ICONE DEVELOPMENT OF COATED GALLIUM ARSENIDE NEUTRON DETECTORS

Proceedings of ICONE 8 8 th International Conference on Nuclear Engineering April 2-6, 2000, Baltimore, MD USA ICONE - 8110 DEVELOPMENT OF COATED GALLIUM ARSENIDE NEUTRON DETECTORS Dr. Raymond T. Klann Argonne National Laboratory 9700 South Cass Ave. Argonne, IL 60439 630-252-4305 630-252-1885 (fax) klann@anl.gov Dr. Douglas S. McGregor University of Michigan 2355 Bonisteel Blvd. Ann Arbor, MI 48109 734-647-8964 dsmcgreg@engin.umich.edu KEYWORDS Gallium Arsenide, Neutron Detection, Semi-conductor ABSTRACT Coated-gallium arsenide detectors are being investigated for use as neutron counters in high radiation environments. First generation detectors have been fabricated and have been tested. This paper discusses the technique used for fabricating devices coated with boron-10 and polyethylene. Initial results from functional tests are also described and indicate that the devices are operational and can detect neutrons. INTRODUCTION Novel neutron detectors are currently being developed that use gallium arsenide coated with selected materials. The coatings serve to convert the incident neutrons into charged particles, which are detected by charge production in the active region of the gallium arsenide wafer. Diodes on the surface of the wafer are used to collect the induced charge. The detectors are rugged and offer many advantages over existing detectors including positional information, directional dependence, gamma discrimination, radiation hardness, and spectral tailoring. In addition, the detectors provide much better spatial resolution than 3 He, fission chambers, and other existing detectors, and much faster timing than existing position sensitive recording devices - such as track-etch and activation foil methods. Primary applications of these detectors in the nuclear industry include neutron radiography, neutron detection in high radiation fields, and radioactive material examination and characterization. THEORY OF OPERATION A unique effect discovered with semi-insulating (SI) bulk GaAs reverse biased diodes, when the bulk GaAs is compensated with the intrinsic deep level defect EL2, is the unusual truncated electric field distribution [1-3]. The electric field is divided into two distinct regions; a high field region and a low to zero field region [4]. The high field region (or active region) is almost constant in magnitude and has been shown to increase linearly with applied electric field, requiring an average of 1-2 volts per :m of active region. The truncated field effect is fortuitous for many reasons when applied to charged particle and neutron detection. First, the required operating voltage is reduced since the active region need be only as wide as the maximum range of the charged particles to be measured. Generally, the required active region width is only a few tens of micrometers. Second, the low electric field region is inactive, hence background gamma ray noise interactions in the active region are significantly reduced. In other words, the devices can self-discriminate between gamma rays and more energetic charged particles and ions. Third, the substrate thickness determines the device capacitance, not the active region thickness. Hence, the actual 1 Copyright 2000 by ASME

device electronic noise can be kept low while minimizing the active region thickness. The basic device structure is a Schottky barrier diode, in which a cleaned and etched GaAs surface has been overlaid with a vacuum evaporated metal. In the present case, the metals consist of a layered structure of Ti, Pt and Au. GaAs is a III-V compound semiconductor, in which many energy states appear in the band gap at the surface. These very same surface states predetermine the formation and barrier height of the Schottky barrier formed at the metal/semiconductor interface [5]. The Schottky barrier provides a rectifying junction, in which the device formed is generally referred to as a Schottky diode. Hence, the simple formation of a metal onto a cleaned and etched GaAs surface will produce a reasonable rectifying diode that can be used for charged particle detection. Furthermore, the metal contact and barrier can be made very thin (on the order of only a few hundred angstroms), thereby attenuating only a small fraction of energy from charged particles entering the detector. It is important to note that the surface state density predetermines the Schottky barrier height, and therefore also predetermines the leakage current and maximum reverse bias voltage. Hence, significant alterations in the surface state density (> 10 16 /cm 3 ) must occur before the device begins to degrade. The effect provides a natural form of radiation hardening, in which very little device degradation is observed over a wide range of gamma ray, charged particle, and neutron fluences [6, 7]. For instance, 10 B coated GaAs detectors have been irradiated with 10 13 thermal neutrons/cm 2 without any noticeable change in the device operating characteristics [8]. The basic GaAs device structure consists of a semiinsulating bulk GaAs substrate which has been polished on both sides. The blocking Schottky contact consists of a series of metals - Ti, Pt and Au. A low resistivity contact (sometimes referred to as an ohmic contact ) is fabricated on the opposite side. Generally, n-type low resistivity contacts are used, consisting of a Au/Ge based eutectic alloy [9] or a multiple level Pd/Ge/Ti/Au contact [10,11]. A neutron reactive film is either evaporated or attached to the Schottky contact. The device is reversed biased and charged particles emanating from the neutron reactive film that enter the GaAs Schottky diode will excite electron-hole pairs. The electrons and holes are separated by the applied electric field in the active region, thereby producing a measurable pulse in the external output device circuitry. COATING STUDIES The coating placed on the Schottky diode of the GaAs substrate will determine the response of the device to neutrons. Boron coated detectors are sensitive to the 7 Li and α-particle reaction products from the 10 B(n,α) 7 Li reaction. High density polyethlyene (HDP) coated detectors are sensitive to recoil protons from elastic scattering from the hydrogen. Based on the material selected, the energy response of the detector will vary according to the cross-section of the interaction of interest. This could be boron-10 which has a high alpha production cross-section for thermal neutrons but a much lower cross-section for epithermal or fast neutrons. Alternatively, hydrogen can be used to produce protons from elastic scattering. The scattering cross-section for hydrogen is larger than the (n,α) cross-section of boron-10 above about 1 kev, but it is much lower for energies below 1 kev. Any material which produces a charged particle is a potential candidate for a coating. This includes any material that is an alpha emitter, proton emitter, ion emitter, and fissionable material. In addition, other options such as scintillators and gamma emitters could also be considered, however, these types of coatings are not feasible for detectors used in high gamma fields. The background radiations would produce as much or more of a signal in the detector than the coating itself, and, therefore would not be a reliable detector for neutrons. This is the main reason why gadolinium screens cannot be used to radiograph spent nuclear fuel or other highly radioactive samples. Because the cross-sections of the numerous potential coating materials vary greatly as a function of the neutron energy from a 1/v response, to a flat response, to a response that has well defined resonances, the intended application of the detector is crucial in the selection of a coating. But this is not the only consideration. The charged particle (or particles) and its energy are also going to effect the efficiency of the detector. If the only factor was the interaction probability of the neutron, then no matter what coating is used, the thickness could be increased until the desired efficiency of interaction is obtained. However, there is a practical limitation based on the range of the charged particle in the coating. Once the coating thickness equals the maximum range of the charged particle, the efficiency of the detector is maximized. Beyond this point, the efficiency will slowly decrease because neutrons are being absorbed without any energy being deposited in the detector. So this maximum range of the charged particle coupled with the neutron cross-section will set an absolute limit on the intrinsic efficiency of the detector. These factors were considered in determining the most efficient detectors for each energy range. For fast neutrons (neutrons above 500 kev) the best choice is a hydrogen-rich coating such as polyethylene. This allows energetic protons to be produced directly from scattering. The proton energy is dependent on energy which is discussed in more detail later. At thermal energies, boron-10 is obviously the best choice because it produces alpha particles and lithium ions with energies well above low-level discriminator settings. This allows good discrimination against the gamma background. Fissionable materials, such as U-235 and Pu-239, can be useful in the energy range from a few ev up to about 1 kev. For example, Pu-239 can offer an improvement of a factor of 10 in efficiency over boron- 10 at 0.3 ev. Most of the other materials reviewed, were quickly discarded because of several reasons. Most did not produce significant charged particle emissions to be of use. Others that 2 Copyright 2000 by ASME

did produce protons or alpha particles had significantly lower cross-sections than boron-10, hydrogen, and fission deposits that there didn t appear to be a useful energy range for these materials. Actually, these materials could be of benefit if one wanted a specific detector response without regard to overall detector efficiency. The existing study was focused on maximizing the detector efficiency for fast neutrons above 1 MeV. A simple code was written to look at the behavior of the coating and proton energy loss in the coating thickness and in the active region of the detector. Data for the energy loss in polyethlyene and gallium was taken from ICRU Report 49 [12]. The first item that was readily apparent is that the cross-section for scattering is so low that the addition of thin layers of polyethylene on the detector did not significantly degrade the beam. The intrinsic efficiency was calculated as a function of neutron energy for different coating thicknesses. The results are shown in Figure 1 along with the estimate of the maximum efficiency. It is observed that there it is not possible to improve the efficiency at any energy (up to 14 MeV) by making the coating thicker than 2200 :m. This is reasonable since it corresponds to the maximum range of a 14 MeV proton in polyethlyene. The protons cannot reach the active region of the detector from interactions in the coating that occur more than 2200 :m away. The other immediately observable phenomenon is that the spectral response of the detector can be tailored by varying the coating thickness. For any given neutron energy, the detector efficiency is maximized for a coating thickness that matches the maximum range of a proton with the same energy. This would occur for a scattering event in which the proton was ejected at a zero degree angle, or directly forward. Intrinsic Efficiency 0.12% 0.10% 0.08% 0.06% 0.04% 0.02% 0.00% 2200 um 1200 um 500 um 200 um 100 um 0 2 4 6 8 10 12 14 Neutron Energy (MeV) Figure 1: Calculated efficiency of polyethylene coated GaAs Schottky barrier detectors as a function of the incident neutron energy for different coating thicknesses. Directional dependence of the HDP coated detectors is readily apparent from the kinematics of elastic scattering from hydrogen. No recoil protons can be scattered in the backward direction. In addition, the proton energy is a strong function of the angle of scattering from full energy in the forward direction to zero energy 90 degrees from the incident neutron. Figure 2 shows the energy of the proton entering the GaAs wafer (from a 14 MeV neutron) as a function of the scattering angle for interactions at different depths in the coating. If the detector is placed in a neutron beam so the coating is behind the GaAs wafer, i.e. backwards, no detector response is expected. Proton Energy (MeV) 14 12 10 8 6 4 2 0 2050 um 1800 um 0 10 20 30 40 50 60 70 80 90 Proton Angle (degrees) 1800 um 2050 um Figure 2: Energy of the proton entering the active region of the GaAs Schottky barrier detector as a function of the scattering angle of the proton. The proton energy is shown for different interaction depths within the polyethlyene coating. (0.5 :m is closest to the active region.) Figure 3 shows the amount of energy deposited in the high field region of the detector (assuming a high field region of 10 :m and 14 MeV neutrons) as a function of the scattering angle for interactions at different depths in the coating. The same depths are shown in Figure 3 that are shown in Figure 2. As the interaction depth increases the solid angle of the emitted protons becomes smaller. This means that as additional coating is added, there is less benefit from doing so, i.e. the effect is not linear. In addition, the directional dependence is exhibited as a function of thickness. The limiting angle of the proton is evidenced by the sharp drop off in energy deposited in the detector. The peak corresponds to the angle at which the full energy of the proton is deposited in the activer region of the detector. It is also evident that as the incident neutron angle deviates from normal that the proton solid angle will decrease and hence a smaller detector response. For example, for 14 3 Copyright 2000 by ASME

MeV neutrons causing an interaction at a depth of 563 :m, the detector response will be cut in half if the incident angle is Energy Deposited (MeV) 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 2050 um 1800 um 0 10 20 30 40 50 60 70 80 90 Incident Proton Angle (degrees) 1800 um 2050 um approximately 75 degrees and will be reduced to zero as the incident angle approaches 90 degrees. Figure 3: Energy deposited in the active region of the GaAs Schottky barrier detector as a function of the incident proton angle. The energy deposition is shown for different interaction depths within the polyethlyene coating. (0.5 :m is closest to the active region.) The active region of the detector is 10 :m. DETECTOR FABRICATION Schottky barrier bulk GaAs diodes were fabricated with either 10 B coatings or high-density polyethylene (HDP) coatings. The GaAs diodes were manufactured in the following manner. Commercial bulk semi-insulating (SI) GaAs wafers were used for the devices. The back surfaces were lapped at 30 rpm with 3 :m calcined aluminum oxide powder/ de-ionized water solution over an optically flat glass plate until 100 microns of GaAs material was removed. Afterwards, the backsides were polished with 0.3 :m calcined aluminum oxide powder mixed in a sodium hypochlorite solution over a chemically resistant polishing cloth at 65 rpm for 10 minutes. A final polish was performed with a 50:50:1 methyl alcohol:glycerol:bromine solution for 10 minutes over a chemically resistant polishing pad at 70 rpm. The wafers were cleaned in a series of solvents and etched in a 1:1:320 H 2 SO 4 :H 2 O 2 :de-ionized water solution for 5 minutes followed by a 2 minute etch in a 1:1 HCl: de-ionized water solution. The wafers were then cleaned in a de-ionized water cascade and blown dry with N 2. The backsides were implanted at an angle of 7 o from normal with 29 Si ions at an average energy of 100 kev at a dose amounting to 5 x 10 13 ions per cm 2. The implants were activated with a rapid thermal anneal in Ar for 30 seconds at 800 o C. Afterwards, a stacked layer of Ge (500 Å): Pd (1300 Å) was vacuum evaporated over the backsides, followed by a low temperature anneal of 250 o C in N 2 for 30 minutes. Vacuum evaporation of a stacked layer of Ti (150 Å): Au (700 Å) completed backside processing of the devices. Front side processing of the devices included lapping and polishing of the samples, in which the initial lapping with 3 :m calcined aluminum oxide powder was used to thin the wafers to 250 :m total thickness. Afterwards, the wafers were polished using the 0.3 :m calcined aluminum oxide powder mixed in a sodium hypochlorite solution followed by the methyl alcohol:glycerol:bromine solution. Again, the wafers were cleaned in a series of solvents and etched in a 1:1:320 H 2 SO 4 :H 2 O 2 :de-ionized water solution for 5 minutes followed by a 2 minutes etch in a 1:1 HCl:de-ionized water solution. Afterwards, the wafers were cleaned in a de-ionized water cascade and blown dry with N 2. Four basic pad area designs were employed for the first batch sets, those being 5mm x 5mm squares, 10mm x 10mm squares, 5.64 mm diameter circles, and 11.12 mm diameter circles. The designs were patterned onto the surfaces with photoresist. A final etch in the circular patterns was performed with the H 2 SO 4 : H 2 O 2 :de-ionized water solution followed by the HCl:deionized water solution. The wafers were washed in a de-ionized water cascade and blown dry with N 2. A stacked layer of Ti:Au was evaporated over the wafer and lifted off in acetone. Other variations used a system of Ti:Pt:Au contacts. Boron coated devices underwent the following steps. Photoresist was used to define patterns that covered and were centered over the Schottky contacts. The wafers were fastened to a chilled plate in a vacuum evaporator and 98% 10 B was vacuum evaporated over the Schottky contacts. Samples received different 10 B deposition thicknesses, being 1 :m and 2.1 :m. After evaporation, the residue was lifted off in acetone, leaving behind a layer of 10 B fastened in intimate contact with the front Schottky contact. A light film of Humiseal was applied to the detector surfaces, amounting to approximately 75 :m of material. Polyethylene-coated devices were manufactured by adhering various thicknesses of (HDP) sheets to the bare Schottky contacts. The HDP sample thicknesses included Humiseal only, 50 :m (0.002 inches), 125 :m (0.005 inches), 250 :m (0.010 inches), 450 :m (0.018 inches), 900 :m (0.035 inches), and 2030 :m (0.080 inches). The individual devices, both HDP coated and 10 B coated, were cleaved from the GaAs wafers, and fastened with silver-based epoxy to 1 mm thick aluminum oxide mounts. INITIAL TESTING The boron coated GaAs Schottky barrier detectors were mounted in light impenetrable Al boxes, in which the wall thickness of the Al boxes was only 1 mm. Aluminum has a very small total microscopic thermal neutron cross section, provides excellent radio frequency (RF) shielding and eliminates photoelectric noise from room lights. The enclosed devices were placed into a collimated and doubly diffracted thermalized neutron beam from the Ford Nuclear Reactor (FNR), a materials test reactor at the University of Michigan. The devices were 4 Copyright 2000 by ASME

tested at reverse bias voltages ranging from 20 to 170 volts. 1000000 100000 Batch 1-25 mm^2 area - bias=-100v Batch 2-100 mm^2 area - bias=-160v 1000 100 shield open, shutter open, bias=-130v shield open, shutter open, bias=-80v Counts 10000 1000 Counts/sec 10 1 100 0.1 10 0 25 50 75 100 125 150 175 200 Channel Number Figure 4 shows the results from various detector batches. Figure 4: Results from Ford Neutron Reactor. GaAs Schottky barrier detectors coated with one :m of 10 B. The small area detectors (5mm x 5mm squares and 5.64 mm diameter circles) clearly showed the best results, as indicated in the figure as Batch 1. The lower noise and cleaner signal for batch 1 detectors over batch 2 detectors are most likely a result from significantly lower capacitance than the larger area devices. All of the detectors demonstrated sensitivity to thermal neutrons, as evidenced by manipulating the neutron beam with various attenuators. For instance, blocking the beam with 1 inch thick HDP almost comp letely eliminated the detector signal, however, blocking the beam with 2 inches of lead hardly changed the signal at all. Since neutrons penetrate lead easily, but scatter in HDP, with the opposite being true for photons, it is clear that the devices are detecting neutrons. Further evidence is provided by the fact that the two main alpha particle reaction energies from the 10 B(n,") 7 Li are clearly visible in the batch 1 detector spectra. A small area GaAs Schottky barrier detector with 2030 :m HDP was mounted in a light impenetrable Al box. The enclosed device was placed in a well collimated direct viewing neutron beam from the Neutron Radiography Reactor (NRAD) at Argonne National Laboratory. This beam is much harder than most reactor beams as it looks directly at the core of NRAD. Because of this direct viewing, there is a strong component of neutrons at fission energies [13]. The detector was tested at reverse bias voltages between 80 and 130 volts. Figure 5 shows the results. The range of the recoil protons exceeds the high field region of the detector as discussed previously. By increasing the bias on the detector, the high field region is increased, leading to higher energy deposition and a larger pulse from the detector. It also allows additional higher energy events to be recorded, because the transmission length is increased. This causes the slope of the response curve to decrease. 0.01 0 200 400 600 800 1000 1200 Figure 5: Results in NRAD North Beam. GaAs Schottky barrier detector (area = 25 mm 2 ) coated with 2030 :m of high-density polyethylene. CONCLUSIONS Gallium-arsenide Schottky barrier detectors have been fabricated with boron-10 and high density polyethlyene coatings. Initial tests confirm that the detectors are functional and are detecting neutrons in a reactor beam. Testing of the devices is currently underway to determine the consistency of the fabrication process, the detector efficiencies, functioning of the HDP coated devices, the radiation hardness, and directional dependence of the devices. ACKNOWLEDGMENTS The submitted manuscript has been authored by a contractor of the U.S. Government under contract No. W-31-109- ENG-38. Accordingly, the U.S. Government retains a nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for U.S. Government purposes. REFERENCES Channel [1] D.S. McGregor, G.F. Knoll, Y. Eisen and R. Brake, IEEE Trans. Nucl. Sci, NS-39 (1992) p. 1226. [2] K. Berwick, M.R. Brozel, C.M. Buttar, M. Cowperthwaite andy. Hou, Inst. Phys. Conf. Series, 135 (1993) p. 305. [3] D.S. McGregor, R.A. Rojeski, G.F. Knoll, F.L. Terry, Jr., J. East and Y. Eisen, Nucl. Instr. and Meth., A343 (1994) p. 527. [4] D.S. McGregor and J.E. Kammeraad, Semiconductors for Room Temperature Nuclear Detector Applications, in Semiconductors and Semimetals, Vol. 43 Chap. 10 (Academic Press, San Diego, 1995) p. 383. 5 Copyright 2000 by ASME

[5] S.M. Sze, Physics of Semiconductor Devices, 2nd. Ed., (Wiley, New York, 1981). [6] G.C. Messenger and M.S. Ash, The Effects of Radiation on Electronic Systems, 2nd Ed., (Van Nostrand Reinhold, New York, 1992). [7] S.P. Beaumont et al., IEEE Trans. Nucl. Sci., NS-40 (1993) p. 1225. [8] D.S. McGregor, J.T. Lindsay, C.C. Brannon and R.W. Olsen, IEEE Trans. Nucl. Sci., NS-43 (1996) p. 1357. [9] D.S. McGregor, G.F. Knoll, Y. Eisen and R. Brake, GaAs Detectors and Electronics for High-Energy Physics, C.del Papa, P.G. Pelfer and K. Smith, eds. (World Scientific, Singapore, 1992) p. 30. [10] M.L. Lovejoy, A.J. Howard, K.R Zavdil, D.J. Rieger, R.J. Shul and P.A. Barnes, J. Vac. Sci. A, 13 (1995) p. 758. [11] D.S. McGregor, et al., IEEE Trans. Nucl. Sci., NS-43 (1996) p. 1397. [12] Stopping Powers and Ranges for Protons and Alpha Particles, International Committee on Radiation Units and Measurements (ICRU) Report No 49D. [13] G.R. Imel and T. Urbatsch, Beam Characterization at the Neutron Radiography Facility (NRAD), Proceedings, 4 th World Conference on Neutron Radiography, Gordon and Breach Publishers, 1994, p. 673. 6 Copyright 2000 by ASME