Department of Physics & Astronomy Experimental Particle Physics Group Kelvin Building, University of Glasgow Glasgow, G12 8QQ, Scotland Telephone: ++44 (0)141 339 8855 Fax: +44 (0)141 330 5881 GLAS-PPE/2002-20 Study of irradiated 3D detectors Patrick Roy a, G. Pellegrini a,b, A. Al-Ajili a, L. Haddad a, J. Melone a, V.O Shea a, K.M. Smith a, V. Wright a, M. Rahman a a Department of Physics and Astronomy University of Glasgow, Scotland b Department of Electronics and Electrical Engineering University of Glasgow, Scotland Email: p.roy@physics.gla.ac.uk Abstract The use of the 3D detector geometry allows the life of semiconductor devices to be extended in harsh radiation environments or the usage of less than optimal material (i.e. with poorer charge collection efficiency). Three different production methods have been investigated: inductively coupled plasma etching, laser machining and photoelectrochemical etching. The electrical characteristics of the resulting test devices made in low resistivity silicon and gallium arsenide have been studied. Some of these 3D detectors were characterised after irradiation by 300 MeV/c pions, up to a fluence of 10 14 π/cm 2 at the Paul Scherrer Institute, Villigen. 1 / 11
1 Introduction The need for radiation tolerant semiconductor detectors for future high luminosity colliders can be met through different means [1]. Usually, four main options are possible. The first and least advantageous option consists in simply producing cheap, non-radiation hard detectors and replacing them as their performance degrades. The second option involves a variation on the detector operation conditions such as using detectors under partially depleted, forward biased or low temperature conditions. The third option exploits novel material engineering. Finally the fourth option is device engineering such as adapting the geometry of the detectors, which is the one investigated with the production of the 3D detectors [2]. In practice, some of those options can be combined for even better results. The electrodes for the 3D detectors are located through the bulk of the semiconductor, not at the surface as is the case for the standard co-planar geometry (Fig. 1). The unit cell considered here consists of an hexagonal structure, where the central anode is surrounded by six cathodes. The collection time and distance involved are thus a function of the electrode spacing, not of the device thickness. For the same active volume compared to the standard co-planar geometry, the charge collection time and operational bias are thus greatly reduced. The faster collection time leads to a reduced probability of trapping, thus improving the charge collection efficiency in presence of traps compared to the co-planar geometry. For those reasons, the useful lifetime of the device is extended. 2 / 11
2 Detector fabrication In order to make a 3D detector, vias have to be created and partially or totally filled in order to produce electrodes. The processes considered for the creation of the vias were dry etching, laser machining and photoelectrochemical (PEC) etching. An important goal is to maximise the aspect ratio (hole depth to diameter). The idea is for the vias to reach the full depth of the sample (typically 200-300 µm) while keeping their volumes to a minimum. It is important to keep in mind that the vias are usually considered as inactive regions as the detecting medium is actually the volume between the electrodes. For this reason, a diameter of less than 2 µm is desirable for the vias. 2.1 Hole creation The strength/weakness of the various fabrication methods are summarised in Table 1 with examples of vias in Fig. 2. See reference [3] for a detailed explanation of the actual fabrication steps. As can be seen in Table 1, the only process that will allow reaching the goal of less than 2 µm hole going through more than 200 µm of silicon is photoelectrochemical etching. Another advantage of that method is that, contrary to the two other methods [4], no defects are created deep on the sidewalls of the vias, which cause problems for Schottky contacts. 2.2 Electrode formation A simple and cost effective method of producing electrodes in a semiconductor consists of evaporating a metal in order to create Schottky contacts. Although this is a valid option for GaAs, in the case of silicon this is only a short-term solution as the goal is to produce p-n junctions in order to decrease the leakage current. Once the electrodes have been formed inside the vias, they have to be connected to the read-out electronics. Eventually flip-chip bump bonding technology will be used in order to read out the 3 / 11
signal from each electrode individually (or that from the central electrode of each unit cell), but for testing purposes two simpler approaches had to be adopted. In the case of current versus voltage (I-V) and capacitance versus voltage (C-V) measurements, the whole array can be studied at the same time by using the connecting scheme shown in Fig. 3 a). Tracks consisting of 150 nm of aluminium connect all the central electrodes to a single pad, while the surrounding electrodes are all connected to a second pad. This method allows the use of wire bonding to ground one pad while reading the signal from the second pad. For spectroscopy measurements, while the surrounding electrodes are all connected to a single ground, one central electrode is directly wire bonded in order to measure a single unit cell as shown in Fig. 3 b). 3 Results and analysis The electrical characteristics of the 3D detectors have been investigated before and after irradiation using the contacting scheme shown in Fig. 3. The semiconductors used for this preliminary study were from low resistivity (~100 Ÿ-cm) material. The silicon 3D detectors were made by dry etching (10 µm diameter x 130 µm deep electrodes with an 85 µm pitch in a 250 µm substrate). The gallium arsenide 3D detectors were made by laser machining (10 µm diameter x 250 µm deep electrodes with an 85 µm pitch in a 500 µm substrate). 3.1 Irradiation Three silicon 3D detectors and one gallium arsenide 3D detector have been irradiated at the Paul Scherrer Institute (PSI, Villigen). Using a beam of 300 MeV/c pions at a flux of 10 14 π/cm 2 /day, fluences between 10 12 and 10 14 π/cm 2 were reached. The 1 MeV neutron equivalent fluences can be obtained by multiplying the 300 MeV/c pion fluences by 0.82 [5]. 4 / 11
3.2 Current-voltage measurements The current versus voltage (I-V) measurements were performed using a Keithley 237 (high voltage source measurement unit) controlled by a LabView program. As can be seen from Fig. 4 a), before and at low irradiation, the I-V characteristics of the silicon 3D detectors are those that can be expected for an ideal diode. Figure 4 b) shows that for fluences of 10 13 and 10 14 π/cm 2, no real distinction between reverse or forward bias can be observed in silicon. The total current has been normalised to a single unit cell. The gallium arsenide sample shows a similar behaviour in Fig. 4 c). 3.3 Capacitance-voltage measurements The capacitance versus voltage (C-V) measurements were performed using a Keithley 237 and a Hewlett-Packard 4274A (Multi-frequency LCR meter) controlled by a LabView program. As can be seen from Fig. 4 d), the capacitance measurements performed at low frequencies on a silicon 3D detector reveal the presence of defects even before irradiation [6]. Those defects are a combination of the initial low resistivity material and the dry etching used to create the vias [4]. By using C-V measurements made at high frequency (100 khz), it is possible to be insensitive to those defects and thus obtain the value of the full depletion bias. As can be seen in Fig. 4 e), for the silicon 3D detectors before irradiation the depletion voltage is around 32 volts and decreases to 18 volts after a fluence of 10 14 π/cm 2. No C-V measurements were possible on the gallium arsenide sample. 3.4 Analysis The full depletion biases of the silicon 3D detectors were extracted from a C-V measurement at each fluence. Those values are reported in Fig. 5 a), where it can be seen that too few irradiation steps were taken to be able to determine without doubt if space charge inversion has been reached, although the figure does seem to indicate that 5 / 11
it is the case. Figure 5 b) shows the value of the leakage current measured at full depletion for the different irradiation steps of the silicon 3D detectors. Even though the I-V curves are moving away from the ideal diode, the values of the leakage current at full depletion are actually decreasing, thus improving the performance of the detectors. 4 Conclusion This preliminary study of 3D detectors, made from low resistivity silicon and gallium arsenide, has shown that they still operate properly after pion fluences up to 10 14 π/cm 2. Furthermore, the performance of those 3D detectors has improved as a function of fluence, requiring a slightly lower bias and having a lower leakage current at full depletion. A more extensive study of 3D detectors made in detector grade semiconductor material will follow this study. Acknowledgements The authors would like to thank M. Moll, M. Glaser and K. Gabathuler for performing the irradiations at PSI and the TOPS facility for the use of the laser. This project was funded in part by the Particle Physics and Astronomy Research Council (PPARC, UK) and by the European Commission as part of the Fifth Framework Programme, G5RD- CT-2001-00536. References [1] "R&D Proposal: Development of radiation hard semiconductor devices for very high luminosity colliders", LHCC 2002-003/P6 (2002). [2] S. Parker, "3D - A Proposed New Architecture for Solid State Radiation Detectors", Nucl. Instr. and Meth. A 395 pp. 328-343 (1997). 6 / 11
[3] G. Pellegrini et al., "Technology Development of 3D Detectors for High Energy Physics and Imaging", Nucl. Instr. and Meth. A 487 pp. 19-26 (2002). [4] M. Rahman, "Channelling and Diffusion in Dry-Etching Damage", J. Appl. Phys. 82 p. 5 (1997). [5] A. Vasilescu and G. Lindstroem, "Displacement damage in silicon", on-line compilation at http://sesam.desy.de/members/gunnar/si-dfuncs.html. [6] Z. Li and H.W. Kraner, "Studies of frequency dependent C-V characteristics of neutron irradiated p + n silicon detectors", IEEE Trans. Nucl. Sci. Vol. 38 No 2, pp. 244-250 (1991). Figures Fig. 1: Cross sections of semiconductor detectors made using: a) standard co-planar geometry and b) the 3D geometry. Fig. 2: Cross sections of vias (10 µm diameter) obtained in silicon by: a) inductively coupled plasma, b) laser machining and c) photoelectrochemical etching. Fig. 3: Circuits created to perform electrical measurements: a) all the central electrodes are connected to a single bond pad while the surrounding electrodes are grounded and b) one central electrode is wire bonded while the surrounding electrodes are grounded. Fig. 4: Current versus voltage results for silicon 3D detectors at a) low and b) high fluences as well as for c) a gallium arsenide 3D detector. Capacitance versus voltage results in silicon 3D detectors for: d) 1 to 100 khz before irradiation and e) 100 khz as a function of fluence. Fig. 5: Evolution of a) the full depletion bias and b) the leakage current as a function of pion fluence for silicon 3D detectors. 7 / 11
Tables Table 1: Summary of the strengths/weaknesses of the different fabrication processes. Process Dry Etching Laser Drilling PEC Etching Table 1 Rate 1-3 µm/min 3-5 sec/hole 250-500 nm/min Aspect ratio Current Expected Sidewall Damage Comment 14:1 <26:1 YES Standard and reliable process. 25:1 <50:1 YES Independent of material. Costly and not easily scalable. 14:1 >100:1 NO Multi-step process. 8 / 11
ionising radiation -ve -ve -ve +ve -ve +ve n p + n + h + e - x 2D h + W 2D E e - x 3D +ve a) b) Fig. 1 W3D E a) b) c) Fig. 2 a) b) Fig. 3 9 / 11
a) b) c) d) e) Fig. 4 10 / 11
Fig. 5 a) b) 11 / 11