Research Article www.acsami.org
Cite This: ACS Appl. Mater. Interfaces 2019, 11, 25140−25146
Sulfur-Enhanced Field-Effect Passivation using (NH4)2S Surface Treatment for Black Si Solar Cells Dae Woong Kim,‡ Jae-Won Song,‡,† Hyun Soo Jin, Bongyoung Yoo,* Jung-Ho Lee,* and Tae Joo Park* Department of Materials Science and Chemical Engineering, Hanyang University, Ansan 15588, Korea
Downloaded via NOTTINGHAM TRENT UNIV on July 18, 2019 at 10:48:37 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: We demonstrated surface passivation of a black Si-based solar cell using an (NH4)2S solution to mitigate surface recombination velocity. Incorporated S at the interface between atomic-layer-deposited Al2O3 and black Si by (NH4)2S solution treatment boosted the density of negative fixed charges, S-enhanced field-effect passivation. Furthermore, NH4OH generated during (NH4)2S solution treatment removed the defective Si phase at the black Si surface, the surface cleaning effect. The optimized (NH4)2S solution treatment significantly enhanced the internal quantum efficiency up to ∼17.2% in the short wavelength region, suggesting suppressed surface recombination. As a result, photoconversion efficiency of the cell increased from 11.6 to 13.5%, by 16% compared to the control cells without (NH4)2S solution treatment. KEYWORDS: black Si, Si photovoltaics, field-effect passivation, S passivation, (NH4)2S
■
INTRODUCTION For high-efficiency crystalline Si solar cells, a nanostructured surface is required to maximize the optical absorption via a light-trapping phenomenon, which increases the optical path of the sunlight. Si nanostructures, such as nanowires and nanoholes, allow the negligible optical reflection of incident light, as their feature size is smaller than the sunlight wavelength.1 Therefore, black Si has been favored as an alternative and simple antireflection strategy, which does not require an antireflection coating layer such as SiN.2,3 Assorted methods to fabricate black Si have been reported, such as metal-assisted chemical etching (MaCE), reactive ion etching, vapor−liquid−solid growth, and so forth.2−7 However, during black Si fabrication, the Si surface is prone to contamination by metal impurity residues as well as damage by amorphization and nanoporous Si (np-Si) formation, resulting in a high interface defect state density. Furthermore, an SiO2 layer containing positive fixed charges is easily grown on the highindex Si planes exposed during the nanostructuring and passivation process.8−10 These enhance the surface recombination of photogenerated carriers, leading to a poor blue response of the internal quantum efficiency (IQE),11 which significantly deteriorates the photoconversion efficiency (PCE). Therefore, conventional chemical and field-effect passivations using a thermal oxide and the atomic-layerdeposited (ALD) Al2O3 layer, respectively, are insufficient for achieving a high PCE in a black Si solar cell.12 Prior to this report, the use of the ALD Al2O3 passivation layer has been the only way to generate the negative fixed © 2019 American Chemical Society
charge at the interface with an Si substrate. In this work, however, we have demonstrated an increase in the density of the negative fixed charge (−Qf), thereby enhancing the fieldeffect passivation via surface S passivation, using a simple and cost-effective (NH4)2S solution treatment prior to ALD of the Al2O3 layer. As a result, a significant increase in the PCE of the black Si solar cell was achieved. Furthermore, surface S passivation removes a defective surface structure from the black Si surface. Compared to a bare black Si solar cell, the IQE increased up to ∼71% in the blue response region, which results in a 33% PCE increase. The Qf at the interface (with a Si substrate) was evaluated using a metal−insulator−semiconductor (MIS) device. The microstructure of the black Si, including the surface/interface structure, was observed by a high-resolution transmission electron microscope (HRTEM) and field-emission scanning electron microscope (FESEM). Chemical analyses with time-of-flight secondary ion mass spectroscopy (ToF-SIMS), Raman spectroscopy, and energydispersive X-ray fluorescence (EDXRF) were performed to monitor the S incorporation behavior at the interface with black Si.
■
RESULTS AND DISCUSSION Due to difficulties in fabrication and characterization of the MIS device containing a black Si substrate, as well as the Received: March 29, 2019 Accepted: June 17, 2019 Published: June 17, 2019 25140
DOI: 10.1021/acsami.9b05589 ACS Appl. Mater. Interfaces 2019, 11, 25140−25146
Research Article
ACS Applied Materials & Interfaces
Figure 1. ToF-SIMS elemental depth profiles of the ALD Al2O3 layer on a flat Si substrate (a) without and (b) with (NH4)2S solution treatment for 5 min, (c) 20 min, and (d) 30 min. The S concentration at the interface increased with (NH4)2S solution treatment time. (e) S areal density of the (NH4)2S solution-treated Si substrate measured by EDXRF increased with treatment time. (f) Typical C−V curves of the MIS devices with and without (NH4)2S solution treatment at 22% for various times, and (g) the negative fixed charge density calculated from the C−V curves as a function of (NH4)2S solution treatment time, which increased up to 20 min of (NH4)2S solution treatment and decreased with further increase in treatment time. (h) Surface roughness of the Si substrate observed by atomic force microscopy (AFM) as a function of (NH4)2S solution treatment time. (i, j) Typical C−V curves of the MIS devices with solution treatments at the lower and higher concentrations (11 and 44%) than 22%, respectively, for various times.
difficulty in estimating the S concentration on the black Si surface, flat Si substrates were treated with (NH4)2S solution for various times prior to their application to black Si. This facilitated the estimation of the S concentration and the corresponding change in −Qf at the interface between the ALD Al2O3 layer and Si. Figure 1a−d shows the ToF-SIMS elemental depth profiles for an ALD Al2O3/flat Si substrate without surface S passivation and with surface S passivation using (NH4)2S solution treatment for 5, 20, and 30 min, respectively. The adsorbed S on the Si substrate by (NH4)2S solution treatment was maintained even after a following ALD of a 10 nm-thick Al2O3 layer at 250 °C. The S concentration at the interface between the Al2O3 and Si substrate (highlighted by a red shadow) increased as (NH4)2S solution treatment time increased from 5 to 30 min. This was confirmed by the EDXRF result in Figure 1e, which show that the S areal density of the (NH4)2S solution-treated Si substrate increased with treatment time. The calculated numbers of equivalent S monolayers were ∼21−24, but it should be considered that the amount of physisorbed S on the surface was included in these values. Thus, (NH4)2S solution treatment can be considered a simple, inexpensive, and effective S incorporation/passivation method, even at room temperature. This solution treatment is
superior to high-temperature annealing in H2S ambient, which deteriorates the interface/surface properties, doping profile in Si emitters, and growth of an interfacial SiOx layer containing positive fixed charges. The MIS device was fabricated with a patterned TiN top electrode using ALD Al2O3/flat Si structures without and with (NH4)2S solution treatment (TiN/Al2O3/p-type Si), from which the C−V curves were obtained for the evaluation of Qf. Figure 1f shows the normalized typical C−V curves for the MIS devices. The VFB increased from ∼0.66 to 1.05 V with the (NH4)2S solution treatment time being up to 20 min, which implies that the −Qf increased from ∼3.1 to ∼4.5 × 1012 cm−2, as summarized in Table 1 and Figure 1g. Qf was calculated from the VFB values using eq 1. Table 1. Flat Band Voltage and Negative Fixed Charge Density Calculated from the C−V Curves Measured Using MIS (TiN/Al2O3/p-Si) Devices with (NH4)2S Solution Treatment for Various Times
25141
(NH4)2S treatment (min)
0
5
10
20
30
VFB (V) −Qf (1012/cm2)
0.658 3.12
0.838 3.83
0.935 4.10
1.046 4.53
0.719 3.95
DOI: 10.1021/acsami.9b05589 ACS Appl. Mater. Interfaces 2019, 11, 25140−25146
Research Article
ACS Applied Materials & Interfaces
Figure 2. Cross-sectional SEM images of the black Si (a) before and after (NH4)2S solution treatment for (b) 5 min, (c) 10 min, (d) 20 min, and (e) 30 min. The inset figures show the corresponding SEM plane-view images. The length and density of Si nanowires decreased with increasing (NH4)2S solution treatment time. (f, g) Length, thickness, relative surface area, and filling factor of Si nanowires in black Si as a function of (NH4)2S solution treatment time.
Figure 3. HRTEM images of Si nanowires in the black Si (a) before and after (NH4)2S solution treatment for (b) 5 min, (c) 10 min, (d) 20 min, and (e) 30 min. The defective surface phase was gradually removed with increasing (NH4)2S solution treatment time, but further increases in treatment time induced the defective surface structure. (f) HRTEM image of Si nanowires after (NH4)2S solution treatment for 30 min and (g) corresponding EDS line profile results confirmed the formation of a damaged surface phase. (h) Photoluminescence (PL) spectra measured at room temperature for the black Si before and after (NH4)2S solution treatment for various times showing that surface carrier recombination behavior by varying the treatment time is consistent with the HRTEM results. (i) Raman spectra and (j, k) relative concentrations of S absorbed on black Si and those per unit surface area of black Si measured by EDXRF as a function of (NH4)2S solution treatment time, respectively. These results indicate that S concentration per unit surface area of black Si increased with solution treatment time.
Qf =
Ci Δ⌀ms − VFB Aq
surface by (NH4)2S solution treatment replaced O in the Al2O3 near the interface. This increased the number of tetrahedrally coordinated Al in Al2O3, because the larger radius of S compared to O15 induced a close-packed array with Al atoms that occupy two-thirds of the tetrahedrally coordinated sites when bonded with Al atoms. This leads to an increase in −Qf; this is called “S-enhanced field-effect passivation”.16−18 However, a further increase in the (NH4)2S solution treatment time to 30 min decreased −Qf due to physicochemical damage to the Si surface via excess (NH4)2S solution treatment time. Figure 1h shows the surface roughness of the
(1)
Here Ci, A, q, and Δ⌀ms represent the capacitance of the insulator, the capacitor area, unit charge, and work function difference, respectively. Therefore, the increase in −Qf with (NH4)2S solution time is attributed to an increase in S concentration at the interface between Al2O3 and Si. In general, negative fixed charges at the interface between Al2O3 and Si are induced by tetrahedrally coordinated Al in Al2O3 near the Si interface.13,14 Here, the S absorbed on the Si 25142
DOI: 10.1021/acsami.9b05589 ACS Appl. Mater. Interfaces 2019, 11, 25140−25146
Research Article
ACS Applied Materials & Interfaces
Figure 4. (a) Typical J−V curves, (b) optical absorbance and EQE of the ALD Al2O3-passivated black Si solar cells with and without (NH4)2S solution treatment for various times. (c) FF and shunt resistance of the cells as a function of (NH4)2S solution treatment are summarized. (NH4)2S solution treatment time was optimized as 10 min. (d) IQE result of the cells with and without (NH4)2S solution treatment for various times suggests that the surface carrier recombination was suppressed by optimized (NH4)2S solution treatment. IQE results of (e) 10 and (f) 30 min(NH4)2S-solution-treated solar cells with and without deionized (DI) water rinse, revealing the level of contributions of S-enhanced field-effect passivation and surface cleaning to IQE.
2f. Concurrently, the Si nanowire filling ratio estimated from the several plan-view SEM images also decreased, from 96.6 to 45.4%. As a result, the surface area of the black Si decreased to 28.8% after (NH4)2S solution treatment for 30 min (Figure 2g and Table S1). This is because NH4OH generated from the (NH4)2S solution etched the Si nanowires, as mentioned above. Because the surface structure of Si nanowires is defective (prone to being etched compared to that of a flat Si substrate), the Si nanowires were obviously etched from the beginning of the (NH4)2S solution treatment. As discussed above, while overetching by NH4OH can cause black Si damage as well as an increase in reflectance, an optimized treatment removes the surface recombination sources, which include the defective surface structure generated during a MaCE process and the surface-agglomerated boron-rich layers ( 2NH4OH + H2S] and etched the Si surface.19 As a result, the high-index Si planes such as (110) and (111) were exposed, increasing the number of positive fixed charges at the interface and thus decreasing −Qf.11,15 Therefore, the optimized (NH4)2S solution treatment time for a flat Si substrate in terms of −Qf was 20 min. This optimization depends on the concentration of (NH4)2S in the solution. For the lower concentration (NH4)2S solution (11%), excessive (NH4)2S was not present to damage the surface, and therefore, the −Qf monotonously increased with treatment time up to 30 min as shown in Figure 1i. Conversely, the higher concentration (NH4)2S solution (44%) treatment induced the damage on the Si surface even after a short time, resulting in decreased −Qf with increasing treatment time (Figure 1j). The (NH4)2S solution treatment was utilized for black Si prepared using a MaCE process. Figure 2a−e shows the crosssectional FESEM images of the black Si substrates before and after (NH4)2S solution treatments for 5−30 min. A black Si contains a number of Si nanowires with the length of ∼560 nm, which decreased to 530, 500, 480, and 380 nm after (NH4)2S solution treatments at 5, 10, 20, and 30 min, respectively. In particular, due to the structural characteristics of the Si nanowire, etching was primarily observed at the tip of the nanowire, because lots of tiny pores were caused at the tip of the Si nanowires during the etching process, which enhanced the etch rate in the direction of length compared to the diameter.20,21 This resulted in a larger decrease in the length of nanowire than in the diameter, which was estimated from several cross-sectional SEM images, as shown in Figure 25143
DOI: 10.1021/acsami.9b05589 ACS Appl. Mater. Interfaces 2019, 11, 25140−25146
Research Article
ACS Applied Materials & Interfaces
ance of the cell rather decreased with increasing (NH4)2S solution treatment time, as shown in Figure 4b. The external quantum efficiency (EQE) of the cells with and without (NH4)2S solution treatment for 10 min was compared in the inset of Figure 4b. The EQE of the cell with (NH4)2S solution treatment was lower in the long wavelength region (>500 nm) due to a decrease in optical absorbance but was considerably higher in the short wavelength region (
