- Email: [email protected]
Diamond sensor arrays for neutron detection: Preamplifier signal dependence on sensor array configuration
Radiation Measurements 73 (2015) 18e25
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: www.elsevier.com/locate/radmeas
Diamond sensor arrays for neutron detection: Preamplifier signal dependence on sensor array configuration Eric Lukosi a, *, Mark Prelas b a b
Department of Nuclear Engineering, 315 Pasqua Nuclear Engineering, University of Tennessee, Knoxville, TN 37996, USA Nuclear Science and Engineering Institute, E2433 Lafferre Hall, University of Missouri, Columbia, MO 65211, USA
h i g h l i g h t s Computational study of diamond detector array designs on output signals. Series detector arrays are intrinsically limited due to sensor impedance. Parallel detector arrays increase detection efficiency and maintain signal integrity. Experimental verification of series and parallel diamond detector arrays.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 30 March 2012 Received in revised form 12 December 2014 Accepted 13 December 2014 Available online 16 December 2014
Experimental verification of simulated results of series and parallel diamond detector arrays is reported. Eight commercially available electronic grade single crystal CVD diamond plates were used in series and parallel array configurations and were characterized through alpha particle and neutron exposures. It was found that a series array of diamond detectors is inherently limited due to the impedances of the diamond plates. Plutoniumeberyllium neutron exposures were conducted with each of these diamond plates and on a parallel array of all eight diamond plates. It was found that the detection efficiency scaled linearly with the number of plates and that the pulse height from the preamplifier was not affected by the parallel array configuration. However, the additional capacitance introduced to the charge sensitive preamplifier indicates that there is a limitation to the total size that can be realized with a parallel diamond sensor array before significant signal degradation occurs. © 2014 Elsevier Ltd. All rights reserved.
Keywords: Diamond detectors Neutron detection Spectroscopy Signal summing Series array Parallel array Detection efficiency
1. Introduction CVD single crystal diamond has been reported in recent years as an emerging option for radiation detection and spectroscopy (Amosov et al., 2011; Isberg et al., 2004, 2002; Lukosi et al., 2013; Oh et al., 2013). One challenge in the use of diamond as a radiation sensor is the available substrate area. Historically, high quality diamond growth suitable for radiation detection requires a diamond substrate, thereby limiting the maximum achievable area. Further, optimum diamond growth for radiation detection is along the <100> direction, which is self-limiting due to the pyramidal formation it creates (Silva et al., 2008). Therefore, large area
* Corresponding author. E-mail address: [email protected] (E. Lukosi). http://dx.doi.org/10.1016/j.radmeas.2014.12.007 1350-4487/© 2014 Elsevier Ltd. All rights reserved.
substrates are required for reasonable detection efficiencies outside of beam tracking applications. Significant effort has been invested into epitaxial diamond growth to overcome this challenge (Regmi et al., 2012), but to-date the adaption of these techniques to produce high quality thick substrates for radiation detection and electronic applications has not yet been achieved. An alternative approach to overcoming the challenge of small diamond substrates is to combine them together in a detector array. Using signal summing, an array of diamond sensors in a series or parallel design may provide a larger detection efficiency without the cost of additional signal processing electronics for each substrate in the array. The evolution of the pulse height spectrum induced by 14.1 MeV neutrons as a function of sensor thickness and signal collection design has already been reported (Lukosi et al., 2012) and indicated that the energy resolution of the detection system was still below experimentally determined values for both
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
19
alpha particles and the 8.4 MeV 12C(n,a)9Be peak from 14.1 MeV neutron interactions (Pillon et al., 2008). In addition, this computational investigation indicated that this technique can achieve an intrinsic detection efficiency of approximately 25 percent with minimal secondary neutron interactions. The purpose of this paper is to report the experimental and computational investigation on the properties of signals generated from diamond detector arrays in series and parallel array configurations. 2. Diamond sample preparation Eight diamond samples were obtained from Element 6, further called here as electronic grade diamond samples (EGDS). These EGDS had quoted dimensions of 2 2 mm with nitrogen and boron concentrations less than five parts per billion and fifty parts per trillion, respectively. The thickness of each sample was measured using a micrometer and are given in Table 1. The EGDS were characterized via Raman and a single peak at 1332 cm1 was observed. The samples were also characterized with an S.I. Photonics Model 440 UV/Visible Spectrophotometer where light absorption did not take place below a photon wavelength above 227 nm, indicating a band gap of 5.47 eV with no perceivable absorption lines due to any energy levels within the band gap. These results were conclusive enough to indicate that the samples were single crystal and that impurity levels were below the detection threshold of the spectrophotometer used. The EGDS were chemically cleaned using a modified RCA-1 process followed by a chemical oxidation to increase surface resistivity. Each sample was then placed in an oxygen plasma to ensure complete surface oxygenation. Electrical contacts were of the broad face design, created through a combination of thermal evaporation and magnetron sputtering to produce a metallization pattern of Ti/Pt/Au (50 nm/25 nm/200 nm) followed by thermal annealing at 600 C for twenty minutes for contact adhesion. The IV characteristics of each sample determined good Ohmic behavior with a current not exceeding 5 nA for any sample at an absolute bias of 500 V, providing a resistance of no less than 100 GU for each plate. A picture of EGDS 1 is given in Fig. 1. 3. Experimental investigation of diamond detector arrays 3.1. Single detector characterization 3.1.1. Alpha particle exposures and charge carrier properties Alpha particle exposures were conducted on each diamond sample to characterize their charge collection properties through charge collection efficiency measurements utilizing Hecht's theory (Leroy and Rancoita, 2009). The pressure pin system shown in Fig. 2 provides a means to use the same test system for all EGDS for alpha particle exposures. The alpha source used was a type A-2 100 210Po source from Eckert and Ziegler. This alpha source is not electroplated, and therefore exhibits a broad alpha emission spectrum
Table 1 Thickness of all EGDS used in this work. EGDS
Thickness (mm)
1 2 3 4 5 6 7 8
521 511 495 495 467 498 470 511
Fig. 1. Picture of fully prepared EGDS ready for experiments.
peaking around 5.05 MeV, verified through a silicon detector in a Canberra Quad Alpha Spectrometer Model 7404, as shown in Fig. 3. Also shown is the calibration 239Pu alpha source from Eberline with an energy resolution of 2.96%. Unfortunately, the emission rate of the 239Pu calibration source was too low to effectively acquire high statistical worth alpha spectra with the detector housing shown in Fig. 3, so the broad energy spectrum 210Po source was used in all measurements. The detector housing base was placed approximately two millimeters away from the surface of the 210Po source during all measurements. The detector chain consisted of an ORTEC 142 PC preamplifier, Canberra model 2026 amplifier set to a gain of 10 and a shaping time of 0.5 ms, and a Canberra Multiport II multichannel analyzer buffer. All data was collected with Genie™ 2000. Hecht's theory describes the charge collection efficiency as a function of the mobility, m, the trapping time constant, t, the applied electric field, E, the detector thickness with respect to the charge collection distance, W, and the location of charge carrier creation with respect to the charge collection electrode (Leroy and Rancoita, 2009).
Qc ¼ Qo
m t E x me te E Wx 1 eme te E þ h h 1 e mh th E W W
(1)
Further, the detector thickness is equal to the charge collection distance at 63.2 percent charge collection efficiency, and the ratio of the detector thickness to the electric field at this collection efficiency is equal to the mobility-trapping time constant product (Berdermann et al., 2004; Schmid et al., 2004), as shown in Equation (2).
me=h te=h ¼
W E
(2)
To obtain the mobility-trapping time constant for each sample, a spectrum was collected on each EGDS with the 210Po source. For reference, spectra collected using EGDS 1 is given in Fig. 4. It can be seen that, as the bias increases, the spectrum shifts to higher channels numbers asymptotically and eventually approaches a constant value, which corresponds to maximum charge collection. The peak location was determined through statistical analysis over a Gaussian distribution fit due to the broad intrinsic alpha particle emission energy spectrum from the 210Po source used.
20
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 2. In part a) the diagram of the EGDS detector housing is displayed and in b) the actual detector housing is shown.
Assuming that the EGDS have a charge collection distance for electrons and holes much larger than 15 mm, approximately equal to the range of a 5.5 MeV alpha particle in diamond, it can be safely assumed that the charge collection properties determined through bias-varying techniques is a function of only one charge carrier. For instance, if the alpha particle enters the diamond through the cathode, then electrons must traverse the diamond thickness and is responsible for nearly the entire generated signal, such that the other term corresponding to hole transport and collection in equation one can be neglected. With this assumption, the mobilitytrapping time constant for the eight EGDS was determined and is provided in Table 2. These results indicate that the EGDS have charge collection properties an order of magnitude lower than that reported in Isberg et al. (2004), but are still of sufficient quality for radiation detection. 3.1.2. Neutron exposures Using the detector housing shown in Fig. 2, each EGDS was exposed to a plutoniumeberyllium (Pu/Be) source. To obtain the highest neutron flux in the diamond samples, the spatially
distributed neutron source was placed directly on top of the detector housing base, providing a nominal separation of 0.013 cm. A moderated 3He detector was placed near the source and diamond detector to calculate a relative fast-to-thermal neutron detection efficiency for the system, defined as the ratio of the total number of fast neutron counts seen by the diamond detector to the total number of thermal neutron counts registered by the 3He detector. Sample spectra collected from this experiment are provided in Fig. 5a. The relative detection efficiency of the Pu/Be neutron exposures on the single plate diamond detection system is given in Table 3. It can be seen that there are some differences seen for the relative detection efficiencies between diamond plates. This variation is attributed to the differences between diamond plate thicknesses and small variations in the electrode contact area between each EGDS. 3.2. Series array of diamond plates Although a thicker active diamond detection medium has no benefit for alpha particle detection, the effect of a series-type array of diamond detectors can provide insight into the expected response if exposed to neutrons. EGDS 1 and EGDS 2 were combined in a series array using the detector housing shown in Fig. 2. EGDS 1 was placed on the bottom and therefore was the substrate exposed to the alpha source. The signal was collected from the top electrode of the EGDS 2. The collected spectra is displayed in Fig. 6. Considering EGDS 1 has the lower charge carrier properties between the two samples, the results indicate that the required bias to obtain maximum charge collection was larger than expected by Table 2 Mobility-trapping time constant products for EGDS 1e8.
Fig. 3. Type A-2 100 mCi Po-210 source spectrum compared to an Eberline Pu-239 alpha particle source using a silicon detector.
EGDS
mete (106 cm2/V)
1 2 3 4 5 6 7 8
90 140 55 61 86 81 49.1 64
± ± ± ± ± ± ± ±
8 12 4 5 4 2 1.8 5
mhth (106 cm2/V) 42 65 51 40 54 30.4 29.8 35.4
± ± ± ± ± ± ± ±
6 11 6 6 5 1.1 2.6 1.7
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 4. The spectral results from
21
210
Po alpha exposures on EGDS 1 with a varying applied voltage is displayed in a) and one of these curve is shown in b).
about 100 V. Further, the maximum channel number reached was 77, a factor of 0.46 ± 0.04 lower than for the single plate exposure results of EGDS 1 and 2. Explanation of this result will be provided in Section 4.2.
change, thereby concluding that the parallel array of diamond plates does not reduce the output signal from the ORTEC 142 PC preamplifier. This result will be further investigated in Section 4.3. 4. Simulated response of diamond detectors
3.3. Parallel array of diamond plates 4.1. Simulation parameters and single plate response EGDS 1e8 were all mounted onto a ceramic chip carrier using a conductive silver epoxy, wire bonded in a parallel configuration, and mounted in an aluminum electronic enclosure insulated from the chamber wall. The surface of the diamond was approximately 1 cm from the top of the electronic enclosure and the cylindrical Pu/ Be source was placed directly over the location of the parallel diamond sensor array. The same single plate exposure operational conditions were used here. The result of this exposure to the parallel diamond sensor array is given in Fig. 5b. The relative efficiency found for the parallel array compared to the 3He detector was found to be 3.14 percent. Summing the data listed in Table 3 yields a total relative detection efficiency of 5.78 percent. Using MCNPX v2.7 (Pelowitz, 2011) to simulate the single plate and parallel array geometries, further described in Lukosi et al. (2012), it was found that the ratio of the total number of neutrons entering the single diamond plates to the parallel array was 51 ± 3 percent. Multiplying this to the summed relative detection efficiencies given in Table 3 yields a total efficiency of 2.95 ± 0.17 percent. This value agrees with the experimentally determined relative detection efficiency of the parallel sensor array of EGDS 1e8 of 3.14 percent. Further, upon inspection of the single plate and parallel plate differential pulse height spectra collected, it can be seen that the peak observed in the collected spectra does not
To quantify the observed effects of series and parallel arrays of diamond detectors exposed to ionizing radiation, the equivalent circuit of the diamond detectors coupled to the simplified block diagram of the ORTEC 142 PC charge sensitive preamplifier was investigated via OrCAD® Capture CIS simulations. The equivalent circuits used in these simulations are provided in Fig. 7. A diamond detector under bias is represented as an RC circuit, as shown in Fig. 7b. The equivalent circuit representation during a radiation interaction is shown in Fig. 7c, where the resistor is replaced by a current source. In Fig. 7d a series array of two diamond sensors is shown with one sensor represented instantaneously after a radiation interaction takes place and the other in a biased condition, equivalent to the experimental condition in Section 3.2. The characteristics of the current pulse was defined through a transient analysis with a large bandwidth oscilloscope. The area of the pulse is constant as long as near complete charge collection takes place and the width is defined through the drift velocity, which is a function of the applied bias. In the simulations conducted, we used a pulse width of 5 ns and a current peak of 13.35 mA, which corresponds to 5.5 MeV of energy deposition using a W-value of 13.2 eV/e-h pair (Balmer et al., 2009). The capacitance of the diamond sensor(s) in the equivalent circuits in Fig. 7aec is
Fig. 5. Spectral response of a) EGDS 7 and 8 and b) parallel array of EGDS 1e8 to a Pu/Be neutron source.
22
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 6. Measured 210Po alpha spectra for a series diamond detector array of EGDS 1 and 2 is shown in a) and the determined charge collection efficiency versus applied electric field is shown in b).
estimated at 5 pF, determined for a parallel plate design and the known permittivity of diamond. The resistance in the diamond plates were chosen to be 100 GU, determined from the currentevoltage analysis reported for the eight diamond samples in Section 2. Using these parameters, the output voltage from the charge sensitive preamplifier for a single diamond detector was determined to be 668 mV. 4.2. Series array of diamond plates Simulation of the circuit shown in Fig. 7d resulted in an output signal of 336 mV, which is 50 percent of the output from a single diamond detector. Comparing this result to the result obtained experimentally in Section 3.3, we find that the two results agree. The reason behind the 50 percent drop in output voltage from the preamplifier in both simulation and experiment is due to signal loading between the two diamond sensors in the series array. When current flows in the equivalent circuit of the diamond sensor, the current is integrated across its capacitance, which is brought back to equilibrium by current flow across the input capacitor of the preamplifier. However, when a second diamond detector (EGDS 2) is placed between the first diamond detector (EGDS 1) and the input to the preamplifier, the image charged developed on the electrode in low resistance contact with EGDS 2 must come from its back electrode. In addition, the resistance of each diamond detector is on the order of 100 GU, as verified through experimentation, resulting in an RC time constant of 0.5 s. Therefore, for two diamond detectors in a series array, the image charged formed between EGDS 1 and 2 from the interaction of ionizing radiation with EGDS 1 will be equalized by the resistive components of each diamond detector, thereby only allowing 50% of the generated charge to be registered by the charge sensitive preamplifier. Considering the equivalent circuit, the current supplied by the radiation interaction across the capacitance of EGDS 1 is split equally across each diamond sensor because the two detectors have equal capacitance and a large RC time constant, or impedance. This conclusion can be inferred from the simulation results shown in Fig. 8a, b. The current across EGDS 1 is a sharp spike followed by a decay to a non-zero current during the duration of the current pulse. This baseline current is 50 percent of the total current flowing in the system because the impedance of EGDS 1 and 2 are equal. The decay time to this baseline is a function of the detector time constant. The remaining current from the current source, or ionizing radiation event, is integrated across EGDS 2. Further, the characteristic decay time is transferred to EGDS 2 as the characteristic charging and de-charging time constant. At large detector array capacitance, the shape of the current pulse changes from a pseudo-square wave to an exponential-like function. The
current profile across EGDS 2 is the same as that across the input capacitor of the preamplifier, and therefore, the feedback capacitor. Finally, simulating a very small resistance for EGDS 2 (~1 kU or less), complete signal loading to the input capacitor of the preamplifier is restored. As the resistance increases, the time constant of EGDS 2 is on the time scale of the integration time on the preamplifier input capacitor, and for even higher resistance values, on the feedback RC circuit in the preamplifier. In this region of time constants for EGDS 2, the signal generated by the preamplifier has a non-ideal tail pulse shape and is reduced in absolute magnitude as the time constant increases. As the resistance of EGDS 2 increases further, its time constant becomes much larger than the amplifier output time constant, and the signal is reduced to 50 percent of the maximum and the ideal tail pulse is restored. Considering these results, it is straightforward to show that the output signal of the preamplifier is a function of the ratio of the two diamond detector capacitances when realistic resistances are used for the diamond detectors. Low resistance diamond detector typically do not have good charge carrier properties because of its highly boron doped nature and therefore, low hole mobility. Running OrCAD® Cadence CIS simulations on the various combination of capacitances with corresponding realistic resistance values using a parallel plate design, a surface plot of the output voltage from the charge sensitive preamplifier as a function of the two diamond detector capacitances was generated, shown in Fig. 8c. This plot indicates that the voltage output is maximized when the capacitance of EGDS 2 is much greater than the capacitance of EGDS 1, or when the impedance of EGDS 2 is much lower than EGDS 1. In addition, the realistic and relatively small small variations in the detector resistance as a function of capacitance in a parallel plate design had no effect on the resultant voltage output, but was entirely controlled by each detector capacitance. It is important to realize that this plot has the same form as the equivalent capacitance of the circuit with one important difference; the output voltage from the preamplifier is not one-half that of the maximum output voltage with minimal input capacitance from a Table 3 Relative detection efficiency of EGDS 1e8 in response to a Pu/Be neutron source versus a moderated 3He detector. EGDS
Relative efficiency (%)
1 2 3 4 5 6 7 8
0.60 0.79 0.84 0.76 0.74 0.70 0.73 0.62
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
single diamond detector when the two diamond detectors have large capacitance values. This important result comes from the fact that the dynamic input capacitance of the preamplifier is not infinite, and the charge integrated across the front detector must be shared with this parallel input capacitance (Spieler, 2003). This effect is quantitatively displayed in Fig. 8d for a single diamond plate with different capacitance values. These results can be extended to a series array of diamond detectors greater than two, where the ratio of the capacitances of each diamond detector will dictate the signal that will be transferred to the next sensor toward the input of the charge sensitive preamplifier. This provides an ultimate limit on the number of detectors that can be placed in the chain, since the capacitance of the diamond detectors can only be controlled by three orders of magnitude in the best case scenario. In addition, it is important to realize that a radiation-induced signal can be generated in either detector and the same rules apply. Specifically, signal optimization when the ionizing radiation interacts with one of the diamond detectors in a given series array results in a highly non-optimized output signal amplitude when the ionizing radiation interacts with any other diamond detector in the array. Therefore, the best case scenario for two diamond detectors in a series array for radiation detection using both diamond detectors is a reduced signal output by approximately a factor of two of a single diamond detector. If this reduction in pulse height can be tolerated, then the increased detection efficiency may be warranted.
4.3. Parallel array of diamond plates When the diamond detectors are connected in parallel, the issue of signal sharing in a series array is no longer present. The output
23
signal is only dependent on the effect of the detector system capacitance as seen by the charge sensitive preamplifier. The output voltage is a function of the integrated charge across the feedback capacitor. The amount of integrated charge, Qi, to generated charge, QG, is described by.
Qi ¼ QG
1 ; 1 þ Cdet=C
(3)
di
which shows that the input dynamic capacitance, Cdi, must be large to ensure maximum signal output. Further, as the detector capacitance, Cdet, to dynamic input capacitance, Cdi, ratio increases, the RC time constant of the system increases, thereby increasing the charge integration time across the feedback capacitor. This results in incomplete charge integration across the feedback capacitor due to current flow across the feedback resistor, yielding a reduced output voltage. However, if the detector capacitance does not change significantly during a measurement when the ratio Cdet/Cdi is not insignificant, then this reduced signal amplitude will still be linearly related to the energy deposition in the detection system. The relationship between the output voltage from the charge sensitive preamplifier and Cdet is provided in Fig. 8d. In Fig. 9, the current across the detectors of a 4 plate parallel diamond detector array is provided for detector capacitances of 7 pF, 5 pF, 3 pF, and 1 pF, all with a resistance of 100 GU. The current profile across all detector elements in the parallel array is determined by the total parallel detector array capacitance (16 pF) rather than any individual detector plate, and the current magnitude across each detector in the array is determined by the relative ratio of the detector element capacitance to the total detector capacitance. Also shown is the current across the input capacitor of the
Fig. 7. The simplified block diagram of the diamond detection system attached to the PC142 preamplifier in a) with b) a single diamond plate in bias, c) a diamond during a radiation interaction and d) two diamond plates in series where the diamond plate between the ground and the other diamond plate is under the influence of a radiation interaction.
24
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 8. a) Current integrated across the parallel capacitor of a diamond detector equivalent circuit undergoing a radiation interaction when in series with a second diamond detector placed between itself and the input to the preamplifier. b) The current seen across EGDS 2 in the configuration described in (a). c) The output voltage as a function of detector capacitance for a two detector series array. d) The output voltage seen from the equivalent circuit ORTEC 142 PC preamplifier as a function of input capacitance.
charge sensitive preamplifier, where the effect of the detector capacitance, or long RC time constant of the system, is apparent. The output voltage profile is also determined by Cdet, where the value of Cdet is determined by the sum of all detector input capacitances. This profile leads to a broadened preamplifier output voltage, as shown in Fig. 9b.
5. Conclusion Experimental and computational results indicate that series arrays of diamond sensors of equal resistance and capacitance are inherently limited by signal amplitude reduction from charge sharing across the detector capacitances. Each diamond plate that is
Fig. 9. a) Current profile across 4 diamond plates in parallel with detector capacitances of 7 pF, 5 pF, 3 pF, and 1 pF. Also shown is the current profile across the input capacitor to the preamplifier. b) Output voltage from the ORTEC 142 PC preamplifier for the system configuration described in (a).
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
added to a series array further reduces the collected charge by the preamplifier and also has the secondary effect of requiring a larger bias to keep near full charge collection in each diamond plate. Optimization of a particular array for a signal generated in one diamond detector in the series array results in non-optimized signal generation for radiation interactions from other diamond detectors in the array. This is important, since signal summing in a detector array is only advantageous when it has a benefit to the detection of the radiation of interest, such as in the detection of neutrons. In the case of small parallel arrays of diamond detectors, the pulse height spectrum observed produces the same preamplifier output response as a single diamond plate of equivalent capacitance. As detector array input capacitance (number of plates in the array) to the charge sensitive preamplifier increases, the amount of charge loaded onto the feedback capacitor decreases, thereby lowering the output signal amplitude. In actuality, this also increases the noise of the system, thereby lowering the energy threshold that can be detected. This indicates that parallel arrays of diamond detectors are limited in their ultimate benefit to increasing detection efficiency, but tens to perhaps a hundred diamond plates are required to reach this limitation. Thicker diamond plates will enable larger arrays because of the reduced capacitance, and for diamond samples with excellent charge carrier properties, large area, high resolution fast neutron detection is feasible for diamond detector elements up 1 cm thick. Acknowledgments This work was conducted through funding provided by the Nuclear Regulatory Commission, grant NRC-38-09-943. Appreciation is given to the staff and faculty of the Nuclear Science and Engineering Institute at the University of Missouri for their support and contributions to this work.
25
References Amosov, V.N., Meshaninov, S.A., Rodionov, N.B., Rodionov, R.N., 2011. Diam. Relat. Mater. 20, 1239e1242. Balmer, R.S., Brandon, J.R., Clewes, S.L., Dhillon, H.K., Dodson, J.M., Friel, I., Inglis, P.N., Madgwick, T.D., Markham, M.L., Mollart, T.P., Perkins, N., Scarsbrook, G.A., Twitchen, D.J., Whitehead, A.J., Wilman, J.J., Woollard, S.M., 2009. J. Phys. Condens. Matter 21. Berdermann, E., Coibanu, M., Connell, S.H., Costa, A.M.O.D.d., FernandezHernando, L.A.Oh, Sellschop, J.P.F., 2004. Charged particle detectors made of single crystal diamond. Phys. Stat. Sol. (a) 201, 2521e2528. Isberg, J., Hammersberg, J., Bernhoff, H., Twitchen, D.J., Whitehead, A.J., 2004. Diam. Relat. Mater. 13, 872e875. Isberg, J., Hammersberg, J., Johansson, E., Wikstrom, T., Twitchen, D.J., Whitehead, A.J., Coe, S.E., Scarsbrook, G.A., 2002. Science 297. Leroy, C., Rancoita, P.-G., 2009. Principles of Radiation Interaction in Matter and Detection, second ed. World Scientific, Singapore. Lukosi, E., Hale, H., Littell, J., Milburn, R., 2013. First Design of a Diamondbased Neutron Spectrometer for Cross Section Measurements, Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC). 2013 IEEE, pp. 1e5. Lukosi, E., Prelas, M., Palsmeier, J., 2012. Monte Carlo simulations of multiplexed electronic grade CVD diamond for neutron detection. Radiat. Meas. 47 (6), 417e425. Oh, A., Caylar, B., Pomorski, M., Wengler, T., 2013. A novel detector with graphitic electrodes in CVD diamond. Diam. Relat. Mater. 38, 9e13. Pelowitz, D.B., 2011. MCNPX User's Manual. Los Alamos National Laboratory. Pillon, M., Angelone, M., Aielli, G., Almaviva, S., Marinelli, M., Milani, E., Prestopino, G., Tucciarone, A., Verona, C., Verona-Rinati, G., 2008. Radiation tolerance of a high quality synthetic single crystal chemical vapor deposition diamond detector irradiated by 14.8 MeV neutrons. J. Appl. Phys. 104, 054513. Regmi, M., More, K., Eres, G., 2012. A narrow biasing window for high density diamond nucleation on Ir/YSZ/Si(100) using microwave plasma chemical vapor deposition. Diam. Relat. Mater. 23, 28e33. Schmid, G.J., Koch, J.A., Lerche, R.A., Moran, M.J., 2004. A neutron sensor based on single crystal CVD diamond. Nucl. Instrum. Methods Phys. Res. A 527, 554e561. Silva, F., Achard, J., Bonnin, X., Brinza, O., Michau, A., Secroun, A., Corte, K.D., Felton, S., Newton, M., Gicquel, A., 2008. Diam. Relat. Mater. 17. Spieler, H., 2003. Front-end electronics and signal processing. In: AIP Conference Proceedings, vol. 674, pp. 76e100.
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: www.elsevier.com/locate/radmeas
Diamond sensor arrays for neutron detection: Preamplifier signal dependence on sensor array configuration Eric Lukosi a, *, Mark Prelas b a b
Department of Nuclear Engineering, 315 Pasqua Nuclear Engineering, University of Tennessee, Knoxville, TN 37996, USA Nuclear Science and Engineering Institute, E2433 Lafferre Hall, University of Missouri, Columbia, MO 65211, USA
h i g h l i g h t s Computational study of diamond detector array designs on output signals. Series detector arrays are intrinsically limited due to sensor impedance. Parallel detector arrays increase detection efficiency and maintain signal integrity. Experimental verification of series and parallel diamond detector arrays.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 30 March 2012 Received in revised form 12 December 2014 Accepted 13 December 2014 Available online 16 December 2014
Experimental verification of simulated results of series and parallel diamond detector arrays is reported. Eight commercially available electronic grade single crystal CVD diamond plates were used in series and parallel array configurations and were characterized through alpha particle and neutron exposures. It was found that a series array of diamond detectors is inherently limited due to the impedances of the diamond plates. Plutoniumeberyllium neutron exposures were conducted with each of these diamond plates and on a parallel array of all eight diamond plates. It was found that the detection efficiency scaled linearly with the number of plates and that the pulse height from the preamplifier was not affected by the parallel array configuration. However, the additional capacitance introduced to the charge sensitive preamplifier indicates that there is a limitation to the total size that can be realized with a parallel diamond sensor array before significant signal degradation occurs. © 2014 Elsevier Ltd. All rights reserved.
Keywords: Diamond detectors Neutron detection Spectroscopy Signal summing Series array Parallel array Detection efficiency
1. Introduction CVD single crystal diamond has been reported in recent years as an emerging option for radiation detection and spectroscopy (Amosov et al., 2011; Isberg et al., 2004, 2002; Lukosi et al., 2013; Oh et al., 2013). One challenge in the use of diamond as a radiation sensor is the available substrate area. Historically, high quality diamond growth suitable for radiation detection requires a diamond substrate, thereby limiting the maximum achievable area. Further, optimum diamond growth for radiation detection is along the <100> direction, which is self-limiting due to the pyramidal formation it creates (Silva et al., 2008). Therefore, large area
* Corresponding author. E-mail address: [email protected] (E. Lukosi). http://dx.doi.org/10.1016/j.radmeas.2014.12.007 1350-4487/© 2014 Elsevier Ltd. All rights reserved.
substrates are required for reasonable detection efficiencies outside of beam tracking applications. Significant effort has been invested into epitaxial diamond growth to overcome this challenge (Regmi et al., 2012), but to-date the adaption of these techniques to produce high quality thick substrates for radiation detection and electronic applications has not yet been achieved. An alternative approach to overcoming the challenge of small diamond substrates is to combine them together in a detector array. Using signal summing, an array of diamond sensors in a series or parallel design may provide a larger detection efficiency without the cost of additional signal processing electronics for each substrate in the array. The evolution of the pulse height spectrum induced by 14.1 MeV neutrons as a function of sensor thickness and signal collection design has already been reported (Lukosi et al., 2012) and indicated that the energy resolution of the detection system was still below experimentally determined values for both
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
19
alpha particles and the 8.4 MeV 12C(n,a)9Be peak from 14.1 MeV neutron interactions (Pillon et al., 2008). In addition, this computational investigation indicated that this technique can achieve an intrinsic detection efficiency of approximately 25 percent with minimal secondary neutron interactions. The purpose of this paper is to report the experimental and computational investigation on the properties of signals generated from diamond detector arrays in series and parallel array configurations. 2. Diamond sample preparation Eight diamond samples were obtained from Element 6, further called here as electronic grade diamond samples (EGDS). These EGDS had quoted dimensions of 2 2 mm with nitrogen and boron concentrations less than five parts per billion and fifty parts per trillion, respectively. The thickness of each sample was measured using a micrometer and are given in Table 1. The EGDS were characterized via Raman and a single peak at 1332 cm1 was observed. The samples were also characterized with an S.I. Photonics Model 440 UV/Visible Spectrophotometer where light absorption did not take place below a photon wavelength above 227 nm, indicating a band gap of 5.47 eV with no perceivable absorption lines due to any energy levels within the band gap. These results were conclusive enough to indicate that the samples were single crystal and that impurity levels were below the detection threshold of the spectrophotometer used. The EGDS were chemically cleaned using a modified RCA-1 process followed by a chemical oxidation to increase surface resistivity. Each sample was then placed in an oxygen plasma to ensure complete surface oxygenation. Electrical contacts were of the broad face design, created through a combination of thermal evaporation and magnetron sputtering to produce a metallization pattern of Ti/Pt/Au (50 nm/25 nm/200 nm) followed by thermal annealing at 600 C for twenty minutes for contact adhesion. The IV characteristics of each sample determined good Ohmic behavior with a current not exceeding 5 nA for any sample at an absolute bias of 500 V, providing a resistance of no less than 100 GU for each plate. A picture of EGDS 1 is given in Fig. 1. 3. Experimental investigation of diamond detector arrays 3.1. Single detector characterization 3.1.1. Alpha particle exposures and charge carrier properties Alpha particle exposures were conducted on each diamond sample to characterize their charge collection properties through charge collection efficiency measurements utilizing Hecht's theory (Leroy and Rancoita, 2009). The pressure pin system shown in Fig. 2 provides a means to use the same test system for all EGDS for alpha particle exposures. The alpha source used was a type A-2 100 210Po source from Eckert and Ziegler. This alpha source is not electroplated, and therefore exhibits a broad alpha emission spectrum
Table 1 Thickness of all EGDS used in this work. EGDS
Thickness (mm)
1 2 3 4 5 6 7 8
521 511 495 495 467 498 470 511
Fig. 1. Picture of fully prepared EGDS ready for experiments.
peaking around 5.05 MeV, verified through a silicon detector in a Canberra Quad Alpha Spectrometer Model 7404, as shown in Fig. 3. Also shown is the calibration 239Pu alpha source from Eberline with an energy resolution of 2.96%. Unfortunately, the emission rate of the 239Pu calibration source was too low to effectively acquire high statistical worth alpha spectra with the detector housing shown in Fig. 3, so the broad energy spectrum 210Po source was used in all measurements. The detector housing base was placed approximately two millimeters away from the surface of the 210Po source during all measurements. The detector chain consisted of an ORTEC 142 PC preamplifier, Canberra model 2026 amplifier set to a gain of 10 and a shaping time of 0.5 ms, and a Canberra Multiport II multichannel analyzer buffer. All data was collected with Genie™ 2000. Hecht's theory describes the charge collection efficiency as a function of the mobility, m, the trapping time constant, t, the applied electric field, E, the detector thickness with respect to the charge collection distance, W, and the location of charge carrier creation with respect to the charge collection electrode (Leroy and Rancoita, 2009).
Qc ¼ Qo
m t E x me te E Wx 1 eme te E þ h h 1 e mh th E W W
(1)
Further, the detector thickness is equal to the charge collection distance at 63.2 percent charge collection efficiency, and the ratio of the detector thickness to the electric field at this collection efficiency is equal to the mobility-trapping time constant product (Berdermann et al., 2004; Schmid et al., 2004), as shown in Equation (2).
me=h te=h ¼
W E
(2)
To obtain the mobility-trapping time constant for each sample, a spectrum was collected on each EGDS with the 210Po source. For reference, spectra collected using EGDS 1 is given in Fig. 4. It can be seen that, as the bias increases, the spectrum shifts to higher channels numbers asymptotically and eventually approaches a constant value, which corresponds to maximum charge collection. The peak location was determined through statistical analysis over a Gaussian distribution fit due to the broad intrinsic alpha particle emission energy spectrum from the 210Po source used.
20
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 2. In part a) the diagram of the EGDS detector housing is displayed and in b) the actual detector housing is shown.
Assuming that the EGDS have a charge collection distance for electrons and holes much larger than 15 mm, approximately equal to the range of a 5.5 MeV alpha particle in diamond, it can be safely assumed that the charge collection properties determined through bias-varying techniques is a function of only one charge carrier. For instance, if the alpha particle enters the diamond through the cathode, then electrons must traverse the diamond thickness and is responsible for nearly the entire generated signal, such that the other term corresponding to hole transport and collection in equation one can be neglected. With this assumption, the mobilitytrapping time constant for the eight EGDS was determined and is provided in Table 2. These results indicate that the EGDS have charge collection properties an order of magnitude lower than that reported in Isberg et al. (2004), but are still of sufficient quality for radiation detection. 3.1.2. Neutron exposures Using the detector housing shown in Fig. 2, each EGDS was exposed to a plutoniumeberyllium (Pu/Be) source. To obtain the highest neutron flux in the diamond samples, the spatially
distributed neutron source was placed directly on top of the detector housing base, providing a nominal separation of 0.013 cm. A moderated 3He detector was placed near the source and diamond detector to calculate a relative fast-to-thermal neutron detection efficiency for the system, defined as the ratio of the total number of fast neutron counts seen by the diamond detector to the total number of thermal neutron counts registered by the 3He detector. Sample spectra collected from this experiment are provided in Fig. 5a. The relative detection efficiency of the Pu/Be neutron exposures on the single plate diamond detection system is given in Table 3. It can be seen that there are some differences seen for the relative detection efficiencies between diamond plates. This variation is attributed to the differences between diamond plate thicknesses and small variations in the electrode contact area between each EGDS. 3.2. Series array of diamond plates Although a thicker active diamond detection medium has no benefit for alpha particle detection, the effect of a series-type array of diamond detectors can provide insight into the expected response if exposed to neutrons. EGDS 1 and EGDS 2 were combined in a series array using the detector housing shown in Fig. 2. EGDS 1 was placed on the bottom and therefore was the substrate exposed to the alpha source. The signal was collected from the top electrode of the EGDS 2. The collected spectra is displayed in Fig. 6. Considering EGDS 1 has the lower charge carrier properties between the two samples, the results indicate that the required bias to obtain maximum charge collection was larger than expected by Table 2 Mobility-trapping time constant products for EGDS 1e8.
Fig. 3. Type A-2 100 mCi Po-210 source spectrum compared to an Eberline Pu-239 alpha particle source using a silicon detector.
EGDS
mete (106 cm2/V)
1 2 3 4 5 6 7 8
90 140 55 61 86 81 49.1 64
± ± ± ± ± ± ± ±
8 12 4 5 4 2 1.8 5
mhth (106 cm2/V) 42 65 51 40 54 30.4 29.8 35.4
± ± ± ± ± ± ± ±
6 11 6 6 5 1.1 2.6 1.7
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 4. The spectral results from
21
210
Po alpha exposures on EGDS 1 with a varying applied voltage is displayed in a) and one of these curve is shown in b).
about 100 V. Further, the maximum channel number reached was 77, a factor of 0.46 ± 0.04 lower than for the single plate exposure results of EGDS 1 and 2. Explanation of this result will be provided in Section 4.2.
change, thereby concluding that the parallel array of diamond plates does not reduce the output signal from the ORTEC 142 PC preamplifier. This result will be further investigated in Section 4.3. 4. Simulated response of diamond detectors
3.3. Parallel array of diamond plates 4.1. Simulation parameters and single plate response EGDS 1e8 were all mounted onto a ceramic chip carrier using a conductive silver epoxy, wire bonded in a parallel configuration, and mounted in an aluminum electronic enclosure insulated from the chamber wall. The surface of the diamond was approximately 1 cm from the top of the electronic enclosure and the cylindrical Pu/ Be source was placed directly over the location of the parallel diamond sensor array. The same single plate exposure operational conditions were used here. The result of this exposure to the parallel diamond sensor array is given in Fig. 5b. The relative efficiency found for the parallel array compared to the 3He detector was found to be 3.14 percent. Summing the data listed in Table 3 yields a total relative detection efficiency of 5.78 percent. Using MCNPX v2.7 (Pelowitz, 2011) to simulate the single plate and parallel array geometries, further described in Lukosi et al. (2012), it was found that the ratio of the total number of neutrons entering the single diamond plates to the parallel array was 51 ± 3 percent. Multiplying this to the summed relative detection efficiencies given in Table 3 yields a total efficiency of 2.95 ± 0.17 percent. This value agrees with the experimentally determined relative detection efficiency of the parallel sensor array of EGDS 1e8 of 3.14 percent. Further, upon inspection of the single plate and parallel plate differential pulse height spectra collected, it can be seen that the peak observed in the collected spectra does not
To quantify the observed effects of series and parallel arrays of diamond detectors exposed to ionizing radiation, the equivalent circuit of the diamond detectors coupled to the simplified block diagram of the ORTEC 142 PC charge sensitive preamplifier was investigated via OrCAD® Capture CIS simulations. The equivalent circuits used in these simulations are provided in Fig. 7. A diamond detector under bias is represented as an RC circuit, as shown in Fig. 7b. The equivalent circuit representation during a radiation interaction is shown in Fig. 7c, where the resistor is replaced by a current source. In Fig. 7d a series array of two diamond sensors is shown with one sensor represented instantaneously after a radiation interaction takes place and the other in a biased condition, equivalent to the experimental condition in Section 3.2. The characteristics of the current pulse was defined through a transient analysis with a large bandwidth oscilloscope. The area of the pulse is constant as long as near complete charge collection takes place and the width is defined through the drift velocity, which is a function of the applied bias. In the simulations conducted, we used a pulse width of 5 ns and a current peak of 13.35 mA, which corresponds to 5.5 MeV of energy deposition using a W-value of 13.2 eV/e-h pair (Balmer et al., 2009). The capacitance of the diamond sensor(s) in the equivalent circuits in Fig. 7aec is
Fig. 5. Spectral response of a) EGDS 7 and 8 and b) parallel array of EGDS 1e8 to a Pu/Be neutron source.
22
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 6. Measured 210Po alpha spectra for a series diamond detector array of EGDS 1 and 2 is shown in a) and the determined charge collection efficiency versus applied electric field is shown in b).
estimated at 5 pF, determined for a parallel plate design and the known permittivity of diamond. The resistance in the diamond plates were chosen to be 100 GU, determined from the currentevoltage analysis reported for the eight diamond samples in Section 2. Using these parameters, the output voltage from the charge sensitive preamplifier for a single diamond detector was determined to be 668 mV. 4.2. Series array of diamond plates Simulation of the circuit shown in Fig. 7d resulted in an output signal of 336 mV, which is 50 percent of the output from a single diamond detector. Comparing this result to the result obtained experimentally in Section 3.3, we find that the two results agree. The reason behind the 50 percent drop in output voltage from the preamplifier in both simulation and experiment is due to signal loading between the two diamond sensors in the series array. When current flows in the equivalent circuit of the diamond sensor, the current is integrated across its capacitance, which is brought back to equilibrium by current flow across the input capacitor of the preamplifier. However, when a second diamond detector (EGDS 2) is placed between the first diamond detector (EGDS 1) and the input to the preamplifier, the image charged developed on the electrode in low resistance contact with EGDS 2 must come from its back electrode. In addition, the resistance of each diamond detector is on the order of 100 GU, as verified through experimentation, resulting in an RC time constant of 0.5 s. Therefore, for two diamond detectors in a series array, the image charged formed between EGDS 1 and 2 from the interaction of ionizing radiation with EGDS 1 will be equalized by the resistive components of each diamond detector, thereby only allowing 50% of the generated charge to be registered by the charge sensitive preamplifier. Considering the equivalent circuit, the current supplied by the radiation interaction across the capacitance of EGDS 1 is split equally across each diamond sensor because the two detectors have equal capacitance and a large RC time constant, or impedance. This conclusion can be inferred from the simulation results shown in Fig. 8a, b. The current across EGDS 1 is a sharp spike followed by a decay to a non-zero current during the duration of the current pulse. This baseline current is 50 percent of the total current flowing in the system because the impedance of EGDS 1 and 2 are equal. The decay time to this baseline is a function of the detector time constant. The remaining current from the current source, or ionizing radiation event, is integrated across EGDS 2. Further, the characteristic decay time is transferred to EGDS 2 as the characteristic charging and de-charging time constant. At large detector array capacitance, the shape of the current pulse changes from a pseudo-square wave to an exponential-like function. The
current profile across EGDS 2 is the same as that across the input capacitor of the preamplifier, and therefore, the feedback capacitor. Finally, simulating a very small resistance for EGDS 2 (~1 kU or less), complete signal loading to the input capacitor of the preamplifier is restored. As the resistance increases, the time constant of EGDS 2 is on the time scale of the integration time on the preamplifier input capacitor, and for even higher resistance values, on the feedback RC circuit in the preamplifier. In this region of time constants for EGDS 2, the signal generated by the preamplifier has a non-ideal tail pulse shape and is reduced in absolute magnitude as the time constant increases. As the resistance of EGDS 2 increases further, its time constant becomes much larger than the amplifier output time constant, and the signal is reduced to 50 percent of the maximum and the ideal tail pulse is restored. Considering these results, it is straightforward to show that the output signal of the preamplifier is a function of the ratio of the two diamond detector capacitances when realistic resistances are used for the diamond detectors. Low resistance diamond detector typically do not have good charge carrier properties because of its highly boron doped nature and therefore, low hole mobility. Running OrCAD® Cadence CIS simulations on the various combination of capacitances with corresponding realistic resistance values using a parallel plate design, a surface plot of the output voltage from the charge sensitive preamplifier as a function of the two diamond detector capacitances was generated, shown in Fig. 8c. This plot indicates that the voltage output is maximized when the capacitance of EGDS 2 is much greater than the capacitance of EGDS 1, or when the impedance of EGDS 2 is much lower than EGDS 1. In addition, the realistic and relatively small small variations in the detector resistance as a function of capacitance in a parallel plate design had no effect on the resultant voltage output, but was entirely controlled by each detector capacitance. It is important to realize that this plot has the same form as the equivalent capacitance of the circuit with one important difference; the output voltage from the preamplifier is not one-half that of the maximum output voltage with minimal input capacitance from a Table 3 Relative detection efficiency of EGDS 1e8 in response to a Pu/Be neutron source versus a moderated 3He detector. EGDS
Relative efficiency (%)
1 2 3 4 5 6 7 8
0.60 0.79 0.84 0.76 0.74 0.70 0.73 0.62
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
single diamond detector when the two diamond detectors have large capacitance values. This important result comes from the fact that the dynamic input capacitance of the preamplifier is not infinite, and the charge integrated across the front detector must be shared with this parallel input capacitance (Spieler, 2003). This effect is quantitatively displayed in Fig. 8d for a single diamond plate with different capacitance values. These results can be extended to a series array of diamond detectors greater than two, where the ratio of the capacitances of each diamond detector will dictate the signal that will be transferred to the next sensor toward the input of the charge sensitive preamplifier. This provides an ultimate limit on the number of detectors that can be placed in the chain, since the capacitance of the diamond detectors can only be controlled by three orders of magnitude in the best case scenario. In addition, it is important to realize that a radiation-induced signal can be generated in either detector and the same rules apply. Specifically, signal optimization when the ionizing radiation interacts with one of the diamond detectors in a given series array results in a highly non-optimized output signal amplitude when the ionizing radiation interacts with any other diamond detector in the array. Therefore, the best case scenario for two diamond detectors in a series array for radiation detection using both diamond detectors is a reduced signal output by approximately a factor of two of a single diamond detector. If this reduction in pulse height can be tolerated, then the increased detection efficiency may be warranted.
4.3. Parallel array of diamond plates When the diamond detectors are connected in parallel, the issue of signal sharing in a series array is no longer present. The output
23
signal is only dependent on the effect of the detector system capacitance as seen by the charge sensitive preamplifier. The output voltage is a function of the integrated charge across the feedback capacitor. The amount of integrated charge, Qi, to generated charge, QG, is described by.
Qi ¼ QG
1 ; 1 þ Cdet=C
(3)
di
which shows that the input dynamic capacitance, Cdi, must be large to ensure maximum signal output. Further, as the detector capacitance, Cdet, to dynamic input capacitance, Cdi, ratio increases, the RC time constant of the system increases, thereby increasing the charge integration time across the feedback capacitor. This results in incomplete charge integration across the feedback capacitor due to current flow across the feedback resistor, yielding a reduced output voltage. However, if the detector capacitance does not change significantly during a measurement when the ratio Cdet/Cdi is not insignificant, then this reduced signal amplitude will still be linearly related to the energy deposition in the detection system. The relationship between the output voltage from the charge sensitive preamplifier and Cdet is provided in Fig. 8d. In Fig. 9, the current across the detectors of a 4 plate parallel diamond detector array is provided for detector capacitances of 7 pF, 5 pF, 3 pF, and 1 pF, all with a resistance of 100 GU. The current profile across all detector elements in the parallel array is determined by the total parallel detector array capacitance (16 pF) rather than any individual detector plate, and the current magnitude across each detector in the array is determined by the relative ratio of the detector element capacitance to the total detector capacitance. Also shown is the current across the input capacitor of the
Fig. 7. The simplified block diagram of the diamond detection system attached to the PC142 preamplifier in a) with b) a single diamond plate in bias, c) a diamond during a radiation interaction and d) two diamond plates in series where the diamond plate between the ground and the other diamond plate is under the influence of a radiation interaction.
24
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
Fig. 8. a) Current integrated across the parallel capacitor of a diamond detector equivalent circuit undergoing a radiation interaction when in series with a second diamond detector placed between itself and the input to the preamplifier. b) The current seen across EGDS 2 in the configuration described in (a). c) The output voltage as a function of detector capacitance for a two detector series array. d) The output voltage seen from the equivalent circuit ORTEC 142 PC preamplifier as a function of input capacitance.
charge sensitive preamplifier, where the effect of the detector capacitance, or long RC time constant of the system, is apparent. The output voltage profile is also determined by Cdet, where the value of Cdet is determined by the sum of all detector input capacitances. This profile leads to a broadened preamplifier output voltage, as shown in Fig. 9b.
5. Conclusion Experimental and computational results indicate that series arrays of diamond sensors of equal resistance and capacitance are inherently limited by signal amplitude reduction from charge sharing across the detector capacitances. Each diamond plate that is
Fig. 9. a) Current profile across 4 diamond plates in parallel with detector capacitances of 7 pF, 5 pF, 3 pF, and 1 pF. Also shown is the current profile across the input capacitor to the preamplifier. b) Output voltage from the ORTEC 142 PC preamplifier for the system configuration described in (a).
E. Lukosi, M. Prelas / Radiation Measurements 73 (2015) 18e25
added to a series array further reduces the collected charge by the preamplifier and also has the secondary effect of requiring a larger bias to keep near full charge collection in each diamond plate. Optimization of a particular array for a signal generated in one diamond detector in the series array results in non-optimized signal generation for radiation interactions from other diamond detectors in the array. This is important, since signal summing in a detector array is only advantageous when it has a benefit to the detection of the radiation of interest, such as in the detection of neutrons. In the case of small parallel arrays of diamond detectors, the pulse height spectrum observed produces the same preamplifier output response as a single diamond plate of equivalent capacitance. As detector array input capacitance (number of plates in the array) to the charge sensitive preamplifier increases, the amount of charge loaded onto the feedback capacitor decreases, thereby lowering the output signal amplitude. In actuality, this also increases the noise of the system, thereby lowering the energy threshold that can be detected. This indicates that parallel arrays of diamond detectors are limited in their ultimate benefit to increasing detection efficiency, but tens to perhaps a hundred diamond plates are required to reach this limitation. Thicker diamond plates will enable larger arrays because of the reduced capacitance, and for diamond samples with excellent charge carrier properties, large area, high resolution fast neutron detection is feasible for diamond detector elements up 1 cm thick. Acknowledgments This work was conducted through funding provided by the Nuclear Regulatory Commission, grant NRC-38-09-943. Appreciation is given to the staff and faculty of the Nuclear Science and Engineering Institute at the University of Missouri for their support and contributions to this work.
25
References Amosov, V.N., Meshaninov, S.A., Rodionov, N.B., Rodionov, R.N., 2011. Diam. Relat. Mater. 20, 1239e1242. Balmer, R.S., Brandon, J.R., Clewes, S.L., Dhillon, H.K., Dodson, J.M., Friel, I., Inglis, P.N., Madgwick, T.D., Markham, M.L., Mollart, T.P., Perkins, N., Scarsbrook, G.A., Twitchen, D.J., Whitehead, A.J., Wilman, J.J., Woollard, S.M., 2009. J. Phys. Condens. Matter 21. Berdermann, E., Coibanu, M., Connell, S.H., Costa, A.M.O.D.d., FernandezHernando, L.A.Oh, Sellschop, J.P.F., 2004. Charged particle detectors made of single crystal diamond. Phys. Stat. Sol. (a) 201, 2521e2528. Isberg, J., Hammersberg, J., Bernhoff, H., Twitchen, D.J., Whitehead, A.J., 2004. Diam. Relat. Mater. 13, 872e875. Isberg, J., Hammersberg, J., Johansson, E., Wikstrom, T., Twitchen, D.J., Whitehead, A.J., Coe, S.E., Scarsbrook, G.A., 2002. Science 297. Leroy, C., Rancoita, P.-G., 2009. Principles of Radiation Interaction in Matter and Detection, second ed. World Scientific, Singapore. Lukosi, E., Hale, H., Littell, J., Milburn, R., 2013. First Design of a Diamondbased Neutron Spectrometer for Cross Section Measurements, Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC). 2013 IEEE, pp. 1e5. Lukosi, E., Prelas, M., Palsmeier, J., 2012. Monte Carlo simulations of multiplexed electronic grade CVD diamond for neutron detection. Radiat. Meas. 47 (6), 417e425. Oh, A., Caylar, B., Pomorski, M., Wengler, T., 2013. A novel detector with graphitic electrodes in CVD diamond. Diam. Relat. Mater. 38, 9e13. Pelowitz, D.B., 2011. MCNPX User's Manual. Los Alamos National Laboratory. Pillon, M., Angelone, M., Aielli, G., Almaviva, S., Marinelli, M., Milani, E., Prestopino, G., Tucciarone, A., Verona, C., Verona-Rinati, G., 2008. Radiation tolerance of a high quality synthetic single crystal chemical vapor deposition diamond detector irradiated by 14.8 MeV neutrons. J. Appl. Phys. 104, 054513. Regmi, M., More, K., Eres, G., 2012. A narrow biasing window for high density diamond nucleation on Ir/YSZ/Si(100) using microwave plasma chemical vapor deposition. Diam. Relat. Mater. 23, 28e33. Schmid, G.J., Koch, J.A., Lerche, R.A., Moran, M.J., 2004. A neutron sensor based on single crystal CVD diamond. Nucl. Instrum. Methods Phys. Res. A 527, 554e561. Silva, F., Achard, J., Bonnin, X., Brinza, O., Michau, A., Secroun, A., Corte, K.D., Felton, S., Newton, M., Gicquel, A., 2008. Diam. Relat. Mater. 17. Spieler, H., 2003. Front-end electronics and signal processing. In: AIP Conference Proceedings, vol. 674, pp. 76e100.