- Email: [email protected]
Radiation monitoring with CVD diamonds in BABAR
ARTICLE IN PRESS
Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182 www.elsevier.com/locate/nima
Radiation monitoring with CVD diamonds in BABAR A.J. Edwardsb,, M. Bruinsmac, P. Burchatb, H. Kagana, R. Kassa, D. Kirkbyc, B.A. Petersenb, T. Pulliama a
Department of Physics, Ohio State University, 174 West 18th Ave., Columbus, OH 43210, USA b Department of Physics, Stanford University, 382 Via Pueblo Mall, Stanford, CA 94305, USA c Department of Physics and Astronomy, University of California at Irvine, 4129 Frederick Reines Hall, Irvine, CA 92697, USA Available online 1 July 2005
Abstract The BABAR experiment has been using two polycrystalline chemical-vapor-deposition (pCVD) diamonds for radiation monitoring for nearly 2 years. In July 2005, an additional 12 diamond based radiation sensors will be installed inside the BABAR detector. These diamonds will take over the function of 12 silicon PIN-diodes that are currently used in the radiation protection and monitoring system. We describe our highly successful experience with using pCVD diamond radiation sensors in a high energy physics experiment. We also detail our findings of persistent signal currents and magnetically suppressed erratic dark currents in pCVD diamond based radiation sensors. r 2005 Elsevier B.V. All rights reserved. PACS: 29.40.Wk; 81.05.Uw Keywords: Diamond; Radiation; Dosimetry
1. Introduction In the BABAR [1] detector at the Stanford Linear Accelerator Center, the silicon vertex tracker (SVT) [2] is the nearest instrument to the interaction point of a 3.1 GeV positron beam and a 9.0 GeV electron beam in the PEP-II storage ring. Because the first layer of silicon is at a radius of 3.3 cm from the interaction point, the SVT is Corresponding author. Tel.: +650 926 5495.
E-mail address: [email protected] (A.J. Edwards).
susceptible to high levels of radiation. The silicon wafers in the SVT and their front-end readout electronics can be damaged and/or made inoperable by both chronic exposure to and sudden bursts of radiation. Inside the SVT, 12 silicon PINdiodes are used as part of the SVT’s radiation monitoring and protection system (SVTRAD) [3]. These diodes are reverse biased at 50 V and their current is monitored. Ionizing radiation incident on the diodes causes an increase in current proportional to the dose rate. SVTRAD monitors the accumulated dose absorbed by the PIN diodes.
0168-9002/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2005.06.028
ARTICLE IN PRESS A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
It also aborts the positron and electron beams, in less than a millisecond, if the dose rate becomes too high. With the SVTRAD system, the SVT is expected to accumulate not more than 40 kGy (4 Mrad) of electromagnetic radiation by the summer of 2005. The 12 silicon PIN-diodes in SVTRAD are themselves damaged by radiation, making monitoring difficult. These diodes have an active volume 1 1 cm2 and 300 mm thick. Based on their size and the amount of energy needed to create electron–hole pairs in silicon, the diodes give a 1 nA signal current for every 52 mGy=s. The unirradiated diodes had leakage currents on the order of 1 nA. However, diodes that have received approximately 20 Gy dose, originating from the positron and electron beams, now have leakage currents above 2 mA. This gives a radiation signal current to intrinsic leakage current ratio of about 1/1000. In addition, the leakage current varies by 10%/1C change in temperature. Variations in leakage current must be closely monitored so they do not obscure the changes in the signal current. In order to correct for temperature variations, thermistors monitor each silicon PIN-diode’s temperature with a precision of 0.011. These temperatures are then used in calculating the portion of total current that is due to radiation. It is expected that the leakage current will continue to increase at its present rate of 1 nA/10 Gy of radiation. At this rate, parts of the SVTRAD system will become inoperable years before the BABAR experiment is planned to end. In order to overcome the difficulty of measuring radiation induced current signals with silicon PINdiodes, an alternative radiation sensor technology, based on polycrystalline chemical-vapor-deposition (pCVD) diamond, has been studied. As discussed below, diamond based sensors have performed remarkably well in the BABAR experiment. Twelve new pCVD diamond based radiation sensors are now being produced for installation inside BABAR during the summer of 2005. These diamond based sensors will be used to perform the function of all 12 silicon PIN-diodes in the SVTRAD system. Some unforeseen effects in pCVD diamond based sensors have been found. These effects are innocuous for our application in
177
BABAR, but they may pose difficulties to other radiation monitoring applications.
2. pCVD diamond and pCVD diamond sensors in BABAR Polycrystalline chemical-vapor-deposition diamond has been produced for many years [4,5]. The quality of pCVD diamond for radiation sensors is typically measured by its charge collection distance (CCD) [6,7]. The CCD is the average separation distance of electron–hole pairs under an electric field before they recombine. Current technology allows the growth of pCVD diamonds with CCD of hundreds of micrometers. A diamond sensor’s CCD is dependent on the bias voltage applied to it and increases with the applied bias voltage. For pCVD diamonds, the CCD typically saturates at approximately 1 V mm [6]. In BABAR, pCVD diamond sensors are operated in the same manner as the silicon PIN-diodes. A bias voltage is applied across a diamond and the current in the circuit is measured. The signal current is proportional to the amount of radiation absorbed by the diamond. The main advantage of pCVD diamond over silicon is its radiation hardness [8–11]. Accumulated radiation dose causes a significant increase in the leakage currents of the silicon PIN-diodes in BABAR. For pCVD diamond sensors, the leakage currents remain small even after accumulating high doses of radiation. Our packaging of pCVD diamond gives leakage currents from a few pA to a few tens of pA. For diamond sensors, there is also no visible change of signal or leakage current due to nominal temperature changes in the SVT operational range of 9–35 1C. Two pCVD diamonds were installed inside BABAR in August 2002 [13]. They were placed in the horizontal plane of the detector, one on either side, about 15 cm from the beams’ interaction point and about 5 cm from the beam line. Their position is similar to that of the silicon PINdiodes. The diamonds are 1 cm 1 cm and 500 mm thick. Metal electrodes are deposited on the two faces of the diamonds making ohmic contacts. The metal contacts are 9 mm 9 mm. Coaxial cables
ARTICLE IN PRESS 178
A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
are connected to the ohmic contacts using indium solder. The diamonds and cables are electrically shielded and insulated. The two diamonds reach a maximum CCD of approximately 200 mm at a bias voltage of 500 V. Both diamonds are operated at 500 V in order to ensure a maximal signal. The same custom electronics and software designed for the SVTRAD system and the silicon PIN-diodes [12] are used to monitor the two pCVD diamonds in BABAR. The system has 20pA current resolution. The signals from the diamonds can also be passed into the same beam-abort-logic circuitry as the silicon diodes. The response of the diamond sensors installed inside BABAR is approximately 1 nA of current for every 70 mGy=s of radiation.
3. Diamond sensor performance in BABAR We have monitored the response of the two pCVD diamonds to radiation in BABAR for approximately 22 months and compared them with the signals from the 12 silicon PIN-diodes in the SVTRAD system. Over this period of time we have observed that the diamond sensors provide a reliable and consistent response to radiation. Fig. 1 shows that a diamond sensor and a silicon diode with similar location in the SVT have comparable responses to dose rates over the course of several hours. This is due to the similar sensor sizes, sensitivities, and radiation environments for the two types of sensors. Leakage currents in the diamond sensors have remained negligible (below the sensitivity of the monitoring electronics), with no detectable temperature sensitivity. After exposure to over 10 kGy, both diamonds have not shown any decrease in sensitivity. The dose rates calculated from similarly located diodes and diamonds have been observed to be proportional to each other over the range of tens of mGy=s to a few mGy/s. The signal currents from both diamond sensors are currently being used to provide prompt feedback to BABAR and PEP-II operators. The response of the two pCVD diamond sensors to radiation in BABAR has also been studied at nanosecond time scales. This is to ensure that the
Fig. 1. Comparison of a pCVD diamond-sensor signal (black curve) and a silicon PIN-diode monitoring radiation in the SVT. Radiation levels in the SVT roughly track the electron beam current. The beam current gradually decreases between successive fills of the PEP-II storage ring. The diamond sensor current is comparable to that of the silicon PIN-diode. At two points the beam is aborted. The first abort is due to an acute spike in the dose rate; the second abort is due to a smaller but longer lived increase in the dose rate. Both increases in radiation are seen in the diamond sensor and the silicon diode sensor.
diamonds can provide a signal in a short enough time to be able to abort the positron and electron beams before severe SVT damage. Using a custom fast amplifier and an oscilloscope, we have observed that the diamond sensors respond to radiation within 20 ns, which is the limit imposed by the speed of the readout electronics. This is well below our required response time of 10 ms. Over the course of several months we have recorded the signals from silicon PIN-diodes and diamond sensors during radiation bursts. Fig. 2 shows an example of the signals from a diamond and a diode during the same large radiation event. Comparisons of the two types of sensors during such events have shown that the diamond sensors can consistently provide signals suitable for use in the SVTRAD system. With both diamond signals and silicon PIN-diode signals supplied to the same SVTRAD abort circuitry, 90% of the time when a diode gave an abort signal a diamond did as well. We do not expect all radiation incidents to trigger an abort in both sensors because of their different locations.
ARTICLE IN PRESS A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
179
Fig. 2. During the same high-radiation event in the SVT, both a diamond sensor (left) and a silicon PIN-diode (right) show comparable responses. It is important for the protection of the SVT that such events be accurately and consistently measured so that beam abort decisions can be made.
All requirements for the SVTRAD system have been shown to be met by the pCVD diamond based radiation sensors. Following these preliminary observations, the installed diamond based sensors have been incorporated into several aspects of detector operation. In particular, both diamond sensors are used in the SVTRAD beam abort system in the place of two silicon PIN-diodes that no longer are able to give adequate measurements. The two diamond sensors now abort the beams if they detect a chronic exposure to over 0.5 mGy/s of radiation. The two diamonds are also used now to monitor the accumulated dose the SVT receives because the systematic errors, from temperature dependencies, for the two corresponding silicon PIN-diodes make the two diamond sensors the favorable choice.
4. Additional effects in installed diamond sensors We observe that when dose rates change suddenly, the diamond signal promptly changes by about 90% of its full response, followed by a nonexponential approach to its final value. This dynamic response to instantaneous radiation change is shown in Fig. 3 over a time interval of minutes. A slow response component is also observed at millisecond time scales for both diamond sensors. The slow component of the
Fig. 3. Response of a diamond sensor in BABAR to loss of positron and electron beams: At time zero PEP-II loses both positron and electron beams. The radiation level immediately drops to zero (as verified by the silicon PIN-diodes). Current from the diamond sensor, however, shows a slow response to the change in dose rate after a fast initial response.
pffiffi response empirically fits a 1= t curve on the seconds time scale. This effect may be due to charge traps in the diamond. Fortunately, this slow component of response accounts for a negligible error for our monitoring purposes. For instance, in Fig. 3, the additional current from the slow component after 0 s corresponds to about 1.5 mGy. The slow component of response also does not affect the diamond’s ability to protect the SVT in the event of a sudden increase in the
ARTICLE IN PRESS 180
A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
radiation level. We have verified this slow component current effect with additional diamond sensors outside BABAR, in tests with a 60 Co source. We have also observed erratic dark currents (EDC) that respond to the magnetic field in the two installed diamond sensors. During normal operation of the BABAR detector, the sensors are in a 1.5-T magnetic field approximately perpendicular to the applied electric field and approximately perpendicular to the diamond grain boundaries as well. As seen in Fig. 4, when the magnetic field is turned off, the dark currents in both diamond sensors increase over the course of several minutes. Once the magnetic field is turned on, the EDC in the diamond sensors return to their stable and low values. This EDC and magnetic effect was first observed in February 2003. The EDC are at different levels for the two diamonds, one being in the nA range while the other is around 100 nA, for periods of several weeks with the magnetic field off. The EDC have been observed to persist. Periods of weeks without a bias voltage appear to reduce the magnitude of the EDC by a significant fraction, but not to eliminate them.
Fig. 4. Diamond sensor response to magnetic field changes in BABAR: At approximately 0.1 h, the 1.5-T magnetic field in BABAR is turned off. A dark current in the diamond sensor appears within minutes and increases to several nA. From about 3 h through 3.6 h, the magnet field is increased linearly from 0 T back up to its nominal value of 1.5 T. During this time the dark current disappears. This phenomenon has been reproduced outside the BABAR detector.
These erratic dark currents are larger than the radiation signals that the sensors measure. If present during normal BABAR operation, the EDC would obscure the signals needed for radiation monitoring. Fortunately, there is always the 1.5-T magnetic field during normal operations and the EDC are always suppressed by this magnetic field.
5. Laboratory tests of erratic dark currents The erratic dark currents shown in Fig. 4 have been verified in laboratory tests. Four additional diamond [13] radiation sensors were made: two with one square contact, and two segmented into four separate square contacts. All four diamond sensors were also metalized with guard rings around the central contacts. No surface currents around the diamond edges were found with the guard rings, so all currents are believed to come from the bulk of the diamonds. Various metalizations, such as CrAu, were used to make the contacts. No variations in EDC between the differently metalized diamonds were observed. Packaging and readout electronics were also verified to not be sources of the EDC. With these 14 metalized regions of diamond, we have found that the EDC are initialized by a combination of radiation and high bias voltage. During pumping and CCD measurements, the four diamonds received approximately 10 Gy from a 90 Sr source. Afterward, three of the 14 channels displayed EDC. All four diamonds were subsequently irradiated with an approximately 1 kGy dose from 60 Co. One of the diamond sensors segmented into four channels was left unbiased during the irradiation process while the other three diamond sensors were biased at 500 V. All biased channels developed EDC while none of the unbiased channels did. A second 60 Co irradiation of the previously unbiased diamond, this time with a 500 V bias, caused that diamond to develop EDC in every channel. All biased channels were monitored during irradiations and it was seen that the EDC developed in each channel independently. After additional irradiations at various bias voltages, 11 of the 14 channels consistently showed EDC. These lab-induced EDC are observed on scales of pA to tens of nA.
ARTICLE IN PRESS A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
181
The EDC in all four test diamond sensors were tested with an electromagnet with magnetic fields up to 1.6 T. We found that with the magnetic field oriented perpendicular to diamond grain boundaries and the electric field, the same as in BABAR, the EDC are always fully suppressed. This is illustrated in Fig. 5 for a subset of diamond channels. Scans of EDC versus magnetic field strength show that the EDC are fully suppressed for fields in the range 0.1–0.6 T. Orienting the magnetic field to be parallel to the diamond grain boundaries has little or no effect on the EDC at magnetic field strengths up to 1.6 T. Further tests at intermediate angles suggest that only the component of magnetic field perpendicular to the grain boundaries suppresses the EDC. The lowering of the bias voltage has also been found to suppress EDC. No EDC has been seen in any diamond with a bias of 100 V. The EDC in the 14 diamond channels are observed to disappear when the voltage is lowered to a range between 500 and 100 V. Both an appropriate magnetic field or lower applied voltage suppress EDC. A magnetic field of adequate strength and direction is not always feasible however. The simple solution is lower applied voltages when an adequate magnetic field is not possible. This would however lower the
operational CCD of the diamond and thus the signal current. The suppression of EDC with a magnetic field does not lower the signal current. These tests lead us to believe that the EDC may be caused by grain boundaries in pCVD diamonds. Future tests with single-crystal chemicalvapor-deposition diamond are planned in order to test this hypothesis.
Fig. 5. Current from five individual electrodes on four diamond sensors, with a bias voltage of 500 V applied. A 1.5 T magnetic field, oriented perpendicular to the diamond grain boundaries, is turned on at the time of 150 s. Erratic dark currents from five different orders of magnitude are seen to be fully suppressed by this applied magnetic field.
Acknowledgements
6. Conclusions CVD diamond radiation sensors are a viable alternative to silicon PIN-diodes for the measurement of ionizing radiation in BABAR. The SVTRAD system must have the ability to monitor large amounts of accumulated dose and also respond quickly to any sudden increases in dose rate. The pCVD diamonds tested in BABAR have shown that they are able to fulfill these requirements. They have also been shown to have advantages over silicon PIN-diodes: namely their radiation hardness, their low leakage currents, and their insensitivity to ambient temperature changes. Based on nearly 2 years of operation in BABAR, we have demonstrated that pCVD diamond sensors can provide fast, reliable, and accurate measurements of the SVT’s radiation dose. In addition, we have found some interesting current effects in our diamond sensors. The slow component of current response to sudden changes in radiation level is found to contribute a negligibly small error in our applications. The erratic dark currents, although large enough to obscure signal currents, are always completely suppressed by the serendipitously oriented 1.5-T magnetic field inside BABAR. Due to the good performance of the pCVD diamond-based radiation sensors that we have tested, we now plan to install an additional 12 diamond sensors for radiation monitoring and protection in BABAR during the summer of 2005.
The authors would like to thank the RD42 collaboration for working with us, sharing information and lending us diamonds. We would also
ARTICLE IN PRESS 182
A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
like to thank Element Six Ltd. for providing diamonds for our research. [6]
References [7] [1] [2] [3] [4]
B. Aubert, et al., Nucl. Instr. and Meth. B 479 (2002) 1. V. Re, et al., IEEE Trans. Nucl. Sci. NS-49 (2002) 3284. T.I. Meyer, et al., Int. J. Mod. Phys. A 16S1C (2001) 1084. M.H. Nazare, et al., Development of diamond tracking detectors for high luminosity experiments at the LHC, RD42 Proposal CERN/DRDC 94-21, 1994. [5] W. Adam, et al., Development of diamond tracking detectors for high luminosity experiments at the LHC,
[8] [9] [10] [11] [12] [13]
CERN/LHCC Status Reports 95-43, 97-3, 2000-015, 2002010. S. Zhao, Characterization of the electrical properties of polycrystalline diamond films, Dissertation, Ohio State University, 1994. D. Meier, CVD diamond sensors for particle detection and tracking, Dissertation, Universitat Heidelberg, 1999. C. Bauer, et al., Nucl. Instr. and Meth. A 367 (1995) 207. D. Meier, et al., Nucl. Instr. and Meth. A 426 (1999) 173. T. Behnke, et al., Nucl. Instr. and Meth. A 489 (2002) 230. R.J. Tapper, Rep. Prog. Phys. 63 (2000) 1273. V. Re, et al., Nucl. Instr. and Meth. A 518 (2004) 290. The pCVD diamonds were made by Element Six Ltd., King’s Ride Park, Ascot, Berkshire SL5 8BP, UK.
Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182 www.elsevier.com/locate/nima
Radiation monitoring with CVD diamonds in BABAR A.J. Edwardsb,, M. Bruinsmac, P. Burchatb, H. Kagana, R. Kassa, D. Kirkbyc, B.A. Petersenb, T. Pulliama a
Department of Physics, Ohio State University, 174 West 18th Ave., Columbus, OH 43210, USA b Department of Physics, Stanford University, 382 Via Pueblo Mall, Stanford, CA 94305, USA c Department of Physics and Astronomy, University of California at Irvine, 4129 Frederick Reines Hall, Irvine, CA 92697, USA Available online 1 July 2005
Abstract The BABAR experiment has been using two polycrystalline chemical-vapor-deposition (pCVD) diamonds for radiation monitoring for nearly 2 years. In July 2005, an additional 12 diamond based radiation sensors will be installed inside the BABAR detector. These diamonds will take over the function of 12 silicon PIN-diodes that are currently used in the radiation protection and monitoring system. We describe our highly successful experience with using pCVD diamond radiation sensors in a high energy physics experiment. We also detail our findings of persistent signal currents and magnetically suppressed erratic dark currents in pCVD diamond based radiation sensors. r 2005 Elsevier B.V. All rights reserved. PACS: 29.40.Wk; 81.05.Uw Keywords: Diamond; Radiation; Dosimetry
1. Introduction In the BABAR [1] detector at the Stanford Linear Accelerator Center, the silicon vertex tracker (SVT) [2] is the nearest instrument to the interaction point of a 3.1 GeV positron beam and a 9.0 GeV electron beam in the PEP-II storage ring. Because the first layer of silicon is at a radius of 3.3 cm from the interaction point, the SVT is Corresponding author. Tel.: +650 926 5495.
E-mail address: [email protected] (A.J. Edwards).
susceptible to high levels of radiation. The silicon wafers in the SVT and their front-end readout electronics can be damaged and/or made inoperable by both chronic exposure to and sudden bursts of radiation. Inside the SVT, 12 silicon PINdiodes are used as part of the SVT’s radiation monitoring and protection system (SVTRAD) [3]. These diodes are reverse biased at 50 V and their current is monitored. Ionizing radiation incident on the diodes causes an increase in current proportional to the dose rate. SVTRAD monitors the accumulated dose absorbed by the PIN diodes.
0168-9002/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2005.06.028
ARTICLE IN PRESS A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
It also aborts the positron and electron beams, in less than a millisecond, if the dose rate becomes too high. With the SVTRAD system, the SVT is expected to accumulate not more than 40 kGy (4 Mrad) of electromagnetic radiation by the summer of 2005. The 12 silicon PIN-diodes in SVTRAD are themselves damaged by radiation, making monitoring difficult. These diodes have an active volume 1 1 cm2 and 300 mm thick. Based on their size and the amount of energy needed to create electron–hole pairs in silicon, the diodes give a 1 nA signal current for every 52 mGy=s. The unirradiated diodes had leakage currents on the order of 1 nA. However, diodes that have received approximately 20 Gy dose, originating from the positron and electron beams, now have leakage currents above 2 mA. This gives a radiation signal current to intrinsic leakage current ratio of about 1/1000. In addition, the leakage current varies by 10%/1C change in temperature. Variations in leakage current must be closely monitored so they do not obscure the changes in the signal current. In order to correct for temperature variations, thermistors monitor each silicon PIN-diode’s temperature with a precision of 0.011. These temperatures are then used in calculating the portion of total current that is due to radiation. It is expected that the leakage current will continue to increase at its present rate of 1 nA/10 Gy of radiation. At this rate, parts of the SVTRAD system will become inoperable years before the BABAR experiment is planned to end. In order to overcome the difficulty of measuring radiation induced current signals with silicon PINdiodes, an alternative radiation sensor technology, based on polycrystalline chemical-vapor-deposition (pCVD) diamond, has been studied. As discussed below, diamond based sensors have performed remarkably well in the BABAR experiment. Twelve new pCVD diamond based radiation sensors are now being produced for installation inside BABAR during the summer of 2005. These diamond based sensors will be used to perform the function of all 12 silicon PIN-diodes in the SVTRAD system. Some unforeseen effects in pCVD diamond based sensors have been found. These effects are innocuous for our application in
177
BABAR, but they may pose difficulties to other radiation monitoring applications.
2. pCVD diamond and pCVD diamond sensors in BABAR Polycrystalline chemical-vapor-deposition diamond has been produced for many years [4,5]. The quality of pCVD diamond for radiation sensors is typically measured by its charge collection distance (CCD) [6,7]. The CCD is the average separation distance of electron–hole pairs under an electric field before they recombine. Current technology allows the growth of pCVD diamonds with CCD of hundreds of micrometers. A diamond sensor’s CCD is dependent on the bias voltage applied to it and increases with the applied bias voltage. For pCVD diamonds, the CCD typically saturates at approximately 1 V mm [6]. In BABAR, pCVD diamond sensors are operated in the same manner as the silicon PIN-diodes. A bias voltage is applied across a diamond and the current in the circuit is measured. The signal current is proportional to the amount of radiation absorbed by the diamond. The main advantage of pCVD diamond over silicon is its radiation hardness [8–11]. Accumulated radiation dose causes a significant increase in the leakage currents of the silicon PIN-diodes in BABAR. For pCVD diamond sensors, the leakage currents remain small even after accumulating high doses of radiation. Our packaging of pCVD diamond gives leakage currents from a few pA to a few tens of pA. For diamond sensors, there is also no visible change of signal or leakage current due to nominal temperature changes in the SVT operational range of 9–35 1C. Two pCVD diamonds were installed inside BABAR in August 2002 [13]. They were placed in the horizontal plane of the detector, one on either side, about 15 cm from the beams’ interaction point and about 5 cm from the beam line. Their position is similar to that of the silicon PINdiodes. The diamonds are 1 cm 1 cm and 500 mm thick. Metal electrodes are deposited on the two faces of the diamonds making ohmic contacts. The metal contacts are 9 mm 9 mm. Coaxial cables
ARTICLE IN PRESS 178
A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
are connected to the ohmic contacts using indium solder. The diamonds and cables are electrically shielded and insulated. The two diamonds reach a maximum CCD of approximately 200 mm at a bias voltage of 500 V. Both diamonds are operated at 500 V in order to ensure a maximal signal. The same custom electronics and software designed for the SVTRAD system and the silicon PIN-diodes [12] are used to monitor the two pCVD diamonds in BABAR. The system has 20pA current resolution. The signals from the diamonds can also be passed into the same beam-abort-logic circuitry as the silicon diodes. The response of the diamond sensors installed inside BABAR is approximately 1 nA of current for every 70 mGy=s of radiation.
3. Diamond sensor performance in BABAR We have monitored the response of the two pCVD diamonds to radiation in BABAR for approximately 22 months and compared them with the signals from the 12 silicon PIN-diodes in the SVTRAD system. Over this period of time we have observed that the diamond sensors provide a reliable and consistent response to radiation. Fig. 1 shows that a diamond sensor and a silicon diode with similar location in the SVT have comparable responses to dose rates over the course of several hours. This is due to the similar sensor sizes, sensitivities, and radiation environments for the two types of sensors. Leakage currents in the diamond sensors have remained negligible (below the sensitivity of the monitoring electronics), with no detectable temperature sensitivity. After exposure to over 10 kGy, both diamonds have not shown any decrease in sensitivity. The dose rates calculated from similarly located diodes and diamonds have been observed to be proportional to each other over the range of tens of mGy=s to a few mGy/s. The signal currents from both diamond sensors are currently being used to provide prompt feedback to BABAR and PEP-II operators. The response of the two pCVD diamond sensors to radiation in BABAR has also been studied at nanosecond time scales. This is to ensure that the
Fig. 1. Comparison of a pCVD diamond-sensor signal (black curve) and a silicon PIN-diode monitoring radiation in the SVT. Radiation levels in the SVT roughly track the electron beam current. The beam current gradually decreases between successive fills of the PEP-II storage ring. The diamond sensor current is comparable to that of the silicon PIN-diode. At two points the beam is aborted. The first abort is due to an acute spike in the dose rate; the second abort is due to a smaller but longer lived increase in the dose rate. Both increases in radiation are seen in the diamond sensor and the silicon diode sensor.
diamonds can provide a signal in a short enough time to be able to abort the positron and electron beams before severe SVT damage. Using a custom fast amplifier and an oscilloscope, we have observed that the diamond sensors respond to radiation within 20 ns, which is the limit imposed by the speed of the readout electronics. This is well below our required response time of 10 ms. Over the course of several months we have recorded the signals from silicon PIN-diodes and diamond sensors during radiation bursts. Fig. 2 shows an example of the signals from a diamond and a diode during the same large radiation event. Comparisons of the two types of sensors during such events have shown that the diamond sensors can consistently provide signals suitable for use in the SVTRAD system. With both diamond signals and silicon PIN-diode signals supplied to the same SVTRAD abort circuitry, 90% of the time when a diode gave an abort signal a diamond did as well. We do not expect all radiation incidents to trigger an abort in both sensors because of their different locations.
ARTICLE IN PRESS A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
179
Fig. 2. During the same high-radiation event in the SVT, both a diamond sensor (left) and a silicon PIN-diode (right) show comparable responses. It is important for the protection of the SVT that such events be accurately and consistently measured so that beam abort decisions can be made.
All requirements for the SVTRAD system have been shown to be met by the pCVD diamond based radiation sensors. Following these preliminary observations, the installed diamond based sensors have been incorporated into several aspects of detector operation. In particular, both diamond sensors are used in the SVTRAD beam abort system in the place of two silicon PIN-diodes that no longer are able to give adequate measurements. The two diamond sensors now abort the beams if they detect a chronic exposure to over 0.5 mGy/s of radiation. The two diamonds are also used now to monitor the accumulated dose the SVT receives because the systematic errors, from temperature dependencies, for the two corresponding silicon PIN-diodes make the two diamond sensors the favorable choice.
4. Additional effects in installed diamond sensors We observe that when dose rates change suddenly, the diamond signal promptly changes by about 90% of its full response, followed by a nonexponential approach to its final value. This dynamic response to instantaneous radiation change is shown in Fig. 3 over a time interval of minutes. A slow response component is also observed at millisecond time scales for both diamond sensors. The slow component of the
Fig. 3. Response of a diamond sensor in BABAR to loss of positron and electron beams: At time zero PEP-II loses both positron and electron beams. The radiation level immediately drops to zero (as verified by the silicon PIN-diodes). Current from the diamond sensor, however, shows a slow response to the change in dose rate after a fast initial response.
pffiffi response empirically fits a 1= t curve on the seconds time scale. This effect may be due to charge traps in the diamond. Fortunately, this slow component of response accounts for a negligible error for our monitoring purposes. For instance, in Fig. 3, the additional current from the slow component after 0 s corresponds to about 1.5 mGy. The slow component of response also does not affect the diamond’s ability to protect the SVT in the event of a sudden increase in the
ARTICLE IN PRESS 180
A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
radiation level. We have verified this slow component current effect with additional diamond sensors outside BABAR, in tests with a 60 Co source. We have also observed erratic dark currents (EDC) that respond to the magnetic field in the two installed diamond sensors. During normal operation of the BABAR detector, the sensors are in a 1.5-T magnetic field approximately perpendicular to the applied electric field and approximately perpendicular to the diamond grain boundaries as well. As seen in Fig. 4, when the magnetic field is turned off, the dark currents in both diamond sensors increase over the course of several minutes. Once the magnetic field is turned on, the EDC in the diamond sensors return to their stable and low values. This EDC and magnetic effect was first observed in February 2003. The EDC are at different levels for the two diamonds, one being in the nA range while the other is around 100 nA, for periods of several weeks with the magnetic field off. The EDC have been observed to persist. Periods of weeks without a bias voltage appear to reduce the magnitude of the EDC by a significant fraction, but not to eliminate them.
Fig. 4. Diamond sensor response to magnetic field changes in BABAR: At approximately 0.1 h, the 1.5-T magnetic field in BABAR is turned off. A dark current in the diamond sensor appears within minutes and increases to several nA. From about 3 h through 3.6 h, the magnet field is increased linearly from 0 T back up to its nominal value of 1.5 T. During this time the dark current disappears. This phenomenon has been reproduced outside the BABAR detector.
These erratic dark currents are larger than the radiation signals that the sensors measure. If present during normal BABAR operation, the EDC would obscure the signals needed for radiation monitoring. Fortunately, there is always the 1.5-T magnetic field during normal operations and the EDC are always suppressed by this magnetic field.
5. Laboratory tests of erratic dark currents The erratic dark currents shown in Fig. 4 have been verified in laboratory tests. Four additional diamond [13] radiation sensors were made: two with one square contact, and two segmented into four separate square contacts. All four diamond sensors were also metalized with guard rings around the central contacts. No surface currents around the diamond edges were found with the guard rings, so all currents are believed to come from the bulk of the diamonds. Various metalizations, such as CrAu, were used to make the contacts. No variations in EDC between the differently metalized diamonds were observed. Packaging and readout electronics were also verified to not be sources of the EDC. With these 14 metalized regions of diamond, we have found that the EDC are initialized by a combination of radiation and high bias voltage. During pumping and CCD measurements, the four diamonds received approximately 10 Gy from a 90 Sr source. Afterward, three of the 14 channels displayed EDC. All four diamonds were subsequently irradiated with an approximately 1 kGy dose from 60 Co. One of the diamond sensors segmented into four channels was left unbiased during the irradiation process while the other three diamond sensors were biased at 500 V. All biased channels developed EDC while none of the unbiased channels did. A second 60 Co irradiation of the previously unbiased diamond, this time with a 500 V bias, caused that diamond to develop EDC in every channel. All biased channels were monitored during irradiations and it was seen that the EDC developed in each channel independently. After additional irradiations at various bias voltages, 11 of the 14 channels consistently showed EDC. These lab-induced EDC are observed on scales of pA to tens of nA.
ARTICLE IN PRESS A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
181
The EDC in all four test diamond sensors were tested with an electromagnet with magnetic fields up to 1.6 T. We found that with the magnetic field oriented perpendicular to diamond grain boundaries and the electric field, the same as in BABAR, the EDC are always fully suppressed. This is illustrated in Fig. 5 for a subset of diamond channels. Scans of EDC versus magnetic field strength show that the EDC are fully suppressed for fields in the range 0.1–0.6 T. Orienting the magnetic field to be parallel to the diamond grain boundaries has little or no effect on the EDC at magnetic field strengths up to 1.6 T. Further tests at intermediate angles suggest that only the component of magnetic field perpendicular to the grain boundaries suppresses the EDC. The lowering of the bias voltage has also been found to suppress EDC. No EDC has been seen in any diamond with a bias of 100 V. The EDC in the 14 diamond channels are observed to disappear when the voltage is lowered to a range between 500 and 100 V. Both an appropriate magnetic field or lower applied voltage suppress EDC. A magnetic field of adequate strength and direction is not always feasible however. The simple solution is lower applied voltages when an adequate magnetic field is not possible. This would however lower the
operational CCD of the diamond and thus the signal current. The suppression of EDC with a magnetic field does not lower the signal current. These tests lead us to believe that the EDC may be caused by grain boundaries in pCVD diamonds. Future tests with single-crystal chemicalvapor-deposition diamond are planned in order to test this hypothesis.
Fig. 5. Current from five individual electrodes on four diamond sensors, with a bias voltage of 500 V applied. A 1.5 T magnetic field, oriented perpendicular to the diamond grain boundaries, is turned on at the time of 150 s. Erratic dark currents from five different orders of magnitude are seen to be fully suppressed by this applied magnetic field.
Acknowledgements
6. Conclusions CVD diamond radiation sensors are a viable alternative to silicon PIN-diodes for the measurement of ionizing radiation in BABAR. The SVTRAD system must have the ability to monitor large amounts of accumulated dose and also respond quickly to any sudden increases in dose rate. The pCVD diamonds tested in BABAR have shown that they are able to fulfill these requirements. They have also been shown to have advantages over silicon PIN-diodes: namely their radiation hardness, their low leakage currents, and their insensitivity to ambient temperature changes. Based on nearly 2 years of operation in BABAR, we have demonstrated that pCVD diamond sensors can provide fast, reliable, and accurate measurements of the SVT’s radiation dose. In addition, we have found some interesting current effects in our diamond sensors. The slow component of current response to sudden changes in radiation level is found to contribute a negligibly small error in our applications. The erratic dark currents, although large enough to obscure signal currents, are always completely suppressed by the serendipitously oriented 1.5-T magnetic field inside BABAR. Due to the good performance of the pCVD diamond-based radiation sensors that we have tested, we now plan to install an additional 12 diamond sensors for radiation monitoring and protection in BABAR during the summer of 2005.
The authors would like to thank the RD42 collaboration for working with us, sharing information and lending us diamonds. We would also
ARTICLE IN PRESS 182
A.J. Edwards et al. / Nuclear Instruments and Methods in Physics Research A 552 (2005) 176–182
like to thank Element Six Ltd. for providing diamonds for our research. [6]
References [7] [1] [2] [3] [4]
B. Aubert, et al., Nucl. Instr. and Meth. B 479 (2002) 1. V. Re, et al., IEEE Trans. Nucl. Sci. NS-49 (2002) 3284. T.I. Meyer, et al., Int. J. Mod. Phys. A 16S1C (2001) 1084. M.H. Nazare, et al., Development of diamond tracking detectors for high luminosity experiments at the LHC, RD42 Proposal CERN/DRDC 94-21, 1994. [5] W. Adam, et al., Development of diamond tracking detectors for high luminosity experiments at the LHC,
[8] [9] [10] [11] [12] [13]
CERN/LHCC Status Reports 95-43, 97-3, 2000-015, 2002010. S. Zhao, Characterization of the electrical properties of polycrystalline diamond films, Dissertation, Ohio State University, 1994. D. Meier, CVD diamond sensors for particle detection and tracking, Dissertation, Universitat Heidelberg, 1999. C. Bauer, et al., Nucl. Instr. and Meth. A 367 (1995) 207. D. Meier, et al., Nucl. Instr. and Meth. A 426 (1999) 173. T. Behnke, et al., Nucl. Instr. and Meth. A 489 (2002) 230. R.J. Tapper, Rep. Prog. Phys. 63 (2000) 1273. V. Re, et al., Nucl. Instr. and Meth. A 518 (2004) 290. The pCVD diamonds were made by Element Six Ltd., King’s Ride Park, Ascot, Berkshire SL5 8BP, UK.