- Email: [email protected]
Application and modeling of single contact electron beam induced current technique on multicrystalline silicon solar cells
Solar Energy Materials & Solar Cells 133 (2015) 143–147
Contents lists available at ScienceDirect
Solar Energy Materials & Solar Cells journal homepage: www.elsevier.com/locate/solmat
Application and modeling of single contact electron beam induced current technique on multicrystalline silicon solar cells L. Meng a,b, A.G. Street a, J.C.H. Phang a, C.S. Bhatia a,b,n a Centre for Integrated Circuit Failure Analysis and Reliability (CICFAR), Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, 117576 Singapore, Singapore b Solar Energy Research Institute of Singapore (SERIS), National University of Singapore, 7 Engineering Drive 1, 117574 Singapore, Singapore
art ic l e i nf o
a b s t r a c t
Article history: Received 1 August 2013 Received in revised form 22 October 2014 Accepted 3 November 2014
The first demonstration of single contact electron beam induced current (SCEBIC) technique on multicrystalline silicon (mc-Si) solar cells is reported. A lumped single-diode analytical model is also proposed to theoretically explain the SCEBIC phenomenon within solar cells as well as the current transient characteristics of the major model parameters, such as shunt resistance Rsh, junction capacitance Cj and parasitic capacitance Cs. The accuracy of the analytical model is then verified using PSPICE simulations, which show a close match with the experimental results. It is found that a large value of parasitic capacitance Cs is necessary to achieve good SCEBIC signal strength with a relatively low signal-to-noise ratio (SNR), and this is realized experimentally by adopting a metal enclosure in the measurement setup. In addition, an advantage of SCEBIC over the conventional double-contact method is also demonstrated by characterizing partially processed solar cells, which clearly illustrates the high degree of flexibility of SCEBIC in solar cell characterization. & 2014 Published by Elsevier B.V.
Keywords: Single contact electron beam induced current SCEBIC EBIC Solar cell Characterization PSPICE simulation
1. Introduction Electron beam induced current (EBIC) has been widely used for characterizing semiconductor materials and microelectronic devices as a useful technique for p–n junction imaging, as well as localization of defects and doping inhomogeneities [1,2]. Conventionally, EBIC requires electrical contacts to both the p- and n-type regions of the p–n junction. Such convention is also adopted in this paper and, unless stated otherwise, EBIC refers to the double-contact technique. To overcome the double-contact EBIC requirement, singlecontact EBIC (SCEBIC) was subsequently developed [3]. As the name suggests, SCEBIC requires only one electrical contact to either the p- or n-type region. It differs from EBIC primarily by being an AC technique that takes advantage of the parasitic capacitance between a floating contact and ground. This single-contact technique proves to be extremely convenient, for instance, for imaging multi-layer integrated circuits (ICs), where both connected and unconnected junctions can be clearly captured [4,5], thus allowing high flexibility for imaging and characterization of semiconductor devices and ICs.
n Corresponding author at: Centre for Integrated Circuit Failure Analysis and Reliability (CICFAR), Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, 117576 Singapore, Singapore. Tel.: þ 65 6516 7216. E-mail address: [email protected] (C.S. Bhatia).
http://dx.doi.org/10.1016/j.solmat.2014.11.003 0927-0248/& 2014 Published by Elsevier B.V.
Recently, there has been increasing popularity in applying EBIC in solar cells to study extended crystallographic defects such as dislocations, grain boundaries, microcracks and breakdown sites in photovoltaic devices [6–9]. EBIC is also employed to characterize local electrical properties of solar cells, such as minority-carrier current collection efficiency, characteristics of the p–n junction, as well as local recombination sites in the semiconductor materials [10,11]. While the double-contact EBIC has been gaining increased traction for characterizing solar cells, application of its singlecontact counterpart, SCEBIC, has remained elusive. In particular, one of the main difficulties in implementing SCEBIC on solar cells is that the time-dependent SCEBIC transient signals are highly sensitive to the junction and parasitic capacitances of the devices [12]. Unlike ICs, where the junction areas and thus the corresponding junction capacitances are relatively small, photovoltaic devices tend to have large junction areas, which in turn result in SCEBIC transient signals with a poor signal-to-noise ratio (SNR). This makes it difficult to separate the real signals from noise. In this work, the first demonstration of SCEBIC measurement on a single p–n junction multicrystalline silicon (mc-Si) solar cell is reported. It is shown that by using a novel technique of employing a metal enclosure in the SCEBIC setup, SCEBIC imaging with accuracy and resolution comparable to the conventional double-contact EBIC can be established on mc-Si solar cells. A theoretical explanation of the SCEBIC phenomenon in solar cells as well as the importance of having a large value of Cs in obtaining good SCEBIC signal strength
144
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
Fig. 1. SCEBIC system experimental setup. The p-type region is left floating and only the n-type region of the sample (top metal finger) is connected to the external circuit.
are also explained using a lumped single-diode analytical model and supported by PSPICE simulations. Lastly, SCEBIC characterization of partially processed solar cells is also demonstrated, clearly illustrating the higher degree of flexibility of SCEBIC in solar cell characterization and an advantage of SCEBIC over the conventional double-contact EBIC.
2. SCEBIC setup and experiment The SCEBIC imaging was carried out in a scanning electron microscope (SEM) with a beam blanker to achieve an intensitymodulated electron beam, (i.e. a pulsed electron beam). Fig. 1 shows the experimental setup of the system. A function generator was used to drive the beam blanking plates and also to provide the reference signals of the lock-in amplifier (LIA). As shown in the setup, the p-type region was left floating and only the n-type region of the sample (top metal finger) was directly connected to the external circuit. As the induced current signals are very weak (in the range of nano-amperes), a low-noise current amplifier was employed to improve the signal-to-noise ratio of the transients. The resulting signals were then fed into the LIA and an imaging acquisition system that synchronized the scan generator with the LIA output. The image acquisition system also controlled the position of the electron beam on the sample and generated the SCEBIC images. An important difference between our experimental setup and previous SCEBIC work is the addition of a metal enclosure that acts as a shield against noise and increases the parasitic capacitance in the SCEBIC circuit. For all the SCEBIC measurements in this work, a 200-mm thick commercial mc-Si solar cell was used. The active n þ –p area of the sample is approximately 1 cm2. The contact to the n þ region was established by using electrically conductive silver paste.
3. Results and discussion Fig. 2a is a secondary electron (SE) image showing the general surface morphology of the mc-Si solar cell. The textured surface is typical of mc-Si solar cells and is a result of the multicrystalline nature of the silicon and the saw-damage etch (SDE) process. Fig. 2b shows a corresponding conventional EBIC image from the same location. This image clearly demonstrates the defect localization capability of EBIC as several defects are visible. When combined with SE and other techniques as reported earlier [13,14], EBIC allows easy correlation and accurate analysis of the nature of the defects.
As also mentioned previously, SCEBIC imaging on solar cells is difficult due to poor SNRs and this is apparent from the very noisy image depicted in Fig. 2c. This image was made by connecting only the n-type region of the sample to the external circuit and leaving the p-type region floating (i.e. using the conventional SCEBIC [3] without a grounded metal enclosure). The addition of a grounded metal enclosure close to the sample surface significantly increases the SNR of the circuit, generating a clear SCEBIC image (Fig. 2d) of comparable quality to the conventional EBIC image. Based on our best knowledge, such successful application of SCEBIC represents the first demonstration of SCEBIC on solar cell characterization. While the capabilities of SCEBIC and the effectiveness of the metal enclosure are clearly demonstrated in Fig. 2, it is important to understand the physical and electrical effects of the metal enclosure. Fig. 3a and b shows the configuration of a typical p–n junction solar cell and its equivalent circuit diagram of the SCEBIC measurements, respectively. A lumped single-diode model is used in this case since it has been used to provide a simple yet reasonably accurate physical picture of device operation under electron-beam excitation [15]. The solar cell is represented by a diode with a shunt resistance Rsh and a junction capacitance Cj in parallel [16]. For ICs, the value of Rsh is typically very high owing to the relatively low density of defects and well-passivated p–n junction edges. Because of this, Rsh is usually omitted in the SCEBIC model of ICs. However, in the case of solar cells, the p–n junction is much leakier, making Rsh a necessary parameter in the model. The series resistance Rs refers to the total contact resistance, and the parasitic capacitance Cs is the capacitance that exists naturally between a floating contact (the p-type region of the solar cell in our particular setup) and ground. When the pulsed high-energy electron beam impinges the sample surface, electron-hole pairs are generated and separated by the internal electric field within the space-charge region of the p–n junction as the holes drift to the floating p-type region and the electrons move to the n-type region. This induced current can be effectively modeled by a current source Ig. In the conventional double-contact EBIC, Ig can be readily measured and is commonly known as the generation current. Although there is no physically closed loop in the SCEBIC model, the parasitic capacitance Cs will charge up and discharge as the pulsed electron beam interacts with the device, leading to a complete AC SCEBIC current path. This allows the pulsed current, ISCEBIC(t), to pass through and to be measured by the current meter between the unconnected p-region and ground. To understand the importance and the effect of the different parameters on the SCEBIC transient response, PSPICE simulations are carried out assuming typical values of each parameter.
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
145
Fig. 2. (a) Secondary electron (SE) image; (b) conventional EBIC image; and SCEBIC image (c) without and (d) with a metal enclosure (at 5.7 kHz electron beam modulation frequency) of a mc-Si solar cell. All the images are taken at 30 keV electron beam energy.
Fig. 3. (a) SCEBIC configuration; (b) its equivalent circuit diagram of a typical p–n junction solar cell; and (c) PSPICE simulation of the SCEBIC transient response ISCEBIC(t) at a modulation frequency of 200 Hz, where Ig ¼ 100 mA, Rsh ¼ 5 kΩ, Rs ¼ 1 Ω, Cj ¼ 200 nF, Cs ¼ 100 pF. The values assigned for each parameter are typical for solar cells with a sample size of about 1 cm2.
The resultant time-dependent current transient ISCEBIC(t) is depicted in Fig. 3c. Typically, when the electron beam is first turned on (at t ¼0), a negative current peak Imax (i.e. ISCEBIC(t ¼0)) is detected at the n-type contact of the sample as the generated
electrons and holes drift to the n- and p-type regions, respectively. The injection of electrons starts to charge up the parasitic capacitor Cs and at the same time drives the p–n junction to a forward bias. This bias lowers the potential barrier of the junction, resulting in a forward-bias current (also known as diffusion current) that increases with time, but in an opposite direction to the generation current Ig. As a result, the measured ISCEBIC(t) gradually decreases in magnitude and finally reaches zero when the forward-bias current and generation current are equal. From this moment onwards, ISCEBIC(t) does not vary as long as the beam is turned on. As the electron beam is turned off (at t ¼2.5 ms), only the positive forward-bias current due to carrier diffusion remains. Such a forward-bias current first peaks at the instant when the electron beam is switched off and slowly decays thereafter as the forward-bias current decreases due to the discharging of the parasitic capacitor. This gives rise to an increase in the potential barrier and at the same time a reduction of the excess electrons and holes in the p- and n-type regions, respectively. The diffusion current finally decreases to zero as a result of the p–n junction slowly returning to its thermal equilibrium state without the driving force of the accelerated electron beam. To fully understand the effect of each parameter (Rsh, Cj, Cs and Ig) on the current transient of SCEBIC ISCEBIC(t), PSPICE simulations were carried out. It is observed that Rsh mainly affects the shape of the decay as the decay rate becomes slower for larger Rsh values (Fig. 4a). Such a phenomenon is expected as the RC time constant increases with larger Rsh values [15]. This is different in the case of Rs, which shows minimal impact on the transient current curve (data not shown) as it has been reported that Rs has negligible effect under low current condition (i.e., Ig ¼100 mA) [17]. Interestingly, it is found that the maximum value of the SCEBIC transient Imax depends strongly on both values of Cs and Cj. While Imax is directly proportional to Cs (inset of Fig. 4b), increasing Cj leads to a smaller Imax (Fig. 4c). Furthermore, it can also be seen from Fig. 4c that the time taken for the SCEBIC current to reach the steady-state level of the current transient (i.e. zero value) is longer for larger Cj. This is expected since a
146
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
Fig. 4. SCEBIC transient characteristics of a typical single-junction solar cell using PSPICE simulations with the same model parameters as Fig. 3, by varying only one parameter each time: (a) shunt resistance Rsh; (b) parasitic capacitance Cs; (c) junction capacitance Cj; and (d) generation current Ig.
junction with a larger area would require a longer time to charge up. Lastly, as illustrated in Fig. 4d, it is also observed that Imax rises proportionally with the increasing Ig, and the ratio of Imax/Ig is approximately equal to the ratio of the parasitic and junction capacitances Cs/Cj in all our simulations. Such observation is consistent with the previous findings [12,18], where the ratio of Imax/Ig was estimated to be equal to Cs/(Cs þCj) and since in this work Cs is much smaller than Cj. For the case of the conventional SCEBIC without the metal enclosure, the Cs/Cj ratio is relatively small owing to the small Cs and large Cj values. Assuming a relatively low value of Cs of 10–20 pF, which is typical in our case given our sample size, a very small Imax (i.e. ISCEBIC(t) at t¼0) in the range of several nano-amperes is expected and such a value is close to the noise level of the circuit. This explains why a conventional SCEBIC setup cannot be directly applied in mc-Si solar cells. By increasing Cs to approximately 100 pF, a well-defined ISCEBIC(t) can be clearly obtained. With these two examples, the physical and electrical effects of the metal enclosure become apparent. When the metal enclosure is connected to the common ground and placed closely above the solar cell sample, it functions as a virtual SEM chamber with a much shorter distance between the floating p-region contact and ground. Cs is therefore much larger in this case, leading to an increased Cs/Cj ratio and thus a much larger and easily detectable Imax. In order to verify the accuracy of our SCEBIC model, ISCEBIC(t) from PSPICE simulations are carefully compared and fit into the actual experimentally measured current transients. Fig. 5 shows one cycle of the measured ISCEBIC(t) transient response of the sample to a pulsed electron beam at the energy of 30 keV with a modulation frequency of 60 Hz. The value of Cj extracted from the simulation is approximately 205 nF. It has been shown that Cj is around 70 nF/cm2 at zero bias for a typical p–n junction solar cell, and the value of Cj increases drastically under even a slightly
Fig. 5. Comparison of experimental (blue stars) and simulated (red-solid line) SCEBIC transient responses to a pulsed electron beam (black dash) at the electron beam energy of 30 keV and the modulation frequency of 60 Hz. The area of the sample is about 1 cm2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
forward bias [19]. Given our sample area of 1 cm2 and the relatively low forward bias due to the injection of electrons, the extracted value of 205 nF is within the expected range. As shown in the figure, the measured current transient curve closely matches with the simulated ISCEBIC(t) except for two noticeable differences: firstly, the steady-state of the induced current transient during beam ON does not return to zero value; and secondly, the positive current peak at the instant when the electron beam is just turned off is slightly smaller. The non-zero (approximately 10 nA) steady-state offset of the current transient during the beam ON time is due to the current from the electron beam itself flowing through the sample and adding to the SCEBIC signal. This also explains the exact match between the pulsed-beam current
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
147
Acknowledgment Funding for this research under the Singapore National Research Foundation Grant # NRF2009EWT-CERP001-037 is acknowledged.
References
Fig. 6. SCEBIC image of the same mc-Si solar cell after removing the bottom contact. The image was taken with a metal enclosure at the electron beam energy of 30 keV, and modulation frequency of 5.7 kHz.
level and the offset of the steady-state (i.e. both at 10 nA). The slightly lower positive current peak (i.e. the maximum diffusion current) can be understood by other types of current loss, such as recombination of the excess minority carriers that regularly occurs in multicrystalline silicon materials [20], which contributes to the SCEBIC transient until thermal equilibrium is established after the beam is switched off. As mentioned previously, SCEBIC can show advantages over the conventional double-contact EBIC technique by offering a higher degree of flexibility in sample configuration. For example, since only one electrical contact is required by SCEBIC, it is possible to perform defect characterization or to examine the charge collection capabilities of a p–n junction in a solar cell without connecting both n- and p-type regions. In other words, SCEBIC can be useful for characterizing partially-processed solar cells with only one region of the junction accessible or analyzing samples with metallization present on only the n- or p-type region. This has been demonstrated by imaging the same sample shown in Fig. 2 with the bottom p-type contact removed. Again, it can be seen that the same defect features are nicely captured in the SCEBIC image (Fig. 6), demonstrating the feasibility and flexibility of SCEBIC on partially processed solar cells. 4. Conclusions A single contact electron beam induced current technique is first demonstrated in mc-Si solar cells. A lumped single-diode analytical model is also proposed to provide the theoretical explanations of the SCEBIC phenomenon in solar cells. The accuracy of the analytical model is verified as PSPICE simulations of the model show a close match with the experimental results. Our analytical model also explains the importance of the metal enclosure in increasing the parasitic capacitance, which is needed for achieving good SCEBIC signal strength. With the metal enclosure, SCEBIC imaging with accuracy and resolution comparable to the conventional double-contact EBIC can be readily established in mc-Si solar cells. In addition, the advantage of SCEBIC over the conventional EBIC is also demonstrated by characterizing a partially processed solar cell, which clearly illustrates the high degree of flexibility of SCEBIC in solar cell characterization.
[1] A. Cavallini, L. Polenta, A. Castaldini, Charge carrier recombination and generation analysis in materials and devices by electron and optical beam microscopy, Microelectron. Reliab. 50 (2010) 1398–1406. [2] H.J. Leamy, Charge collection scanning electron microscopy, J. Appl. Phys. 53 (1982) R51–R80. [3] D.S.H. Chan, J.C.H. Phang, W.S. Lau, V.K.S. Ong, V. Sane, S. Kolachina, New developments in beam induced current methods for the failure analysis of VLSI circuits, Microelectron. Eng. 31 (1996) 57–67. [4] D.S.H. Chan, J.C.H. Phang, J.M. Chin, S. Kolachina, Single contact beam induced current phenomena – a review, Beam Injection Assessment of Microstructures in Semiconductors, vol. 78–79, 2000, pp. 11–18. [5] J.C.H. Phang, D.S.H. Chan, V.K.S. Ong, S. Kolachina, J.M. Chin, M. Palaniiappan, G. Gilfeather, Y.X. Seah, Single contact beam induced current phenomenon for microelectronic failure analysis, Microelectron. Reliab. 43 (2003) 1595–1602. [6] K. Arafune, T. Sasaki, F. Wakabayashi, Y. Terada, Y. Ohshita, M. Yamaguchi, Study on defects and impurities in cast-grown polycrystalline silicon substrates for solar cells, Physica B 376–377 (2006) 236–239. [7] J. Chen, B. Chen, T. Sekiguchi, M. Fukuzawa, M. Yamada, Correlation between residual strain and electrically active grain boundaries in multicrystalline silicon, Appl. Phys. Lett. 93 (2008) 112105. [8] Z. Hameiri, T. Puzzer, L. Mai, A.B. Sproul, S.R. Wenham, Laser induced defects in laser doped solar cells, Prog. Photovolt.: Res. Appl. 19 (2011) 391–405. [9] J. Bauer, D. Lausch, H. Blumtritt, N. Zakharov, O. Breitenstein, Avalanche breakdown in multicrystalline solar cells due to preferred phosphorous diffusion at extended defects, Prog. Photovolt.: Res. Appl. (2012), http://dx. doi.org/10.1002/pip.2220. [10] D. Lausch, M. Werner, V. Naumann, J. Schneider, C. Hagendorf, Investigation of modified p–n junctions in crystalline silicon on glass solar cells, J. Appl. Phys. 109 (2011) 084513. [11] T. Sasaki, K. Arafune, W. Metzger, M.J. Romero, K. Jones, M. Al-Jassim, Y. Ohshita, M. Yamaguchi, Characterization of carrier recombination in lattice-mismatched InGaAs solar cells on GaAs substrates, Sol. Energy Mater. Sol. Cells 93 (2009) 936–940. [12] S. Kolachina, J.C.H. Phang, D.S.H. Chan, Single contact electron beam induced currents (SCEBIC) in semiconductor junctions. Part I: quantitative verification of SCEBIC model, Solid-State Electron. 42 (1998) 957–962. [13] L. Meng, D. Nagalingam, C.S. Bhatia, A.G. Street, J.C.H. Phang, Distinguishing morphological and electrical defects in polycrystalline silicon solar cells using scanning electron acoustic microscopy and electron beam induced current, Sol. Energy Mater. Sol. Cells 95 (2011) 2632–2637. [14] W. Kwapil, J. Nievendick, A. Zuschlag, P. Gundel, M.C. Schubert, W. Warta, Influence of surface texture on the defect-induced breakdown behavior of multicrystalline silicon solar cells, Prog. Photovolt.: Res. Appl. (2012), http: //dx.doi.org/10.1002/pip.1226. [15] A. Romanowski, D.B. Wittry, J.M. Tsaur, Analysis of the short-circuit current of a polycrystalline solar cell with excitation by a gated electron beam, J. Appl. Phys. 59 (1986) 951–957. [16] D.S.H. Chan, J.C.H. Phang, A method for the direct measurement of solar cell shunt resistance, IEEE Trans. Electron Devices 31 (1984) 381–383. [17] M. Wolf, H. Rauschenbach, Series resistance effects on solar cell measurements, Adv. Energy Convers. 3 (1963) 455–479. [18] E.I. Rau, A.V. Gostev, S. Zhu, D. Phang, D. Chan, D. Thong, W. Wong, Comparative analysis of scanning electron microscopy techniques for semiconductors: electron-beam-induced potential method, single-contact electron-beam-induced current method, and thermoacoustic detection, Russ. Microelectron. 30 (2001) 207–218. [19] F. Recart, A. Cuevas, Application of junction capacitance measurements to the characterization of solar cells, IEEE Trans. Electron Devices 53 (2006) 442–448. [20] A. Cuevas, F. Recart, Capacitive effects in quasi-steady-state voltage and lifetime measurements of silicon devices, J. Appl. Phys. 98 (2005) 074507.
Contents lists available at ScienceDirect
Solar Energy Materials & Solar Cells journal homepage: www.elsevier.com/locate/solmat
Application and modeling of single contact electron beam induced current technique on multicrystalline silicon solar cells L. Meng a,b, A.G. Street a, J.C.H. Phang a, C.S. Bhatia a,b,n a Centre for Integrated Circuit Failure Analysis and Reliability (CICFAR), Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, 117576 Singapore, Singapore b Solar Energy Research Institute of Singapore (SERIS), National University of Singapore, 7 Engineering Drive 1, 117574 Singapore, Singapore
art ic l e i nf o
a b s t r a c t
Article history: Received 1 August 2013 Received in revised form 22 October 2014 Accepted 3 November 2014
The first demonstration of single contact electron beam induced current (SCEBIC) technique on multicrystalline silicon (mc-Si) solar cells is reported. A lumped single-diode analytical model is also proposed to theoretically explain the SCEBIC phenomenon within solar cells as well as the current transient characteristics of the major model parameters, such as shunt resistance Rsh, junction capacitance Cj and parasitic capacitance Cs. The accuracy of the analytical model is then verified using PSPICE simulations, which show a close match with the experimental results. It is found that a large value of parasitic capacitance Cs is necessary to achieve good SCEBIC signal strength with a relatively low signal-to-noise ratio (SNR), and this is realized experimentally by adopting a metal enclosure in the measurement setup. In addition, an advantage of SCEBIC over the conventional double-contact method is also demonstrated by characterizing partially processed solar cells, which clearly illustrates the high degree of flexibility of SCEBIC in solar cell characterization. & 2014 Published by Elsevier B.V.
Keywords: Single contact electron beam induced current SCEBIC EBIC Solar cell Characterization PSPICE simulation
1. Introduction Electron beam induced current (EBIC) has been widely used for characterizing semiconductor materials and microelectronic devices as a useful technique for p–n junction imaging, as well as localization of defects and doping inhomogeneities [1,2]. Conventionally, EBIC requires electrical contacts to both the p- and n-type regions of the p–n junction. Such convention is also adopted in this paper and, unless stated otherwise, EBIC refers to the double-contact technique. To overcome the double-contact EBIC requirement, singlecontact EBIC (SCEBIC) was subsequently developed [3]. As the name suggests, SCEBIC requires only one electrical contact to either the p- or n-type region. It differs from EBIC primarily by being an AC technique that takes advantage of the parasitic capacitance between a floating contact and ground. This single-contact technique proves to be extremely convenient, for instance, for imaging multi-layer integrated circuits (ICs), where both connected and unconnected junctions can be clearly captured [4,5], thus allowing high flexibility for imaging and characterization of semiconductor devices and ICs.
n Corresponding author at: Centre for Integrated Circuit Failure Analysis and Reliability (CICFAR), Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, 117576 Singapore, Singapore. Tel.: þ 65 6516 7216. E-mail address: [email protected] (C.S. Bhatia).
http://dx.doi.org/10.1016/j.solmat.2014.11.003 0927-0248/& 2014 Published by Elsevier B.V.
Recently, there has been increasing popularity in applying EBIC in solar cells to study extended crystallographic defects such as dislocations, grain boundaries, microcracks and breakdown sites in photovoltaic devices [6–9]. EBIC is also employed to characterize local electrical properties of solar cells, such as minority-carrier current collection efficiency, characteristics of the p–n junction, as well as local recombination sites in the semiconductor materials [10,11]. While the double-contact EBIC has been gaining increased traction for characterizing solar cells, application of its singlecontact counterpart, SCEBIC, has remained elusive. In particular, one of the main difficulties in implementing SCEBIC on solar cells is that the time-dependent SCEBIC transient signals are highly sensitive to the junction and parasitic capacitances of the devices [12]. Unlike ICs, where the junction areas and thus the corresponding junction capacitances are relatively small, photovoltaic devices tend to have large junction areas, which in turn result in SCEBIC transient signals with a poor signal-to-noise ratio (SNR). This makes it difficult to separate the real signals from noise. In this work, the first demonstration of SCEBIC measurement on a single p–n junction multicrystalline silicon (mc-Si) solar cell is reported. It is shown that by using a novel technique of employing a metal enclosure in the SCEBIC setup, SCEBIC imaging with accuracy and resolution comparable to the conventional double-contact EBIC can be established on mc-Si solar cells. A theoretical explanation of the SCEBIC phenomenon in solar cells as well as the importance of having a large value of Cs in obtaining good SCEBIC signal strength
144
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
Fig. 1. SCEBIC system experimental setup. The p-type region is left floating and only the n-type region of the sample (top metal finger) is connected to the external circuit.
are also explained using a lumped single-diode analytical model and supported by PSPICE simulations. Lastly, SCEBIC characterization of partially processed solar cells is also demonstrated, clearly illustrating the higher degree of flexibility of SCEBIC in solar cell characterization and an advantage of SCEBIC over the conventional double-contact EBIC.
2. SCEBIC setup and experiment The SCEBIC imaging was carried out in a scanning electron microscope (SEM) with a beam blanker to achieve an intensitymodulated electron beam, (i.e. a pulsed electron beam). Fig. 1 shows the experimental setup of the system. A function generator was used to drive the beam blanking plates and also to provide the reference signals of the lock-in amplifier (LIA). As shown in the setup, the p-type region was left floating and only the n-type region of the sample (top metal finger) was directly connected to the external circuit. As the induced current signals are very weak (in the range of nano-amperes), a low-noise current amplifier was employed to improve the signal-to-noise ratio of the transients. The resulting signals were then fed into the LIA and an imaging acquisition system that synchronized the scan generator with the LIA output. The image acquisition system also controlled the position of the electron beam on the sample and generated the SCEBIC images. An important difference between our experimental setup and previous SCEBIC work is the addition of a metal enclosure that acts as a shield against noise and increases the parasitic capacitance in the SCEBIC circuit. For all the SCEBIC measurements in this work, a 200-mm thick commercial mc-Si solar cell was used. The active n þ –p area of the sample is approximately 1 cm2. The contact to the n þ region was established by using electrically conductive silver paste.
3. Results and discussion Fig. 2a is a secondary electron (SE) image showing the general surface morphology of the mc-Si solar cell. The textured surface is typical of mc-Si solar cells and is a result of the multicrystalline nature of the silicon and the saw-damage etch (SDE) process. Fig. 2b shows a corresponding conventional EBIC image from the same location. This image clearly demonstrates the defect localization capability of EBIC as several defects are visible. When combined with SE and other techniques as reported earlier [13,14], EBIC allows easy correlation and accurate analysis of the nature of the defects.
As also mentioned previously, SCEBIC imaging on solar cells is difficult due to poor SNRs and this is apparent from the very noisy image depicted in Fig. 2c. This image was made by connecting only the n-type region of the sample to the external circuit and leaving the p-type region floating (i.e. using the conventional SCEBIC [3] without a grounded metal enclosure). The addition of a grounded metal enclosure close to the sample surface significantly increases the SNR of the circuit, generating a clear SCEBIC image (Fig. 2d) of comparable quality to the conventional EBIC image. Based on our best knowledge, such successful application of SCEBIC represents the first demonstration of SCEBIC on solar cell characterization. While the capabilities of SCEBIC and the effectiveness of the metal enclosure are clearly demonstrated in Fig. 2, it is important to understand the physical and electrical effects of the metal enclosure. Fig. 3a and b shows the configuration of a typical p–n junction solar cell and its equivalent circuit diagram of the SCEBIC measurements, respectively. A lumped single-diode model is used in this case since it has been used to provide a simple yet reasonably accurate physical picture of device operation under electron-beam excitation [15]. The solar cell is represented by a diode with a shunt resistance Rsh and a junction capacitance Cj in parallel [16]. For ICs, the value of Rsh is typically very high owing to the relatively low density of defects and well-passivated p–n junction edges. Because of this, Rsh is usually omitted in the SCEBIC model of ICs. However, in the case of solar cells, the p–n junction is much leakier, making Rsh a necessary parameter in the model. The series resistance Rs refers to the total contact resistance, and the parasitic capacitance Cs is the capacitance that exists naturally between a floating contact (the p-type region of the solar cell in our particular setup) and ground. When the pulsed high-energy electron beam impinges the sample surface, electron-hole pairs are generated and separated by the internal electric field within the space-charge region of the p–n junction as the holes drift to the floating p-type region and the electrons move to the n-type region. This induced current can be effectively modeled by a current source Ig. In the conventional double-contact EBIC, Ig can be readily measured and is commonly known as the generation current. Although there is no physically closed loop in the SCEBIC model, the parasitic capacitance Cs will charge up and discharge as the pulsed electron beam interacts with the device, leading to a complete AC SCEBIC current path. This allows the pulsed current, ISCEBIC(t), to pass through and to be measured by the current meter between the unconnected p-region and ground. To understand the importance and the effect of the different parameters on the SCEBIC transient response, PSPICE simulations are carried out assuming typical values of each parameter.
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
145
Fig. 2. (a) Secondary electron (SE) image; (b) conventional EBIC image; and SCEBIC image (c) without and (d) with a metal enclosure (at 5.7 kHz electron beam modulation frequency) of a mc-Si solar cell. All the images are taken at 30 keV electron beam energy.
Fig. 3. (a) SCEBIC configuration; (b) its equivalent circuit diagram of a typical p–n junction solar cell; and (c) PSPICE simulation of the SCEBIC transient response ISCEBIC(t) at a modulation frequency of 200 Hz, where Ig ¼ 100 mA, Rsh ¼ 5 kΩ, Rs ¼ 1 Ω, Cj ¼ 200 nF, Cs ¼ 100 pF. The values assigned for each parameter are typical for solar cells with a sample size of about 1 cm2.
The resultant time-dependent current transient ISCEBIC(t) is depicted in Fig. 3c. Typically, when the electron beam is first turned on (at t ¼0), a negative current peak Imax (i.e. ISCEBIC(t ¼0)) is detected at the n-type contact of the sample as the generated
electrons and holes drift to the n- and p-type regions, respectively. The injection of electrons starts to charge up the parasitic capacitor Cs and at the same time drives the p–n junction to a forward bias. This bias lowers the potential barrier of the junction, resulting in a forward-bias current (also known as diffusion current) that increases with time, but in an opposite direction to the generation current Ig. As a result, the measured ISCEBIC(t) gradually decreases in magnitude and finally reaches zero when the forward-bias current and generation current are equal. From this moment onwards, ISCEBIC(t) does not vary as long as the beam is turned on. As the electron beam is turned off (at t ¼2.5 ms), only the positive forward-bias current due to carrier diffusion remains. Such a forward-bias current first peaks at the instant when the electron beam is switched off and slowly decays thereafter as the forward-bias current decreases due to the discharging of the parasitic capacitor. This gives rise to an increase in the potential barrier and at the same time a reduction of the excess electrons and holes in the p- and n-type regions, respectively. The diffusion current finally decreases to zero as a result of the p–n junction slowly returning to its thermal equilibrium state without the driving force of the accelerated electron beam. To fully understand the effect of each parameter (Rsh, Cj, Cs and Ig) on the current transient of SCEBIC ISCEBIC(t), PSPICE simulations were carried out. It is observed that Rsh mainly affects the shape of the decay as the decay rate becomes slower for larger Rsh values (Fig. 4a). Such a phenomenon is expected as the RC time constant increases with larger Rsh values [15]. This is different in the case of Rs, which shows minimal impact on the transient current curve (data not shown) as it has been reported that Rs has negligible effect under low current condition (i.e., Ig ¼100 mA) [17]. Interestingly, it is found that the maximum value of the SCEBIC transient Imax depends strongly on both values of Cs and Cj. While Imax is directly proportional to Cs (inset of Fig. 4b), increasing Cj leads to a smaller Imax (Fig. 4c). Furthermore, it can also be seen from Fig. 4c that the time taken for the SCEBIC current to reach the steady-state level of the current transient (i.e. zero value) is longer for larger Cj. This is expected since a
146
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
Fig. 4. SCEBIC transient characteristics of a typical single-junction solar cell using PSPICE simulations with the same model parameters as Fig. 3, by varying only one parameter each time: (a) shunt resistance Rsh; (b) parasitic capacitance Cs; (c) junction capacitance Cj; and (d) generation current Ig.
junction with a larger area would require a longer time to charge up. Lastly, as illustrated in Fig. 4d, it is also observed that Imax rises proportionally with the increasing Ig, and the ratio of Imax/Ig is approximately equal to the ratio of the parasitic and junction capacitances Cs/Cj in all our simulations. Such observation is consistent with the previous findings [12,18], where the ratio of Imax/Ig was estimated to be equal to Cs/(Cs þCj) and since in this work Cs is much smaller than Cj. For the case of the conventional SCEBIC without the metal enclosure, the Cs/Cj ratio is relatively small owing to the small Cs and large Cj values. Assuming a relatively low value of Cs of 10–20 pF, which is typical in our case given our sample size, a very small Imax (i.e. ISCEBIC(t) at t¼0) in the range of several nano-amperes is expected and such a value is close to the noise level of the circuit. This explains why a conventional SCEBIC setup cannot be directly applied in mc-Si solar cells. By increasing Cs to approximately 100 pF, a well-defined ISCEBIC(t) can be clearly obtained. With these two examples, the physical and electrical effects of the metal enclosure become apparent. When the metal enclosure is connected to the common ground and placed closely above the solar cell sample, it functions as a virtual SEM chamber with a much shorter distance between the floating p-region contact and ground. Cs is therefore much larger in this case, leading to an increased Cs/Cj ratio and thus a much larger and easily detectable Imax. In order to verify the accuracy of our SCEBIC model, ISCEBIC(t) from PSPICE simulations are carefully compared and fit into the actual experimentally measured current transients. Fig. 5 shows one cycle of the measured ISCEBIC(t) transient response of the sample to a pulsed electron beam at the energy of 30 keV with a modulation frequency of 60 Hz. The value of Cj extracted from the simulation is approximately 205 nF. It has been shown that Cj is around 70 nF/cm2 at zero bias for a typical p–n junction solar cell, and the value of Cj increases drastically under even a slightly
Fig. 5. Comparison of experimental (blue stars) and simulated (red-solid line) SCEBIC transient responses to a pulsed electron beam (black dash) at the electron beam energy of 30 keV and the modulation frequency of 60 Hz. The area of the sample is about 1 cm2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
forward bias [19]. Given our sample area of 1 cm2 and the relatively low forward bias due to the injection of electrons, the extracted value of 205 nF is within the expected range. As shown in the figure, the measured current transient curve closely matches with the simulated ISCEBIC(t) except for two noticeable differences: firstly, the steady-state of the induced current transient during beam ON does not return to zero value; and secondly, the positive current peak at the instant when the electron beam is just turned off is slightly smaller. The non-zero (approximately 10 nA) steady-state offset of the current transient during the beam ON time is due to the current from the electron beam itself flowing through the sample and adding to the SCEBIC signal. This also explains the exact match between the pulsed-beam current
L. Meng et al. / Solar Energy Materials & Solar Cells 133 (2015) 143–147
147
Acknowledgment Funding for this research under the Singapore National Research Foundation Grant # NRF2009EWT-CERP001-037 is acknowledged.
References
Fig. 6. SCEBIC image of the same mc-Si solar cell after removing the bottom contact. The image was taken with a metal enclosure at the electron beam energy of 30 keV, and modulation frequency of 5.7 kHz.
level and the offset of the steady-state (i.e. both at 10 nA). The slightly lower positive current peak (i.e. the maximum diffusion current) can be understood by other types of current loss, such as recombination of the excess minority carriers that regularly occurs in multicrystalline silicon materials [20], which contributes to the SCEBIC transient until thermal equilibrium is established after the beam is switched off. As mentioned previously, SCEBIC can show advantages over the conventional double-contact EBIC technique by offering a higher degree of flexibility in sample configuration. For example, since only one electrical contact is required by SCEBIC, it is possible to perform defect characterization or to examine the charge collection capabilities of a p–n junction in a solar cell without connecting both n- and p-type regions. In other words, SCEBIC can be useful for characterizing partially-processed solar cells with only one region of the junction accessible or analyzing samples with metallization present on only the n- or p-type region. This has been demonstrated by imaging the same sample shown in Fig. 2 with the bottom p-type contact removed. Again, it can be seen that the same defect features are nicely captured in the SCEBIC image (Fig. 6), demonstrating the feasibility and flexibility of SCEBIC on partially processed solar cells. 4. Conclusions A single contact electron beam induced current technique is first demonstrated in mc-Si solar cells. A lumped single-diode analytical model is also proposed to provide the theoretical explanations of the SCEBIC phenomenon in solar cells. The accuracy of the analytical model is verified as PSPICE simulations of the model show a close match with the experimental results. Our analytical model also explains the importance of the metal enclosure in increasing the parasitic capacitance, which is needed for achieving good SCEBIC signal strength. With the metal enclosure, SCEBIC imaging with accuracy and resolution comparable to the conventional double-contact EBIC can be readily established in mc-Si solar cells. In addition, the advantage of SCEBIC over the conventional EBIC is also demonstrated by characterizing a partially processed solar cell, which clearly illustrates the high degree of flexibility of SCEBIC in solar cell characterization.
[1] A. Cavallini, L. Polenta, A. Castaldini, Charge carrier recombination and generation analysis in materials and devices by electron and optical beam microscopy, Microelectron. Reliab. 50 (2010) 1398–1406. [2] H.J. Leamy, Charge collection scanning electron microscopy, J. Appl. Phys. 53 (1982) R51–R80. [3] D.S.H. Chan, J.C.H. Phang, W.S. Lau, V.K.S. Ong, V. Sane, S. Kolachina, New developments in beam induced current methods for the failure analysis of VLSI circuits, Microelectron. Eng. 31 (1996) 57–67. [4] D.S.H. Chan, J.C.H. Phang, J.M. Chin, S. Kolachina, Single contact beam induced current phenomena – a review, Beam Injection Assessment of Microstructures in Semiconductors, vol. 78–79, 2000, pp. 11–18. [5] J.C.H. Phang, D.S.H. Chan, V.K.S. Ong, S. Kolachina, J.M. Chin, M. Palaniiappan, G. Gilfeather, Y.X. Seah, Single contact beam induced current phenomenon for microelectronic failure analysis, Microelectron. Reliab. 43 (2003) 1595–1602. [6] K. Arafune, T. Sasaki, F. Wakabayashi, Y. Terada, Y. Ohshita, M. Yamaguchi, Study on defects and impurities in cast-grown polycrystalline silicon substrates for solar cells, Physica B 376–377 (2006) 236–239. [7] J. Chen, B. Chen, T. Sekiguchi, M. Fukuzawa, M. Yamada, Correlation between residual strain and electrically active grain boundaries in multicrystalline silicon, Appl. Phys. Lett. 93 (2008) 112105. [8] Z. Hameiri, T. Puzzer, L. Mai, A.B. Sproul, S.R. Wenham, Laser induced defects in laser doped solar cells, Prog. Photovolt.: Res. Appl. 19 (2011) 391–405. [9] J. Bauer, D. Lausch, H. Blumtritt, N. Zakharov, O. Breitenstein, Avalanche breakdown in multicrystalline solar cells due to preferred phosphorous diffusion at extended defects, Prog. Photovolt.: Res. Appl. (2012), http://dx. doi.org/10.1002/pip.2220. [10] D. Lausch, M. Werner, V. Naumann, J. Schneider, C. Hagendorf, Investigation of modified p–n junctions in crystalline silicon on glass solar cells, J. Appl. Phys. 109 (2011) 084513. [11] T. Sasaki, K. Arafune, W. Metzger, M.J. Romero, K. Jones, M. Al-Jassim, Y. Ohshita, M. Yamaguchi, Characterization of carrier recombination in lattice-mismatched InGaAs solar cells on GaAs substrates, Sol. Energy Mater. Sol. Cells 93 (2009) 936–940. [12] S. Kolachina, J.C.H. Phang, D.S.H. Chan, Single contact electron beam induced currents (SCEBIC) in semiconductor junctions. Part I: quantitative verification of SCEBIC model, Solid-State Electron. 42 (1998) 957–962. [13] L. Meng, D. Nagalingam, C.S. Bhatia, A.G. Street, J.C.H. Phang, Distinguishing morphological and electrical defects in polycrystalline silicon solar cells using scanning electron acoustic microscopy and electron beam induced current, Sol. Energy Mater. Sol. Cells 95 (2011) 2632–2637. [14] W. Kwapil, J. Nievendick, A. Zuschlag, P. Gundel, M.C. Schubert, W. Warta, Influence of surface texture on the defect-induced breakdown behavior of multicrystalline silicon solar cells, Prog. Photovolt.: Res. Appl. (2012), http: //dx.doi.org/10.1002/pip.1226. [15] A. Romanowski, D.B. Wittry, J.M. Tsaur, Analysis of the short-circuit current of a polycrystalline solar cell with excitation by a gated electron beam, J. Appl. Phys. 59 (1986) 951–957. [16] D.S.H. Chan, J.C.H. Phang, A method for the direct measurement of solar cell shunt resistance, IEEE Trans. Electron Devices 31 (1984) 381–383. [17] M. Wolf, H. Rauschenbach, Series resistance effects on solar cell measurements, Adv. Energy Convers. 3 (1963) 455–479. [18] E.I. Rau, A.V. Gostev, S. Zhu, D. Phang, D. Chan, D. Thong, W. Wong, Comparative analysis of scanning electron microscopy techniques for semiconductors: electron-beam-induced potential method, single-contact electron-beam-induced current method, and thermoacoustic detection, Russ. Microelectron. 30 (2001) 207–218. [19] F. Recart, A. Cuevas, Application of junction capacitance measurements to the characterization of solar cells, IEEE Trans. Electron Devices 53 (2006) 442–448. [20] A. Cuevas, F. Recart, Capacitive effects in quasi-steady-state voltage and lifetime measurements of silicon devices, J. Appl. Phys. 98 (2005) 074507.