Doping-Free Intrinsic Amorphous Silicon Thin-Film Solar Cell Having ...

3 downloads 281 Views 645KB Size Report
silicon (i-a-Si) thin-film solar cells having a structure of glass/SnO2/MoO3/i-a-Si/LiF/Al. The short-circuit current density of the cell markedly increased while the ...
96

IEEE ELECTRON DEVICE LETTERS, VOL. 35, NO. 1, JANUARY 2014

Doping-Free Intrinsic Amorphous Silicon Thin-Film Solar Cell Having a Simple Structure of Glass/SnO2/MoO3/i-a-Si/LiF/Al Ji-Hwan Yang, Sang Jung Kang, Yunho Hong, and Koeng Su Lim

Abstract— We fabricated doping-free intrinsic amorphous silicon (i-a-Si) thin-film solar cells having a structure of glass/SnO2 /MoO3 /i-a-Si/LiF/Al. The short-circuit current density of the cell markedly increased while the open-circuit voltage and fill factor were low due to a lower work-function of the MoO3 than that of a conventional amorphous silicon carbide film. To solve these drawbacks, we UV-treated on the MoO3 layer, obtaining a greatly enhanced conversion efficiency of 6.42%. Index Terms— Doping free, MoO3 , LiF, i-a-Si solar cells.

I. I NTRODUCTION

H

YDROGENATED amorphous silicon (a-Si) thin-film solar cells are still promising candidates for future photovoltaic markets due to their strong advantages such as low cost production and easy large-area module fabrication. However, for fabricating conventional a-Si-based thin-film solar cells, very toxic gases such as trimethylborane (TMB) and diborane (B2 H6 ) for p-type doping and phosphine (PH3 ) for n-type doping must be used. Removing these toxic doping gases in fabrication process, no safety facilities needs in order to manage them. Therefore, doping-free intrinsic (i ) a-Si (i -a-Si) solar cell fabrication process simplifies production facility and reduces production cost. We have already reported two kinds of the solar cells without using any one of the p- and n-doping gases. One is the cell having a MoO3 window layer [1] instead of a boron-doped p-type silicon carbide ( p-SiC) layer in the conventional cell. The other is the cell having a lithium fluoride/aluminum (LiF/Al) rear Schottky junction [2] instead of a phosphorus-doped n-type a-Si (n-a-Si) layer. However, both types of the cells showed lower Voc compared to the conventional cell without a p/i buffer and ZnO/Ag back reflector. In order to perfectly remove the doping gases and to obtain higher performance, especially high Voc , we have investigated doping-free i -a-Si solar cells having both the MoO3 window layer and the LiF/Al rear Schottky junction at the same time. In this letter, we report on the characteristics of the doping-free i -a-Si solar cells having the MoO3 and LiF/Al, Manuscript received September 21, 2013; revised October 26, 2013; accepted October 30, 2013. Date of publication November 20, 2013; date of current version December 20, 2013. This work was supported by the R&D Program through the National Fusion Research Institute of Korea funded by the Government funds. The review of this letter was arranged by Editor A. Ortiz-Conde. The authors are with the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology, Daejeon 305-701, Korea (e-mail: [email protected]). Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LED.2013.2289309

which can be easily deposited by thermal evaporation process without heating substrate. II. E XPERIMENTAL P ROCEDURE MoO3 and LiF films were thermally evaporated by using tantalum boats under a high vacuum condition (1-3 ×10−6 Torr). Products of MoO3 (Alfa Aesar 99.5%) and LiF powders (Sigma Aldrich 99.99%) were used for depositing the MoO3 and LiF films, respectively. The thicknesses of the films were monitored during evaporation using a thickness monitor (SQM-160, Sigma). After depositing MoO3 films, additional UV treatment was performed on some MoO3 films under a high vacuum condition. A low-pressure mercury (LP Hg) lamp with resonance lines of 184.9 nm and 273.7 nm was used as a UV light source. The work function (WF) values of the UV irradiated MoO3 films were calculated from UPS spectra obtained using a Sigma probe system under an ultrahigh vacuum condition (1×10−10 Torr) with a He I (21.2 eV) gas-discharge lamp. Asahi-U type SnO2 :F-coated (SnO2 ) glass was used as a substrate. MoO3 , i -a−Si, LiF, and Al layers were consecutively deposited on the SnO2 glass. The cell area was 0.09 cm2 . Photo current density versus voltage (photo J -V ) measurements were performed under air mass (AM) 1.5 (100 mW/cm2 ) illumination using a Keithley 2611A source meter. A spectral response system (CEP-25ML, Bunkoukeiki) was used for external quantum efficiency (EQE) measurements. III. R ESULTS AND D ISCUSSION We fabricated four kinds of a-Si-based solar cells: a glass/SnO2 / p-a-SiC(10 nm)/i -a-Si(200 nm)/ n-a-Si(40 nm)/Al(150 nm) structure solar cell (device A); a glass/SnO2 /MoO3 (10 nm)/i -a-Si(200 nm)/n-a-Si(40 nm)/ Al(150 nm) cell (device B); a doping-free glass/SnO2 /MoO3 (10 nm)/i -a-Si(200 nm)/LiF(2.1 nm)/Al(150 nm) cell (device C); a doping-free cell having the same structure as the device C except for the MoO3 window layer in C was 60-min-UV treated (device D). Fig. 1 presents the J -V characteristics of the four kinds of the i -a-Si solar cells. The untreated 10-nm-thick MoO3 window layer in B has a larger optical band gap (E opt ) of 2.83 eV, which was evaluated using spectroscopic ellipsometry, than that of the p-a-SiC (∼2.0 eV) in A. Thus, B shows less absorption loss in the short wave length region. Therefore, the Jsc was improved from 10.3 mA/cm2 for A to

0741-3106 © 2013 IEEE

YANG et al.: DOPING-FREE i-a-Si THIN-FILM SOLAR CELL

Fig. 1.

J -V characteristics of four kinds of i-a-Si based solar cells.

11.8 mA/cm2 for B. However, the Voc (0.754 V) of B was smaller than that (0.815 V) of A due to a relatively lower WF of the MoO3 film than that of the p-a-SiC and its n-type nature. The Voc (0.748 V) of C was significantly degraded compared to that of A and slightly degraded compared to that of B. However, the Jsc was largely increased compared to that of A due to its large E opt but slightly decreased compared to that of B. The FF was also greatly decreased from 0.687 for A to 0.574 for C. Consequently, the cell performance of C markedly decreased compared to those of A and B. The Voc (0.829 V) of D was larger than that (0.815 V) of A due to the elevated WF of the 60-min UV treated MoO3 film. From the UPS spectra, the WF of MoO3 film was increased from 5.17 eV to 5.37 eV after 60-min UV treatment. The WF values were calculated from the difference between the vacuum level and the Fermi level. Furthermore, the Jsc was significantly increased compared to those of A and C due to its large E opt and the elevated WF. The FF was also largely decreased from 0.687 for A to 0.605 but increased compared to 0.574 for C. Thus, we obtained a higher conversion efficiency of 6.42 % in D, which surpass that of 5.76 % in A. Compared to the state-of-the-art a-Si solar cells, the reference A showed the lower conversion efficiency, because very thin i -layer (200 nm) was used without any light trapping scheme such as ZnO/Ag reflector at the backside. Besides, there was no buffer layer inserted at the p/i interface in order to compare only the MoO3 window layer used cell with the reference cell. Therefore, the Voc showed the lower value. For reference, the electrical conductivities of the p-a-SiC and n-a-Si films were 1×10−6 S/cm and 1×10−2 S/cm, respectively. The Rs and Rsh of A were 5.38 cm2 and 2.10 kcm2, respectively. Fig. 2 shows the EQE of the four solar cells. Note that all the cells having the MoO3 window layer show very much enhanced EQE in the short wave length region than that of A because of its wide E opt . Note also that D shows the highest EQE in the whole wave length region and markedly enhanced EQE in the long wave length region compared to those of A, B, and C. First, we investigate the behavior of the i -a-Si/LiF/Al Schottky junction in both C and D. In order to clarify the current transport mechanism through the Schottky barrier, we fabricated an Al(150 nm)/n-a-Si(40 nm)/i -a-Si(200 nm)/

97

Fig. 2.

External quantum efficiencies of i-a-Si based solar cells.

Fig. 3. J -V characteristic of a test device having a structure of Al(150 nm)/ n-a-Si(40 nm)/i-a-Si(200 nm)/LiF(2.1 nm)/Al(150 nm).

LiF (2.1 nm)/Al(150 nm) sandwich structure which is identical structure included in both C and D except for the front Al/n-a-Si structure. In this structure, the Al/n-a-Si heterojunction represents ohmic contact property [3]. So, we can investigate the i -a-Si/LiF/Al Schottky barrier property using the sandwich device. Fig. 3 shows the dark J -V characteristics of the device. The J -V graph is almost symmetric with respect to the origin. In the low applied voltage region between +0.3 and −0.3 V, the current through the barrier is nearly proportional to applied voltage. Therefore, we can calculate an ohmic resistance in this region, thus obtaining the value of 1.57-cm2. On the other hand, the current in the higher applied voltage region than 0.3 V is exponentially proportional to applied voltage. Note that for the applied voltage of 0.3 V, the current density is about 0.25 A/cm2 , which is around twenty-times larger than those of C and D. Therefore, we can conclude that the LiF/Al Schottky barrier junction shows a very low ohmic resistance in the low applied voltage region, resulting in easy current flowing through the junction. The reason for this can be explained by a strong dipole moment creation at the LiF/Al interface, helping electrons transfer through the barrier. A. J. Mozer et al. [4] explained about the dipole induced transport mechanism that well-ordered dipoles might create “highways” of hopping conduction. Next, we discuss the remarkable performance enhancement of D as shown in Figs. 1 and 2. The Voc of the cell was increased compared to that of C. In A, the WF deference between the p-a-SiC window and n-a-Si layers is one of

98

Fig. 4. Band diagrams of the devices (a) untreated cell on the MoO3 window layer and (b) UV-treated cell.

the main elements determining Voc of the cell. Therefore, we can conclude from Fig. 3 that the Voc increase in D resulted from the WF elevation of the 60-min UV treatment on the MoO3 . TheVoc improvement indicates that the internal electrical field (E int ) in the i -a-Si absorption layer becomes strong, resulting in helping photo-generated carriers easily transport into the MoO3 layer and be collected. Therefore, the FF and Jsc were increased. Let us consider this in more detail. Fig. 4 represents band diagrams of C (dashed blue line) and D (solid line) under short circuit condition. Compared to the untreated MoO3 , the UV-treated MoO3 film showed the elevated WF by 0.20 eV and the reduced E opt by 0.26 eV (from 2.83 eV to 2.57 eV). Because the WF difference was reflected in the lowered conduction band edge, the valence band edge should be elevated by only 0.06 eV. Path (1) denotes that hole drifts in the direction of the E int and then it enters variable-range hopping (VRH) conduction [5], [6] in the MoO3 film. Path (2) denotes that hole goes up along the interface states and then it is trapped at some MoO3 /i -a-Si interface defect state, recombining with electron. Path (3) denotes that electron diffuses backwards against the E int and then it is trapped at some interface defect state, recombining with hole. Path (4) denotes that electron drifts in the opposite direction of the E int . Each total current flowing through C or D consists of two components: photo current (I ph ) due to photo-generated carriers under illumination and dark current (Id ). By the way, the I ph through each device flows in the opposite direction to its Id under illumination condition. However, each photo shunt recombination current (I ps,rec ) [7] through the defect states at the MoO3 /i -a-Si interface flows in the same direction to its Id . Therefore, for D, the I ps,rec can be decreased due to the increased E int as aforementioned. This means that the Rsh increases, resulting in the increased FF as seen in Fig. 1 and the increased EQE in the long wavelength region as seen in Fig. 2. In fact, the value of Rsh for D determined from the slope of the photo J -V curve near 0 V in Fig. 1 was 811 cm2 , while that of the Rsh for C was 542 cm2 . Furthermore, we found that a T −1/4 dependence of the reverse saturation current density was found in C and D at temperatures between 300 and 328 K, which means that the current transport mechanism can be described by the VRH conduction mechanism. The density of states at (or near) the Fermi level, N(E F ), was found

IEEE ELECTRON DEVICE LETTERS, VOL. 35, NO. 1, JANUARY 2014

from Mott parameters analysis [5], [6]. The N(E F )s of C and D were 3.27×1016 eV−1 cm−3 and 8.65×1015 eV−1 cm−3 , respectively. Therefore, we can conclude that the UV treatment on the MoO3 window layer could reduce the defect density at the MoO3 /i -a-Si interface and in the MoO3 film, resulting in decreasing I ps,rec and in increasing Rsh . The reduced recombination loss at the MoO3 /i -a-Si interface due to the reduced defect density can be clarified by comparing the EQE curves at the wavelengths shorter than 500 nm shown in Fig. 2. Furthermore, the enhanced E int can decrease recombination through the paths (2) and (3) at the MoO3 /i -a-Si interface when electron-hole pairs are generated in the i -layer near the MoO3 /i -a-Si interface. The enhanced E int can also decrease recombination in the i -a-Si layer near the i -a-Si/LiF interface and help easy carriers transfer through the LiF/Al barrier as aforementioned in Fig. 3. This is the reason why the EQE of D is higher at the wavelengths longer than 500 nm than those of B and C with untreated MoO3 window layers. Consequently, the I ph increases and the Rs decreases. In fact, the Rs of D determined from the slope of the photo J -V curve at J = 0 mA/cm2 in Fig. 1 was 9.1 cm2 , while the Rs was 12.9 cm2 for C. Therefore, we can conclude that the FF enhancement resulted from the increased Rsh and decreased Rs . That is, the remarkable performance enhancement of D originated from the increased Voc (E int ) due to the UV treatment on the MoO3 window layer. However, the 60-minUV treatment can cause the interruption for simple fabrication and cost reduction which are originally aimed. Therefore, we have to reduce UV treatment time using a higher intensity LP Hg lamp or develop alternative structures using a buffer like SiOx between the MoO3 and i -a-Si. IV. C ONCLUSION We fabricated glass/SnO2 /MoO3 (10 nm)/i -a-Si(200 nm)/ LiF(2.1 nm)/Al solar cells and performed UV treatment on the MoO3 window layer. As a result, the Voc , Jsc , and FF markedly increased due to the elevated work-function of the UV-treated MoO3 . A maximum efficiency of 6.42 % was obtained. R EFERENCES [1] S. I. Park, S. J. Baik, J.-S. Im, et al., “Towards a high efficiency amorphous silicon solar cell using molybdenum oxide as a window layer instead of conventional p-type amorphous silicon carbide,” Appl. Phys. Lett., vol. 99, no. 6, pp. 063504-1–063504-3, Aug. 2011. [2] L. Fang, S. J. Baik, S. Lim, et al., “Amorphous Si rear Schottky junction solar cell with a LiF/Al back electrode,” IEEE Trans. Electron. Devices, vol. 58, no. 9, pp. 3048–3052, Sep. 2011. [3] J. Kanicki, “Contact resistance to undoped and phosphorus-doped hydrogenated amorphous silicon films,” Appl. Phys. Lett., vol. 53, no. 20, pp. 1943–1945, Sep. 1988. [4] A. J. Mozer, P. Denk, M. C. Scharber, et al., “Novel regiospecific MDMO–PPV copolymer with improved charge transport for bulk heterojunction solar cells,” J. Phys. Chem. B, vol. 108, no. 17, pp. 5235–5242, Feb. 2004. [5] L. Fang, S. J. Baik, K. S. Lim, et al., “Tungsten oxide as a buffer layer inserted at the SnO2 /p-a-SiC interface of pin-type amorphous silicon based solar cells,” Appl. Phys. Lett., vol. 96, no. 19, pp. 193501-1– 193501-3, May 2010. [6] I. G. Austin and N. F. Mott, “Polarons in crystalline and non-crystalline materials,” Adv. Phys., vol. 18, no. 71, pp. 41–102, 1969. [7] A. Shah, “Module fabrication and performance,” in Thin-Film Silicon Solar Cells, 1st ed. Laussanne, Switzerland: EPEL, 2010, ch. 6, pp. 346–348.