3E.2 Intrinsic Reliability of Amorphous Silicon Thin ... - Semantic Scholar
Intrinsic Reliabilitty of Amorphous Silicon Thin Film Solar S Cells M. A. Alam, S. Donngaonkar, Karthik Y., S. Mahapatra, D. Wang*, M. Frei* Purdue Univversity, School of Electrical and Computer Engineeringg 465 Northw western Avenue, West Lafayette, Indiana 47907, USA * Applied Materials, Santa Clara, CA Phoone: 765-494-5988 E-mail: [email protected]
1.
/
Introduction and Motivation
Although solar cells made of single crystalline c silicon (c-Si) have long dominated the market for fo their efficiency and reliability, there has been a growing interest in exploring other (somewhat) less efficient materials (e.g., amorphous Si or α-Si:H, CIGs, CdTe) thhat offer improved price/kWh, through significantly lower manufacturing m costs. If the reliability challenges associated wiith these materials could be understood and addressed, the economic e viability of these PV options would improve dramaatically. e be extrinsic The reliability issues of α-Si PV could either (e.g., humidity, glass breakage, wiring faault)[1] or intrinsic (e.g., light induced degradation). In this article, we will discuss the physics and technological origgin of three intrinsic reliability concerns for α-Si solar cells: (i) shunt conduction, (ii) shadow degradation and hoot-spot generation, and (iii) light induced degradation. Oveer the years, these issues have been discussed in the literaturre by many groups, although the context of the discussion was w often isolated, and the approach frequently empirical. Here we take a comprehensive approach to show that thhese reliability issues are usually not independent, but can and should be understood in a broader framework. Moreeover, many of the reliability concerns of α-Si:H solar cells are a shared by other low-cost, thin-film PV materials (e.g. CIIGS, polymer) deposited by solution processing or chemicaal vapor deposition. Therefore, a good understanding of the reeliability issues for a-Si solar cells will also address broader questions regarding the reliability of other thin film technoologies. 2.
1
1 ;
is rate of generationn of electron-hole pairs (after where accounting for various opticall losses due to reflection and free carrier absorption), iss the series resistance, and n is the ideality factor. ( 1 for classical diode, while 2 for diodes dominateed by recombination in the space charge region). , the maximum m power delivered to For a variable load the load is given by (see Fig. 1b 1 for definitions)
; where the open circuit voltagee given by
2
(as shown in Fig. 1b) is
1
,
3
refllects the power lost to series and the fill factor resistance, and other efffects. (see Fig. 1b, bottom). Typically one would string 1000s of such diodes in series as shown in Fig. 2 to increase thee output voltage.
Background Information
2. 1 Basics of PV Operation The textbook description of the operationn of a solar cell is o as follows: The Sun radiates at 6000 C and a the part of the o ‘6000 C’ blackbody spectrum that is not absorbed by various molecules in the atmosphere and reaaches earth surface can generate electron-hole pairs within thee p-n junction of a solar cell (see Fig. 1a). If the charges cann be separated before they self-annihilate through recom mbination, the collected photo-induced carriers can deliverr power to an external load. Assuming superposition is sattisfied [2],
978-1-4244-5431-0/10/$26.00©2010 IEEE
3E.2.1
(a)
(b)
Fig. 1: (a) Classical diode as solar cell, and equivalent circuit of actual solar cell. (b) Schematic diagram showing light and dark IV characteristiics. Superposition is assumed. The red-dot signifies maximum m power point.
IRPS10-312
2.2 Basics of PV Reliability Regardless of the material used for thee solar cells, i.e., crystalline or amorphous, the reliability issues i can be analyzed using Eq. (1). For a given value off , Eq. (3) suggests that (and therefore the outpuut power) decreases with increase in the ideality factor n, ‘dark current’ pre-factor . A time dependent shift in thhese parameters as a consequence of operating conditions coonstitutes the reliability issues of a technology. Obviously,, the reliability issues that influence these parameters depend on the material type and the configuration of the solar ceell. As a result, the reliability concerns appear very differentt for various technologies; e.g., while light induced degraddation is an important reliability concern for a-Si:H solar cells, the issue is less important for c-Si solar cells. The gooal of this paper is to understand how various reliability issuees of a-Si:H affect Eq. (2) and how they relate to the commoon reliability issues for other solar cell technologies. 2.3 Thin Film Technology l area macroeThe fabrication and installation costs of large lectronic devices like solar cells can be minimized m only if they are processed with low energy input in high-efficiency equipment using inexpensive and durablle substrate to reduce the use of raw materials to the minimum. While crystalline silicon requires 100s of μm thicck active layer[3], thin film solar cells use only 100s of nm of o intrinsic material sandwiched between 10-20nm thick, heeavily doped p and n layers. Fluorinated Tin Oxide (SnO:F) or Indium Tin e while Al Oxide (ITO) is used as transparent front electrode, acts as back electrode. The cells are coonnected in series using laser scribes on TCO and a-Si:H laayers (see Fig. 2). This long string of series-connected cellss produces a module output of ~100V at ~1A. The qualityy and thickness of the ITO and Al determine the optical prooperties and series resistance of the system.
Fig. 2: Schematic diagram showing thee series connected solar cells supplying a load in an equivallent circuit picture (top) and the actual device layout schemaatic (bottom), with the arrows indicating the current paths. The economical use of materials in thin film technologies,
IRPS10-313
3E.2.2
like a-Si:H or CIGS, is not duue to any fundamental innovation, but rather related to the generic g wavelength dependent absorption in semiconductors. Typically, a-Si:H has larger bandgap (~1.72 eV) comparedd to c-Si (~1.1 eV); given that absorption in silicon scales with w density of states and is higher at shorter wavelengths, only a thin film of crystalline or amorphous silicon is neccessary to completely absorb (E > 1.72 eV) part of the solarr spectrum (discussed in more detail in Fig. 5). While the manufacturing m cost is greatly reduced by low-temperature vapor phase deposition of thin-films on inexpensive glaass substrate, there are some unique reliability concerns of thin film solar cells like light induced degradation, shunt leaakage, etc. 3.
Intrinsic Reliability Issu ues of a-Si:H Solar Cells
3.1 Shunt Leakage
Since the features of the darkk current (I-V characteristics without sunlight shining on) dictate (Eq. (3)), measurement of this characteristiccs offers insight into ultimate device efficiency. While darkk I-V of thin film PV at sufficiently large forward biases appears a purely classical (with 1 2 ), one of the most m interesting and universal feature is an anomalous leakage component at low forward A this feature has long and reverse bias conditions. Although been accounted for in equivallent circuit models with a parallel ohmic shunt resistance, (see Fig. 1(a), bottom), and has often been eliminatedd/reduced by so called ‘shunt busting’, the physical origin of o the phenomena and the universality of leakage across various v thin-film technologies have remained unclear [4]. c is defined by four chaWe find that this anomalous current racteristics; namely, c current-voltage symmetry ( ); voltage nonlinearity ( ~ ); temperature insensitivity ~ ), annd large statistical fluctuation ( in the leakage current magnittude from sample to sample. These observations cannot be explained by intrinsic device properties (e.g., defect distribbutions), but rather lead us to attribute this anomalous leakage to generalized space-charge limited currennt through localized metal-semiconductor-metal strucctures, and described by of ⁄ ~ , where ~1 2 (to be published). These localized structures shhould not be confused with pin-holes related to inadequaate deposition of Si, but are likely to arise from surface noon uniformity of TCO coated glass substrates. Recent mapping m of the localized light-spots by lock-in thermometry appears to support the conclusion[5]. Other expperiments involving metal-a-Si-metal resistive memoories are consistent with the hypothesis[6]. Indeed, this leeakage must be reduced to a level so that it does not degraade and becomes a reliability concern for thin film m technologies. Fortunately, however, shunt conductance may m be reduced by improved
surface planarization[7], blocking layerrs[8], shunt busting[4], etc. The choice of a specific appproach depends on application.
Initial experiments with “shadow stress” demonstrate that degradation may be describbed by a power law, i.e., Δ ~ with ~1/4 (see ( Fig. 4(b)), and the voltage acceleration is exponenttial (to be published). These two equations are sufficient too predict the effect of shadow degradation in large scale solaar cell installations. The problem of shadow deggradation might be addressed both at device as well as circuuit levels. At device level, the reduction of Zener voltage reduces r reverse stress on the affected cell and reduces defeect generation for a given duration of shadow. If cell redesign is difficult or expensive, circuit/system solutions are apppropriate.
Fig. 3: Dark IV showing anomalous leakaage currents in (a) a-Si:H p-i-n solar cells; (b) Organic BHJ cells c from [7]. 3.2 Shadow Degradation Shadow degradation is related to the reequirement that N solar cells are series connected in a module to achieve large . If the various structuraal elements of the assembly (e.g., antennas, solar paddles, booms, b etc. in satellite applications) or other natural elem ments (e.g. clouds, leaves, trees, dirt, etc. for terrestrial appplications) cast a shadow on one or several elements of thee solar module, the current in the affected cells is reduced. Current C continuity dictates the voltage be redistributed in suuch a way among the N-cells so that , where is the voltage across the load, is the voltage generated illuminated cells of the array across each of the and is the voltage developed across the shadowed part ( cells). The negative sign of reflects the fact that the affected cells are reversed biased in thhe Zener tunneling or Avalanche breakdown mode (see Fig 4(a)). 4 Indeed, shadowing not only eliminates the cells from m power generation, but dramatically degrades the power output from the rest of the system by reducing the output voltagee. One may wonder if shadowing, like radiation induced soft errors in CMOS circuits, is a transient effect, and if the system might be restored to its pristine performance once the shadow is lifted (e.g., with change in orientation of the satellite with respect to the Sun, or cleaaning of the solar panels). Indeed, most papers in the literatture treat shadowing as a power management problem[9], with w no discussion of long term consequences. The large revverse bias endured by the shadowed α-Si:H cells however wiill depassivate SiH and/or weak Si-Si bonds, create mid-gaap defects, reduce recombination lifetime , and increase . The corresponding shift in reflects the integratted duration of the shadowing (and the corresponding stresss) endured by the cells. The change in the maximum poweer point would reduce power output of the system. Details of the system level effects will be discussed elsewhere.
3E.2.3
(aa)
(bb) Fig. 4: (a) Schematics showinng the operating points before and after shading; identifyinng voltages across individual cells (V1), reverse voltage devveloped across shaded device (VS), and the corresponding load l voltage (VL). (b) Power law time dependent degradatioon of solar cells under different reverse bias stresses that has been scaled by constant factors to highlight the power--exponent of degradation. Recall that for SRAM memorry, use of redundant arrays in post-Si phase dramatically im mproves yield by allowing remapping of defective cells (due to process related f leading to READ or fluctuation or fluctuation in for WRITE failures)[10]. Similar approaches may be appropriate for PV modules as well[11]. For example, a significant
IRPS10-314
reduction in (see Eq. (3)) would indicate onset of shadowing; a sweep across the elements of the module would determine the affected row; and a standby redundant array would then be switched in to bypass the affected cell. Obviously, once the shadow is lifted, the same redundant cell can be released and can be reconfigured such that they can be used by other shadowed cells. Simulations show that small degree of redundancy and management overhead result in large improvements [11]. In some cases, rapid defect generation in cells affected by intense shadows may lead to catastrophic failure of solar cells by a process called the “hot-spot” formation[12]. Hot spots may delaminate the mirrors and destroy the cells. Like ‘hard breakdown’ in thick gate dielectrics, hot-spot formation and propagation is likely to involve a positive feedback and interplay between temperature and current[13, 14]. Like the early literature on gate dielectric breakdown, there is considerable debate regarding whether the hot-spot is related to pre-existing defects like etch pits as discussed in Section 3.1 (thermal images seems to indicate such possibility) or intrinsic defect generation terminated by a runaway process. Further work is needed to identify the mechanism of hot-spot generation.
Si/SiO2 interface of c-Silicon solar cells (See Fig. 5). These low energy photons ~1 that reach the back-interface cannot dissociate SiH bonds. In contrast, the distance over which the high energy photons are absorbed in a-Si solar cells are replete with SiH bonds, and therefore it is hardly surprising that Si-H bonds dissociation is a concern. In short, film thickness of solar cells, bandgap of the PV material and the light-induced degradation are related by fundamental consideration and must be understood within this comprehensive framework. It is important to note that LID is not absent in c-Si, but involves dissociation processes (e.g. of B-H complexes) that can be initiated by lower energy photons.
3.3 Light Induced Degradation Light induced degradation (LID) involves time dependent reduction in output current of a solar cell, under solar illumination. Since the early definitive studies of LID in 1977s by Staebler and Wronski[15], this notorious reliability issue has been studied in depth; and several features are known thereof; (i) the degradation follows a power law of the form with ~1/3 as the power exponent [16] ∆ (see Fig. 6a), (ii) there is evidence of some recovery once light is removed, (iii) the degradation increases with light intensity, (iv) many studies suggest dissociation of Si-H complexes during light soaking, (v) degradation is reduced if H is replaced by isotope Deuterium, and finally, (vi) and most intriguingly, thin-films are susceptible to it, while the effect, although present, is considerably suppressed in crystalline silicon solar cells[17]. The last three observations appear to be contradictory; as the backplane of most c-Si solar cells involve large Si-SiO2 interface passivated by SiH bonds, otherwise minority carand decrease and rier recombination will increase hence the efficiency. If light induced dissociation of SiH bonds creates defects in thin -Si:H solar cells, why isn’t LID degradation a similar concern for c-Si solar cells? After all, in c-Si cells, light bounces many times between the electrodes for full absorption[3]. To resolve this puzzle, recall that absorption length of high energy photons are relatively small (~10s of nm) and it is low-energy, near gap, radiation that reaches the bottom
IRPS10-315
3E.2.4
Fig. 5: Surface plot showing absorbance of crystalline silicon vs. thickness ( ) and photon energy (eV). There have been several models proposed to explain the time dependent kinetics of LID issue. Some involve time dependent recombination of electron-hole pairs breaking weak Si-Si bonds[16], while others have explicitly attributed the defect generation to light induced dissociation of SiH bonds[18, 19]. LID continues to be an active topic of research, therefore more work is needed before a robust theory is developed. If the time-exponent of LID is robust, as several group have reported, a reaction-diffusion based model might be appropriate. Below we offer a simple derivation and numerical solution of the concept. Detailed prediction of the model needs to be explored by experiments. Let us assume that the small wavelength photons of flux have sufficient energy to dissociate the weak Si-H ) at the rate of . The bonds (initial concentration remain fixed in space to act as broken dangling bonds recombination centers, while the freshly released atomic H diffuse laterally through the film. These two populations of mobile H and static dangling bonds interact throughout the volume of the thin film, with opportunities for repassiva). In sum, therefore, we can write the tion (rate constant
rate of defect creation as .
4
In addition to repassivation, H can react with each other and be lost from the kinetics of the problem by forming . Therefore the evolution of H is given by molecular .
5
The coupled equations can be solved numerically (Fig. 6), although the following analytical solution provides additional insight. After an initial transient, the rate of change of the system can be presumed to be slow, compared to the fluxes sustaining them. Therefore, the forward dissociation and reverse repassivation of the Si-H bonds are evenly balanced as ~ ; 6 ⁄ ~ and together, we find
/
so that
3
/ ~
. Taken
.
7
Reassuringly, this theoretical result compares well with the experimental results from literature, as shown in Fig. 6. The scaling of degradation rate (observation 2 above) as a function of flux G (for concentrator PV applications, for
and has been noted by many groups. This should motivate a NBTI like thorough study of solar cell degradation, with the full understanding that in addition to SiH bond-dissociation, other bond dissociation processes (e.g. weak Si-Si bonds, B-O complexes) may also contribute to LID. 4.
Conclusion
In this paper, we have discussed three intrinsic reliability issues of thin-film -Si:H solar cells; space charge limited shunt conduction through localized metal-semiconductor-metal structures; shadow degradation in series connected cells in a module, and light induced degradation. Despite their distinct external manifestation, these intrinsic reliability issues appear to share common physical phenomena. For example, the light induced and the shadow degradation may be related because they are described by very similar time-exponents (see Fig. 4c and 6a). While the physics of G are different (e.g. photon induced dissociation for LID and (possibly) electron-hole recombination induced dissociation for shadow degradation), it is likely that they both break SiH bonds and are subsequently follow similar diffusive kinetics. Finally, analogies to CMOS reliability; e.g., shunt conduction related to non uniform conduction through oxides, shadow degradation to bulk defect generation and TDDB in gate dielectric, and light induced degradation to NBTI in PMOS transistors; may help illuminate many aspects of the degradation processes.
Acknowledgements Fig. 5 is taken from unpublished work by Mohammad Ryyan Khan, a graduate student in Prof. Alam’s group. We gratefully acknowledge Applied Materials for samples, the Birck Nanotechnology Center for characterization facilities, and the Network of Computational Nanotechnology for computational resources. The work is supported by grants from Applied Materials and Columbia EFRC.
References [1]
Fig. 6: The numerical solution of reaction diffusion equations anticipates the ⁄ time dependence trends (right -3 axis in a.u.) of measured DB density (left axis in cm ) from [18, 19, 21]. The theoretical predictions need to be confirmed with additional experiments. example) is confirmed and once the Sun is down, relaxation is expected. The analogy of LID and NBTI is obvious[20],
3E.2.5
[2]
[3]
C. R. Osterwald and T. J. McMahon, "History of Accelerated and Qualification Testing of Terrestrial Photovoltaic Modules: A Literature Review," PROGRESS IN PHOTOVOLTAICS, vol. 17, pp. 11-33, Jan 2009. F. A. Lindholm, J. G. Fossum, and E. L. Burgess, "Application of the Superposition Principle to Solar-Cell Analysis," IEEE Transactions on Electron Devices, vol. 26, pp. 165-171, 1979. M. A. Green, J. H. Zhao, A. H. Wang, and S. R. Wenham, "Very high efficiency silicon solar cells -
IRPS10-316
[4]
[5]
[6]
[7]
[8]
[9] [10]
[11]
[12]
[13]
[14]
IRPS10-317
Science and technology," IEEE Transactions on Electron Devices, vol. 46, pp. 1940-1947, Oct 1999. T. R. Johnson, G. Ganguly, G. S. Wood, and D. E. Carlson, "Investigation of the causes and variation of leakage currents in amorphous silicon p-i-n diodes," Warrendale, PA, USA, 2003, pp. 381-6. O. Kunz, J. Wong, J. Janssens, J. Bauer, O. Breitenstein, and A. G. Aberle, "Shunting Problems Due to Sub-Micron Pinholes in Evaporated Solid-Phase Crystallised Poly-Si Thin-Film Solar Cells on Glass," Progress In Photovoltaics, vol. 17, pp. 35-46, Jan 2009. J. W. Seo, S. J. Baik, S. J. Kang, Y. H. Hong, J.-H. Yang, L. Fang, and K. S. Lim, "Evidence of Al induced conducting filament formation in Al/amorphous silicon/Al resistive switching memory device," Applied Physics Letters, vol. 96, pp. 053504-3. M. D. Irwin, J. Liu, B. J. Leever, J. D. Servaites, M. C. Hersam, M. F. Durstock, and T. J. Marks, "Consequences of Anode Interfacial Layer Deletion. HCl-Treated ITO in P3HT:PCBM-Based Bulk-Heterojunction Organic Photovoltaic Devices," Langmuir, 2009. J. D. Hwang and C. H. Chou, "On the origin of leakage current reduction in TiO[sub 2] passivated porosus silicon Schottky-barrier diode," Applied Physics Letters, vol. 96, pp. 063503-3. J. Feldman, S. Singer, and A. Braunstein, "Solar-Cell Interconnections and the Shadow Problem," Solar Energy, vol. 26, pp. 419-428, 1981. A. Agarwal, B. C. Paul, H. Mahmoodi, A. Datta, and K. Roy, "A process-tolerant cache architecture for improved yield in nanoscale technologies," IEEE Transactions on Very Large Scale Integration (Vlsi) Systems, vol. 13, pp. 27-38, Jan 2005. D. Nguyen and B. Lehman, "A reconfigurable solar photovoltaic array under shadow conditions," Apec 2008: Twenty-Third Annual Ieee Applied Power Electronics Conference and Exposition, Vols 1-4, pp. 980-986, 2008. J. Bauer, J. M. Wagner, A. Lotnyk, H. Blumtritt, B. Lim, J. Schmidt, and O. Breitenstein, "Hot spots in multicrystalline silicon solar cells: avalanche breakdown due to etch pits," Physica Status Solidi-Rapid Research Letters, vol. 3, pp. 40-42, Mar 2009. M. A. Alam, B. E. Weir, and P. J. Silverman, "A study of soft and hard breakdown - Part I: Analysis of statistical percolation conductance," IEEE Transactions on Electron Devices, vol. 49, pp. 232-238, Feb 2002. M. A. Alam, B. E. Weir, and P. J. Silverman, "A study of soft and hard breakdown - Part II: Principles of area, thickness, and voltage scaling," IEEE Transactions on Electron Devices, vol. 49, pp. 239-246, Feb 2002.
3E.2.6
[15]
[16]
[17] [18]
[19]
[20]
[21]
D. L. Staebler and C. R. Wronski, "Reversible Conductivity Changes in Discharge-Produced Amorphous Si," Applied Physics Letters, vol. 31, pp. 292-294, 1977. M. Stutzmann, W. B. Jackson, and C. C. Tsai, "Light-Induced Metastable Defects in Hydrogenated Amorphous-Silicon - a Systematic Study," Physical Review B, vol. 32, pp. 23-47, 1985. J. Schmidt, "Light-Induced Degradation in Crystalline Silicon Solar Cells," Brandenburg, Germany, 2004, pp. 187-196. H. R. Park, J. Z. Liu, and S. Wagner, "Saturation of the Light-Induced Defect Density in Hydrogenated Amorphous-Silicon," Applied Physics Letters, vol. 55, pp. 2658-2660, Dec 18 1989. C. Godet, "Metastable hydrogen atom trapping in hydrogenated amorphous silicon films: A microscopic model for metastable defect creation," Philosophical Magazine B-Physics of Condensed Matter Statistical Mechanics Electronic Optical and Magnetic Properties, vol. 77, pp. 765-777, Mar 1998. M. A. Alam, "A critical examination of the mechanics of dynamic NBTI for PMOSFETs," 2003 IEEE International Electron Devices Meeting, Technical Digest, pp. 345-348, 2003. T. Shimizu, "Staebler-Wronski effect in hydrogenated amorphous silicon and related alloy films," Japanese Journal of Applied Physics Part 1-Regular Papers Brief Communications & Review Papers, vol. 43, pp. 3257-3268, Jun 2004.
1.
/
Introduction and Motivation
Although solar cells made of single crystalline c silicon (c-Si) have long dominated the market for fo their efficiency and reliability, there has been a growing interest in exploring other (somewhat) less efficient materials (e.g., amorphous Si or α-Si:H, CIGs, CdTe) thhat offer improved price/kWh, through significantly lower manufacturing m costs. If the reliability challenges associated wiith these materials could be understood and addressed, the economic e viability of these PV options would improve dramaatically. e be extrinsic The reliability issues of α-Si PV could either (e.g., humidity, glass breakage, wiring faault)[1] or intrinsic (e.g., light induced degradation). In this article, we will discuss the physics and technological origgin of three intrinsic reliability concerns for α-Si solar cells: (i) shunt conduction, (ii) shadow degradation and hoot-spot generation, and (iii) light induced degradation. Oveer the years, these issues have been discussed in the literaturre by many groups, although the context of the discussion was w often isolated, and the approach frequently empirical. Here we take a comprehensive approach to show that thhese reliability issues are usually not independent, but can and should be understood in a broader framework. Moreeover, many of the reliability concerns of α-Si:H solar cells are a shared by other low-cost, thin-film PV materials (e.g. CIIGS, polymer) deposited by solution processing or chemicaal vapor deposition. Therefore, a good understanding of the reeliability issues for a-Si solar cells will also address broader questions regarding the reliability of other thin film technoologies. 2.
1
1 ;
is rate of generationn of electron-hole pairs (after where accounting for various opticall losses due to reflection and free carrier absorption), iss the series resistance, and n is the ideality factor. ( 1 for classical diode, while 2 for diodes dominateed by recombination in the space charge region). , the maximum m power delivered to For a variable load the load is given by (see Fig. 1b 1 for definitions)
; where the open circuit voltagee given by
2
(as shown in Fig. 1b) is
1
,
3
refllects the power lost to series and the fill factor resistance, and other efffects. (see Fig. 1b, bottom). Typically one would string 1000s of such diodes in series as shown in Fig. 2 to increase thee output voltage.
Background Information
2. 1 Basics of PV Operation The textbook description of the operationn of a solar cell is o as follows: The Sun radiates at 6000 C and a the part of the o ‘6000 C’ blackbody spectrum that is not absorbed by various molecules in the atmosphere and reaaches earth surface can generate electron-hole pairs within thee p-n junction of a solar cell (see Fig. 1a). If the charges cann be separated before they self-annihilate through recom mbination, the collected photo-induced carriers can deliverr power to an external load. Assuming superposition is sattisfied [2],
978-1-4244-5431-0/10/$26.00©2010 IEEE
3E.2.1
(a)
(b)
Fig. 1: (a) Classical diode as solar cell, and equivalent circuit of actual solar cell. (b) Schematic diagram showing light and dark IV characteristiics. Superposition is assumed. The red-dot signifies maximum m power point.
IRPS10-312
2.2 Basics of PV Reliability Regardless of the material used for thee solar cells, i.e., crystalline or amorphous, the reliability issues i can be analyzed using Eq. (1). For a given value off , Eq. (3) suggests that (and therefore the outpuut power) decreases with increase in the ideality factor n, ‘dark current’ pre-factor . A time dependent shift in thhese parameters as a consequence of operating conditions coonstitutes the reliability issues of a technology. Obviously,, the reliability issues that influence these parameters depend on the material type and the configuration of the solar ceell. As a result, the reliability concerns appear very differentt for various technologies; e.g., while light induced degraddation is an important reliability concern for a-Si:H solar cells, the issue is less important for c-Si solar cells. The gooal of this paper is to understand how various reliability issuees of a-Si:H affect Eq. (2) and how they relate to the commoon reliability issues for other solar cell technologies. 2.3 Thin Film Technology l area macroeThe fabrication and installation costs of large lectronic devices like solar cells can be minimized m only if they are processed with low energy input in high-efficiency equipment using inexpensive and durablle substrate to reduce the use of raw materials to the minimum. While crystalline silicon requires 100s of μm thicck active layer[3], thin film solar cells use only 100s of nm of o intrinsic material sandwiched between 10-20nm thick, heeavily doped p and n layers. Fluorinated Tin Oxide (SnO:F) or Indium Tin e while Al Oxide (ITO) is used as transparent front electrode, acts as back electrode. The cells are coonnected in series using laser scribes on TCO and a-Si:H laayers (see Fig. 2). This long string of series-connected cellss produces a module output of ~100V at ~1A. The qualityy and thickness of the ITO and Al determine the optical prooperties and series resistance of the system.
Fig. 2: Schematic diagram showing thee series connected solar cells supplying a load in an equivallent circuit picture (top) and the actual device layout schemaatic (bottom), with the arrows indicating the current paths. The economical use of materials in thin film technologies,
IRPS10-313
3E.2.2
like a-Si:H or CIGS, is not duue to any fundamental innovation, but rather related to the generic g wavelength dependent absorption in semiconductors. Typically, a-Si:H has larger bandgap (~1.72 eV) comparedd to c-Si (~1.1 eV); given that absorption in silicon scales with w density of states and is higher at shorter wavelengths, only a thin film of crystalline or amorphous silicon is neccessary to completely absorb (E > 1.72 eV) part of the solarr spectrum (discussed in more detail in Fig. 5). While the manufacturing m cost is greatly reduced by low-temperature vapor phase deposition of thin-films on inexpensive glaass substrate, there are some unique reliability concerns of thin film solar cells like light induced degradation, shunt leaakage, etc. 3.
Intrinsic Reliability Issu ues of a-Si:H Solar Cells
3.1 Shunt Leakage
Since the features of the darkk current (I-V characteristics without sunlight shining on) dictate (Eq. (3)), measurement of this characteristiccs offers insight into ultimate device efficiency. While darkk I-V of thin film PV at sufficiently large forward biases appears a purely classical (with 1 2 ), one of the most m interesting and universal feature is an anomalous leakage component at low forward A this feature has long and reverse bias conditions. Although been accounted for in equivallent circuit models with a parallel ohmic shunt resistance, (see Fig. 1(a), bottom), and has often been eliminatedd/reduced by so called ‘shunt busting’, the physical origin of o the phenomena and the universality of leakage across various v thin-film technologies have remained unclear [4]. c is defined by four chaWe find that this anomalous current racteristics; namely, c current-voltage symmetry ( ); voltage nonlinearity ( ~ ); temperature insensitivity ~ ), annd large statistical fluctuation ( in the leakage current magnittude from sample to sample. These observations cannot be explained by intrinsic device properties (e.g., defect distribbutions), but rather lead us to attribute this anomalous leakage to generalized space-charge limited currennt through localized metal-semiconductor-metal strucctures, and described by of ⁄ ~ , where ~1 2 (to be published). These localized structures shhould not be confused with pin-holes related to inadequaate deposition of Si, but are likely to arise from surface noon uniformity of TCO coated glass substrates. Recent mapping m of the localized light-spots by lock-in thermometry appears to support the conclusion[5]. Other expperiments involving metal-a-Si-metal resistive memoories are consistent with the hypothesis[6]. Indeed, this leeakage must be reduced to a level so that it does not degraade and becomes a reliability concern for thin film m technologies. Fortunately, however, shunt conductance may m be reduced by improved
surface planarization[7], blocking layerrs[8], shunt busting[4], etc. The choice of a specific appproach depends on application.
Initial experiments with “shadow stress” demonstrate that degradation may be describbed by a power law, i.e., Δ ~ with ~1/4 (see ( Fig. 4(b)), and the voltage acceleration is exponenttial (to be published). These two equations are sufficient too predict the effect of shadow degradation in large scale solaar cell installations. The problem of shadow deggradation might be addressed both at device as well as circuuit levels. At device level, the reduction of Zener voltage reduces r reverse stress on the affected cell and reduces defeect generation for a given duration of shadow. If cell redesign is difficult or expensive, circuit/system solutions are apppropriate.
Fig. 3: Dark IV showing anomalous leakaage currents in (a) a-Si:H p-i-n solar cells; (b) Organic BHJ cells c from [7]. 3.2 Shadow Degradation Shadow degradation is related to the reequirement that N solar cells are series connected in a module to achieve large . If the various structuraal elements of the assembly (e.g., antennas, solar paddles, booms, b etc. in satellite applications) or other natural elem ments (e.g. clouds, leaves, trees, dirt, etc. for terrestrial appplications) cast a shadow on one or several elements of thee solar module, the current in the affected cells is reduced. Current C continuity dictates the voltage be redistributed in suuch a way among the N-cells so that , where is the voltage across the load, is the voltage generated illuminated cells of the array across each of the and is the voltage developed across the shadowed part ( cells). The negative sign of reflects the fact that the affected cells are reversed biased in thhe Zener tunneling or Avalanche breakdown mode (see Fig 4(a)). 4 Indeed, shadowing not only eliminates the cells from m power generation, but dramatically degrades the power output from the rest of the system by reducing the output voltagee. One may wonder if shadowing, like radiation induced soft errors in CMOS circuits, is a transient effect, and if the system might be restored to its pristine performance once the shadow is lifted (e.g., with change in orientation of the satellite with respect to the Sun, or cleaaning of the solar panels). Indeed, most papers in the literatture treat shadowing as a power management problem[9], with w no discussion of long term consequences. The large revverse bias endured by the shadowed α-Si:H cells however wiill depassivate SiH and/or weak Si-Si bonds, create mid-gaap defects, reduce recombination lifetime , and increase . The corresponding shift in reflects the integratted duration of the shadowing (and the corresponding stresss) endured by the cells. The change in the maximum poweer point would reduce power output of the system. Details of the system level effects will be discussed elsewhere.
3E.2.3
(aa)
(bb) Fig. 4: (a) Schematics showinng the operating points before and after shading; identifyinng voltages across individual cells (V1), reverse voltage devveloped across shaded device (VS), and the corresponding load l voltage (VL). (b) Power law time dependent degradatioon of solar cells under different reverse bias stresses that has been scaled by constant factors to highlight the power--exponent of degradation. Recall that for SRAM memorry, use of redundant arrays in post-Si phase dramatically im mproves yield by allowing remapping of defective cells (due to process related f leading to READ or fluctuation or fluctuation in for WRITE failures)[10]. Similar approaches may be appropriate for PV modules as well[11]. For example, a significant
IRPS10-314
reduction in (see Eq. (3)) would indicate onset of shadowing; a sweep across the elements of the module would determine the affected row; and a standby redundant array would then be switched in to bypass the affected cell. Obviously, once the shadow is lifted, the same redundant cell can be released and can be reconfigured such that they can be used by other shadowed cells. Simulations show that small degree of redundancy and management overhead result in large improvements [11]. In some cases, rapid defect generation in cells affected by intense shadows may lead to catastrophic failure of solar cells by a process called the “hot-spot” formation[12]. Hot spots may delaminate the mirrors and destroy the cells. Like ‘hard breakdown’ in thick gate dielectrics, hot-spot formation and propagation is likely to involve a positive feedback and interplay between temperature and current[13, 14]. Like the early literature on gate dielectric breakdown, there is considerable debate regarding whether the hot-spot is related to pre-existing defects like etch pits as discussed in Section 3.1 (thermal images seems to indicate such possibility) or intrinsic defect generation terminated by a runaway process. Further work is needed to identify the mechanism of hot-spot generation.
Si/SiO2 interface of c-Silicon solar cells (See Fig. 5). These low energy photons ~1 that reach the back-interface cannot dissociate SiH bonds. In contrast, the distance over which the high energy photons are absorbed in a-Si solar cells are replete with SiH bonds, and therefore it is hardly surprising that Si-H bonds dissociation is a concern. In short, film thickness of solar cells, bandgap of the PV material and the light-induced degradation are related by fundamental consideration and must be understood within this comprehensive framework. It is important to note that LID is not absent in c-Si, but involves dissociation processes (e.g. of B-H complexes) that can be initiated by lower energy photons.
3.3 Light Induced Degradation Light induced degradation (LID) involves time dependent reduction in output current of a solar cell, under solar illumination. Since the early definitive studies of LID in 1977s by Staebler and Wronski[15], this notorious reliability issue has been studied in depth; and several features are known thereof; (i) the degradation follows a power law of the form with ~1/3 as the power exponent [16] ∆ (see Fig. 6a), (ii) there is evidence of some recovery once light is removed, (iii) the degradation increases with light intensity, (iv) many studies suggest dissociation of Si-H complexes during light soaking, (v) degradation is reduced if H is replaced by isotope Deuterium, and finally, (vi) and most intriguingly, thin-films are susceptible to it, while the effect, although present, is considerably suppressed in crystalline silicon solar cells[17]. The last three observations appear to be contradictory; as the backplane of most c-Si solar cells involve large Si-SiO2 interface passivated by SiH bonds, otherwise minority carand decrease and rier recombination will increase hence the efficiency. If light induced dissociation of SiH bonds creates defects in thin -Si:H solar cells, why isn’t LID degradation a similar concern for c-Si solar cells? After all, in c-Si cells, light bounces many times between the electrodes for full absorption[3]. To resolve this puzzle, recall that absorption length of high energy photons are relatively small (~10s of nm) and it is low-energy, near gap, radiation that reaches the bottom
IRPS10-315
3E.2.4
Fig. 5: Surface plot showing absorbance of crystalline silicon vs. thickness ( ) and photon energy (eV). There have been several models proposed to explain the time dependent kinetics of LID issue. Some involve time dependent recombination of electron-hole pairs breaking weak Si-Si bonds[16], while others have explicitly attributed the defect generation to light induced dissociation of SiH bonds[18, 19]. LID continues to be an active topic of research, therefore more work is needed before a robust theory is developed. If the time-exponent of LID is robust, as several group have reported, a reaction-diffusion based model might be appropriate. Below we offer a simple derivation and numerical solution of the concept. Detailed prediction of the model needs to be explored by experiments. Let us assume that the small wavelength photons of flux have sufficient energy to dissociate the weak Si-H ) at the rate of . The bonds (initial concentration remain fixed in space to act as broken dangling bonds recombination centers, while the freshly released atomic H diffuse laterally through the film. These two populations of mobile H and static dangling bonds interact throughout the volume of the thin film, with opportunities for repassiva). In sum, therefore, we can write the tion (rate constant
rate of defect creation as .
4
In addition to repassivation, H can react with each other and be lost from the kinetics of the problem by forming . Therefore the evolution of H is given by molecular .
5
The coupled equations can be solved numerically (Fig. 6), although the following analytical solution provides additional insight. After an initial transient, the rate of change of the system can be presumed to be slow, compared to the fluxes sustaining them. Therefore, the forward dissociation and reverse repassivation of the Si-H bonds are evenly balanced as ~ ; 6 ⁄ ~ and together, we find
/
so that
3
/ ~
. Taken
.
7
Reassuringly, this theoretical result compares well with the experimental results from literature, as shown in Fig. 6. The scaling of degradation rate (observation 2 above) as a function of flux G (for concentrator PV applications, for
and has been noted by many groups. This should motivate a NBTI like thorough study of solar cell degradation, with the full understanding that in addition to SiH bond-dissociation, other bond dissociation processes (e.g. weak Si-Si bonds, B-O complexes) may also contribute to LID. 4.
Conclusion
In this paper, we have discussed three intrinsic reliability issues of thin-film -Si:H solar cells; space charge limited shunt conduction through localized metal-semiconductor-metal structures; shadow degradation in series connected cells in a module, and light induced degradation. Despite their distinct external manifestation, these intrinsic reliability issues appear to share common physical phenomena. For example, the light induced and the shadow degradation may be related because they are described by very similar time-exponents (see Fig. 4c and 6a). While the physics of G are different (e.g. photon induced dissociation for LID and (possibly) electron-hole recombination induced dissociation for shadow degradation), it is likely that they both break SiH bonds and are subsequently follow similar diffusive kinetics. Finally, analogies to CMOS reliability; e.g., shunt conduction related to non uniform conduction through oxides, shadow degradation to bulk defect generation and TDDB in gate dielectric, and light induced degradation to NBTI in PMOS transistors; may help illuminate many aspects of the degradation processes.
Acknowledgements Fig. 5 is taken from unpublished work by Mohammad Ryyan Khan, a graduate student in Prof. Alam’s group. We gratefully acknowledge Applied Materials for samples, the Birck Nanotechnology Center for characterization facilities, and the Network of Computational Nanotechnology for computational resources. The work is supported by grants from Applied Materials and Columbia EFRC.
References [1]
Fig. 6: The numerical solution of reaction diffusion equations anticipates the ⁄ time dependence trends (right -3 axis in a.u.) of measured DB density (left axis in cm ) from [18, 19, 21]. The theoretical predictions need to be confirmed with additional experiments. example) is confirmed and once the Sun is down, relaxation is expected. The analogy of LID and NBTI is obvious[20],
3E.2.5
[2]
[3]
C. R. Osterwald and T. J. McMahon, "History of Accelerated and Qualification Testing of Terrestrial Photovoltaic Modules: A Literature Review," PROGRESS IN PHOTOVOLTAICS, vol. 17, pp. 11-33, Jan 2009. F. A. Lindholm, J. G. Fossum, and E. L. Burgess, "Application of the Superposition Principle to Solar-Cell Analysis," IEEE Transactions on Electron Devices, vol. 26, pp. 165-171, 1979. M. A. Green, J. H. Zhao, A. H. Wang, and S. R. Wenham, "Very high efficiency silicon solar cells -
IRPS10-316
[4]
[5]
[6]
[7]
[8]
[9] [10]
[11]
[12]
[13]
[14]
IRPS10-317
Science and technology," IEEE Transactions on Electron Devices, vol. 46, pp. 1940-1947, Oct 1999. T. R. Johnson, G. Ganguly, G. S. Wood, and D. E. Carlson, "Investigation of the causes and variation of leakage currents in amorphous silicon p-i-n diodes," Warrendale, PA, USA, 2003, pp. 381-6. O. Kunz, J. Wong, J. Janssens, J. Bauer, O. Breitenstein, and A. G. Aberle, "Shunting Problems Due to Sub-Micron Pinholes in Evaporated Solid-Phase Crystallised Poly-Si Thin-Film Solar Cells on Glass," Progress In Photovoltaics, vol. 17, pp. 35-46, Jan 2009. J. W. Seo, S. J. Baik, S. J. Kang, Y. H. Hong, J.-H. Yang, L. Fang, and K. S. Lim, "Evidence of Al induced conducting filament formation in Al/amorphous silicon/Al resistive switching memory device," Applied Physics Letters, vol. 96, pp. 053504-3. M. D. Irwin, J. Liu, B. J. Leever, J. D. Servaites, M. C. Hersam, M. F. Durstock, and T. J. Marks, "Consequences of Anode Interfacial Layer Deletion. HCl-Treated ITO in P3HT:PCBM-Based Bulk-Heterojunction Organic Photovoltaic Devices," Langmuir, 2009. J. D. Hwang and C. H. Chou, "On the origin of leakage current reduction in TiO[sub 2] passivated porosus silicon Schottky-barrier diode," Applied Physics Letters, vol. 96, pp. 063503-3. J. Feldman, S. Singer, and A. Braunstein, "Solar-Cell Interconnections and the Shadow Problem," Solar Energy, vol. 26, pp. 419-428, 1981. A. Agarwal, B. C. Paul, H. Mahmoodi, A. Datta, and K. Roy, "A process-tolerant cache architecture for improved yield in nanoscale technologies," IEEE Transactions on Very Large Scale Integration (Vlsi) Systems, vol. 13, pp. 27-38, Jan 2005. D. Nguyen and B. Lehman, "A reconfigurable solar photovoltaic array under shadow conditions," Apec 2008: Twenty-Third Annual Ieee Applied Power Electronics Conference and Exposition, Vols 1-4, pp. 980-986, 2008. J. Bauer, J. M. Wagner, A. Lotnyk, H. Blumtritt, B. Lim, J. Schmidt, and O. Breitenstein, "Hot spots in multicrystalline silicon solar cells: avalanche breakdown due to etch pits," Physica Status Solidi-Rapid Research Letters, vol. 3, pp. 40-42, Mar 2009. M. A. Alam, B. E. Weir, and P. J. Silverman, "A study of soft and hard breakdown - Part I: Analysis of statistical percolation conductance," IEEE Transactions on Electron Devices, vol. 49, pp. 232-238, Feb 2002. M. A. Alam, B. E. Weir, and P. J. Silverman, "A study of soft and hard breakdown - Part II: Principles of area, thickness, and voltage scaling," IEEE Transactions on Electron Devices, vol. 49, pp. 239-246, Feb 2002.
3E.2.6
[15]
[16]
[17] [18]
[19]
[20]
[21]
D. L. Staebler and C. R. Wronski, "Reversible Conductivity Changes in Discharge-Produced Amorphous Si," Applied Physics Letters, vol. 31, pp. 292-294, 1977. M. Stutzmann, W. B. Jackson, and C. C. Tsai, "Light-Induced Metastable Defects in Hydrogenated Amorphous-Silicon - a Systematic Study," Physical Review B, vol. 32, pp. 23-47, 1985. J. Schmidt, "Light-Induced Degradation in Crystalline Silicon Solar Cells," Brandenburg, Germany, 2004, pp. 187-196. H. R. Park, J. Z. Liu, and S. Wagner, "Saturation of the Light-Induced Defect Density in Hydrogenated Amorphous-Silicon," Applied Physics Letters, vol. 55, pp. 2658-2660, Dec 18 1989. C. Godet, "Metastable hydrogen atom trapping in hydrogenated amorphous silicon films: A microscopic model for metastable defect creation," Philosophical Magazine B-Physics of Condensed Matter Statistical Mechanics Electronic Optical and Magnetic Properties, vol. 77, pp. 765-777, Mar 1998. M. A. Alam, "A critical examination of the mechanics of dynamic NBTI for PMOSFETs," 2003 IEEE International Electron Devices Meeting, Technical Digest, pp. 345-348, 2003. T. Shimizu, "Staebler-Wronski effect in hydrogenated amorphous silicon and related alloy films," Japanese Journal of Applied Physics Part 1-Regular Papers Brief Communications & Review Papers, vol. 43, pp. 3257-3268, Jun 2004.
