- Email: [email protected]
The primed state of CVD diamond under blue light illumination
Diamond & Related Materials 15 (2006) 1362 – 1368 www.elsevier.com/locate/diamond
The primed state of CVD diamond under blue light illumination C. Manfredotti *, F. Fizzotti, Y. Garino Experimental Physics Department, University of Torino, Italy and Center of Excellence on Nanostructured Surfaces and Interfaces, University of Torino, Italy Received 26 May 2005; received in revised form 2 September 2005; accepted 10 October 2005 Available online 11 January 2006
Abstract The primed state of a detector grade CVD diamond sample has been investigated with respect to light illumination during alpha particles (5.5 MeV) detection. The measurements have been carried out as a function of elapsed time after beta-rays priming both in short term (1 h) and in long term (33 h) conditions. The contribution of holes and electrons to the average charge collection efficiency and to total number of counts above a threshold has been qualitatively separated by selecting bias polarity. The behaviour of electron collection after X-ray priming during blue light or UV illumination is improved while the hole one is worsened, confirming previous data. Linearity of the average collection efficiency for alpha particle is observed only at very low doses and in the hole case. Long term stability of both collection efficiency and of total number of counts is observed both in the primed state for holes and during blue light illumination for electrons. D 2005 Elsevier B.V. All rights reserved. Keywords: CVD diamond; Priming; Illumination; Stability
1. Introduction There are several controversies about the effect of light illumination on the primed state of CVD diamond as obtained by X- or beta-rays irradiation at doses around 20– 30 Gy: according to different authors, the effect could be either positive or negative, i.e. it can be used in order either to improve the performances of CVD diamond as nuclear detector [1] or to erase the improvement obtained by priming [2]. According to our experience, blue light illumination (around 400 nm) transfers the main contribution to charge collection from holes to electrons, and this conclusion has been largely demonstrated by using alpha particles in order to discriminate between electrons and holes contribution simply by changing bias polarity [3,4]. Illumination below 550 – 600 nm has been proved not to give rise to any effect on the primed state or at least to any improvement in holes and electrons collection. The investigation of primed state can be carried out also by Below Gap PhotoCurrent (BGPC) measurements [3] which can give some insight into transient or dynamic phenomena. In effect, the primed state is not stable under illumination for photon
* Corresponding author. Tel.: +39 0116707306. E-mail address: [email protected] (C. Manfredotti). 0925-9635/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.diamond.2005.10.006
energies below 2.7 eV, but it decays probably because of an optically stimulated detrapping of trapped holes generated by priming, which are responsible [5] of what has been called Persistent PhotoConductivity (PPC). As a matter of fact, the integrated photocurrent is linearly proportional to the cumulated priming dose, as it was the case of ThermoLuminescence (TL) which starts to be used to measure X-ray doses. What is very strange, at least in our case, is that the linearity [5] is restricted to very low doses (below 1 Gy for instance) with respect to TL. Above 2.7 eV the situation changes because probably the main contribution to photocurrent is given by electrons and a good long_term stability is reached. In order to throw some light on all these phenomena, we started a much longer investigation by using alpha particles spectra in the same way as BGPC as a function of time after priming and blue light illumination, which is a quite long and difficult task, but which gives a better understanding of electrons and holes behaviour. In order to follow the time behaviour, we extracted from alpha multichannel spectra both the centroid or the average value of charge collection efficiency (cce) and the total number of counts (ci), after subtracting the background. This latter parameter can in effect decrease in time because of polarization effects due to trapped charges that locally lower electric field and may cause charge pulses to fall below the electronic threshold. As a consequence, this
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
1363
parameter takes into a larger account regions of lower cce, which, according to our experience, should be more affected by blue light. 2. Experimental The CVD diamond sample used for alpha spectroscopy measurements was grown by MWCVD and it was previously used as a strip detector for tracking purposes. With minimum ionising particles (mips), cce turned out to be about 10%. The total sensitive surface area is 5.5 5.5 mm2 and thickness 600 Am. Strip contacts (width 19.8 Am and pitch 47.6 Am) were deposited on the growth side, while, in order to be illuminated, the sample was equipped with four parallel Cr/Au contacts (20 nm Cr, 100 nm Au, obtained by vacuum deposition) at the substrate side. Fig. 1 shows a SEM micrograph of the growth side of the sample equipped with strip contacts. Alpha particles measurements were performed by using a calibrated Am-241 alpha source (mean energy = 5.48 MeV, penetration depth in diamond = 13 Am including contacts), placed in front of the growth surface at a distance of about 1 mm from the specimen. The alpha particles were suitably collimated to a 2-mm diameter spot. Pulses were recorded by means of a standard charge sensitive electronic chain composed by amplifier and preamplifier (Ortec mods. 142 and 572 respectively). The sample was primed with a strong Sr-90 beta source for larger doses and with a weaker one for lower doses, being the doses measured at the sample position by standard TLD100 dosimeters read by a mod. 3500 Harshaw reader. With respect to mips, cce measured with alpha particles is much lower, not only because of the much lower bias voltage used in our case, but also because of the strong polarization due to the short range of alpha particles and to the consequent space-charge build-up in front of the detector. Before each measurement, the sample was annealed at 360 -C in order to completely recover the virgin state of the sample. Between the measurements and during the priming, the sample was kept in the dark. As already mentioned, because of the short penetration depth of alpha particles, spectroscopy measurements allow to discriminate between electrons and holes by imposing a positive or negative bias voltage, in the assumption that the main contribution is given by the carrier having at its disposal a longer path. This assumption is quite understandable, even if the obtained charge collection distances (ccd) are of the same order of magnitude of the penetration depth of alpha particles. In effect, carriers moving towards front or irradiated electrode have a higher chance of recombination with the opposite charge carriers and because the plasma generated along alpha particle track shields and lowers locally the electrical field. In any case, it is obvious that all the quoted results referring to different carriers are to be assumed in a semiquantitative way. Since the main topic of the paper is the time behaviour of charge collection after priming and during illumination (or darkness), the applied bias voltage was fixed at 200 V, i.e. + 200 V for electrons contribution and 200 V for holes
Fig. 1. SEM micrograph (7.0 kV, original magnification: 60) of the growth side of the CVD diamond sample.
contribution, even if the sample itself could support more than 600 V. In the following we will refer not to the voltage polarity, but only to the carriers that are mainly involved. The sample was annealed and subsequently primed. The sample was placed in the experimental apparatus exactly 5 min after priming, in order to obtain the same starting conditions for all the measurements. A series of consecutive spectra was then acquired for a total time of about 1 h. The time required for each spectrum was 4 min. The electronic threshold was calibrated in order to eliminate the electronic noise. The average charge collection efficiency and the integral of counts of each spectrum were calculated. The charge collection
1364
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
efficiency (cce) g was evaluated by calibration via a silicon surface barriers detector (surface area 25 mm2) with a 100% cce and energy resolution of 14 keV. For normalization the pair creation energies for Si (3.6 eV) and CVD diamond (13.2 eV) were taken into account. Since cce is relatively low, Hecht’s relationship, which is adopted because the penetration depth of alpha particles is much shorter than detector thickness, can be approximated to a linear behaviour g = d / L where L is the detector thickness and d is what is defined as charge collection distance (ccd). 3. Results and discussion 3.1. Short time measurements — holes behaviour The measurements were performed either in dark conditions or under constant illumination at a selected wavelength. We used a tungsten incandescent lamp (150 W) equipped with an interference filter in order to select a 400 nm light (blue light), which was laterally illuminating the sample (Fig. 2). We estimate the light power onto the sample to be approximately 10 AW. The behaviour of average cce (centroid) of the spectra as a function of time in different priming conditions is reported for holes in Fig. 3. First of all, we can observe that under blue light the average cce is lowered in agreement with previous data [4]. Secondly, in the case of blue light illumination there is a slight sensitivity with respect to primed dose, while in the dark the situation is unclear. The third observation is that in the primed case there is a small decay in time, while under illumination cce is constant. In the case of ci, the situation is just the opposite: under illumination ci decays. Since cce, being an average, is more sensitive to the higher efficiency part of the multichannel spectrum, and ci, on the other side, is more sensitive to the lower efficiency one, the conclusion could be quite easy. Priming, by improving cce, is more important in higher efficiency regions, which, being less in volume, are not important for ci, while illumination gives a larger contribution to lower efficiency regions, much wider and therefore more important in the case of ci. In all the cases the decays seem to be due to polarization effects which, in the case of ci, may push some charge pulses below the electronic threshold.
Fig. 3. Average charge collection efficiency (g) and charge collection distance (ccd) versus time for holes in different experimental conditions.
unprimed case, but also with respect to the primed case. In particular the effect of illumination on ci is very large. The second observation is that now ci under illumination is almost constant, while in the dark it decays, even being lower. The third observation is that cce decays in all the cases. The first conclusion is therefore that illumination avoids polarization effects, in the sense that the event rate is kept constant (what in effect has been observed in other kinds of measurements). The second conclusion is that in the electron the homogeneity of the response improves with time, while in the hole case illumination has the effect to eliminate lower cce regions and as a consequence to worsen the homogeneity with time. This
3.2. Short time measurements — electrons behaviour In the electron case (see Fig. 4) the first observation is that light improves both cce and ci with respect not only to the
Fig. 2. Schematic representation of the experimental apparatus.
Fig. 4. Integral of spectra (ci, top) and average charge collection efficiency (g) and charge collection distance (ccd) (bottom) versus time for electrons in different experimental conditions.
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
1365
contribution to holes collection in the dark after priming become the main suppliers of electron events under illumination. A further observation is that in the hole case there is, as already noticed, a small sensitivity to the priming dose, while for electrons this sensitivity disappears completely. 3.3. Short time measurements: dependence on the kind of illumination
Fig. 5. Integral of spectra (ci, top) and average charge collection efficiency (g) and charge collection distance (ccd) (bottom) versus time for electrons subjected to different kinds of illumination.
conclusion is in agreement with previous data [5]. In effect, looking at the equilibrium values of ci, one notices a curious interchange between electrons and holes, under dark and under illumination: if this is not just a coincidence, a possible conclusion could be that regions from which there is a strong
We studied also the effects due to different kinds of light illumination in the electron case. We considered different experimental conditions: a filtered W lamp (blue light), an unfiltered W lamp (white light) and a deuterium lamp (deuterium light, 30 W). Fig. 5 shows the behaviour of cce and ci, respectively, as a function of time. It is clear that deuterium lamp improves cce while it is not able to avoid an initial, small decay. In the case of ci, apart from its constancy in time, as observed before, the three kinds of illumination display the same behaviour both in time and in the absolute values. The conclusion could be that deuterium lamp (average wavelength 200 nm) has a larger effect on average or high cce regions. Of course the conclusion is only qualitative, since light intensities were not measured. Looking at multichannel spectra in different illumination conditions at different times after priming, it can be observed that, in dark conditions, the spectrum rapidly decreases with time and completely vanishes about 12 min after the first measurement. For sake of clearness, Fig. 6 shows multichannel spectra in different illumination conditions at four different times after priming: 5, 9, 13 and 17 min (5 min is the minimum interval between priming and measurement under illumination). The comparison between blue light (blue line) and deuterium light (green line) under-
Fig. 6. Consecutive spectra (elapsed time from previous spectrum: 4 min) for electrons subjected to different kinds of illumination: (A) first spectrum (0 – 4 min), (B) second spectrum (4 – 8 min), (C) third spectrum (8 – 12 min), (D) fourth spectrum (12 – 16 min).
1366
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
lines how, even if their integrals are comparable (see Fig. 5top), cce results higher for deuterium light, since in this case a larger number of counts on higher channels with respect to blue light can be observed. On lower channels the number of counts shows an opposite trend. 3.4. Hole case under priming alone: low priming doses and persistent effects In the hole case, priming effects at low doses were also carefully investigated. A fixed delay time (5 min) was used between the end of priming and the application of bias voltage to the sample. The results concerning cce and ci, respectively at different priming doses (no illumination) are reported in Fig. 7 (the same acquisition procedure as reported above was used). A y = y o + A 1e x/s fit of experimental curves of Fig. 7 (top) has been tried in order to get some conclusions and the results of the fits are reported in the same figure, while the behaviours of coefficient s and y 0 as a function of dose are shown in Fig. 8. Data referring to the highest dose of Fig. 7 (top) are not reported since the decay is very slow and the fit is affected by a large error. In any case, we can observe that s increases exponentially with dose, while y 0 increases linearly with the dose quite nicely. In other words, the equilibrium value of ci ( y 0) is linearly proportional to the dose, while the dependence of ci on time becomes weaker and weaker with the dose, approaching to a constant value at a dose value of 630 mGy.
Fig. 8. Behaviour of decay time constant s (a) and of initial value of integral counts y 0 (b) versus dose obtained from fittings reported in Fig. 7 (top).
In effect, the proportionality of ci with respect to priming dose is very clear, but, as it can be clearly noticed in Fig. 9 that shows the integral of the last recorded spectrum (after 3400 s) versus dose, the priming effect on the integral of spectra saturates already at about 10 mGy. This behaviour can be compared to what has been observed in photoconductivity [5], and it is a very large effect with respect to what is generally reported for priming, which – as it has been reported – seems to need for doses above 20 Gy in order to saturate. Needless to emphasize, the behaviour of y 0 seems to indicate a persistent effect of priming – like PPC (Persistent
Fig. 7. Total counts of spectra (ci, top, exponential fits are also shown) and (bottom) average charge collection efficiency (g) and charge collection distance (ccd) versus time for holes corresponding to different X-ray priming. No illumination.
Fig. 9. Integral of last recorded spectra (see Fig. 6) versus dose (X-ray priming) in the case of holes collection. The linearity behaviour is better shown in the insert.
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
Fig. 10. Integral of spectra (ci, top) and (bottom) average charge collection efficiency (g) and charge collection distance (ccd) versus time for electrons (blue light illumination) and for holes (dark condition), both of them after X-ray priming. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
PhotoConductivity [5]) – on equilibrium conditions (i.e. for measurement times approaching to infinite). We point out again that the virgin conditions of the sample have been restored by a controlled annealing procedure before each measurement. 3.5. Long term measurements Two 33 h long measurements at the most investigated conditions in short term experiments have been carried out. In such a way, we studied the persistency of the effect of priming on holes in dark condition and on electrons under blue light illumination. The bias voltage was continuously applied to the sample during all the measurement time, as well as the blue light illumination (for electrons only). The spectra have been acquired using the same experimental procedure reported for short term experiment during the first hour, while during the subsequent hours we acquired three consecutive spectra at longer time intervals. We report the collection efficiency and the integral counts of spectra in Fig. 10. The behaviour during the first hour is similar to that observed in short term measurements, while in the subsequent hours the collection efficiency and to some extent also counts integral remains constant at increasing times. 4. Conclusions Measurements with alpha particles have been carried out both after X-ray priming and under blue light illumination on a CVD diamond sample used for tracking applications, trying to
1367
differentiate the contributions of electrons and holes by using different bias polarity and following the behaviour of both cce (charge collection efficiency) and ci (counts integral) with time for short and long term conditions. The main conclusion is that light helps to avoid long term polarization effects for electrons, and to some extent also short term polarization. After a strong priming (4.2 Gy), long term polarization is avoided also for holes. Short term polarization reduces cce by about 50% both for holes and electrons. To give some details, in short term conditions (i.e. during one h), cce is constant only for holes and under illumination, while ci is constant for electrons and under illumination and for holes in the dark. Illumination seems to affect more lower cce regions, while priming is likely to be more active in the higher ones. In the case of holes, illumination lowers cce, by erasing completely the effect of priming (and it keeps cce constant in time), while in the electron case it improves ci above the value obtained with priming (and it keeps ci constant in time). An interpretation of all the observed phenomena according to some model is out of the scope of the present paper, but some speculations are possible. Models of CVD diamond have been introduced by several authors [6 –9]: what is presently accepted is that there are hole traps in the grains and electron traps in the grain boundaries or in the more defective regions. The general view is that priming is a trap filling process: in our view [4,5,10] this trapping should involve only holes (but also electrons can be captured in the defective regions). By this way polarization due to holes is reduced. Light illumination should give rise to an optical detrapping of holes, erasing the effect of priming. What about electrons? Light has been proved to increase cce in these regions, which should be affected by space-charge effects: as a consequence, detrapping of electrons [11] should occur in these regions, but this detrapping should be kept constant in time by illumination. So, in order to avoid polarization due to electrons, shining light onto the sample must be kept constant in time. A particular study has been carried out on holes in the primed state without illumination: their response in terms of ci is found to be linear with X-ray doses below about 60 mGy, it decays for lower doses but it is stable in time (short terms conditions) when in saturation (about 600 mGy dose). The saturation of hole response (ci) at about 60 mGy (i.e. for a hole generation of only about 8 1012 cm 3) implies a very small concentration of hole trapping centers which, according to Refs. [7 – 10], should be distributed in the bulk, with a presumably a very large capture cross-section. The behaviour of ci with respect to dose may be compared with that of the photocurrent [5,9,10] and as a matter of fact the equilibrium value of ci at very long term displays a persistent effect like PPC. The fact that we observed this only for holes could imply that BGPC data should be referred to holes. In long term conditions, the primed state, after a decay during the first hour, is constant in time up to 33 h both under blue light illumination for electrons and in the dark for holes. As it was already observed [5,10], blue light illumination seems to improve the homogeneity of the response, by eliminating to some extent the effects generated by space-charge at grain boundaries.
1368
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
Acknowledgements The authors gratefully acknowledge the supply of the sample used in this work by H. Kagan, Ohio University, USA, in the framework of RD42 collaboration. References [1] B. Gudden, R. Pohl, Zeitschrift fur Physik 16 (1923) 170. [2] D. Meier, CVD diamond sensors for particle detection and tracking, Thesis, Duesseldorf 1999, p. 93. [3] C. Manfredotti, F. Fizzotti, A. Lo Giudice, C. Paolini, P. Olivero, E. Vittone, Diamond and Related Materials 12 (2003) 662.
[4] C. Manfredotti, E. Vittone, F. Fizzotti, A. Lo Giudice, C. Paolini, Diamond and Related Materials 11 (2002) 446. [5] C. Manfredotti, F. Fizzotti, E. Vittone, C. Paolini, P. Olivero, A. Lo Giudice, Diamond and Related Materials 13 (2004) 914. [6] P. Gonon, S. Prawer, D. Jamieson, Applied Physics Letters 70 (1997) 2996. [7] C.E. Nebel, et al., Semiconductor Science and Technology 18 (2003) S1. [8] C.E. Nebel, A. Waltenspiel, M. Stutzmann, M. Paul, L. Schaefer, Diamond and Related Materials 9 (2000) 404. [9] J. Alvarez, J.P. Kleider, P. Bergonzo, D. Tromson, E. Snidero, C. Mer, Diamond and Related Materials 12 (2003) 546. [10] C. Manfredotti, Diamond and Related Materials 14 (2005) 531. [11] S.M. Hearne, The induction and collection of charge in wide gap semiconductors, Thesis, School of Physics, Melbourne 2003, p. 138.
The primed state of CVD diamond under blue light illumination C. Manfredotti *, F. Fizzotti, Y. Garino Experimental Physics Department, University of Torino, Italy and Center of Excellence on Nanostructured Surfaces and Interfaces, University of Torino, Italy Received 26 May 2005; received in revised form 2 September 2005; accepted 10 October 2005 Available online 11 January 2006
Abstract The primed state of a detector grade CVD diamond sample has been investigated with respect to light illumination during alpha particles (5.5 MeV) detection. The measurements have been carried out as a function of elapsed time after beta-rays priming both in short term (1 h) and in long term (33 h) conditions. The contribution of holes and electrons to the average charge collection efficiency and to total number of counts above a threshold has been qualitatively separated by selecting bias polarity. The behaviour of electron collection after X-ray priming during blue light or UV illumination is improved while the hole one is worsened, confirming previous data. Linearity of the average collection efficiency for alpha particle is observed only at very low doses and in the hole case. Long term stability of both collection efficiency and of total number of counts is observed both in the primed state for holes and during blue light illumination for electrons. D 2005 Elsevier B.V. All rights reserved. Keywords: CVD diamond; Priming; Illumination; Stability
1. Introduction There are several controversies about the effect of light illumination on the primed state of CVD diamond as obtained by X- or beta-rays irradiation at doses around 20– 30 Gy: according to different authors, the effect could be either positive or negative, i.e. it can be used in order either to improve the performances of CVD diamond as nuclear detector [1] or to erase the improvement obtained by priming [2]. According to our experience, blue light illumination (around 400 nm) transfers the main contribution to charge collection from holes to electrons, and this conclusion has been largely demonstrated by using alpha particles in order to discriminate between electrons and holes contribution simply by changing bias polarity [3,4]. Illumination below 550 – 600 nm has been proved not to give rise to any effect on the primed state or at least to any improvement in holes and electrons collection. The investigation of primed state can be carried out also by Below Gap PhotoCurrent (BGPC) measurements [3] which can give some insight into transient or dynamic phenomena. In effect, the primed state is not stable under illumination for photon
* Corresponding author. Tel.: +39 0116707306. E-mail address: [email protected] (C. Manfredotti). 0925-9635/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.diamond.2005.10.006
energies below 2.7 eV, but it decays probably because of an optically stimulated detrapping of trapped holes generated by priming, which are responsible [5] of what has been called Persistent PhotoConductivity (PPC). As a matter of fact, the integrated photocurrent is linearly proportional to the cumulated priming dose, as it was the case of ThermoLuminescence (TL) which starts to be used to measure X-ray doses. What is very strange, at least in our case, is that the linearity [5] is restricted to very low doses (below 1 Gy for instance) with respect to TL. Above 2.7 eV the situation changes because probably the main contribution to photocurrent is given by electrons and a good long_term stability is reached. In order to throw some light on all these phenomena, we started a much longer investigation by using alpha particles spectra in the same way as BGPC as a function of time after priming and blue light illumination, which is a quite long and difficult task, but which gives a better understanding of electrons and holes behaviour. In order to follow the time behaviour, we extracted from alpha multichannel spectra both the centroid or the average value of charge collection efficiency (cce) and the total number of counts (ci), after subtracting the background. This latter parameter can in effect decrease in time because of polarization effects due to trapped charges that locally lower electric field and may cause charge pulses to fall below the electronic threshold. As a consequence, this
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
1363
parameter takes into a larger account regions of lower cce, which, according to our experience, should be more affected by blue light. 2. Experimental The CVD diamond sample used for alpha spectroscopy measurements was grown by MWCVD and it was previously used as a strip detector for tracking purposes. With minimum ionising particles (mips), cce turned out to be about 10%. The total sensitive surface area is 5.5 5.5 mm2 and thickness 600 Am. Strip contacts (width 19.8 Am and pitch 47.6 Am) were deposited on the growth side, while, in order to be illuminated, the sample was equipped with four parallel Cr/Au contacts (20 nm Cr, 100 nm Au, obtained by vacuum deposition) at the substrate side. Fig. 1 shows a SEM micrograph of the growth side of the sample equipped with strip contacts. Alpha particles measurements were performed by using a calibrated Am-241 alpha source (mean energy = 5.48 MeV, penetration depth in diamond = 13 Am including contacts), placed in front of the growth surface at a distance of about 1 mm from the specimen. The alpha particles were suitably collimated to a 2-mm diameter spot. Pulses were recorded by means of a standard charge sensitive electronic chain composed by amplifier and preamplifier (Ortec mods. 142 and 572 respectively). The sample was primed with a strong Sr-90 beta source for larger doses and with a weaker one for lower doses, being the doses measured at the sample position by standard TLD100 dosimeters read by a mod. 3500 Harshaw reader. With respect to mips, cce measured with alpha particles is much lower, not only because of the much lower bias voltage used in our case, but also because of the strong polarization due to the short range of alpha particles and to the consequent space-charge build-up in front of the detector. Before each measurement, the sample was annealed at 360 -C in order to completely recover the virgin state of the sample. Between the measurements and during the priming, the sample was kept in the dark. As already mentioned, because of the short penetration depth of alpha particles, spectroscopy measurements allow to discriminate between electrons and holes by imposing a positive or negative bias voltage, in the assumption that the main contribution is given by the carrier having at its disposal a longer path. This assumption is quite understandable, even if the obtained charge collection distances (ccd) are of the same order of magnitude of the penetration depth of alpha particles. In effect, carriers moving towards front or irradiated electrode have a higher chance of recombination with the opposite charge carriers and because the plasma generated along alpha particle track shields and lowers locally the electrical field. In any case, it is obvious that all the quoted results referring to different carriers are to be assumed in a semiquantitative way. Since the main topic of the paper is the time behaviour of charge collection after priming and during illumination (or darkness), the applied bias voltage was fixed at 200 V, i.e. + 200 V for electrons contribution and 200 V for holes
Fig. 1. SEM micrograph (7.0 kV, original magnification: 60) of the growth side of the CVD diamond sample.
contribution, even if the sample itself could support more than 600 V. In the following we will refer not to the voltage polarity, but only to the carriers that are mainly involved. The sample was annealed and subsequently primed. The sample was placed in the experimental apparatus exactly 5 min after priming, in order to obtain the same starting conditions for all the measurements. A series of consecutive spectra was then acquired for a total time of about 1 h. The time required for each spectrum was 4 min. The electronic threshold was calibrated in order to eliminate the electronic noise. The average charge collection efficiency and the integral of counts of each spectrum were calculated. The charge collection
1364
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
efficiency (cce) g was evaluated by calibration via a silicon surface barriers detector (surface area 25 mm2) with a 100% cce and energy resolution of 14 keV. For normalization the pair creation energies for Si (3.6 eV) and CVD diamond (13.2 eV) were taken into account. Since cce is relatively low, Hecht’s relationship, which is adopted because the penetration depth of alpha particles is much shorter than detector thickness, can be approximated to a linear behaviour g = d / L where L is the detector thickness and d is what is defined as charge collection distance (ccd). 3. Results and discussion 3.1. Short time measurements — holes behaviour The measurements were performed either in dark conditions or under constant illumination at a selected wavelength. We used a tungsten incandescent lamp (150 W) equipped with an interference filter in order to select a 400 nm light (blue light), which was laterally illuminating the sample (Fig. 2). We estimate the light power onto the sample to be approximately 10 AW. The behaviour of average cce (centroid) of the spectra as a function of time in different priming conditions is reported for holes in Fig. 3. First of all, we can observe that under blue light the average cce is lowered in agreement with previous data [4]. Secondly, in the case of blue light illumination there is a slight sensitivity with respect to primed dose, while in the dark the situation is unclear. The third observation is that in the primed case there is a small decay in time, while under illumination cce is constant. In the case of ci, the situation is just the opposite: under illumination ci decays. Since cce, being an average, is more sensitive to the higher efficiency part of the multichannel spectrum, and ci, on the other side, is more sensitive to the lower efficiency one, the conclusion could be quite easy. Priming, by improving cce, is more important in higher efficiency regions, which, being less in volume, are not important for ci, while illumination gives a larger contribution to lower efficiency regions, much wider and therefore more important in the case of ci. In all the cases the decays seem to be due to polarization effects which, in the case of ci, may push some charge pulses below the electronic threshold.
Fig. 3. Average charge collection efficiency (g) and charge collection distance (ccd) versus time for holes in different experimental conditions.
unprimed case, but also with respect to the primed case. In particular the effect of illumination on ci is very large. The second observation is that now ci under illumination is almost constant, while in the dark it decays, even being lower. The third observation is that cce decays in all the cases. The first conclusion is therefore that illumination avoids polarization effects, in the sense that the event rate is kept constant (what in effect has been observed in other kinds of measurements). The second conclusion is that in the electron the homogeneity of the response improves with time, while in the hole case illumination has the effect to eliminate lower cce regions and as a consequence to worsen the homogeneity with time. This
3.2. Short time measurements — electrons behaviour In the electron case (see Fig. 4) the first observation is that light improves both cce and ci with respect not only to the
Fig. 2. Schematic representation of the experimental apparatus.
Fig. 4. Integral of spectra (ci, top) and average charge collection efficiency (g) and charge collection distance (ccd) (bottom) versus time for electrons in different experimental conditions.
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
1365
contribution to holes collection in the dark after priming become the main suppliers of electron events under illumination. A further observation is that in the hole case there is, as already noticed, a small sensitivity to the priming dose, while for electrons this sensitivity disappears completely. 3.3. Short time measurements: dependence on the kind of illumination
Fig. 5. Integral of spectra (ci, top) and average charge collection efficiency (g) and charge collection distance (ccd) (bottom) versus time for electrons subjected to different kinds of illumination.
conclusion is in agreement with previous data [5]. In effect, looking at the equilibrium values of ci, one notices a curious interchange between electrons and holes, under dark and under illumination: if this is not just a coincidence, a possible conclusion could be that regions from which there is a strong
We studied also the effects due to different kinds of light illumination in the electron case. We considered different experimental conditions: a filtered W lamp (blue light), an unfiltered W lamp (white light) and a deuterium lamp (deuterium light, 30 W). Fig. 5 shows the behaviour of cce and ci, respectively, as a function of time. It is clear that deuterium lamp improves cce while it is not able to avoid an initial, small decay. In the case of ci, apart from its constancy in time, as observed before, the three kinds of illumination display the same behaviour both in time and in the absolute values. The conclusion could be that deuterium lamp (average wavelength 200 nm) has a larger effect on average or high cce regions. Of course the conclusion is only qualitative, since light intensities were not measured. Looking at multichannel spectra in different illumination conditions at different times after priming, it can be observed that, in dark conditions, the spectrum rapidly decreases with time and completely vanishes about 12 min after the first measurement. For sake of clearness, Fig. 6 shows multichannel spectra in different illumination conditions at four different times after priming: 5, 9, 13 and 17 min (5 min is the minimum interval between priming and measurement under illumination). The comparison between blue light (blue line) and deuterium light (green line) under-
Fig. 6. Consecutive spectra (elapsed time from previous spectrum: 4 min) for electrons subjected to different kinds of illumination: (A) first spectrum (0 – 4 min), (B) second spectrum (4 – 8 min), (C) third spectrum (8 – 12 min), (D) fourth spectrum (12 – 16 min).
1366
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
lines how, even if their integrals are comparable (see Fig. 5top), cce results higher for deuterium light, since in this case a larger number of counts on higher channels with respect to blue light can be observed. On lower channels the number of counts shows an opposite trend. 3.4. Hole case under priming alone: low priming doses and persistent effects In the hole case, priming effects at low doses were also carefully investigated. A fixed delay time (5 min) was used between the end of priming and the application of bias voltage to the sample. The results concerning cce and ci, respectively at different priming doses (no illumination) are reported in Fig. 7 (the same acquisition procedure as reported above was used). A y = y o + A 1e x/s fit of experimental curves of Fig. 7 (top) has been tried in order to get some conclusions and the results of the fits are reported in the same figure, while the behaviours of coefficient s and y 0 as a function of dose are shown in Fig. 8. Data referring to the highest dose of Fig. 7 (top) are not reported since the decay is very slow and the fit is affected by a large error. In any case, we can observe that s increases exponentially with dose, while y 0 increases linearly with the dose quite nicely. In other words, the equilibrium value of ci ( y 0) is linearly proportional to the dose, while the dependence of ci on time becomes weaker and weaker with the dose, approaching to a constant value at a dose value of 630 mGy.
Fig. 8. Behaviour of decay time constant s (a) and of initial value of integral counts y 0 (b) versus dose obtained from fittings reported in Fig. 7 (top).
In effect, the proportionality of ci with respect to priming dose is very clear, but, as it can be clearly noticed in Fig. 9 that shows the integral of the last recorded spectrum (after 3400 s) versus dose, the priming effect on the integral of spectra saturates already at about 10 mGy. This behaviour can be compared to what has been observed in photoconductivity [5], and it is a very large effect with respect to what is generally reported for priming, which – as it has been reported – seems to need for doses above 20 Gy in order to saturate. Needless to emphasize, the behaviour of y 0 seems to indicate a persistent effect of priming – like PPC (Persistent
Fig. 7. Total counts of spectra (ci, top, exponential fits are also shown) and (bottom) average charge collection efficiency (g) and charge collection distance (ccd) versus time for holes corresponding to different X-ray priming. No illumination.
Fig. 9. Integral of last recorded spectra (see Fig. 6) versus dose (X-ray priming) in the case of holes collection. The linearity behaviour is better shown in the insert.
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
Fig. 10. Integral of spectra (ci, top) and (bottom) average charge collection efficiency (g) and charge collection distance (ccd) versus time for electrons (blue light illumination) and for holes (dark condition), both of them after X-ray priming. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
PhotoConductivity [5]) – on equilibrium conditions (i.e. for measurement times approaching to infinite). We point out again that the virgin conditions of the sample have been restored by a controlled annealing procedure before each measurement. 3.5. Long term measurements Two 33 h long measurements at the most investigated conditions in short term experiments have been carried out. In such a way, we studied the persistency of the effect of priming on holes in dark condition and on electrons under blue light illumination. The bias voltage was continuously applied to the sample during all the measurement time, as well as the blue light illumination (for electrons only). The spectra have been acquired using the same experimental procedure reported for short term experiment during the first hour, while during the subsequent hours we acquired three consecutive spectra at longer time intervals. We report the collection efficiency and the integral counts of spectra in Fig. 10. The behaviour during the first hour is similar to that observed in short term measurements, while in the subsequent hours the collection efficiency and to some extent also counts integral remains constant at increasing times. 4. Conclusions Measurements with alpha particles have been carried out both after X-ray priming and under blue light illumination on a CVD diamond sample used for tracking applications, trying to
1367
differentiate the contributions of electrons and holes by using different bias polarity and following the behaviour of both cce (charge collection efficiency) and ci (counts integral) with time for short and long term conditions. The main conclusion is that light helps to avoid long term polarization effects for electrons, and to some extent also short term polarization. After a strong priming (4.2 Gy), long term polarization is avoided also for holes. Short term polarization reduces cce by about 50% both for holes and electrons. To give some details, in short term conditions (i.e. during one h), cce is constant only for holes and under illumination, while ci is constant for electrons and under illumination and for holes in the dark. Illumination seems to affect more lower cce regions, while priming is likely to be more active in the higher ones. In the case of holes, illumination lowers cce, by erasing completely the effect of priming (and it keeps cce constant in time), while in the electron case it improves ci above the value obtained with priming (and it keeps ci constant in time). An interpretation of all the observed phenomena according to some model is out of the scope of the present paper, but some speculations are possible. Models of CVD diamond have been introduced by several authors [6 –9]: what is presently accepted is that there are hole traps in the grains and electron traps in the grain boundaries or in the more defective regions. The general view is that priming is a trap filling process: in our view [4,5,10] this trapping should involve only holes (but also electrons can be captured in the defective regions). By this way polarization due to holes is reduced. Light illumination should give rise to an optical detrapping of holes, erasing the effect of priming. What about electrons? Light has been proved to increase cce in these regions, which should be affected by space-charge effects: as a consequence, detrapping of electrons [11] should occur in these regions, but this detrapping should be kept constant in time by illumination. So, in order to avoid polarization due to electrons, shining light onto the sample must be kept constant in time. A particular study has been carried out on holes in the primed state without illumination: their response in terms of ci is found to be linear with X-ray doses below about 60 mGy, it decays for lower doses but it is stable in time (short terms conditions) when in saturation (about 600 mGy dose). The saturation of hole response (ci) at about 60 mGy (i.e. for a hole generation of only about 8 1012 cm 3) implies a very small concentration of hole trapping centers which, according to Refs. [7 – 10], should be distributed in the bulk, with a presumably a very large capture cross-section. The behaviour of ci with respect to dose may be compared with that of the photocurrent [5,9,10] and as a matter of fact the equilibrium value of ci at very long term displays a persistent effect like PPC. The fact that we observed this only for holes could imply that BGPC data should be referred to holes. In long term conditions, the primed state, after a decay during the first hour, is constant in time up to 33 h both under blue light illumination for electrons and in the dark for holes. As it was already observed [5,10], blue light illumination seems to improve the homogeneity of the response, by eliminating to some extent the effects generated by space-charge at grain boundaries.
1368
C. Manfredotti et al. / Diamond & Related Materials 15 (2006) 1362 – 1368
Acknowledgements The authors gratefully acknowledge the supply of the sample used in this work by H. Kagan, Ohio University, USA, in the framework of RD42 collaboration. References [1] B. Gudden, R. Pohl, Zeitschrift fur Physik 16 (1923) 170. [2] D. Meier, CVD diamond sensors for particle detection and tracking, Thesis, Duesseldorf 1999, p. 93. [3] C. Manfredotti, F. Fizzotti, A. Lo Giudice, C. Paolini, P. Olivero, E. Vittone, Diamond and Related Materials 12 (2003) 662.
[4] C. Manfredotti, E. Vittone, F. Fizzotti, A. Lo Giudice, C. Paolini, Diamond and Related Materials 11 (2002) 446. [5] C. Manfredotti, F. Fizzotti, E. Vittone, C. Paolini, P. Olivero, A. Lo Giudice, Diamond and Related Materials 13 (2004) 914. [6] P. Gonon, S. Prawer, D. Jamieson, Applied Physics Letters 70 (1997) 2996. [7] C.E. Nebel, et al., Semiconductor Science and Technology 18 (2003) S1. [8] C.E. Nebel, A. Waltenspiel, M. Stutzmann, M. Paul, L. Schaefer, Diamond and Related Materials 9 (2000) 404. [9] J. Alvarez, J.P. Kleider, P. Bergonzo, D. Tromson, E. Snidero, C. Mer, Diamond and Related Materials 12 (2003) 546. [10] C. Manfredotti, Diamond and Related Materials 14 (2005) 531. [11] S.M. Hearne, The induction and collection of charge in wide gap semiconductors, Thesis, School of Physics, Melbourne 2003, p. 138.