Charge collection efficiency degradation on Si diodes irradiated with high energy protons

Accepted Manuscript Charge collection efficiency degradation on Si diodes irradiated with high energy protons J. Garcia Lopez, M.C. Jimenez-Ramos PII: S0168-583X(14)00322-X DOI: http://dx.doi.org/10.1016/j.nimb.2014.02.065 Reference: NIMB 60212 To appear in: Nucl. Instr. and Meth. in Phys. Res. B Please cite this article as: J. Garcia Lopez, M.C. Jimenez-Ramos, Charge collection efficiency degradation on Si diodes irradiated with high energy protons, Nucl. Instr. and Meth. in Phys. Res. B (2014), doi: http://dx.doi.org/ 10.1016/j.nimb.2014.02.065 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. 1 Charge collection efficiency degradation on Si diodes irradiated with high energy protons. J. Garcia Lopez1,2* and M. C. Jimenez-Ramos1 1CNA ( U. Sevilla, J. Andalucia, CSIC), Av. Thomas A. Edison 7, 41092 Sevilla, Spain 2Dpto. Física Atómica, Molecular y Nuclear, Universidad de Sevilla. 41080 Sevilla, Spain. The charge collection efficiency (CCE) of several p-type Si diodes has been determined by the Ion Beam Induced Charge (IBIC) technique with 4 MeV protons. In addition, the time evolution of the collected carriers has been recorded as a function of the reverse bias voltage. The diodes were irradiated in our cyclotron with 17 MeV protons and fluences ranging from 3.3x1011 to 1.65x1013 p/cm2. The high energy irradiation was selected because of the practically constant value of the proton stopping power across the samples, leading to a uniform vacancy profile with depth. It is observed that the CEE decreases linearly with radiation fluence while the leakage current increases with ion dose. From these results, the diffusion length of minority carriers, the damage constant and the damage coefficient of p-type Si diodes have been evaluated. Keywords Si diodes, IBIC, proton irradiation, charge collection efficiency, transient analysis *Corresponding author Tel. : +34 954460553, fax: +34 954460145, email: [email protected] 2 Introduction The study of radiation effects in semiconductor electronics and detectors is fundamental to evaluate the lifetime and performance deterioration of the semiconductor devices working in high radiation environments like nuclear reactors, particle accelerators and outer space [1,2]. While the effect of radiation induced defects on properties of semiconducting and insulating materials and devices has been studied for 50 years, there are still important gaps in the understanding of what types of defects are formed, how they can be detected and their effects on electrical and structural properties. The present work has been done in the framework of the IAEA Coordinated Research Project (CRP) No: F11016, whose main objective is to obtain a deeper theoretical knowledge and experimental data on defects created in semiconductor and insulator materials by the use of accelerator-based light and heavy ion irradiation. In this paper we present a study by the Ion Beam Induced Charge (IBIC) technique of the damage produced by 17 MeV protons (the higher proton energy available among the participants of this CRP) in silicon diodes at low ion fluencies. A review of the IBIC theory and applications can be found in [3]. Experimental The samples studied in this work consist of a 6 x 6 mm2 Floating Zone p-type Si diodes fabricated by the Helsinki Institute of Physics (HIP). They are formed by a 3 μm n+ top layer, followed by a 300 μm intrinsic substrate with doping concentration of 1012/cm3 and a 7 μm p+ back layer. Ohmic contacts were deposited on the top n+ and the back p+ regions by Al evaporation. The thickness of the Al contacts was 422±10 nm, as determined by RBS using a 2 MeV He2+ beam. Fig. 1 shows a scheme of the diode with the doping profile extracted from the C-V curve determined at the Sandia National Laboratory (SNL) [4]. Also included in Fig. 1 is the ionization profile produced by the 4 MeV proton beam used for the IBIC measurements (see below). The inset represents the width of the depletion region vs. reverse bias voltage, as found out in [4]. The external beam of our cyclotron was used to irradiate the diodes at several fluences. The samples were placed in air at 15 cm from the 150 μm thick kapton exit window. Although the cyclotron delivers 18 MeV protons, the actual proton energy at the sample’s surface after traversing the kapton foil and the air was 17 MeV, as calculated using the SRIM code [5]. Since the proton beam current can not be measured directly on 3 the diode, the proton fluence was indirectly determined by knowing the ratio between the amount of beam impinging on a 10 mm diameter carbon collimator placed right after the exit window and connected to a Brookhaven 1000c current integrator, and the amount of beam passing through this collimator. The proton flux was typically 3x1010 p/cm2s. A second Al collimator with 0.55 mm diameter was placed in front of the diodes in order to limit the irradiated volume, which otherwise would increase the leakage current to unacceptable levels. The diodes were irradiated to fluences of 3.3 x 1011 p/cm2, 1.65 x 1012 p/cm2, 3.3 x 1012 p/cm2 and 1.65 x 1013 p/cm2, respectively. The IBIC measurements were performed in our microbeam line using a 4.07 MeV proton beam. To avoid creating additional damage during the measurements, the beam was slightly focused to a spot of 50x50 μm2 and the proton rate was kept below 500 Hz. The signal height was recorded as a function of the applied bias voltage using a Canberra 2003BT preamplifier, a Tennelec TC245 amplifier with a shaping time of 2 μs and the OMDAQ ADC/MCA system from Oxford Microbeams Ltd. A triple alpha source (244Cm, 241Am and 239Pu) with about 1 μCi activity was placed inside the vacuum chamber, in that way the alpha spectrum was simultaneously recorded together with the IBIC signals providing an absolute calibration of the full electronic chain. Moreover, in order to correct for the possible changes on the overall electronic gain due to the variation of the detector capacitance at different bias, a calibrated pulser was connected to the “Test” input of the preamplifier. We did not observe any significant change in the position of the pulser signal for bias voltages between 42 V and 0.5 V. Finally, in order to obtain further information about the transport of free carriers to the collecting electrodes, the time dependence of the signal current was measured by connecting the “Energy output” of the preamplifier to the input of a 2 GHz LeCroy 204Mxi-A digital oscilloscope. Results and discussion Fig. 2 shows the vacancy profile induced by the 17 MeV proton beam, as evaluated by SRIM, assuming a displacement energy in Si of 21 eV. As observed, the high energy protons create a vacancy profile that is practically constant all along the Si diode, leading to a uniform distribution of active defects in the detector volume. In these conditions, the increase in reverse diode bias current due to the creation of mid-gap states after irradiation can be written as [6]: 4 AdIId⋅⋅ + = φ α 0 (1) where I0 is the leakage current before irradiation, α is a damage coefficient dependent on particle type and energy, φ is the particle fluence and the product of detector area and thickness Ad is the irradiated detector volume. At 40 volts, where the leakage current saturates, the damage coefficient for 17 MeV protons is α ≈ 7.8·10-17 A/cm, which is about 2.6 times higher that the value found in the ATLAS inner silicon detector irradiated with 650 MeV protons [7].The displacement energy is linked to nonionizing energy loss (NIEL), which is not proportional to the total energy absorbed. The displacement damage vs. energy for neutrons, protons, pions and electrons, relative to 1 MeV neutrons, has been compiled by Lindstrom [8]. From this data, the relative damage for 17 MeV protons is 2.9 larger compared to 650 MeV protons, in good agreement with our experimental results. The reverse bias voltage dependence of the total collected charges measured for a non-irradiated diode is shown in Fig.3. As observed, for voltages from 40 V to 7 V, the CCE is practically 100% and this value decreases for lower voltages until reaching 55% at zero volts. If trapping and recombination are negligible within the depletion zone, the CCE can be calculated as [3]: )( 100 0dxe dx dE dx dx dE E CCE diff pL wx R w w i − − ∫∫ += (2) where Ei = 4.07 MeV and Rp = 156 μm are the initial energy and the projected range of the incident protons in the material of the diode, respectively, dE/dx is the ionizing energy loss that we have estimated using SRIM, w is the width of the depletion layer and Ldiff is the diffusion length of minority carriers (electrons) within the electroneutral region. The first term in eq. (2) represents the drift current and the second term corresponds to the diffusion current. Fig. 3 shows the calculation of the collected charge efficiency for Ldiff values between 0 and 100 μm, where it becomes apparent that the diffusion length for electrons in the pristine diodes must be of the order of 60 μm. It is also observed that, for very low voltages, the experimental CCE is lower than the calculated one, reflecting the fact that in these conditions it may be some recombination of the minority carriers before they are collected. Further information about the transport of free carriers to the 5 collecting electrodes was obtained with the cumulative plots of the signal current. The time evolution of the total collected charge for the pristine diode biased at 7 volts and zero volts is depicted in Fig. 4a and 4b, respectively. To rule out the possibility of artificial variations in the transient times due to changes in the capacitance of the full system with the applied bias, the response of a pulser signal was also measured. At zero volts, the pulser signal was fully collected in 100 ns. Note that at 7 volts the width of the depletion region (153 μm) coincides approximately with the range of the 4.07 MeV protons, and therefore there is no diffusion term. We observe two different time regimes, a faster stage until ~64 ns, where 72% of the charge is collected, and a slower component that extend up to ~200 ns before 100% of the CCE is reached. A simple analysis of the transient behaviour of created carriers in our diodes was done by approximating the samples by an abrupt PN junction, where the doping of the p-type substrate is much lower than the doping of the n+ region, and therefore the depletion region extends predominantly into the p-type bulk. In these conditions, the time required for a charge originated at point x0 to reach a point x is [6]: 0 2 ln )(2 )( xw xw VV w xt bib− − + = μ (3) where Vb is the source reverse bias voltage, Vbi is the built-in voltage (typically 0.5 volts in Si) and w is the depletion width. The mobility μ in Si at 300 K at low fields (up to about 104 V/cm) is 1350 cm2/Vs for electrons and 450 cm2/Vs for holes. Considering Vb=7 V and w = 153 μm, the longest drift time for electrons will correspond to those carriers formed at the end of the depletion layer (x0 = 153 μm) drifting towards the high field region and reaching the n-electrode situated at x = 0. In that case the electron will have traversed 99.5% of the depletion layer in a time nste62≈, which agrees well with the fast time step of Fig.4a. Similarly, the holes formed at the surface of the diode (x0 = 0) and drifting toward the low field electrode x = w, will have traversed 99.5% of the detector in a time nsth185≈, in good conformity with our experimental results. On the other hand, when no bias is applied to the diode, the CCE falls to about 55% and the maximum signal is reached only after 2000 ns (Fig. 4b). In that case w ≈ 80 μm and the longest drift times for electrons 6 and holes according to eq. (3) are nste250≈ and nsth750≈, respectively. Moreover, the minority carriers must diffuse a maximum of ~70 μm until the junction. As this distance is of the order of the diffusion length (Ldiff = 60 μm), some carriers are recombined before they reach the electrode, as we found experimentally. After irradiation of the diodes with 17 MeV protons, the samples were introduced in the microbeam line for IBIC measurements. In order to locate the irradiated spots, beam scans of 2x2 mm2 were carried out to obtain CCE mappings from regions with different charge pulse height. Fig. 5 shows various CCE maps recorded around the spot irradiated with the highest fluence (1.65 x 1013 p/cm2) at several bias voltages. As observed, when the electric field is sufficiently strong (Fig. 5a), the CCE is higher than 92% in all the diode and the map presents no contrast. At lower bias (Fig. 5b and c) the electric field can not prevent some recombination, thus the charge collection decreases at the irradiated spot and the damaged region can be easily located. Further reduction of the voltage (Fig. 5d, e and f) shows an annular region whose CCE presents intermediate values between the pristine and the irradiated point. This effect is probably due to charge carriers created in a non-defective region which may diffuse to the damaged zone before reaching the collecting electrodes. This behavior is enhanced at zero volts (not shown), when the charge collection takes longer and, therefore, the diffusion of carriers is relatively more important compared to drift. In fact, in the case of the spot with the lower fluence (3.3 x 1011 p/cm2), contrast was obtained only when the CCE map was measured with no external bias applied, which means that this is approximately the smallest fluence (using 17 MeV protons) that we can analyze by IBIC. Point measurements at different bias where carried out in the middle of each irradiated spot to study the decrease in CCE with the proton fluence. The obtained values are depicted in Table 1. It is important to remark that, as the leakage current of the diodes after irradiation is not negligible, the actual bias voltage applied to the diode, VR, has to be calculated taking into account the voltage drop through the preamplifier load resistance. VR (volts) Pristine (CCE %) 3.3 x 1011 p/cm2 (CCE %) 1.65 x 1012 p/cm2 (CCE %) 3.3 x 1012 p/cm2 (CCE %) 1.65 x 1013 p/cm2 (CCE %) 7 38 100 99.8 99.8 99.4 97.5 12 99.7 99.4 99.2 99.2 94.0 8.5 99.4 99.2 98.6 98.1 89.0 5.1 95.2 95.3 94.3 94.0 74.4 3.8 90.1 90.3 88.9 88.7 65.4 2.5 82.4 82.4 79.8 79.6 53.4* 1.0 65.1 62.5 57.2 56.3 29.1* Table 1. CCE vs proton fluence measured at different bias. For the data marked with (*) the collection times were longer than the shaping time of the amplifier and therefore the amplitude of the signal is not correct The fitting of the CCE vs. bias voltage curve using eq. (2) for the sample irradiated to the highest fluence 1.65 x 1013 p/cm2 shows that the diffusion length for electrons drops to a value ≈ 2-3 μm. Moreover, as observed in Table 1, for fluence values up to 3.3 x 1012 p/cm2 the CCE decreases linearly with the proton fluence independently of the bias voltage applied during the measurements. The efficiency of the diode irradiated to 1,65 x 1013 p/cm2 drops more rapidly, except for the measurements performed at 38 volts, where the CCE behavior can also be considered to be linear. Note that for bias voltages below 3.8 volts, the charge collecting time for this diode exceeds 2 μs and therefore the experimental efficiencies shown in Table 1 are lower than the actual values, as they are influenced by the shaping time of the amplifier. From the phenomenological relationship [9]: Φ⋅−= K CCE CCE 1 0 (4) where the suffix 0 stands for the pristine diode and Φ is the particle fluence, the calculated damage constant ranges from K ≈ 1,5 x 10-15 cm2 at VR=38 volts to K ≈ 4,1 x 10-14 cm2 at VR=1 volt. Conclusions 8 The charge collection efficiency of p-type Si diodes has been measured by IBIC using 4 MeV protons. In the case of pristine samples, the fitting of the CCE curve vs. bias voltage using the simple drift-diffusion model, indicates that the diffusion length of minority carriers (electrons) is ≈ 60 μm. This value drops a factor of 20 for the sample irradiated with 17 MeV protons to fluences of 1.65 x 1013 p/cm2. After irradiation, the increase of the leakage current shows that the damage coefficient for 17 MeV protons is α ≈ 7.8·10-17 A/cm. The IBIC results demonstrate that the efficiency of the diodes decreases linearly with fluence for values up to 3.3 x 1012 p/cm2. The variation of the damage constant K with the voltage bias indicates that the electrical defects created by the high energy proton irradiation are considerably more effective at low voltages, when the electric field inside the diode is not sufficiently high to prevent trapping and recombination. References [1] Kohriki et al. IEEE Trans. Nucl. Sci. NS43/3 (1996) 1200-1202 [2] C.J. Bebek et al. SPIE 4669 (2002) 161-171 [3] M.B.H. Breese, E. Vittone, g. Vizkelethy and P.J. Sellin. NIM B 264 (2007) 345-360 [4] G. Vizkelethy, “Annual report of Sandia National Laboratories (SNL) contribution to IAEA CRP F11016“ 28 April 2013. [5] J.F. Ziegler, J.P. Biersack, M.D. Ziegler, “SRIM – The Stopping and Range of Ions in Matter”, Ion Implantation Press, 2008. http://www.lulu.com/content/1524197. [6] Semiconductor Detector Systems. Helmuth Spieler. Oxford University Press. ISBN 978-0-19-852784-8. (2011). [7] A. Chilingarov et al. Nucl. Instr. and Meth. A360 (1995) 432. [8] Lindstrom, G. Displacement Damage in Silicon (2000). http://sesam.desy.de/members/gunnar/Sidfuncs.html. [9] Z. Pastuovic, M. Jaksic, G. Kalinka, M. Novak, A. Simon, IEEE Trans. Nucl. Sci. 56, No.4, 2467 (2009). Acknowledgements