Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES
Flexible Solar Cells Using Doped Crystalline Si Film Prepared by Selfbiased Sputtering Solid Doping Source in SiCl/H Microwave Plasma 4
Ping-Yen Hsieh, Chi-Young Lee, and Nyan-Hwa Tai ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.5b11151 • Publication Date (Web): 27 Jan 2016 Downloaded from http://pubs.acs.org on February 6, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ABSTRACT An innovative approach known as self-biased sputtering solid doping source process to synthesize doped crystalline Si film on flexible polyimide (PI) substrate via microwave plasma enhanced chemical vapor deposition (MWPECVD) using SiCl4/H2 mixture was developed. In this process, P and/or B dopants were introduced by sputtering the solid doping target through charged-ion bombardment in situ during high-density microwave plasma deposition. A strong correlation between the number of solid doping targets and the characteristics of doped Si films was investigated in detail. The results show that both P- and B-doped crystalline Si films possessed a dense columnar structure, and the crystallinity of these structures decreased with increasing the number of solid doping targets. The films also exhibited a high growth rate (>4.0 nm/s). Under optimal conditions, the maximum conductivity and corresponding carrier concentration were, respectively, 9.48 S/cm and 1.2 × 1020 cm−3 for P-doped Si film, and 7.83 S/cm and 1.5 × 1020 cm−3 for B-doped Si film. Such high values indicate that the incorporation of dopant with high doping efficiency (around 40%) into the Si films was achieved regardless of solid doping sources used. Furthermore, a flexible crystalline Si film solar cell with substrate configuration was fabricated by using the structure of PI/Mo film/n-type Si film/i-type Si film/p-type Si film/ITO film/Al grid film. The best solar cell performance was obtained with an open-circuit voltage of 0.54 V, short-circuit current density of 19.18 mA/cm2, fill factor of 0.65, and high energy conversion of 6.75%. According to the results of bending tests, the critical radius of curvature (RC) was 12.4 mm, and the loss of efficiency was less than1% after the cyclic bending test for 100 cycles at RC, indicating superior flexibility and bending durability. These results represent important steps toward a low-cost approach to high-performance flexible crystalline 2 ACS Paragon Plus Environment
1. INTRODUCTION To meet the demand for high-efficiency flexible Si film solar cells (Si-SCs), the fabrication of undoped and doped crystalline Si films is essential, as it involves construction of an n–i–p structure on the polymer substrate using a low-temperature technique. The crystalline Si film exhibits unique electronic and photovoltaic properties, such as high carrier mobility, enhanced absorption in the red/infrared wavelength range, and high doping efficiency compared with its amorphous counterpart.1,2 Because of its highly ordered structure, crystalline Si film is also highly stable against light-induced degradation.2 The most important characteristic of doped crystalline Si film is high electrical conductivity, which reduces the series resistance during conduction of photocurrent through an electrode and creates a built-in electric field that separates photogenerated charge carriers.3 Chemical vapor deposition (CVD) accompanying with in situ doping process is the most widely adopted technique for growing doped Si films, as it does not require post-annealing and is economically feasible.4 In such a process, a decomposed Si-based precursor is reacted with the doping source through a series of gas-phase reactions in a range of deposition methods to synthesize doped Si film. These methods include plasma-enhanced chemical vapor deposition (PECVD), hot-wire chemical vapor deposition (HWCVD), and photochemical vapor deposition (photo-CVD).4-6 Dopants for acceptor impurities belong group III (B) and donor impurities belong group V (P) are extensively used to form p-type and n-type characters, respectively. Various doping sources for B include diborane (B2H6)5, trimethylboron (B(CH3)3)7, and boron trifluoride (BF3)7, and sources for P include phosphane (PH3)6 and phosphorus trichloride (PCl3)8. However, the most commonly used gaseous doping sources for B and/or P are explosive, highly hazardous and difficult to store. Moreover, 4 ACS Paragon Plus Environment
added elements in the doping sources, e.g., H, C, F, and Cl, which may drastically degrade the structural and electrical properties of the doped Si films, may be incorporated into the synthesized films.9 The ability to develop a solid doping source with low toxicity and easy handling is therefore desirable and significant. Although solid doping methods using the floating zone technique in the Si wafer manufacturing industry are commercially available,10 to the best of our knowledge, there are few reports on the use of solid doping sources in the CVD process for growing doped Si films. In our previous study, we demonstrated the feasibility of using microwave plasma enhanced chemical vapor deposition (MWPECVD) with SiCl4/H2 mixture to prepare crystalline intrinsic Si film onto flexible polyimide (PI) polymer substrates at low temperature. This method does not use recrystallization technique or transfer process in one step. The crystal characteristics, carrier mobility, flexibility, and thin film transistor performance have been investigated in detail.11 It is well known that self-bias is generated between the electrode and plasma in a conventional glow discharge system, leading to impingement of ionized particles (known as ion bombardments) on the substrate and on the growing surface of the films.12 Herein, we propose an innovative approach based on the self-biased effect for doping films, known as self-biased sputtering solid doping source (SSSDS) process. This approach involves bombarding charged ions on solid doping targets to sputter out the dopant atoms in situ during high-density plasma deposition. Through this approach, it has become possible to dope the Si film efficiently by incorporating the synthesis of Si film using plasma-chemical reduction of SiCl4/H2 and generation of dopant atoms using the SSSDS process of plasma-ionized H2 bombardment in microwave plasma at low temperature. The solid doping sources were fabricated by hot pressing the targets 5 ACS Paragon Plus Environment
B and/or P powder and then embedding them in the substrate holder. Solid B and/or P powder are much cheaper and safer raw materials compared with gaseous doping sources. The present study focuses on the effect of dopant atoms, which is determined by the number of solid doping targets, on the microstructure of the synthesized doped Si film. Electronic properties and doping efficiency were extensively investigated. Flexible crystalline Si-SCs with single junction n–i–p structure on PI substrates were fabricated for device applications and their characteristics were evaluated.
2. EXPERIMENTAL METHODS 2.1 Fabrication of the Solid Doping Source After the adsorbed moisture of the B and/or P powder was removed by annealing at a temperature of 150oC for 1 hr in vacuum, the powder was hot pressed at a pressure of 150 MPa and temperature of 100oC for 30 min to form a solid doping target with 2 mm diameter and 2 mm thickness. Subsequently, the solid doping targets were inserted into the surrounding area of the substrate holder surface with symmetric geometry to generate doping atoms uniformly.
2.2 Preparation of the Doped Crystalline Si Film A flexible PI substrate (150 µm in thickness) with a size of 20 × 20 mm2 was used in this study. It was ultrasonically cleaned in an ethanol bath and then dried. Deposition of the doped Si film was carried out by glow discharge decomposition of a H2-diluted SiCl4 mixture in a MWPECVD system composed of a microwave generator (2.45 GHz, Muegge MH30000S-2100BB) and a cylindrical resonator with an annular slot structure. After the substrate holder with inserted solid doping targets was placed in the chamber, the chamber was pumped down to a base pressure of 1.3 × 6 ACS Paragon Plus Environment
10-2 mbar. To efficiently control the flow rate of liquid SiCl4 precursor, a cryostat apparatus was employed to maintain the temperature of SiCl4 at –55oC. The saturated vapor was then carried into the vacuum chamber by H2 at a flow rate of 10 sccm. The corresponding SiCl4 flow rate was 2.9 × 10−2 sccm. During deposition, the SiCl4 vapor and additional H2 (dilution gas) at 100 sccm flow rate were channeled separately into the chamber. Both of these flow rates were chosen on the basis of the optimal conditions for synthesis of intrinsic crystalline Si film in our previous study.11 Upon ignition of the SiCl4/H2 microwave plasma, SiCl4 decomposed and then underwent reduction by H2 plasma; at the same time, the dopants were generated from the SSSDS process by H+ ion bombardment. Subsequently, the doped Si film was deposited onto the PI substrate for 3 min under a fixed pressure of 6.7 mbar and a MW power of 750 W. The system of SiCl4/H2 MW plasma integrated with the SSSDS process and the reaction mechanism for deposition of doped crystalline Si films are schematically presented in Fig. 1 (a) and locations of solid doping targets on holder are shown in Fig. 1 (b). Details of the deposition conditions used in this study are shown in Table 1. Here, the number of solid doping targets for depositing high-quality doped crystalline Si film on PI substrate was systematically investigated. Besides, the byproduct HCl from the reaction of SiCl4/H2 mixture was trapped in a liquid nitrogen cold trap for preventing damage to the vacuum pumping system.
2.3 Characterization of the Doped Crystalline Si Film The crystallinity characteristics of the doped Si films were evaluated by X-ray diffractometry (XRD, Shimadzu XRD6000) using Cu Kα1 radiation (1.5405 Å) and by micro-Raman spectroscopy (Horiba Jobin Yvon HR800UV) using an excitation source of a 514 nm Ar+ laser. Raman spectra were deconvoluted into three peaks: one 7 ACS Paragon Plus Environment
for the crystalline phase near 520 cm−1, the other for the intermediate (nanocrystalline) phase around 500 cm−1, and the last for the amorphous phase at 480 cm−1. The crystalline volume fraction (XC) of the doped Si film was calculated according to its integrated intensity. A field emission scanning electron microscope (FESEM, JEOL JSE-6500F) was used to examine the surface and cross-sectional morphologies and to determine the thickness of the doped Si film. To characterize detailed crystallographic structure of the deposited Si films, transmission electron microscopy (TEM, JEOL JEM-ARM200F) was utilized to obtain the cross-sectional bright field images (BFI) of the whole film and the selected area electron diffraction (SAED) patterns at 200 kV. Meanwhile, high-resolution (HR) mode and fast Fourier transformation (FFT) mode were also employed to conduct lattice imaging analysis and consequent structural identification. To quantify the dopant concentration (CD) and to confirm the uniformity of the doped Si film, depth profile analysis using secondary-ion mass spectrometry (SIMS, ION-TOF TOF-SIMS IV) was performed. To measure conductivity (σ) and carrier concentration (Cc) through the Hall-effect measurement system (Ecopia HMS-3000) at a magnetic field of 0.55 T, Ohmic contact was assured by using a sputtered Al film (~100 nm) with a coplanar configuration on the Si film.
2.4 Preparation of Flexible Crystalline Si-SCs Most developments in flexible Si-SCs used substrate configuration because of their wide range of substrates, stable back contact, and suitable monolithic integration.13 In the present study, a single-junction n–i–p solar cell with substrate configuration was fabricated on flexible PI substrates. After deposition of sputtered 8 ACS Paragon Plus Environment
Mo metal film as the back contact with thickness of 1 µm, a series stack of doped and undoped crystalline Si films were consecutively deposited to form an n–i–p junction by using SiCl4/H2 microwave plasma with and without SSSDS process, respectively. The thicknesses of the p- and n-type Si films in the solar cell were both 50 nm; however, the thickness of the undoped Si films was varied to investigate the effect of the absorber layer thickness on the performance of the solar cell. A sputtered indium tin oxide (ITO) film with a thickness of 80 nm was deposited as a transparent conducting material to enhance lateral conductivity, and an antireflection layer to reduce the loss of incident light.14 Using a shadow mask, a 500 nm Al grid metal film was sputtered on the ITO surface for the front contact. Post-annealing was performed at 250oC for 1 hr in vacuum to crystallize the ITO film (possessing a transmittance of 90.3%@550 nm and a sheet resistance of 43.2 Ω/sq.) and stabilize the interface between each layer for enhancing the solar cell performance. The structure of the flexible crystalline Si-SC employed in this study was PI/Mo film/n-type Si film/i-type Si film/p-type Si film/ITO film/Al grid film, as shown in Fig. 2. Also shown is a photograph of bent crystalline Si-SC on flexible PI substrate. The active area was set at 1.0 cm2.
2.5 Characterization of the Flexible Crystalline Si-SCs The performance of flexible Si-SCs under AM 1.5G illumination with an intensity of 100 mW/cm2 at 25oC was measured by using a solar simulator (Newport Corporation, 91160A, Oriel class A). The current density–voltage (J–V) characteristics were obtained by using a source-measure unit (Keithley 2400). The open-circuit voltage (VOC), short-circuit current density (JSC), fill factor (FF), and efficiency (η) were then estimated. In addition, the external quantum efficiency (EQE) 9 ACS Paragon Plus Environment
characteristics of the crystalline Si-SC was measured using spectral response measurement system (ENLI EQE-D-3011). Furthermore, the flexibility of the Si-SCs was evaluated by mechanical static bending tests. The flexible Si-SCs were bent to generate tensile strain at a designated radius of curvature (R) by controlling the stroke distance in a bending tester that was fabricated in-house. Moreover, the dynamic bending tests were also performed by repeating bending and releasing the Si-SC 100 times to a critical radius of curvature (RC) of 12.4 mm, which was chosen for the static bending test. The performance in terms of VOC, JSC, FF, and η of the flexible crystalline Si-SCs was recorded during both bending tests.
3. RESULTS AND DISCUSSION During deposition of the doped Si film, the substrate temperature gradually increased to 185oC after deposition for 3 min because of continuous H+ ion bombardment under high-density plasma. The substrate temperature, however, was far below the glass transition temperature (~260oC) and melting temperature (~350oC) of the PI polymer.15 Thus, the PI polymer did not suffer degradation by thermal exposure during Si film deposition.
3.1 Microstructure Characteristics of the Doped Si Films Figures 3 and 4 show the XRD patterns and Raman spectra of the doped Si films prepared with different numbers of solid doping targets. Characteristic peaks of crystalline Si can be observed irrespective of deposition condition used. The feasibility of one-step synthesis of doped crystalline Si film onto thermosensitive polymer PI substrates was thus validated at low temperature by using the MWPECVD system with the SiCl4/H2 mixture and the SSSDS process. 10 ACS Paragon Plus Environment
Detailed observation of XRD patterns in Fig. 3 reveal a strong correlation between the number of solid doping targets and crystallization characteristics, such as crystallinity, and crystallite size. The intrinsic Si film deposited without using the SSSDS process, i.e., P(0) or B(0) condition, was crystalline, exhibiting three distinct peaks at 2θ values of about 28.4o, 47.3o, and 56.1o. The corresponding crystal orientations were (111), (220), and (311) respectively. After the SSSDS process was introduced in SiCl4/H2 microwave plasma to synthesize the doped Si film, the diffraction peaks of all crystallographic planes decreased in intensity and broadened in full width at half maximum. The crystallinity and crystalline size further deteriorated with further increase in the number of solid doping targets. Similar tendencies of crystallization characteristics can also be seen in the Raman spectra of Fig. 4. Raman spectra for the intrinsic Si film reveals a sharp peak centered at 520.4 cm−1 and a high XC value, 98.3%, as determined from the integrated intensity of deconvoluted Raman peaks. By comparison, when dopants were incorporated into the Si film, the characteristic peak of the crystalline phase shifted toward lower wavenumber; concurrently, a broad shoulder centered around 480 cm−1 ascribed to the amorphous Si phase emerged. With further increase in the number of solid doping targets for the conditions P(12) and/or B(12), both crystalline size and XC decreased, as confirmed by the substantial shift in the dominant peak to the nanocrystalline phase region and by the appearance of a predominant amorphous phase in the Raman spectra. Therefore, it is concluded that both XC and crystallite size were reduced when the dopant was introduced in the Si films. These results agree with amorphization in the doped Si network due to the local deformation caused by dopant atoms.2
Variation of the deposition rate of the doped Si film with the number of solid doping targets is presented graphically in Fig. 5. The deposition rate for the intrinsic Si film reached 4.0 nm/s, which is substantially higher than that of many other MWPECVD processes using SiH4/H2 or SiH2Cl2/H2 as precursor.16-17 On the other hand, the deposition rate of both P- and B-doped Si films increased linearly with the increase in the number of solid doping targets after adopting the SSSDS process. Maximum deposition rates of 4.66 and 4.45 nm/s obtained in the present study for the conditions P(12) and B(12), respectively, are much higher than those of other CVD processes using SiH4 as precursors with various gaseous doping sources.4-6 Enhancement of the deposition rate for the doped Si film may be attributed to the incorporation of dopants in the Si film. Moreover, the morphology markedly changed from columnar morphology to featureless non-columnar when the number of solid doping targets increased. Therefore, results of microstructure observation on the doped Si film agree well with findings for the crystallinity characteristics from XRD and Raman analyses (Figs. 3 and 4). To investigate the dopant distribution in the doped Si film at various doping conditions, SIMS depth profile analyses were performed, as shown in Fig. 6. According to the results, B and P atoms can be detected in the B- and P-doped Si film. These confirm that the SSSDS process could be used to dope Si film in SiCl4/H2 microwave plasma. The growth mechanism for the doped crystalline Si film is due to the synergistic effects of three predominant models in the SiCl4/H2 plasma-chemical reaction, namely, preferential etching, self-cleaning, and chemical annealing11 accompanied by efficient incorporation of dopant atoms in the Si network during the SSSDS process in high-density plasma deposition, MWPECVD.
Figure 6 shows that the concentration of the incorporated dopant atoms depends strongly on the number of solid doping targets. This dependence shows that the CD value monotonically increased with the increase in the number of solid doping targets. The P-CD value for P-doped Si film ranged from 7.2 × 1017 to 9.3 × 1021 cm−3, and the B-CD value for B-doped Si film ranged from 4.9 × 1016 to 1.1 × 1021 cm−3. Therefore, the CD value of the doped Si film prepared through the SSSDS process could be facilely controlled within a wide range by varying the number of solid doping targets. Figure 6 also suggests no obvious difference in homogeneity of the CD values across both P- and B-doped Si films, indicating that incorporation of dopant in the Si film was uniform entirely. From comparison with both doped Si films, it is of interest that the P-CD value was higher than B-CD by about one order of magnitude when the same number of solid doping target conditions was adopted. In the field of using gaseous doping source, the bond strength, dissociation probability, and other possible gas-phase reactions should always be taken into consideration for the differences in the incorporation of the dopant atoms into the Si phase.18 By contrast, a probable explanation for the doped Si film prepared through the SSSDS process is the sputtering yield and impurity formation energy. Dopants generated by sputtering the solid doping target using charged ion bombardment (i.e., self-biased effect) in a glow-discharge system is a critical part of this synthetic method. Using experimental calculations based on Ne sputtering gas at 500 eV, the National Physical Laboratory showed that the sputtering yields of B and P are 0.299 and 0.427 atom/ion, respectively.19 Their finding implies that the amount of sputtered B atoms is less than that of P atoms during the SSSDS process. In addition, the proposed theoretical prediction of impurity formation energy indicates that the formation energy of 13 ACS Paragon Plus Environment
substitutional B is larger than that of substitutional P in the case of doped Si film growth.20 The two mechanisms, therefore, support our finding that B is less efficiently incorporated than P in Si film during the SSSDS process.
3.2 Electrical Properties of the Doped Si Films Figure 7 shows the variation of CC and σ values with the number of solid doping targets for the doped Si film. The σ and CC values for the undoped Si film as measured from the Hall effect are 3.5 × 10−9 S/cm and 5.3 × 1011 cm−3, respectively. Increasing the number of solid P doping target to P(6) resulted in an abrupt increase in σ value by over nine orders of magnitude to 9.48 S/cm, which corresponds to a CC value of 1.2 × 1020 cm−3 for the P-doped Si film. With further increase in the number of solid P doping targets, the σ decreased whereas CC approached a saturation value. Electrical properties for the B-doped Si film showed an identical trend. The highest value of σ was 7.83 S/cm, with a CC value of 1.5 × 1020 cm−3 at the condition B(10). Such a high value of CC indicates that the dopants were activated effectively in the Si films. It is well accepted that the σ value of the Si film is determined by the crystallinity and CC value.2 The initial increase in σ of doped Si film prepared through the proposed method is mainly attributed to the increase in the CC. Thus, it is due to the rise in the number of active dopants in the doped Si film. Further increase in the number of solid doping targets beyond the optimum value, i.e., P(6) and/or B(12), drastically reduced the σ values because of reduction of the crystalline phase in the doped Si film. Results for the XRD patterns and Raman spectra in Figs. 3 and 4 confirm these findings.
The variation of CC with CD for the doped Si film, what is usually called as “doping efficiency”, is may be evaluated from the results in Figs. 6 and 7 and is depicted in Fig. 8. Generally, the doping efficiency of Si films prepared by HWCVD is about 10–20%; in contrast, a higher doping efficiency (10–50%) can be obtained for Si films synthesized by PECVD.18 Si film with a crystalline phase exhibits higher doping efficiency compared with the amorphous phase, which is also a well-known fact.2 In our study, the doping efficiency for both doped Si films under the appropriate conditions was >40%, consistent with the trend of XC and with the independence of the two solid doping sources used. Such high doping efficiency is ascribed to the MWPECVD system, which is known to have high-density plasma and a high ionization rate21-22; the process was also adopted to synthesize ultrananocrystalline diamond23-24. It thus stimulated the development of high-quality crystalline Si film and efficiently activated dopant atoms in the Si network even under a low substrate temperature (185°C). Reduction of XC in the doped Si film caused the decline in doping efficiency with further increase in the number of solid doping targets beyond the optimum value, i.e., P(6) and/or B(10).
3.3 Performance of the Flexible Crystalline Si-SCs To demonstrate the photovoltaic properties of the Si films, an n–i–p single junction crystalline Si-SC with substrate configuration was fabricated by using the structure of PI/Mo film/n-type Si film/i-type Si film/p-type Si film/ITO film/Al grid film. It is known that doped Si films exhibiting high CC and σ values are required for achieving high efficiency of crystalline Si-SCs. The high CC value can result in a sufficiently large internal electric field across the intrinsic Si film, which ensures separation of photogenerated carriers.3 A high σ value results in low-loss ohmic 15 ACS Paragon Plus Environment
electrical contacts between the Si part of the solar cell and external electrodes for the photogenerated carriers.3 On the basis of these two factors, the optimal conditions of P(6) and B(10) were used to respectively synthesize n- and p-type Si films with a thickness of 50 nm. On the other hand, light absorption and photogeneration of carriers in the n–i–p structure mainly takes place in the intrinsic part of the Si-SC. The J–V characteristics of crystalline Si-SCs with various thicknesses of intrinsic Si films are shown in Fig. 9 (a), and the performance in terms of VOC, JSC, FF, and η is summarized in Fig. 9 (b). According to these results, the intrinsic Si film thickness significantly affects the JSC and η values of crystalline Si-SC with VOC and FF values of approximately 0.54 V and 0.65, respectively. With the increase in thickness of intrinsic Si films from 0.3 to 2 µm, the JSC value increased abruptly from 3.58 to 19.18 mA/cm2, and, consequently, reached a maximum η value of 6.75%. Furthermore, the EQE characteristics of the crystalline Si film solar cell using the intrinsic Si film of 2 µm is revealed in Fig. 10. The results depicting the significant spectral response in long-wavelength region were observed, which resulted from the lower band gap of crystalline Si film (typically 1.1 eV) in nature. 25 To further investigate the effect of the intrinsic Si film thickness on the Si-SC performance, the characteristics of the doped crystalline Si film with 50 nm thickness should be verified first. The microstructure characteristics of both the thin doped crystalline Si films were characterized and the results are show in Fig. S1 and Fig. S2. Based on these results, the characteristic peaks of crystalline Si in XRD pattern, a sharp peak centered around 517 cm-1 in Raman spectrum, and a high XC over 90% demonstrate the attainment of a highly crystalline Si film even under such short deposition time and low temperature. Furthermore, the cross-sectional BFI images 16 ACS Paragon Plus Environment
with their SAED patterns and corresponding HRTEM images of both the doped crystalline Si films are also presented in Fig. S1 (c) and Fig. S2 (c). Both (111) and (220) planes are co-existing as evidenced from the lattice spacing throughout the thickness. It is also found that doped Si film directly crystallizes onto PI substrate without an incubation layer of amorphous phase even at the initial growth stage by very careful HRTEM examinations. Such microstructure features are the responsible for the measured high σ value of 8.81 S/cm and CC value of 9.8 × 1019 cm−3 for P-doped Si film, as well as 7.01 S/cm and 1.0 × 1020 cm−3, respectively, for B-doped Si film. The results demonstrated that a thinner doped crystalline Si film such as 50 nm, regardless of different doping sources used, can also possess excellent performance as compared with its thicker one. The intrinsic layer of crystalline Si-SCs is the main photovoltaically active layer, i.e., the photogeneration layer or absorber layer. To be a usable intrinsic layer in crystalline Si-SC with high efficiency, requirements for microstructure characteristics and electrical properties must be reached.26 The former is linked to growth conditions such as the deposition technique, reactive precursor, or operating parameters. Beside the requirement of high crystallinity in intrinsic Si film, columnar growth of the grains with a cone-shaped conglomerate, and a (220) preferential orientation are always observed. Such structure leads to an electrical transport path along the length direction of the columnar grains, which are perpendicular to the substrate and thus bypass the clustered defects between the columns. This results in highly efficient transport of photogenerated carriers in the solar cell.25-26 Thus, this explanation agrees with the microstructure observation on the deposited intrinsic Si film with 2 µm thickness in this study (Fig. S3). On the other hand, a sufficiently high mobility (µ) and lifetime (τ) for photogenerated carrier in the intrinsic Si film are also important parameters. 17 ACS Paragon Plus Environment
Therefore, the product µ × τ was used to determine the ability of the materials for carrier transport. A value of at least 4 × 10−7 cm2/V has been proposed.26 In our previous study, intrinsic crystalline Si film prepared by using SiCl4/H2 microwave plasma showed an extremely high Hall carrier mobility (170 cm2/Vs).11 An average carrier lifetime of around 20 µs was measured by microwave photoconductive decay (Semilab WT-2000PVN), as shown in Fig. S4. Hence, the intrinsic crystalline Si film exhibits a µ × τ value as high as 3.4 × 10−3 cm2/V, which exceeds significantly the proposed minimum. This result indicates that the intrinsic crystalline Si film had the capability of collecting photogenerated carriers without obvious recombination losses. In view of the above findings, we conclude that the prepared intrinsic and doped crystalline Si films on PI substrate using SiCl4/H2 microwave plasma without and with SSSDS process respectively, are satisfactory and promising SC-grade materials. Therefore, as compared with the published data in performance of flexible Si-SCs using a single junction n–i–p structure (Fig. S5),27–34 this study demonstrated a relatively large area and high efficiency flexible Si film-based solar cell. In our study, the crystalline Si film solar cell on PI substrate with η value of 6.75% in an active area of 1 cm2 was obtained by using an intrinsic Si film of 2 µm and a doped Si film of 50 nm. Further enhancement in performance of crystalline Si film solar cell is expected if postmodification techniques are applied. These techniques include integration of amorphous Si film to form a tandem structure for increasing the light absorption range and VOC, surface-textured treatment to enhance light trapping, and surface passivation for decreasing carrier recombination.
3.4 Flexibility of the Flexible Crystalline Si-SCs
To explore the applicability of the flexible crystalline Si-SCs to soft electronics, their mechanical flexibility were examined. As revealed in the previous sections, the best value of η for Si-SC with intrinsic Si film of 2 µm and doped Si film of 50 nm, was used for the test sample. Figure 11 (a) shows the η value of the bent crystalline Si-SC evaluated by using a static bending test at various R values. With progressive decrease in the R value, the η value of the Si-SC declined slowly at first and then suddenly dropped. The R value corresponding to such a sudden drop is denoted as RC, which is the minimum R value (or the largest bending strain) that the flexible crystalline Si-SC can tolerate. The overall test results indicate that the RC value of the crystalline Si-SC prepared on flexible PI substrate is 12.4 mm. However, further reduction of the R value (i.e., increasing the bending strain for the Si-SC) led to an abrupt decrease in η value to 1.89% at an R value of 3.7 mm. This decrease is most likely due to the formation of perpendicular microcracks in each layer and the detachment of layers in the flexible crystalline Si-SC under a high bending strain, which result in disruption of conductive paths of carriers. To understand better the bending durability of the flexible crystalline Si-SC, dynamic bending tests at an R value of 12.4 mm (i.e., RC) were performed, as shown in Fig. 11 (b). We found limited changes in η values of Si-SC, showing that the loss in η is less than 1% after the cyclic bending test at RC for 100 cycles. This indicates that the crystalline Si-SC prepared on the PI substrate exhibits high flexibility and bending durability under severe bending conditions.
4. CONCLUSIONS Doped crystalline Si film with high Cc and σ values were grown on flexible PI substrate at a low temperature by using microwave glow-discharge decomposition of 19 ACS Paragon Plus Environment
H2-diluted SiCl4 with the SSSDS process. The growth mechanism for the doped crystalline Si film was due to the synergistic effects of three predominant models in SiCl4/H2 plasma-chemical reaction, namely, preferential etching, self-cleaning, and chemical annealing accompanied by efficient incorporation of dopant atoms into the Si network through the SSSDS process during high-density plasma deposition, MWPECVD. The number of solid doping targets significantly affected the characteristics of the doped Si film. Optimal conditions for P- and B-doped Si film were P(6) and B(10), respectively. Both doped Si films possessed a dense crystalline columnar morphology that grow at a rate of >4.0 nm/s. A high doping efficiency (>40%) was also obtained in this study. This resulted in high σ value of 9.48 S/cm and Cc value of 1.2 × 1020 cm−3 for P-doped Si film, as well as 7.83 S/cm and 1.5 × 1020 cm−3, respectively, for B-doped Si film. Doped and intrinsic crystalline Si films could be incorporated into a flexible crystalline Si-SC with a PI/Mo film/n-type Si film/i-type Si film/p-type Si film/ITO film/Al grid film structure. They showed the best performance in VOC of 0.54 V, JSC of 19.18 mA/cm2, FF of 0.65, and η of 6.75% values. No obvious deterioration in η values was observed after the cyclic bending test at RC for 100 cycles. Therefore, the features of these flexible crystalline Si-SCs demonstrate their great potential in industrial and commercial applications, such as building-integrated photovoltaics and portable electronic products.
ASSOCIATED CONTENT Supporting information: The microstructures of both the doped crystalline Si films with a thickness of 50 nm and the intrinsic crystalline Si film with a thickness of 2 µm were characterized. The carrier lifetime of the intrinsic crystalline Si film was measured and the comparisons of performance of flexible Si-SCs using a single 20 ACS Paragon Plus Environment
Properties of nc-Si:H Deposited in Low Frequency Inductively Coupled Plasma. J. Appl. Phys. 2011, 110, 023517. 
Waman, V. S.; Kamble, M. M.; Ghosh, S. S.; Mayabadi, A.; Sathe, V. G.; Pathan, H. M.; Shinde, S. D.; Adhic, K. P.; Jadkar, S. R. Highly Conducting Phosphorous Doped n-type nc-Si:H Films by HW-CVD for c-Si Heterojunction Solar Cells. RSC Adv. 2012, 2, 9873–9880.
Van Aken, B. B.; Duchamp, M.; Boothroyd, C. B.; Dunin-Borkowski, R. E.; Soppe, W. J. EELS Measurements of Boron Concentration Profiles in p-a-Si and nip a-Si Solar Cells. J. Non-Cryst. Solids 2012, 358, 2179–2182.
Mao, H. Y.; Lo, S. Y.; Wuu, D. S.; Wu, B. R.; Ou, S. L.; Hsieh, H. Y.; Horng, R. H. Hot-wire Chemical Vapor Deposition and Characterization of p-type Nanocrystalline Si Films for Thin Film Photovoltaic Applications. Thin Solid Films 2012, 520, 5200–5205.
Yan, W. S.; Wei, D. Y.; Xu, S.; Sern. C. C.; Zhou, H. P. Highly Doped p-type Microcrystalline Si Films Fabricated by a Low-frequency Inductively Coupled Plasma at a Low Temperature. J. Phys. D-Appl. Phys. 2011, 44, 35401.
Alonso, J. C.; Oshima, T.; Yamada, A.; Konagai, M.; Takahashi, K. Low-temperature Epitaxial Growth of Undoped and n-doped Silicon by Photochemical Vapor Deposition Using SiH4/SiH2Cl2/H2/PH3 Mixtures. Thin Solid Films 1994, 237, 98–104.
Kumar, P; Kupich, M.; Grunsky, D.; Schroeder, B. Microcrystalline B-doped Window Layers Prepared Near Amorphous to Microcrystalline Transition by HWCVD and Its Application in Amorphous Silicon Solar Cells. Thin Solid Film 2006, 501, 260–263.
Benvenuto, A. G.; Buitrago, R. H.; Schmidt, J. A. Doped Polycrystalline Silicon 22 ACS Paragon Plus Environment
Thin Films Deposited on Glass from Trichlorosilane. Chem. Vapor Depos. 2015, 21, 54–62. 
Huang, S. Y.; Cheng, Q. J.; Xu, S.; Ostrikov, K. Inductively Coupled Plasma-assisted RF Magnetron Sputtering Deposition of Boron-doped Microcrystalline Si Films. J. Alloy. Compd. 2010, 499, 166–170.
 Flemish, J. R.; Tressler, R. E. Phosphorus Doping of Silicon Using a Solid Planar Diffusion Source at Reduced Pressures. J. Electrochem. Soc. 1991, 138, 233–238.  Hsieh, P. Y.; Lee, C. Y.; Tai, N. H. A High Carrier-mobility Crystalline Silicon Film Directly Grown on Polyimide Using SiCl4/H2 Microwave Plasma for Flexible Thin Film Transistors. J. Mater. Chem. C 2015, 3, 7513–7522.  Nishida, S.; Tasaki, H.; Konagai, M.; Takahashi, K. Highly Conductive and Wide Band Gap Amorphous-microcrystalline Mixed-phases Silicon Films Prepared by Photochemical Vapor Deposition. Appl. Phys. Lett. 1985, 58, 1427.  Zeman, M. In Thin film solar cells: Fabrication, characterization and applications; Poortmans, J.; Arkhipov, V., Eds.; John Wiley & Sons Ltd: England; Chapter 5, pp 173–236.  Pláa, J.; Tamasi, M.; Rizzoli, R.; Losurdo, M.; Centurioni, E.; Summonte, C.; Rubinelli, F. Optimization of ITO layers for applications in a-Si/c-Si heterojunction solar cells. Thin Solid Films 2003, 425, 185–192.  Sroog, C.E. Polyimides. J. Polymer Sci. Macromol. Rev. 1976, 11, 161–208.  Zhong, C.J.; Tanaka, H.; Sugawa, S.; Ohmi, T. Effect of Power Density on the Structure Properties of Microcrystalline Silicon Film Prepared by High-density Low-ion-energy Microwave Plasma. Thin Solid Films 2005, 493, 54–59.  Saha, J.K.; Ohse, N.; Hamada, K.; Matsui, H.; Kobayashi, T.; Jia, H.; Shirai, H. 23 ACS Paragon Plus Environment
Fast Deposition of Microcrystalline Si Films from SiH2Cl2 Using a High-density Microwave Plasma Source for Si Thin-film Solar Cells. Sol. Energy Mater. Sol. Cells 2010, 94, 524–530.  Kumar, P.; Schroeder, B. Electrical Properties/Doping Efficiency of Doped Microcrystalline Silicon Layers Prepared by Hot-wire Chemical Vapor Deposition. Thin Solid Films 2008, 516, 580–583.  http://www.npl.co.uk/science-technology/surface-and-nanoanalysis/services/spu tter-yield-values  Pi, X.D.; Gresback, R.; Liptak, R.W.; Campbell, S.A.; Kortshagen, U. Doping Efficiency, Dopant Location, and Oxidation of Si Nanocrystals. Appl. Phys. Lett. 2008, 92, 123102.  Sugai, H.; Ghanashev, I.; Nagatsu, M. High-density Flat Plasma Production Based on Surface Waves. Plasma Sources Sci. Technol. 1998, 7, 192–205.  Werner, F.; Kotzec, D.; Engemann, J. Slot Antenna 2.45 GHz Microwave Plasma Source. Plasma Sources Sci. Technol. 1994, 3, 473–481.  Sankaran, K.J.; Chen, H.C.; Panda, K.; Sundaravel, B.; Lee, C.Y.; Tai, N.H.; Lin, I.N.
Ultrananocrystalline Diamond Films After Cu and Au Ion Implantation. ACS Appl. Mater. Interfaces 2014, 6, 4911–4919.  Sankaran, K.J.; Kalpataru, P.; Balakrishnan, S.; Tai, N.H.; Lin, I.N. Catalytically Induced Nanographitic Phase by a Platinum-ion Implantation/Annealing Process
Ultrananocrystalline Diamond Films. J. Mater. Chem. C 2015, 3, 2632–2641.  Rath, J.K. Low Temperature Polycrystalline Silicon: A Review on Deposition, Physical Properties and Solar Cell Applications. Sol. Energy Mater. Sol. Cells 24 ACS Paragon Plus Environment
 Filonovich, S.A.; Alpuim, P.; Rebouta, L.; Bourée, J.E.; Soro, Y.M. Hydrogenated amorphous and nanocrystalline silicon solar cells deposited by HWCVD and RF-PECVD on plastic substrates at 150oC. J. Non-Cryst. Solids 2008, 354, 2376–2380.  Wang, J.; Tao, K.; Cai, H.; Zhang, D.; Li, G. A study of the i/p interface for flexible n-i-p a-Si:H thin film solar cells. Mater. Sci. Semicond. Process 2014, 25, 186–189.
Figure Captions Fig. 1 (a) Schematic of SiCl4/H2 microwave plasma integrated with the SSSDS process and the reaction mechanism for deposition of doped crystalline Si films, and (b) definition of N correlated with different number of solid doping targets on holder. Fig. 2 Photograph and schematic cross-sectional image of the fabricated flexible crystalline Si-SC. Fig. 3 XRD patterns of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. Fig. 4 Raman spectra of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. The integrated XC value of each doped Si film is also presented. Fig. 5 Deposition rate of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. Corresponding cross-sectional morphologies of the doped Si films at a deposition time of 3 min are also shown. Scale bars in the FESEM images are 250 nm. Fig. 6 Dopant concentration of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. The depth profiles of P in P-doped Si film under condition P(6) and those of B in the B-doped Si film under condition B(10) are also given in (a) and (b), respectively. Fig. 7 Carrier concentration and conductivity of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. Fig. 8 Doping efficiency of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. Fig. 9 (a) J–V characteristics and (b) SC performance in terms of VOC, JSC, FF, and η of Si-SC with varied thicknesses of intrinsic Si films. The inset in (a) shows the
FESEM cross-sectional morphology corresponding to a Si-SC with intrinsic Si film thickness of 2 µm. Fig. 10 EQE characteristics of flexible Si-SC using a single junction n–i–p structure with intrinsic crystalline Si film of 2 µm and doped crystalline Si film of 50 nm. Fig. 11 Change in η value of Si-SC during (a) mechanical static bending tests using various radii of curvature and (b) mechanical dynamic bending tests at RC value of 12.4 mm for various bending cycles. Insets in (a) and (b) show photographs of crystalline Si-SC during the bending state.
Fig. 1 (a) Schematic of SiCl4/H2 microwave plasma integrated with the SSSDS process and the reaction mechanism for deposition of doped crystalline Si films, and (b) definition of N correlated with different number of solid doping targets on holder.
Fig. 5 Deposition rate of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. Corresponding cross-sectional morphologies of the doped Si films at a deposition time of 3 min are also shown. Scale bars in the FESEM images are 250 nm.
Fig. 6 Dopant concentration of (a) P-doped and (b) B-doped Si films at various numbers of solid doping targets. The depth profiles of P in P-doped Si film under condition P(6) and those of B in the B-doped Si film under condition B(10) are also given in (a) and (b), respectively.
Fig. 9 (a) J–V characteristics and (b) SC performance in terms of VOC, JSC, FF, and η of Si-SC with varied thicknesses of intrinsic Si films. The inset in (a) shows the FESEM cross-sectional morphology corresponding to a Si-SC with intrinsic Si film thickness of 2 µm.
Fig. 11 Change in η value of Si-SC during (a) mechanical static bending tests using various radii of curvature and (b) mechanical dynamic bending tests at RC value of 12.4 mm for various bending cycles. Insets in (a) and (b) show photographs of crystalline Si-SC during the bending state.
Tables Table 1 Deposition parameters for the doped Si film. Deposition parameters Value Precursor SiCl4 Precursor temperature (oC) −55 Flow rate of carrier gas H2 (sccm) 10 Flow rate of dilution gas H2 (sccm) 100 6.7 Working pressure (mbar) Deposition time (min) 3 Max. substrate temperature (oC)* 185 Microwave power (W) 750 Number of solid doping targets 0, 2, 4, 6, 8, 10, 12 * The substrate temperature was measured at the end of deposition.