1999) 4025. w2x J.W. Steeds, S.J. Charles, J.E. Butler, B-distributions, b-related photoluminescence and SEM contrast of b-doped CVD and HPHT diamond specimens, de Beers Diamond Conference 2001, Bristol, 2001. w3x J.W. Steeds, S.J. Charles, J. Davies, I. Griffin, Photoluminescence microscopy of TEM irradiated diamond, Diamond. Relat. Mater. 9 (2000) 397–403. w4x H. Kawarada, H. Matsuyama, Y. Yokota, T. Sogi, A. Yamaguchi, A. Hiraki, Excitonic recombination radiation in undoped and boron-doped chemical-vapor-deposited diamonds, Phys. Rev. B 47 (1993) 3633. w5x F. Maier, M. Riedel, B. Mantel, J. Ristein, L. Ley, Origin of surface conductivity in diamond, Phys. Rev. Lett. 85 (2000) 3472.
Characterisation of electron irradiated boron-doped diamond S.J. Charlesa,*, J.W. Steedsa, D.J.F. Evansb, S. Lawsonb, J.E. Butlerc a
Department of Physics, University of Bristol, Bristol, UK b DTC Research Centre, Bristol, UK c Naval Research Laboratory, Washington, WA, USA
Abstract Various boron-doped polycrystalline chemical vapour deposition (CVD) and single crystal high pressure–high temperature (HPHT) diamonds of different doping levels have been investigated by transmission electron microscopy (TEM) electron irradiation and subsequent study by low-temperature photoluminescence spectroscopy in a Renishaw microspectrometer. Several new optical centres have been discovered that have only been seen in irradiated B-doped diamonds that are evidently closely related, with zero phonon lines at 635.7, 651.2 and 666.0 nm. UV laser irradiation causes inter-centre conversion. In the presence of the 666.0-nm centre, the 635.7-nm centre grows enormously at the expense of the 666.0-nm centre at temperatures above 100 K and the initial relative intensities of these centres determine the details of the temperature dependence. Other unrelated centres are also normally obtained with zero phonon lines at 506.0, 648.0 and 732 nm. The 648.0-nm centre evidently has an associated local mode. Polycrystalline CVD samples of B-doped diamond were examined at DTC Maidenhead to measure the free and boron-bound excitons of the samples and surface maps of the intensity were created. These quantitative results show that the boron level varies from grain to grain by up to 10 times presumably as a result of facet orientation on each crystallite. This detail corresponds exactly with the contrast shown under a scanning electron microscope (SEM) under low beam currents; however, after a long exposure to the electron beam the contrast in the SEM changes to show much larger structures that are believed to be related to grain clusters with a common growth direction (while the exciton map remains constant). 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Diamond defects; Boron doping; Photoluminescence; Vacancy complexes
1. Introduction Boron is a well-established dopant, that controls the (p-type) electrical conductivity of diamond. Its use has enabled the production of a variety of electronic devices, sensors and electrodes. The uptake of the boron during growth has been found to rely on surface facet orientation and on the presence of oxygen in the growth chamber. Others indicate that there may be B depletion at grain boundaries or a grain size or boundary dependence on the uptake, but our previous work does not support these hypotheses w1x. The study of point defects is facilitated by their introduction via irradiation, usually done with electrons. *Corresponding author. H.H. Wills Physics Laboratory, Tyndall Avenue, Bristol BS8 1TL, UK. Tel.: q44-117-928-8750; fax: q44117-925-5624. E-mail address: [email protected] (S.J. Charles).
When irradiating with electrons, simple point defects are formed, mainly interstitials and vacancies. Borondoped diamonds have an abundance of holes and could be a candidate to find the elusive positive vacancy (Vq) in diamond. To study these defects, microscopic photoluminescence (PL) is carried out, which enables local studies of the spectra, and hence boron levels in different regions of the diamond. The polycrystalline chemical vapour deposition (CVD) samples covered a wide range of B concentrations and were typically 100–200 mm in thickness and had occasional large grains of up to 50 mm in diameter, while the high pressure–high temperature (HPHT) samples had boron levels of approximately 1018 cm3. The growth directions of the CVD samples were either N110M or N112M. All samples had been mechanically polished to give smooth surfaces. Electron irradiation of the samples was performed in an ion-free electron beam in a transmission electron
0925-9635/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 9 6 3 5 Ž 0 1 . 0 0 5 5 2 - 0
682
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
microscope (TEM). The usual procedure used here was to irradiate the bulk samples with a 200-mm (or less) diameter area at 200 keV to electron doses of between 1018 –1021 e y cmy2 with dose rates of approximately 1016 –1018 e y cmy2 sy1. After irradiation the samples were transferred to a Renishaw micro-Raman spectrometer fitted with an Oxford Instruments Microstat that allowed microscopic PL examination down to 6.4 K with a spatial and depth resolution of approximately 4 mm. The PL was excited using the 488-nm line of an Arq laser (although a few experiments were carried out using the 325-nm line of a HeCd laser). CL in the visible region was carried out on a JEOL 6400 scanning electron microscope (SEM) using exciting electrons of 20 keV at currents of 10y6 amperes and photographs were taken of the results using an Olympus camera. Also CL in the UV region (in the region of the free and boron-bound excitons) was carried out at DTC Maidenhead. Exciton maps of the CVD samples were obtained and these results were used to get quantitative values for the boron level of each grain facet and hence show its distribution throughout the samples. Secondary electron images on the surface of the CVD diamond samples were taken on a modified Coates and Welter field emission gun SEM operating at low voltage (1 keV). 2. Experimental results 2.1. Electron irradiation experiments Expanding on previous work done in this group on electron irradiated B-doped diamond w1x, further studies have been carried out on polycrystalline CVD diamond and now also on HPHT diamond. These results show numerous centres that have previously only been reported by some of these authors and only in B-doped diamonds. However, as a result of the use of the homogeneous HPHT diamonds, systematic and internally consistent results have been obtained, which would have been almost impossible on CVD samples. The zero phonon lines (ZPLs) seen include 635.7 nm, 651.2 nm, 666.1 nm, 506.0 nm, 648.0 nm and 732 nm, the GR1 (neutral vacancy) luminescence centre is also seen in Fig. 1. Some of the centres seen have interesting properties and these will be discussed in the sections below. 2.1.1. Luminescence: 635.7 nm and 666.1 nm One or other of these two centres occur in all irradiated B-doped diamonds and generally both occur together w2x. They seem to have a close relationship with each other. It is usual to find that, initially, the 635.7-nm centre is quite weak and the 666.1-nm centre is relatively stronger (as in Fig. 1). Investigating the
Fig. 1. Spectra showing the 600–690-nm region of a boron-doped CVD diamond sample at 7 K. Note in particular the 635.7-nm and 666.1-nm centres.
intensity of these centres with temperature shows an increase in the intensity of the 635.7-nm centre and a decrease in that of the 666.1-nm centre. At 220 K the 635.7-nm centre reaches its maximum, as we have shown previously w3x. It is also noticeable that the 666.1-nm centre reduces in intensity as the 635.7-nm ZPL increases, and this relationship can be shown, as in Fig. 2. Laser power dependence results were also taken on the irradiated areas and it is possible to see a difference with power of these centres and of the GR1. This particular work was done on an irradiated HPHT sample, to eliminate the possibility that the changes are caused by slight changes of beam position in homogeneous material. As can be seen in Fig. 2, the intensity, relative to the Raman intensity, of the 635.7-nm and 666.1-nm centres increases with decreasing power, but the relative intensity of the GR1 decreases. The change in intensity has no hysteresis and returning to 100% or 1% power restores the original spectrum. This phenomenon is unique to B-doped diamond and there is generally no power dependence of the GR1 centre in undoped or nitrogen doped specimens. Looking at the spectra at 1% power (0.04 mW) the effect is enhanced even more dramatically and it is clear that at the lower laser powers the GR1 is greatly reduced relative to the other two centres. As well as the power and temperature dependence, there is also an occasional positional dependence shown between the 666.1-nm ZPL and the GR1. One particular series of scans across a sample, at 7 K, with 10-mm steps between each spectrum, shows that at one point in the sample the intensity of the GR1 is reduced, while at the same point, the intensity of the ZPL at 666.1 nm has increased. This is attributed to a local change, perhaps of the Fermi level, within the diamond lattice. One CVD sample was found to have only GR1 in the irradiated area when excited with 488-nm laser light, and this sample was subsequently exposed to 325-nm (near UV) excitation and then re-examined with 488-
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
683
Fig. 2. The temperature dependence of the 635.7-nm, 666.1-nm and GR1 centres showing how they change with applied laser power. 100% power is equivalent to 4 mW on the sample. Also note that the 100% power on the GR1 is only applicable at above 100 K, as below that the detector on the Renishaw spectrometer is saturated.
nm excitation. After the UV treatment the irradiated area showed 635.7-nm luminescence, where there had previously been none and a lack of GR1 where it had previously been detected, but only in the area that had been exposed to the ultraviolet laser. This centre does not decay over short time scales, and persists for months afterwards. A local mode has also been discovered, associated with the 635.7-nm centre, it occurs at 187meV from the ZPL neither the energy of this nor that of ZPL depends on the change of the boron isotope to 10 B.
phonon replica of the 635.7-nm centre and also displays unusual temperature dependence. Unlike the 635.7-nm centre, however, this centre has a maximum intensity at 150 K. It has not yet been seen in HPHT boron-doped diamonds, and only occurs infrequently in CVD Bdoped diamonds. The unusual temperature dependence can only be investigated in the absence of the 635.7-nm and the 666.1-nm centres due to the overlap that otherwise occurs.
2.1.2. Other luminescence centres Although the forgoing has concentrated on the two centres that have been studied most in this work, there are also several other centres that occur in electronirradiated boron-doped diamond. The 648.0-nm centre occurs in all of these diamonds, as does the 506.0-nm centre. These two have not been studied in detail yet, but they do not show any unusual temperature dependence. Also a 732-nm centre has been encountered quite recently w1x. This has an irregular shape and seems to only occur in localised regions of a sample. In one sample it was seen only in a 20-mm-wide strip across a 200-mm irradiated area. The final centre of that we have studied has a ZPL at 651.2 nm. It occurs just at the point of the first
Maps of free and bound exciton luminescence intensity, taken on B-doped polycrystalline CVD diamond samples, show clear inhomogeneities between different crystal facets. If spot spectra are taken at different points of the sample, in areas with different luminescence intensities it is possible to see different ratios of the free to bound exciton peaks. These ratios can be used to determine the boron levels within each region of the sample w4x. If these maps are correlated with visible cathodoluminescence topography images then it is possible to see that the different colours in these images correspond to the different boron levels, shown on the exciton maps (Fig. 3). In fact the dark grey areas correspond to approximately 1 ppm or less of boron, while the mostly light grey areas relate to levels of 7
2.2. Cathodoluminescence spectroscopy
684
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
contrast only reverts after the sample has been brought to atmosphere and then returned to the SEM. 3. Discussion of results As has already been reported, electron irradiation, followed by microscopic PL in CVD B-doped diamond produces several unique centres in the visible spectral region. These centres have now also been seen in Bdoped HPHT diamond, leading to the conclusion that it is the boron involvement in the samples that produces these centres. Furthermore, not only are the 635.7-nm centre and the 666.1-nm centre related to each other (as demonstrated by their temperature dependence), they also maintain a very close relationship with GR1. This relationship can be explained if it is assumed that these two centres represent a positive vacancy (Vq). The laser power dependencies with the 488-nm laser indicate that at higher powers an electron is being removed from the valance band and put onto the Vq, converting it into the neutral vacancy. When the illumination is removed the V0 captures a hole and converts back into the Vq. This process is facilitated by the fact that Bdoped diamond is p-type and has an abundance of holes
Fig. 3. A free and bound exciton intensity map, shown with a correlated area in visible CL. The different greyscale areas correlate with different B-doping levels within the diamond lattice.
ppm or above. The mid-grey areas give values somewhere in the middle. These values for the boron doping and the size of the grains also correspond accurately with the SIMS results that we have shown previously w1 x . 2.3. SEM contrast of B-doped CVD diamonds Studying the surface of the diamond in a low voltage (1 keV) SEM shows grain contrasts that also corresponds with the boron level and grain size seen in the CL maps above, see Fig. 4. However, if the voltage is increased to above 5 keV then the contrast can be seen to change to that corresponding to larger area grains (approx. 100 mm in size) of approximately the size of the grain clusters seen in the TEM (Fig. 4). In this work the beam current was approximately 0.5 nA. The contrast can also be changed with extended exposure to higher beam currents, of the order of mA. Additionally, even after reducing the beam voltage to 1 keV (after changing the contrast at 5 keV) the contrast remains in its changed form with large grain clusters evident. This
Fig. 4. (a) SEM image of CVD B-doped diamond, studied at 1 keV and 0.5 nA beam current; (b) SEM image of CVD B-doped diamond studied at 5 keV and 0.5 nA beam.
S.J. Charles et al. / Diamond and Related Materials 11 (2002) 681–685
in the lattice. The UV result supports this conclusion. Thu UV photons have sufficient energy to ionise the V0 centre by removal of an electron which is subsequently trapped, leaving the centre in a Vq state. The lack of electrons in the conduction band prevents its back conversion to V0. This explains why the positive vacancy persists, even after the UV excitation has been removed. An explanation of the positional variation of the GR1 and 666.1-nm centre can be attributed to a greater density of boron acceptors at this position in the lattice. More acceptors at this point will pin the Fermi level lower in the band gap at this point and, by Boltzmann theory, make the population of V0 lower while increasing the population of Vq. This information combined with the fact that the local mode of the 635.7nm centre is boron isotope independent indicates that we may indeed have identified the positive vacancy in diamond. From the CL exciton maps it can be seen that the levels of boron within each crystallite can vary by an order of magnitude, on a scale of 5 mm or less. This inhomogeneity can also be closely linked to the colours shown in the visible CL topographs and the secondary electron contrast in the SEM. The higher levels of boron give brighter contrast in the SEM and more yellow colours in the CL topographs. This gives the opportunity of a simple method of quickly gauging the doping levels from crystallite to crystallite in a CVD sample. The SEM contrast, as has been mentioned, can correlate exactly to the distribution of the boron within the sample. However, it is possible to lose this contrast with prolonged exposure to an electron beam. The only explanation possible for this is that the electron beam is somehow changing the character of the diamond surface, and in fact this fits a reported theory of the surface conductivity in diamond w5x. The diamond surface, when exposed to air attracts impurities, such as water, to it. These then form a layer on the surface of the diamond which, when combined with band bending at the surface of the sample, produces enhanced electron emission. As the electron beam probes the surface this adsorbed water
685
layer is being removed and eventually the crystallite contrast goes, to replaced with channelling related contrast from the much larger ‘grain clusters’ which have a precise, common growth direction. The original contrast does not return while the sample is left in the vacuum, but does after it has been to atmosphere. Again this shows that a surface impurity layer is likely the cause of this phenomenon, and is related to the surface conductivity of the diamond. 4. Conclusions We have found the first evidence of a likely candidate for the positive vacancy in diamond, as can be seen by its clear relationship to the GR1 centre. Another important point is that now it is possible to measure the variation in boron doping levels from grain to grain in diamonds doped to the 1018 –1019 cmy3, where previously this was thought not possible. This level can also be correlated with simple ‘quick-look’ methods to get an idea of boron levels. Finally we have shown that there is a distinction between crystal facets and grain clusters in the SEM, which is evident from the behaviour of the diamond surface under an electron beam. References w1x J.W. Steeds, A. Gilmore, S. Charles, P. Heard, B. Howarth, J.E. Butler, Use of novel methods for the investigation of the boron distribution in CVD diamond, Acta Mater. 47 (1999) 4025. w2x J.W. Steeds, S.J. Charles, J.E. Butler, B-distributions, b-related photoluminescence and SEM contrast of b-doped CVD and HPHT diamond specimens, de Beers Diamond Conference 2001, Bristol, 2001. w3x J.W. Steeds, S.J. Charles, J. Davies, I. Griffin, Photoluminescence microscopy of TEM irradiated diamond, Diamond. Relat. Mater. 9 (2000) 397–403. w4x H. Kawarada, H. Matsuyama, Y. Yokota, T. Sogi, A. Yamaguchi, A. Hiraki, Excitonic recombination radiation in undoped and boron-doped chemical-vapor-deposited diamonds, Phys. Rev. B 47 (1993) 3633. w5x F. Maier, M. Riedel, B. Mantel, J. Ristein, L. Ley, Origin of surface conductivity in diamond, Phys. Rev. Lett. 85 (2000) 3472.