High‐Performance and Stable Dopant‐Free Silicon Solar Cells with Magnesium Acetylacetonate Electron‐Selective Contacts

One of the challenges in fabricating high‐performance n‐type crystalline silicon (n‐type c‐Si) solar cells is the high‐quality n‐type c‐Si/metal contact. Schottky barriers are commonly found on the n‐type c‐Si/metal contact, which suppresses electron transportation. Herein, novel stacks of magnesium acetylacetonate (Mg(Acac)2)/magnesium (Mg)/silver (Ag) to form electron‐selective contacts for n‐type c‐Si solar cells are presented, which enables a dopant‐free process. An ohmic contact on n‐type c‐Si is formed using the Mg(Acac)2/Mg/Ag stacks. The transmission spectrum and ultraviolet photoelectron spectroscopy measurements show negligible conduction‐band offset and large valence‐band offset between Mg(Acac)2 and n‐type c‐Si, which indicates the electron‐transporting and hole‐blocking properties of Mg(Acac)2/n‐type c‐Si heterocontacts. Moreover, the contact resistivities (ρc) between the Mg(Acac)2/Mg/Ag electron‐selective heterocontacts and n‐type c‐Si substrates are lower than 10 mΩ cm2, which demonstrates the good electrode properties of the Mg(Acac)2/Mg/Ag stacks. The Mg(Acac)2/Mg/Ag electron‐selective stacks are applied on n‐type c‐Si solar cells with partial rear contact, and >20% efficiency is achieved, which is higher than that in a reference cell with only Ag contact. The stability of the n‐type c‐Si solar cell performance equipped with Mg(Acac)2/Mg/Ag contacts is verified under ambient conditions. This novel low‐temperature contact technique offers a reliable alternative for high‐performance n‐type c‐Si solar cells.

high-temperature treatments, which give rise to a high energy consumption. Moreover, heavy doping in Si results in η loss due to extra Auger recombination and free carrier absorption. [10,11] Another solution for n-type c-Si contacts is the application of electron-selective materials. Low-work-function materials (e.g., lithium fluoride (LiF) [12,13] and magnesium (Mg) [14] ) or materials with small conduction-band offsets with Si (TaN x ) are applied as electron-selective materials in n-type c-Si contacts. [15] In addition, oxides play a vital role in selective contacts, such as TiO x and ZnO. [16,17] N-type c-Si solar cells using TiO x /LiF x /Al electron-selective heterocontacts exhibit a low contact resistivity of %5 mΩ cm 2 and achieved 23% energy conversion efficiency in a recent study. [18] However, low-workfunction materials, such as LiF, suffer from instability in ambient environments. Thus, a stable electron-selective contact for n-type c-Si with a simple fabrication process is still of great interest. Bullock et al. reported dopant-free asymmetric heterocontact Si solar cells with efficiencies above 20% using a TiO x insert layer between c-Si and LiF/Ag. This triple layer ensures a thermal stability of c-Si solar cells after undergoing 300 C post-annealing treatment. [19] Metal acetylacetonates have been used as stable electronselective heterocontacts for perovskite solar cells. [20] In this work, we present a novel contact technique for electron-selective contacts on n-type c-Si solar cells using a magnesium acetylacetonate (Mg(Acac) 2 ) insert layer between n-type c-Si and Ag that features ambient stability. The Mg(Acac) 2 organic material serving as an insert layer between n-type c-Si and Mg is introduced here to achieve stable and high-quality electron-selective contacts for the first time. Moreover, the electronic band structure and application of n-type c-Si solar cells in the thermal-evaporated Mg(Acac) 2 /Mg/Ag stacks are discussed. The stability of proofof-concept contact structure and performance discrepancy with different contacts is revealed. This study aims at closing the gap between high-performance n-type c-Si solar cell devices and stable dopant-free contacts with a simplified process and presents the first results of Mg(Acac) 2 on n-type c-Si solar cells.
The atomic structure of Mg(Acac) 2 is shown in Figure 1a. Mg(Acac) 2 films (12 nm) were evaporated on n-type c-Si (100) substrates (1-3 Ω cm) for the X-ray photoelectron spectroscopy (XPS) characterization to obtain the bonding information of Mg(Acac). The chemical states of Mg, C, and O in Mg(Acac) 2 studied by XPS are shown in Figure 1b-d. The Mg1s spectrum exhibits a typical peak at 1303.1 eV. [21] The peak analysis of the C1s spectrum shows the presence of four bonding types of the carbon atoms. Two predominant peaks at 285.1 and 284.2 eV are attributed to sp 3 C─C and sp 2 C═C bonds, and the peaks at %289.0 and 286.7 eV are attributed to the C═O and C─O bonds. [22] In the O1s XPS spectrum, the peaks of the O─Mg (532.4 eV), C═O (531.3 eV), and C─O (530.1 eV) bonds are observed. [23] All of the bonds in Mg(Acac) 2 are found in the XPS spectra. The atomic ratio of Mg:C:O is %5:16:12, which is not close to that of the raw Mg(Acac) 2 powder, which is 1:5:2. Some of the missing C may be oxidized by rare O 2 in the chamber during the Mg(Acac) 2 thermal evaporation process.
To study the mechanisms underlying the electron-selective transport of the Mg(Acac) 2 /n-type c-Si heterocontact, the band alignment at the Mg(Acac) 2 /n-type c-Si interface was investigated. A fit was applied in the transmittance spectrum of Mg(Acac) 2 , giving an optical bandgap value (E g ) of 3.9 eV, as shown in  www.advancedsciencenews.com www.pss-rapid.com Figure 2a. Figure 2b shows the ultraviolet photoelectron spectroscopy (UPS) spectrum of the Mg(Acac) 2 thin film using a He I excitation (21.22 eV). The work function (Φ) of 4.2 AE 0.05 eV was calculated by considering the onset (16.9 AE 0.05 eV) in the UPS spectrum at high binding energy. [15] The gap between the Fermi level and valence band maximum (E f À E v ) in Mg(Acac) 2 , obtained from the linear fit in the UPS spectrum with the inset, is 3.8 AE 0.05 eV. The work function and (E f À E v ) values of the Si substrate are 4.4 and 1.0 eV, respectively, obtained from the UPS spectrum. [15] The band alignment diagram at the Mg(Acac) 2 /n-type c-Si interface shows carrier selectivity, as shown in Figure 2c. Two desirable properties of this Mg(Acac) 2 insert layer are demonstrated: 1) a negligible conduction-band offset between the n-type c-Si and Mg(Acac) 2 layer, which enables the free transport of electrons and 2) a large valence-band offset (ΔE v ¼ 2.6 eV), which effectively blocks the hole transportation. The efficient electron transportation of the Mg(Acac) 2 /Mg/Ag structure can be addressed by its contact resistivity via the transfer length method (TLM). To evaluate the performance of the n-type c-Si/Mg(Acac) 2 heterocontact, the contact resistivity (ρ c ) with the different Mg(Acac) 2 thicknesses is measured by TLM, as shown in Figure 2d. The results show that ρ c increases as the thickness of the Mg(Acac) 2 layer increases. ρ c is below 10 mΩ cm 2 when the Mg(Acac) 2 layer is ≤5 nm, showing a promising contact for n-type c-Si. A high electron selectivity was demonstrated by the low contact resistivity.
N-type c-Si solar cells with a partial rear contact were fabricated to demonstrate the quality of the Mg(Acac) 2 /Mg/Ag heterocontact on the device. Thermal SiO 2 growth and plasmaenhanced chemical vapor deposition (PECVD) of a SiN x passivation/dielectric spacer were patterned with 30 μm-diameter holes to the c-Si surface on the rear side. Through these holes, the Mg(Acac) 2 /Mg/Ag stack contacted the n-type c-Si surface, directly forming the electron-selective contact shown in Figure 3a. The Ag/Mg/Mg(Acac) 2 /n-type c-Si contact was observed via scanning transmission electron microscopy (STEM) acquired with a high-angle annular dark field (HAADF) in combination with energy-dispersive X-ray spectroscopy (EDX), as shown in Figure S1, Supporting Information. From the elemental mapping in Figure S1b, Supporting Information, a Mg(Acac) 2 /Mg interlayer stack between Si and Ag is observed. Mg(Acac) 2 is too thin, making it difficult to distinguish with the Mg layer. The results highlight that the Mg and Mg(Acac) 2 layers separate the Si from the Ag layer. The oxygen signal in Figure S1c, Supporting Information, in the EDX mapping came from Mg(Acac) 2 as the material has a molecular structure, as shown in Figure 1a, which is consistent with the STEM observation.   Figure 2. a) Determination of the optical bandgap from the transmittance spectrum for the Mg(Acac) 2 . b) UPS spectrum of the Mg(Acac) 2 film using He I excitation; a zoom for the E f -E v value linear fit is included in the inset. c) Band alignment at the n-type c-Si/Mg(Acac) 2 interface. d) ρ c as a function of the Mg(Acac) 2 thickness; TLM pattern structure is included in the inset.
www.advancedsciencenews.com www.pss-rapid.com properties for solar cells. The solar cells with Mg(Acac) 2 /Mg/Ag heterocontacts were investigated at 25 C in an ambient environment. The J-V curves were periodically measured to extract device parameters. In the inset of Figure 3b, no significant change in the Mg(Acac) 2 /Mg/Ag heterocontact solar cell performance, which maintained more than 97.8% of η, can be observed over the specified period. As a comparison, the performance of the reference cell using LiF/Al electrodes rapidly degraded to 83.6% of η. This finding indicates that the Mg(Acac) 2 /Mg/Ag contact shows good stability in an ambient environment. Figure 3c shows the external quantum efficiency (EQE), solar cell reflection spectrum (R), and integrated J sc . In the EQE spectrum, a good carrier collection in the whole wavelength range of the n-type c-Si solar cells applying Mg(Acac) 2 /Mg electron-selective contacts was demonstrated. Moreover, an integrated J sc of 42.36 mA cm À2 was obtained from the EQE that was measured using a spot exclude grid, which shows a slight discrepancy with the values obtained via the J-V measurement considering %4% coverage of the front Ag contact. The actual J sc of the solar cells was close to 40.67 mA cm À2 ; hence, an efficiency of champion cell is below 21.3% and above 20%. A possible reason for this discrepancy is attributed to the calibration issue of the currentvoltage tool. The measured J-V results are still suitable for the comparison of different cell performances with varying rear contact stacks as they are performed under the same test conditions. Table 1 shows the champion solar cell performance with V oc , FF, J sc , and η using different rear contact stacks. Wan et al.
believed that a poor electron transport to electrode is suffered in the n-type c-Si/Ag contact due to the rectifying contact at the rear side, although the cell without the Mg layer has a similar carrier generation and recombination. [14] Moreover, a low J sc was obtained in the direct Ag contact in this study. Mg was inserted between n-type c-Si and Ag to achieve an ohmic contact, resulting in a high J sc of 41.26 mA cm À2 . Mg has one of the lowest work functions in metals, making it very suitable in forming an electron-conductive contact for Si solar cells. [14,21] Wu et al. suggested that recombination can be suppressed by introducing low-work-function metals to obtain a high J sc . [24] High-quality electron selectivity was achieved by reducing recombination from the direct metal-semiconductor contact and high-performance n-type c-Si solar cells with a Mg(Acac) 2 insert layer. Mg(Acac) 2 /Mg/Ag stacks improved solar cells' η from 20.6% to 21.3% compared with the direct Ag contact, which resulted in a higher J sc . A higher EQE in the 400-900 nm wavelength region with the Mg(Acac) 2 /Mg/Ag stack was compared with  www.advancedsciencenews.com www.pss-rapid.com the Ag contact shown in Figure S2c, Supporting Information, which demonstrates that a better carrier collection was performed by the Mg(Acac) 2 /Mg/Ag electron-selective contact even though the exact mechanism is still unclear. Chee revealed that the electron affinity, surface band bending, and charge transfer properties are changed because of the strong surface work function dependence on the Si surface. [25] Therefore, carriers are more easily extracted from the Si substrate by applying low-work-function Mg(Acac) 2 /Mg/Ag contacts that gave a low contact resistivity, high J sc , and high EQE.
In conclusion, we present stable dopant-free Mg(Acac) 2 /Mg/Ag stacks as electron-selective heterocontacts for n-type c-Si. Fabricated in room temperature, the n-type c-Si/Mg(Acac) 2 /Mg/Ag contacts showed a low contact resistivity below 10 mΩ cm 2 . The band alignment analysis results extracted from the transmittance spectrum and UPS measurements show an electron-selective property between n-type c-Si and Mg(Acac) 2 . Moreover, ambient stability was demonstrated in the solar cells with the Mg(Acac) 2 / Mg/Ag electron-selective heterocontacts. These results prove that the Mg(Acac) 2 /Mg/Ag stacks can be alternatives for an n-type c-Si contact in high-performance solar cells because of their good contact, simple process, and good stability.

Experimental Section
XPS (Escalab 250Xi, Thermo Fisher Scientific, Waltham, MA, USA) was performed to characterize the components of Mg(Acac) 2 . UPS measurements (Escalab 250Xi using the monochromated He I radiation [21.2 eV]) were used to determine the valence band and work function of Mg(Acac) 2 . For the XPS and UPS characterization, 12 nm Mg(Acac) 2 films were evaporated on n-type c-Si (100) substrates (1-3 Ω cm). Ar etch (2 min) for the sample surface was conducted to clean the sample surface before the UPS and XPS measurements. Mg(Acac) 2 (30 nm) was deposited on 1.1 mm-thick Corning Eagle glass for the transmission measurement. The optical properties were obtained using a UV-vis-NIR spectrophotometer (U4100, HITACHI), and the wavelength step was 1 nm from 300 to 1200 nm. To fabricate the electron-selective heterocontact, Mg(Acac) 2 films with different thicknesses were deposited on the n-type Cz c-Si surface (1.0 Ω cm) at room temperature via thermal evaporation. A growth rate of %0.1 Å s À1 was used for the deposition. Following this, Mg (%10 nm)/Ag (%300 nm) metal stacks were evaporated in the same system. The purity of Mg(Acac) 2 used in this study is 99.9%. Then, 99.99% Mg and Ag particles were used to deposit the metal electrode. Thermal evaporation depositions were conducted with a base chamber pressure of %8 Â 10 À4 Pa. The distance from the source to the substrate is 38 cm. Deposition rates of 0.50 and 0.50 nm s À1 were used for Mg and Ag, respectively. The film thickness of Mg(Acac) 2 was determined by fitting polarized reflectance using the Tauc-Lorentz model in the spectroscopic ellipsometry measurement (SENTECH SE 800 PV ellipsometer). [26] The TLM for the contact resistivity measurement was carried out with a Keithley 2000 source meter.
N-type Cz c-Si wafers with a resistivity of 1.0 Ω cm and a thickness of 180 μm were used for solar cell fabrications. A homogeneous p þ emitter was formed by boron diffusion furnace after texturing and cleaning. The n-type c-Si solar cells were fabricated with double-boron diffusion on the front to create localized heavily doped p þþ regions under the front metal contacts. One-side wet bench was conducted for removing the rear p þ region. An Al 2 O 3 /SiN x passivation and antireflection stack was deposited on the front side via atomic layer deposition and PECVD, respectively. The front contacts, which only took up %4% of the front surface, were conducted by screen-printing Ag paste and firing. A peak temperature of %770 C was used for the firing process of the cells. Then, different stacks, such as direct Ag, Mg/Ag, and Mg(Acac) 2 /Mg/Ag, were deposited by thermal evaporations with a partial rear contact via 30 μm-diameter holes opened by a picosecond laser. The fabrication process of the n-type solar cells with the Mg(Acac) 2 /Mg/Ag selective contact is shown in Figure S3, Supporting Information. We measured the illuminated J-V behavior of the solar cells under standard 1 sun conditions (100 mW cm À2 , AM1.5 spectrum, 25 C) using a solar simulator (Class AAA, Oriel Sol3A, Newport). Reflection measurements were performed using a double-beam spectrophotometer (Cary 5000, Agilent Technologies) in the range of 300-1200 nm. The EQE spectrum was measured in the direct current mode on a spectrum corresponding system (Enlitech QE-R). For the stability research, the solar cells were exposed in ambient air at room temperature without illumination and heating for 120 h, and the J-V measurement was performed after every 24 h. The cross section of the n-type c-Si/ Mg(Acac) 2 /Mg/Ag contact stack on the rear opening area was observed with a FEI (Field Electron and Ion Company) Titan Themis G2 200 STEM. STEM microscopy images were acquired with HAADF in combination with EDX.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.