In-situ deposition of tungsten oxide hole-contact by Hot-Wire CVD and its application in dopant-free heterojunction solar cells

The parasitic absorption of the p-type or n-type doped amorphous silicon layers in silicon heterojunction solar cells limits the further increase of photon generated current density [1, 2]. The dopant-free passivating contacts are of highly scientific and technical interest for silicon heterojunction solar cells due to their reduced parasitic absorption, as well as ease-of-deposition [3, 4]. Metal oxides and fluorides of low work function (WF) such as lithium fluoride [5], magnesium fluoride [6], tin oxide [7], and niobium oxide [8], are the candidates for electron selective contacts. On the other hand transition metal oxide of high WF like molybdenum oxide (MoO_x ) [9–11], tungsten oxide (WO_x ) [12, 13], vanadium oxide [14, 15] can be the candidates for hole selective contacts. The hole-selective contact is often used as the window layer, where materials of both wide bandgap and high WF is on the top in the candidate list. So far a cell conversion efficiency of 23.5% with MoO_x hole selective contact has been reported [9]. Yet the evaporated MoO_x tends to be unstable under elevated temperature during the subsequent processes of metallization and post annealing [16, 17]. Amongst the reported hole-selective contacts, WO_x has the largest optical band gap and can remain stable with the processing temperature up to 200 °C [18–20].

The stoichiometry value x of WO_x is crucial for its opto-electrical property influencing the cell performance [21]. Several deposition techniques have been reported for the preparation of WO_x films including magnetron sputtering, vacuum thermal vapor deposition [20, 22–24]. It is difficult to tune the composition of WO_x using thermal vapor deposition. Magnetron sputtering causes sputtering damage to the WO_x /Si interface [25]. Known as ion bombardment free and conformal deposition, the Hot-Wire chemical vapor deposition (HWCVD) is a relatively mild deposition process and the composition of the deposited film can be adjusted by adjusting the gas flow rate. Some literature has reported the use of similar deposition methods and studied the structure of as-deposited WO_x and tried to apply it in electrochemistry [26–28]. However, less attention has been paid to the effect on their opto-electrical properties. So far there are few reports on the WO_x thin films prepared by HWCVD for solar cells as the hole selective contact.

In this study, the WO_x films were prepared by a novel hot filament oxidation-sublimation process in a high-vacuum chamber. The oxygen flow ratio has been varied to moderate the material property of WO_x films. The composition, surface morphology and opto-electrical properties of these films were investigated. These WO_x hole selective layers were applied in dopant-free silicon heterojunction solar cells, resulting in high photon current densities of 39.6 mA·cm⁻² and conversion efficiencies of 14.9%.

All WO_x films were deposited by direct oxidation of the tungsten wire in a HWCVD chamber. The purity of the tungsten wire is 99.99%, and the wire diameter is 0.5 mm. Five straight tungsten wires were arranged in parallel with a length of about 22 cm. The distance between the tungsten wire and the substrate is about 8 cm. A gas mixture of oxygen and argon was adopted as the deposition gases. The oxygen flow ratio r_O2, as defined as the flow rate of oxygen (f_O2) divided by total flow rate (f_O2 + f_Ar), namely r_O2 = f_O2/(f_O2 + f_Ar), was varied in a range of 1.7%–6.7%. The total flow rate was kept at 80 sccm. The working pressure was 5 Pa. The hot wire current is 48 A, leading to an estimated filament temperature of about 1600 °C. All WO_x samples were co-deposited on glass and silver coated glass substrates at room substrate temperature. The sample on silver coated glass substrate was used for the Raman measurement (Thermo Scientific, USA) with the excitation laser wavelength of 473 nm. The film thickness was measured by a step profiler (Brucker DektakXT). The surface topography was measured by scanning electron microscopy (SEM, ZEISS EV0 MA15). The root-mean-square (RMS) roughness and surface potential of WO_x thin films were measured using atomic force microscopy (AFM, KEYSIGHT Technologies 7500) via tapping mode. The surface potential was determined by the contact potential difference between the Pt-coated conductive cantilever probe and the samples. The WF of the sample can be calculated. Further details on the calculation of the WF can be found elsewhere [11]. The optical transmittance of WO_x films was measured by a UV–VIS–NIR spectrophotometer (SHIMADZU UV-3600 Plus). The composition of samples was measured by the x-ray photoelectron spectroscopy (XPS, Thermo Fisher ESCALAB XI+). The sample surface was etched using clusters with an etching energy of 4 keV for 50 s. All core level spectra were calibrated using the C1s (284.8 eV) peak. The conductivity of WO_x films on a glass substrate was measured at room temperature with two silver pads in parallel.

The WO_x films prepared by HWCVD were applied in silicon heterojunction solar cells as the hole selective contact. The schematic drawing of the cell configuration and its photo are shown in figure 1. The solar cell area is about 4 cm². The n-type silicon wafer based structure of a-Si:H(i)/n-c-Si/a-Si:H(i)/a-Si:H(n⁺) was adopted. The wafers were soaked in hydrofluoric acid solution for 3 min before deposition. We tested the implied Voc for WO_x /a-Si:H(i)/n-Si contact with the Photoconductance Lifetime Tester (WCT-120, USA). The indium tin oxide film with a thickness of about 90 nm were prepared by magnetron sputtering. On top of the WO_x layer, the 100 nm-thick indium zinc oxide (IZO) films were prepared by sputtering. The cell was finished with silver contacts, and then annealed at 120 °C in air for 30 min. The current–voltage performance of the solar cells was measured with a sun simulator (WAVELABS Sinus-220, Germany) under standard conditions. The external quantum efficiency (EQE) of the solar cell was measured at room temperature.

Zoom In Zoom Out Reset image size

3.1. Material property of WO_x thin films

Figure 2 shows the deposition rate of WO_x films with the oxygen flow ratio r_O2 ranging from 1.7% to 6.7%. The film thickness of all samples is about 150 nm. The deposition rate increases linearly from about 5 nm·min⁻¹ at r_O2 = 1.7% to 50 nm·min⁻¹ at r_O2 = 6.7%, which indicates that high oxygen concentration favors for the growth of WO_x .

Zoom In Zoom Out Reset image size

Figure 3 shows the Raman spectra of WO_x films deposited with different r_O2. The Raman signals in the wavenumber region of 200–450 cm⁻¹ can be assigned to the bending vibrations of O–W–O bond vibration modes [29]. A peak band in the wavenumber range of 550–1000 cm⁻¹ can be observed. The peak centered at 714 cm⁻¹ and 804 cm⁻¹ can be assigned to the asymmetric and symmetric stretching vibrations of W⁶⁺−O bond, respectively, corresponding to the WO₃ phase [30]. These two signals tend to enhance with increasing oxygen flow, indicating the increasing of WO₃ composition. A peak centered at around 950 cm⁻¹ could be the stretching mode of unsaturated W=O bond [29], and it shifts to lower wavenumbers at higher r_O2. One could argue that this is due to the stoichiometric value x increases.

Zoom In Zoom Out Reset image size

The RMS roughness of these WO_x films measured by AFM increases generally from about 6.1 nm to 15.2 nm by increasing the r_O2. Figure 4 shows the SEM images of the corresponding WO_x samples. It is shown that small cauliflower-like clusters on the surface at r_O2 = 1.7% and the diameter of the clusters is about 50 nm. When the r_O2 increases to 3.4%, the cluster diameter increases to about 100 nm and the gap between clusters increases. As the r_O2 increases further to 5.1% and 6.7%, a number of larger clusters with the diameter over 200 nm seem to accumulate. This could be correlated with the high deposition rates, where the abundant WO_x particles combine in the gas phase prior to the deposition on the substrate. It is noted that the room substate temperature was applied. The largely combined particles may not integrate sufficiently on the substrate. Hence the films prepared at high oxygen flow ratio tend to show higher porosity [31].

Zoom In Zoom Out Reset image size

Figure 5 shows the XPS signal of the W4f core level of the WO_x samples with r_O2 = 1.7%–6.7%. The main peaks at the binding energies of 36.0 eV and 38.1 eV are attributed to the W⁶⁺ 4f_7/2 and 4 f_5/2, respectively. The additional shoulder at 34.8 eV can be the W⁵⁺ 4f_7/2, which is paired with the W⁵⁺4f_5/2 at 37.0 eV [20, 32, 33]. The presence of W⁵⁺ in all four samples indicates the existence of non-stoichiometric WO_x and possible oxygen vacancy. The W 4 f signal was deconvoluted as indicated by the dashed line in figure 5 for the calculation of W⁶⁺ and W⁵⁺ content ratio. When r_O2 = 1.7%, the W⁶⁺ ratio is 76.1%. As the r_O2 increases the W⁶⁺ ratio increases to 91.4%, while the W5⁺ ratio decreases. This is in agreement with the weakening of the Raman signal at 950 cm⁻¹ as shown in figure 3.

Zoom In Zoom Out Reset image size

As shown in figure 6(a), the surface potential (SP) of the WO_x film decreases from −0.5 V to about −1.1 V, as the oxygen percentage increases from 1.7% to 6.7%. The WF of these WO_x samples were calculated given the surface potential measured by Kelvin probe force microscopy (KPFM) [11]. The WF of the samples increases from about 5.5 eV to 6.1 eV with r_O2 increasing from 1.7% to 6.7%. The higher WF could be mainly caused by the more oxygen content [24, 34].

Zoom In Zoom Out Reset image size

The transmittance of WO_x thin films on the glass substrate are included in figure 7(a). We used air as the test baseline. The transmittance of WO_x /glass is higher than 80% in the wavelength range of 370–1200 nm. The optical bandgap E_tauc of these WO_x samples were evaluated using the Tauc plot model [35]. It is shown in figure 7(b) that all samples are highly transparent with E_tauc above 3.2 eV. By increasing the r_O2, the E_tauc increases slightly by about 0.1 eV. This could by mainly due to the increases stoichiometry value as indicated by the XPS and Raman results.

Zoom In Zoom Out Reset image size

The conductivity of these WO_x samples is in the order of 10⁻⁵ S·cm⁻¹ as shown in figure 8. The conductivity of the WO_x films prepared by HWCVD was about 2.1 × 10⁻⁵ S·cm⁻¹ when r_O2 = 1.7%. The conductivity decreases by half when the r_O2 increases to 6.7%.

Zoom In Zoom Out Reset image size

3.2. Dopant-free heterojunction solar cells with WO_x

The WO_x films deposited with oxygen flow ratio r_O2 = 1.7%–6.7% are used as the hole selective contact layer in dopant free silicon heterojunction solar cells. Given the deposition rate in figure 2 the thickness of WO_x hole selective contact is about 15 nm by controlling the deposition time. Figure 9 shows the current–voltage (JV) parameters of these solar cells with WO_x hole selective contact. As the r_O2 increases from 1.7% to 6.7%, the short-circuit current density (J_SC) increases from 37.5 mA·cm⁻² to nearly 38.0 mA·cm⁻². This is due to that the higher optical band gap with increasing r_O2. The implied open circuit voltage (V_OC) of WO_x films with different oxygen flow ratio r_O2 was tested on the silicon wafer. We found that the implied V_OC of all samples lightly increases from 694 mV to 699 mV with increasing the oxygen flow ratio. Yet the measured V_OC shows a significant decrease from about 640 mV to 550 mV with increasing r_O2. This is similar to what has been reported in other literatures, in which the V_OC of cells with either MoO_x or WO_x hole-selective transport layers decreases with increasing oxygen flow ratio [24, 36]. Such decrease of V_OC could be related to contact interface, band alignment with the transparent conductive oxide contact, and material quality. The FF decreases from 63% to about 45% with increasing r_O2. One of the reasons can be the lower conductivity of the WO_x layer, which leads to higher series resistance. As a result, the cell conversion efficiency (η) decreases from about 14.0% to about 10.0% with an increasing r_O2.

Zoom In Zoom Out Reset image size

Figure 10 shows (JV) parameters of cells as a function of WO_x layer thicknesses (5–30 nm). The oxygen flow ratio was 1.7% for this cell series. As the WO_x film thickness is less than 10 nm, the cell suffers from low JV parameters like J_SC = 37.0 mA·cm⁻², V_OC = 560 mV and FF = 53%. This is mainly due to the poor band bending when the hole selective contact is too thin. Additionally, the following sputtered IZO layer could cause a decrease in the WF of WO_x thin films, leading to the deterioration of hole selective transport in WO_x thin films [25, 35]. As the WO_x film thickness in the range of 10–20 nm, both J_SC and V_OC tend to maintain at 37.5 mA·cm⁻² and 640 mV, respectively, despite of a certain scattering in the data. While the FF shows a slight decrease from 63% to 55%. The JV parameters deteriorates severely as the WO_x film thickness increases further above 25 nm. Overall, the optimum WO_x film thickness is in the range of 10–20 nm as taking the cell conversion efficiency into account.

Zoom In Zoom Out Reset image size

Figure 11 shows the EQE of the best cell shown with 15 nm-thick WO_x hole selective contact as well as the reflection on the cell surface. In the 500–1000 nm wavelength range the average EQE of the cell reaches 96.6%. The optical losses of the best device is analyzed using the EQE and 100%-R data, as proposed by Holman et al [37]. The evaluated photon current density J_QE by EQE is about 39.6 mA·cm⁻² and the integrated current density of the reflected light is about 1.5 mA·cm⁻². Considering the wide bandgap (>3.0 eV) of WO_x and IZO films on the front side of the solar cell [38], it is suggested that the parasitic absorption is mainly caused by the intrinsic amorphous silicon passivation layer in the 300–600 nm band. The current loss of the backside is caused by the poor reflection effect of the non-mirrored rear silver contact [14].

Zoom In Zoom Out Reset image size

The deposition of WO_x films by the HWCVD technique with oxygen flow ratio r_O2 range of 1.7%–6.7% was investigated. The opto-electronical properties of WO_x films were analyzed in correlation with the material composition and structure. The WO_x films with WF of 5.5–6.1 eV, the conductivity of 10⁻⁵–10⁻⁶ S·cm⁻¹, and the optical band gap of 3.2–3.4 eV were obtained. These WO_x layers were applied in dopant free silicon heterojunction solar cells as the hole selective layer. Thanks to the highly transparent WO_x layer, a high photon current density of 39.6 mA·cm⁻² in EQE measurement was achieved. It is found that the optimum cell efficiency of 14.9% was obtained with the WO_x layer thickness of 10–20 nm and low oxygen flow ratio r_O2 = 1.7%. Despite of the high oxygen flow ratio favoring the formation of stoichiometric WO_x , our study shows that the solar cell JV parameters especially open circuit voltage and fill factor deteriorates at higher oxygen flow ratio. The poor interface between the intrinsic passivation layer and the WO_x contact could be the main cause. One might improve the interface property by adjusting the filament temperature or substrate temperature. Nevertheless, this work shows the feasibility of WO_x hole selective contact by HWCVD for silicon heterojunction solar cells and good potential for its industrial application.

This work is supported by the National Natural Science Foundation of China (Nos. 51702269 and 61904154), Innovation Research Team Project of Southwest Petroleum University (No. 2019CXTD10), Sichuan Science and Technology Program (No. 2022YFG0229), Chengdu Science and Technology Project (Nos. 2022-YF05-00384-SN and 2020-GH02-00014-HZ). The authors would like to thank Xianhui Cui, Yan Xiang and Haichuan Zhang for their assistance with the measurements.

All data that support the findings of this study are included within the article (and any supplementary files).