Yimao Wan, Chris Samundsett, James Bullock, Di Yan, Thomas Allen, Peiting Zheng, Xinyu Zhang, Jie Cui,. Josephine McKeon, and Andres Cuevas. Research ...
Magnesium Fluoride Based Electron–Selective Contact Yimao Wan, Chris Samundsett, James Bullock, Di Yan, Thomas Allen, Peiting Zheng, Xinyu Zhang, Jie Cui, Josephine McKeon, and Andres Cuevas Research School of Engineering, The Australian National University, Canberra, ACT 0200 Australia Abstract — This work demonstrates a simple process for fabricating an electron–selective contact with a contact resistivity of 76 mΩ∙cm2 and a contact recombination current density of 10 fA/cm2 on n-type crystalline silicon. The contact is facilitated via an amorphous silicon (~ 6.5 nm) / magnesium fluoride (~ 1 nm) / aluminium stack. The application of this novel electron–selective contact enables fabrication of a 20.1%– efficient n-type front-junction silicon solar cell. This contact is made to the full area of the rear surface, omitting the need for (i) high-temperature phosphorus diffusion, and (ii) patterning of the rear dielectrics by photolithography or laser processing.
I. INTRODUCTION Alkali/Alkaline earth metals and salts have been exploited extensively in organic devices as electron-selective materials. Some of these materials are starting to be explored on crystalline silicon (c-Si), due to their potential to form an ohmic electron contact between a metal electrode and the c-Si substrates, as well as reducing surface recombination. These materials, deposited at low temperature, offer exciting technological possibilities for simplifying solar cell fabrication and upgrading their performance. In particular, they represent an alternative to high temperature phosphorus (n+) doping, which is required both for conventional silicon solar cells and for advanced passivating contacts based on doped polycrystalline silicon. For example, the insertion of only a ~1 nm thick lithium fluoride (LiF) film between the aluminium (Al) electrode and c-Si has been demonstrated to dramatically enhance the electron injection/extraction compared to a direct Al contact structure [1]. It was also recently demonstrated that such a layer can be combined with an amorphous silicon (a-Si:H) layer to simultaneously provide low resistance and recombination, suitable for high efficiency silicon solar cells [1]. Magnesium fluoride (MgF2) shares similar characteristics with these materials, yet its electrical contact behavior has not been explored on any photovoltaic absorber, and its use within photovoltaics has been limited to antireflection applications. In this work, we first study the recombination and resistive properties of the MgF2/Al contact on n-type c-Si (n-Si), with or without an a-Si:H buffer layer. We then apply the a-Si:H/MgF2/Al contact to the full back surface of an ntype front-junction silicon solar cell (hereafter referred to as full area contact, FRC MgF2 cell), and compare to a conventional cell with partial rear contacts (PRC) that are formed by patterning a silicon nitride (SiNx) layer (hereafter referred as PRC control cell). Spectral response and
reflectance measurements are undertaken to provide insight into the physical mechanisms for the differences in cell performance of the two rear contact technologies. II. SELECTIVITY OF MGF2 BASED CONTACTS A. Sample preparation All test structures were fabricated on float-zone, (100) oriented, n-Si substrates. The wafer resistivities for contact and lifetime measurements were 0.5 Ω∙cm and 10 Ω∙cm, respectively. All wafers were RCA cleaned and dipped in 1% HF acid prior to PECVD a-Si:H and/or MgF2 deposition. MgF2 films were thermally evaporated at a rate of 0.25 Å/s from a 3N-purity MgF2 powder source, with a base pressure of < 3 × 10–6 Torr. B. Contact resistivity and recombination The positive selectivity towards majority carriers, that is, the efficient transport of electrons, of the a-Si:H/MgF2/Al contact structure can be evaluated via its contact resistivity ρc. The negative selectivity, that is, the “blocking action”, of the a-Si:H/MgF2/Al contact structure towards holes can be evaluated via its interface recombination parameter J0c (as determined from effective minority carrier lifetime test structures). A highly selective contact is achieved through a simultaneous reduction in both parameters. Figure 1 shows representative current voltage (I–V) measurements of samples with and without an 1 nm MgF2 film. As we can see, the sample with Al directly on n-Si (i.e., without MgF2) exhibits a rectifying behavior, preventing an accurate extraction of contact resistivity. A high contact resistance, or a rectifying behavior, between Al metal and nSi is attributable to the presence of a high surface potential barrier. In contrast, the insertion of a thin MgF2 (~1 nm) film dramatically improves the contact performance, leading to the formation of an ohmic contact (i.e., linear I–V relation) between the Al electrode and the n-Si substrate. The extracted contact resistance ρc for the structure with 1 nm MgF2 is determined to be 35 mΩ∙cm2. The high electron conductivity provided by the n-Si/MgF2/Al contact structure can be attributed to a reduction in the work function, compared to that of Al. In a complementary study [2], we have measured that the work function of this structure is approximately 3.5 eV, significantly lower than the 4.2 eV of Al. This modification of the work function, and the resulting low contact resistivity, could be attributable to (i) reaction between MgF2 film and Al metal layer during thermal evaporation, and/or (ii) electron tunneling through a reduced barrier width, which is similar to the mechanism of the ohmic
contact between the heavily phosphorus doped n+ c-Si and the direct Al metal. 40
J0c(fA/cm2)
MgF2/Al
30 n-type 0.5 cm
10
FZ 10 cm n-type c-Si a-Si:H/1 nm MgF2/10 nm Al
102
101
Al 100
0
(b)
-10 -20
Extracted c:
-30
35 5 mcm2
c (mcm2)
Current (mA)
20
(a) 103
102
-40 -0.15
1 nm MgF2
-0.10
-0.05
0.00
0.05
0.10
2.5 nm MgF2
0.15
101
Voltage (V) Fig. 1: Representative current voltage (I–V) measurements of n-Si samples with and without a thin (~1 nm) MgF2 film.
Although it achieves a low ρc, the MgF2/Al stack provides little passivation to n-Si, exhibiting a J0c ~1500 fA/cm2 (i.e., equivalent to the J0c of Al metal directly on n-Si). This motivated us to explore the insertion of a thin passivating film such as intrinsic a-Si:H between MgF2 and c-Si. As shown in Figure 2(a), the surface passivation quality is markedly enhanced, showing a two orders of magnitude reduction in J0c. As the a-Si:H interlayer thickness increases, the J0c first decreases sharply and then saturates to ~10 fA/cm2 when the a-Si:H thickness exceeds 6.5 nm. This J0c value is comparable to those reported for n+ polysilicon/SiOx tunneling contacts [3, 4]. A possible drawback of inserting an intrinsic a-Si:H interlayer is that it can introduce additional resistances to current transport through the contact structure. Figure 2(b) presents the dependence of the measured ρc on a-Si:H thickness for two different MgF2 films, 1 nm and 2.5 nm. It can be seen that for both MgF2 thicknesses, ρc increases as the a-Si:H thickness increases. For the 1 nm MgF2 samples, ρc exhibits approximately one order of magnitude increase as the a-Si:H layer increases from 0 to 9.6 nm. When a-Si:H thickness exceeds 10 nm, the contact behaves in a rectifying fashion. Moreover, the ρc of the samples with 2.5 nm MgF2 are generally higher than those of samples with 1 nm MgF2. The increase in ρc for thicker a-Si:H or MgF2 films is likely due to the bulk resistivity of the a-Si:H and MgF2 materials. Although higher than that reported for n+ doped polysilicon contacts (~15 mΩ∙cm2) [4], the contact resistivity of all the samples in Figure 2(b) is low enough to permit the fabrication of efficient solar cells. For the proof-of-concept cell design, a 6.5 nm a-Si:H / 1 nm MgF2 / 300 nm Al stack was employed, as it provides a combination of reasonably low ρc (~76 mΩ∙cm2) and low J0c (~10 fA/cm2).
0
1
2
3
4
5
6
7
8
9 10 11 12
a-Si:H thickness (nm)
Fig. 2: Dependence of (a) ρc and (b) J0c on a-Si:H thickness for aSi:H/MgF2/Al electron-selective contacts.
III. DEVICE PERFORMANCE A. Device structure and fabrication
(a)
(b) SiNx Al2O3 p+
n-base a-Si:H n+ MgF2 SiNx Ag Al Partial rear contact (PRC) Full-area rear contact (FRC) MgF2 cell control cell Alkaline texturing Boron diffusion Phosphorus diffusion
Front ALD Al2O3 and PECVD SiNx Thermal activation of Al2O3
Rear PECVD SiNx
Rear PECVD a-Si:H
Rear contact patterning Rear thermally evaporated Ag
Rear thermal evaporated MgF2/Al
Front contact patterning and metallisation (evaporated and eletroplated Ag)
Fig. 3: Schematic cross-section and fabrication sequences of two front-junction n-type silicon solar cells: (a) partial rear contact (PRC) control cell, and (b) full-area rear contact (FRC) MgF2 cell.
Figure 3 depicts the schematic structures and fabrication sequences of the front-junction n-type silicon solar cells with two different types of rear electron contacts: (a) partial rear contact (PRC) to a full-area phosphorus (n+) diffusion, and (b) full-area rear contact (FRC) consisting of a 6.5 nm a-Si:H /
B. Device results and discussion Table I presents cell results of the two rear contact schemes. The FRC MgF2 cell has a high VOC at 687 mV and a reasonable fill factor (FF) of 77.3%, demonstrating that the good electron–selective characteristics of the contact (i.e., low recombination and low resistance) can be translated to the fabrication of efficient solar cells. The 20 mV higher VOC of the FRC MgF2 cell over the PRC control cell can be partly attributable to (i) a lower substrate resistivity, and therefore higher pn products, and (ii) a lower J0c associated with the aSi:H/MgF2/Al electron contact (i.e., ~10 fA/cm2). For the PRC control cell, the total J0 at the back surface is made up of ~25 fA/cm2, corresponding to the 99% area passivated with SiNx, plus ~15 fA/cm2 corresponding to the 1% metalcontacted area, adding up to a total J0 of ~40 fA/cm2 at back surface. The lower FF in the MgF2 cell is partly due to a significant shunt resistance loss in this particular device. It can further be attributed to the relatively high contact resistance of the a-Si:H/MgF2/Al structure (~76 mΩ∙cm2). In comparison, the PRC cell has an estimated contact resistance of 1–10 mΩ∙cm2 (0.01–0.1 mΩ∙cm2 for the Al direct contact to n+ surfaces, divided by a contact fraction of 1%). The short-circuit current (JSC) of the FRC MgF2 cell is 2.4 mA/cm2 lower than that of the control cell, primarily due to a significantly lower reflectance at the back surface, as to be discussed below. As a result, the final MgF2 cell has an efficiency of 20.1%, which is 1.4% lower than that of the PRC control cell.
Cell
TABLE I. Summary of cell results Substrate VOC JSC FF (Ω∙cm) (mV) (mA/cm2) (%)
η (%)
PRC control
2.3
667
40.2
80.1
21.5
FRC MgF2
0.5
687
37.8
77.3
20.1
To further investigate the differences between the two cell technologies, which are particularly significant in JSC, we performed reflectivity and spectral response measurements. In addition to the external quantum efficiency (EQE) and reflectance, Figure 4 shows the simulated absorption in the rear SiNx layer for the PRC control cell, and in the rear aSi:H/MgF2/Al stack for the FRC MgF2 cell. The absorption was simulated using the “Wafer Ray Tracer” available at the PVLighthouse website. A strong difference between the measured EQE of both cells can be noted in the infrared region; this is largely attributable to a poor back surface reflectance by the MgF2 structure, and probably to a significant absorption in the a-Si:H and/or Al layers [5]. The results indicate that the back surface mirror created by the aSi:H/MgF2/Al based contact is not as good as that provided
by the SiNx/Ag system. 100
EQE, reflectance, absorption (%)
1 nm MgF2 / Al structure. As Figure 3 indicates, an advantage of the MgF2 contact approach is the removal of (i) the hightemperature phosphorus diffusion, and (ii) the patterning of the rear dielectrics (by photolithography or laser processing). Although not shown in Figure 3, the process using two dopant diffusions requires additional masking steps, and poses restrictions on the process sequence.
90 80 70 60 50 40
PRC control cell: EQE Reflectance Absorption FRC MgF2 cell: EQE Reflectance Absorption
30 20 10 0 400 500 600
700 800 900 1000 1100 1200
Wavelength (nm) Fig. 4: Measured EQE, reflectance and simulated absorption of the partial rear contact device and full-area MgF2 rear contact device.
IV. CONCLUSION We have developed a novel a-Si:H/MgF2/Al electron– selective contact for silicon solar cells. Optimized contact structures present a ρc < 100 mΩ∙cm2, and a J0c ~ 10 fA/cm2 on n-Si. These parameters are appropriate for full rear contacts, and have enabled the fabrication of a 20.1% n-type front-junction proof-of-concept silicon solar cell. There is scope for significantly higher cell performance, particularly in terms of current and fill factor. Although quite different from current prevalent fabrication technology, the simplicity and low temperature of such a carrier–selective contact structure as presented here, opens up new possibilities for silicon solar cells. REFERENCES [1] J. Bullock, M. Hettick, J. Geissbühler, A. J. Ong, T. Allen, C. M. SutterFella, et al., "Efficient silicon solar cells with dopant-free asymmetric heterocontacts," Nature Energy, vol. 2, 2016. [2] Y. Wan, C. Samundsett, J. Bullock, D. Yan, T. Allen, P. Zheng, et al., "Magnesium Fluoride based electron contact for silicon solar cells," Nano Letters, In preparation 2016. [3] F. Feldmann, M. Bivour, C. Reichel, M. Hermle, and S. W. Glunz, "Passivated rear contacts for high-efficiency n-type Si solar cells providing high interface passivation quality and excellent transport characteristics," Solar Energy Materials and Solar Cells, vol. 120, Part A, pp. 270-274, 1// 2014. [4] D. Yan, A. Cuevas, J. Bullock, Y. Wan, and C. Samundsett, "Phosphorus-diffused polysilicon contacts for solar cells," Solar Energy Materials and Solar Cells, vol. 142, pp. 75-82, 11// 2015. [5] A. Uruena, L. Tous, F. Duerinckx, I. Kuzma-Filipek, E. Cornagliotti, J. John, et al., "Understanding the Mechanisms of Rear Reflectance Losses in PERC Type Silicon Solar Cells," Energy Procedia, vol. 38, pp. 801806, // 2013.
