Solid-State Electronics 45 (2001) 20172021

Comparative study of metalsemiconductor contact degradation by current pulses on silicon solar cells with two contact types
I. Perez-Quintana a, A. Martel a,b, L. Hern andez a,*,1
b

Facultad de F sica, Universidad de La Habana, Colina Universitaria, 10400 La Habana, Cuba CICATA-IPN Unidad Altamira, Km 14.5 Carretera Tampico-Puerto Industrial 89600 Altamira, Mexico Received 17 March 2001; received in revised form 21 April 2001; accepted 14 May 2001

Abstract This work accomplishes a comparative study of metalsemiconductor contact degradation on two dierent types of silicon solar cells contacts. One of them was a thermally vacuum-evaporated Ti/Pd/Ag contact, and the other one was a screen-printed contact. An experimental and theoretical methodology was applied in order to study the degradation due to periodic hot/cool switching and knowledge about all fundamental parameters from IV characteristics of both types of solar cells was obtained. The periodic hot/cool process was carried out by current pulses and the double exponential model of IV characteristic was used to acquire all fundamental parameters of the solar cells. We found that all fundamental parameters of both types of cells were degraded with the application of current pulses in the time studied, but in any case, the screen-printed contacts were degraded more smoothly than the thermally vacuum-evaporated front contacts of Ti/Pd/Ag. 2001 Elsevier Science Ltd. All rights reserved.
Keywords: IV solar cells characteristics; Metalsemiconductor contact

1. Introduction Contact resistance losses occur at the interface between the silicon solar cell and the metal contact, and it is one of the most important parameters in the solar cell performance. Contact resistance has also a signicant eect on the currentvoltage characteristic (IV) since it modies the series resistance values. The main impact of series resistance increase is to reduce the ll factor (FF), although excessively high values may also reduce the short-circuit current. The IV characteristic is the most important theoretical and experimental technique to research any solar cell independently of its interface, of utilized materials on fabrication, of technological

Corresponding author. E-mail address: luisman@.oc.uh.cu (L. Hern andez). 1 Also at Departamento de F sica. CINVESTAV. A.P. 14740 Mexico, D.F. Mexico.

design, of illumination levels and so on. The theoretical models that describe the experimental currentvoltage curve are of particular interest on the research of these devices, because from the experimental measurements it is possible to obtain immediate information about the essential parameters, which characterize the solar cell performance. The other physical parameters can also be obtained from measured IV characteristic, but it is necessary to use extraction mathematic algorithms. Several models have been developed to describe experimentally the currentvoltage characteristic in solar cells. A complete physic-mathematical approach to the IV characteristic can be represented by the two-diode model where the diusion current and the recombination current are represented by two diodes with dierent exponential behavior and become more closely related to the physical phenomena. This model was proposed by Wolf [1] to silicon solar cells working under low illumination conditions, and the relation IV is expressed, by the implicit equation

0038-1101/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 3 8 - 1 1 0 1 ( 0 1 ) 0 0 2 0 3 - 9

2018

I. Perez-Quintana et al. / Solid-State Electronics 45 (2001) 20172021

  qV IRs 1 I V IL Id exp nd kT   qV IRs V IRs Ir exp 1 Rsh nr kT

2. Experimental details 1 Two types of contact silicon solar cells were processed: thermally vacuum-evaporated front contacts of Ti/ Pd/Ag (cell 1 or C1); screen-printed front contacts of Ag paste (cell 2 or C2). Starting from a 1.5 X cm resistivity p-type silicon wafer a standard 3 in. diameter n p silicon solar cell was processed like C1 cell. Front radial contacts were fabricated of Ti(0.5 lm)/Pd(0.5 lm)/Ag(3 lm) thermally vacuum-evaporated and a SnO2 layer of 80 nm thickness was deposited as antireection coating. The C2 cell, was also n p silicon solar cell. The preparation of C2 cell involved the use of a standard 10 10 cm2 area and 1 X cm resistivity p-type silicon wafer. The ohmic contacts were fabricated by screenprinted technique and the following technological steps were taken: deposition of silver/aluminum paste like back contact; deposition of silver paste like front contact; deposition of aluminum paste on the back contact. This contact requires subsequent thermal annealing, for this reason after each deposition, the paste was dried at 100C on air atmosphere and nally a thermal annealing between 700C and 800C was made. To avoid the strong eects on leakage currents of environmental factors, electrical isolation of the solar cells were achieved by edge sealing with transparent polymer materials, ethylvinyl acetate, with high elasticity and exibility properties. The sealed cells were mounted on berglass and the temperature of each cell was monitored with a cromelalumel thermocouple placed in good thermal contact with the back surface of each one of both cells. Current pulses were applied on the cells in forward bias. These pulses were formed applying current to the circuit for 5 min and followed by 5 min without current. During the test time, we applied an equivalent current to 10 times the short-circuit current at 100 mW/cm2 . In this regime, the stationary temperature increased to 30C and 3C above room temperature with on/o current, respectively. Because of the thermal conductivity dierence between metallic contacts and silicon semiconductor, the thermal cycles causes the degradation of the series and shunt resistances. The loss of metalsemiconductor adherence by current pulses, increases the contact resistances to the front and back as well as the surface leakages along the cell edges, by diusion spikes along dislocations or possibly by ne metallic bridges

In this equation, IL is the photo-current, Id exponential term is the Shockley diusion current which includes the electronic conduction phenomena in the quasi-neutral region of the junction and the Ir exponential term corresponds to the carrier recombination trough deep levels in the spacecharge region of the junction and in the device surface. In addition, the model includes the series and shunt resistance, Rs and Rsh respectively, and other classical diusion and recombination diode ideality factors, nd and nr respectively. It has been demonstrated that the values of the solar cell parameters calculated by double-exponential models agree well with the measured ones [24]. Alternative analytical expressions for the IV characteristic have been reported [5,6]. In all cases, the accuracy of the expressions has been examined as a function of the voltage drop through each cell constituent, the electrical and geometrical parameters, the concentration factor of light, and so on. A recent report [7] describes the use of two new techniques for obtaining photovoltaic solar cells parameters by ts to IV characteristics. The rst one, which consists of the simultaneous tting of more than one IV characteristic of the same or dierent type, allows more reliable parameters to be obtained than those secured by means of a single t of a single characteristic. The second technique consists of making a series of single or simultaneous ts to dierent IV characteristics, transferring the parameters resulting from one t to the next one to be used as the starting or reference data. This technique is valid for those cases in which a parameter diverges to a large extent from its normal values and where single or simultaneous ts are not very reliable or even fail to converge during tting process. In a previous work [8], we have presented a method to degrade the metalsemiconductor contacts by current pulses and searched the inuence of the series and shunt resistances in dierent analytical models. In particular the two-exponential model was applied for its simplicity, reliability and extended use, together with the technique for the determination of series resistance, suggested by Rajkanan and Shewchun [9]. The shunt resistance was evaluated by the expression reported in Ref. [10]. The purpose of this work is to extend our reported experimental and theoretical methodology [8] to degrade the metalsemiconductor contacts in solar cells by current pulses and then to make a comparative study in two dierent types of silicon solar cell contacts.

I. Perez-Quintana et al. / Solid-State Electronics 45 (2001) 20172021

2019

along microcracks, grain boundaries, or crystal defects as stacking faults leading to a reduction in the shunt resistance values. Systematically, with a common experimental setup, the IV characteristics under illumination were measured. A solar simulator tungsten lamp with infrared lter was used together with a reference cell. A signal proportional to illumination level was measured by a reference cell in the light beam and was used as feedback to the lamp power supply that was in a current circuit with the solar cell. In this way accurate monitoring of the lamp stability during a measurement was achieved. The IV output characteristic was registered by analog to digital converts to monitor the voltage across the cell. An automatized system oers graphical and tabular presentation of the data besides data storage. The maximum power point was calculated directly, by searching the maximum IV product from the measured data set, and compared with the expression reported by [11]. A complete measurement including evaluation of the cell parameters and plotting of the IV curve was achieved within a minute and the cell temperature was kept constant in a 0.5C range. The dark IV characteristics were obtained by forward biasing the cells with a regulated power supply. Also, the spectral response of both cells were measured using a GS0895 lamp at 100 mW/cm2 , interference lters from 350 to 1100 nm wavelength from 50 to 50 nm and a lock-in equipment.

to consider that the current pulses kept the solar cells volumetric characteristics invariant. To determine the values of the two-exponential model parameters, iterative methods were used to solve Eq. (1). The open-circuit voltage (Voc ), short-circuit current (Isc ) and maximum power conditions (Po ) in the experimental IV characteristic were used for the solution. It was assumed nd constant and nr constant, then ve iterative equations were necessary to solve Eq. (1). Four of them are the explicit expressions for Rsh , Rs , IL and Id . The fth equation was obtained from: I Ir ; Vo Io 0 2

3. Results and discussion No change between the initial and nal values of the spectral response for both silicon solar cells was observed with the applied current pulses. Similarly, no variations in the maximum of internal quantum eciency Qk, always measured at wavelength 850 nm and whose values were equal to 0.75 for the C1 cell and 0.69 for the C2 cell during the 18 144 current pulses. These results allow to consider assert that for both solar cells the photo-current generated is not aect by the current pulses and in a good approximation it is possible

where Vo , Io are the experimental values of the voltage and current at the maximum or optimum power point. As a convergence criterion, the squares mean of the dierence of the calculated and a set experimental current values on IV curves was taken. IV characteristic were measured before application of the current pulses to the both silicon solar cells and the experimental values of the fundamental parameters Isc , Voc , Vo , Io , Po , FF, conversion eciency (g), Rs and Rsh were extracted. Using the iterative method a simulated curve was generated in order to reproduce experimental values. The extracted parameters obtained by this method (Ir , Id , nr and nd ), were used to generate a set of the simulated IV curves using Eq. (1) which allowed to search the tted values shown in the Tables 1 and 2. The time variations of the measured parameters (Voc , Isc , FF, g) and their tted values, as a function of number of pulses for C1 cell and C2 cell are shown in Tables 1 and 2 , respectively. In the evaluation of these tted values, IL was kept constant, supported by independence of the maximum of internal quantum eciency with a numbers of pulses. Over the examined time, all fundamental parameters were degraded with the applied current pulses. IV characteristics conrm these results, because at 13 104 and more at 18 144 pulses, the increase of the series resistance Rs associated to the loss of metalsemiconductor

Table 1 Behavior of the measured fundamental parameters with the applied current pulses for C1 cell Number of pulses 1008 4032 9072 10080 13104 16128 18144 Voc (V) Measured 0.560 0.555 0.552 0.556 0.558 0.561 0.560 Fitted 0.560 0.555 0.551 0.557 0.559 0.562 0.561 Isc (A) Measured 0.152 0.152 0.151 0.146 0.130 0.128 0.124 Fitted 0.152 0.152 0.151 0.146 0.130 0.128 0.124 FF Measured 0.559 0.562 0.553 0.464 0.407 0.390 0.384 Fitted 0.555 0.561 0.554 0.463 0.407 0.390 0.383 g (%) Measured 8.1 8.05 7.8 6.37 5.0 4.7 4.5 Fitted 8.0 8.0 7.8 6.4 5.0 4.8 4.5 0.048 0.047 0.046 0.037 0.029 0.028 0.026 0.414 0.460 0.488 0.746 0.892 1.16 1.24 25.0 16.6 11.7 10.0 9.4 9.1 8.8 Po (W) Rs (X) Rsh (X)

Only for Voc , Isc , FF and g the tted values are shown beside the measured values.

2020

I. Perez-Quintana et al. / Solid-State Electronics 45 (2001) 20172021

Table 2 Behavior of the measured fundamental parameters with the applied current pulses for C2 cell Number of pulses 1008 2016 6048 7056 10080 13104 18144 Voc (V) Measured 0.554 0.556 0.554 0.551 0.547 0.547 0.546 Fitted 0.554 0.555 0.551 0.547 0.547 0.547 0.546 Isc (A) Measured 0.098 0.099 0.098 0.098 0.098 0.099 0.098 Fitted 0.098 0.099 0.099 0.098 0.098 0.099 0.097 FF Measured 0.674 0.676 0.675 0.673 0.670 0.669 0.668 Fitted 0.666 0.672 0.669 0.668 0.664 0.668 0.668 g (%) Measured 8.7 8.8 8.7 8.6 8.5 8.5 8.4 Fitted 8.9 8.8 8.7 8.6 8.5 8.5 8.3 0.036 0.037 0.036 0.036 0.035 0.035 0.035 0.561 0.569 0.645 0.714 0.726 0.740 0.750 21.0 20.0 12.5 10.0 7.80 5.6 5.5 Po (W) Rs (X) Rsh (X)

Only for Voc , Isc , FF and g the tted values are shown beside the measured values.

metalsemiconductor and this zone the current pulses eect in the screen-printed solar cell is less important than the evaporated contact cell. The starting square series resistances 0.109 and 0.070 X/cm2 , for C1 and C2 cells respectively, suggest that the screen-printed contacts have inferior quality than the other one, then the collection charge eciency and the injection (current pulses) is also lower, and therefore the eect of the current pulses is lower accentuated than the thermally vacuum-evaporated contacts type.

4. Conclusion
Fig. 1. Behavior of the IV characteristics as a function of applied current pulses for C1, Ti/Pd/Ag contact and C2, Ag paste screen-printed contact.

adherence, it is an evident fact because a strong reduction of the slope of the IV curve can be observed at high voltage. Table 2 also displays the time variations of the measured and tted parameters as a function of number of pulses for the screen-printed silicon solar cell C2. Again in the evaluation of the tted values Voc , Isc , FF and g, IL was kept constant supported by independence of the maximum of internal quantum eciency with a number of pulses. All fundamental parameters of the C2 screen-printed cell were degraded more smoothly than the thermally vacuum-evaporated contact C1 cell. Notice in Fig. 1 that no substantial variation of the shortcircuit current was observed even at 18 144 current pulses, conserving its atting IV curve in the low values voltage, indicating small variation for the shunt resistance Rsh . On the other hand, small variation of the series resistance Rs was observed around the open circuit voltage zone of this cell, because the IV curve remain with similar slope at high values of voltage. These results can be explained by supposing that the outcome of the current pulses is only on the interface

We found that all fundamental parameters of both types of cells were degraded with the application of current pulses in the time studied, but in any case, the screen-printed contacts were degraded more smoothly than the thermally vacuum-evaporated front contacts of Ti/Pd/Ag. Also, in both cases, there was kept a correspondence between the experimental values of IV curves and the values calculated by the double-exponential model. It was shown that the contact degradation is the unique factor deteriorating the fundamental parameters of these types of silicon solar cells. In both types of cells, the diusion, recombination and photo-current terms remained unchanged in spite of the adherence loss between metal contact and silicon semiconductors. The use of current pulses to simulate the conditions of normal operation of solar cells does not degrade neither the electrical characteristic of the junction, nor the volume characteristics. Acknowledgements I. Perez-Quintana thanks Ferrara University and ICTP (Trieste), L. Hern andez and A. Martel thank CONACyT and CICATA IPN (Altamira) respectively for the partial support in this work.

I. Perez-Quintana et al. / Solid-State Electronics 45 (2001) 20172021

2021

References
[1] Wolf HF. Silicon semiconductors data. In: Henisch HK, editor. International series of monographs on semiconductors, vol. 9. Oxford: Pennsylvania State University, Pergamon Press; 1969. p. 526. [2] Charles J-P, Bordure G, Khoury A, Mialhe P. Consistency of double exponential model with the physical mechanisms of conduction for solar cells under illumination. J Phys D Appl Phys 1985;18:22618. [3] Charles J-P, Mekkaqui-Alaqui I, Bordure G, Mialhe P. A critical study of the eectiveness of the single and double exponential models for IV characterization of solar cells. Solid-State Electron 1985;28(8):80720. [4] Enebish N, Agchbayar D, Dorjkhand S, Baatar D, Ylemj I. Numerical analysis of solar cell currentvoltage characteristics. Sol Energy Mater Solar Cells 1993;29:2018. jo GL. Mathematical analysis of the [5] S anchez E, Arau eciency-concentration characteristic of a solar cell. Solar Cells 1984;12:26376.

[6] Akbaba M, Alattawi MAA. New model for IV characteristic of solar cell generators and its applications. Sol Energy Mater Solar Cells 1995;37:12332. [7] Mart nez V, Jimeno JC. Simultaneous t to IV characteristics of solar cells. 2nd World Conference and Exhibition on Photovoltaic Solar Energy Conversion, Vienna, Austria, 610 July, 1998. p. 13246. [8] Hern andez L, P erez-Quintana I, Martel A. Modication of fundamental parameters in the silicon solar cells by current pulses. Sol Energy Mater Solar Cells 1996;44: 36773. [9] Rajkanan K, Shewchun J. A better approach to the evaluation of the series resistance of solar cells. Solid-State Electron 1979;22(2-E):1937. [10] Charles J-P, Abdelkrim M, Muoy YH, Mialhe P. A practical method of analysis of the currentvoltage characteristics of solar cells. Solar Cells 1981;4:16978. [11] Hern andez L, Soler R, Martel A. N dopant exponential proles for silicon solar cells. Sol Energy Mater Solar Cells 1993;30:291300.