Core–Shell Heterojunction Solar Cells Based on Disordered Silicon

Subscriber access provided by CMU Libraries - http://library.cmich.edu

Article

Core-Shell Heterojunction Solar Cells Based on Disordered Silicon Nanowire Arrays Alienor Svietlana Togonal, Martin Foldyna, Wanghua CHEN, Jian Xiong Wang, Vladimir Neplokh, Maria Tchernycheva, Joaquim Nassar, Pere Roca i Cabarrocas, and * Rusli J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.5b09618 • Publication Date (Web): 21 Dec 2015 Downloaded from http://pubs.acs.org on December 26, 2015

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.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Core-Shell Heterojunction Solar Cells Based on Disordered Silicon Nanowire Arrays Aliénor Svietlana Togonal1, Martin Foldyna2, Wanghua Chen2, Jian Xiong Wang1, Vladimir Neplokh3, Maria Tchernycheva3, Joaquim Nassar2, Pere Roca i Cabarrocas2* and Rusli1 1

School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798 CNRS, Laboratoire de Physique des Interfaces et Couches Minces (LPICM), Ecole polytechnique, 91128 Palaiseau, France 3 Institut d'Electronique Fondamentale, UMR 8622 CNRS, University Paris-Sud, 91405 Orsay, France

2

* Corresponding author: Tel: +33 1 69 33 43 14, Fax: +33 1 69 33 43 33, [email protected]

Abstract Heterojunction with intrinsic thin layer (HIT) solar cells are still costly due to the use of silicon wafers that contribute to ∼30% of the final module cost. To reduce the costs, thinner wafers can be used but to the detriment of the optical absorption. This loss has to be compensated by an efficient light trapping scheme. In this paper we study HIT devices based on Si nanowire (SiNW) core-shell structures to enhance light absorption. The SiNWs are fabricated by a wet etching technique and the heterojunction is formed using an optimized low temperature plasma enhanced chemical vapor deposition (PECVD) process at 200 °C. The solar cells are characterized via carrier lifetime and electron beam induced current (EBIC) measurements to understand their electrical properties at nanoscale. The impact of the SiNW length on the cell performance is also investigated. The solar cells show a good performance reaching average fill factor of 81 %, Voc of 0.525 V and Jsc of 29.27 mA/cm2 giving rise to an average efficiency of 12.43%. Finally, the fabrication procedure is applied to epitaxial Si thin films and the performance of such devices is discussed.

Complete list of affiliations: Alienor Svietlana Togonal1,2,4, Rusli1,4 4 CINTRA UMI CNRS/NTU/THALES 3288, Research Techno Plaza, 50 Nanyang Drive, Singapore 637553 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1. Introduction Solar cells based on crystalline silicon (c-Si) wafers have dominated the photovoltaic market since the very beginning of the PV industry back in the 1950’s, keeping a market share of over 80 %. Recently, Panasonic has adapted the HIT technology pioneered by Sanyo to a back contact architecture. This has brought a new world efficiency record of 25.6% 1, making HIT solar cells the most efficient technology and keeping pace in the strong competition of PV markets. While HIT technology uses only low temperature processes to form the p-n heterojunction, in contrast with the high-temperature diffusion process used in passivated emitter rear locally-diffused (PERL) technology, it still relies on the use of expensive c-Si wafers, which account for ∼30% of the final module cost 2. Thinner wafers have been used to reduce the cost of the c-Si modules 3, but with reduced optical absorption. Therefore, to cut costs through the reduced thickness, new low-cost and efficient light trapping schemes have to be developed. Recent progress in nanostructure fabrication has permitted to apply more efficient light trapping schemes 4,5 compared to the standard surface texturization of single crystal silicon by random pyramids. In particular, silicon nanowire (SiNW) arrays have been shown to have outstanding optical properties with a very low reflectance from the visible range to the near infrared spectral range

6–9

. Optimally designed nanowires may also significantly increase the

absorption in the active material beyond the theoretical limits of ray optics by a good control of the light diffraction and the localization of the electromagnetic field. Therefore, HIT solar cells based on SiNW arrays are an attractive and exciting approach to reduce the wafer thickness for the standard HIT technology and thus cut the cost further.

ACS Paragon Plus Environment

Page 2 of 34

Page 3 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Several works have been reported on the use of SiNWs for solar cell applications 10. Our group has investigated thin film hydrogenated amorphous silicon (a-Si:H) radial junction solar cells built on VLS-grown SiNWs with a highest reported efficiency of 9.2% 11,12. This architecture uses an ultrathin a-Si:H layer (100 nm) as the active material. In this work, we are investigating the potential of SiNWs in a core-shell structure where c-SiNWs are used as the active material. In this type of devices, the p-n junction is typically formed by the high temperature phosphorus diffusion process. As a consequence, the NWs might be fully converted to n-type (or p-type) materials, making the p-n junction axial rather than radial. In this case the SiNWs are used for their good anti-reflection and enhanced light absorption, but the benefit of the radial junction is lost

4,13–19

. On the other hand, SiNWs with larger diameter fabricated by deep reactive ion

etching (DRIE) have been successfully used in solar cells with a core shell structure fabricated using the phosphorus diffusion process with a reported efficiency of 10.8 % for a few micrometer thick c-Si films

20

. Moreover an efficiency of 18.2% was also achieved for black

silicon solar cells nanostructured by a wet etching process on a c-Si wafer. The use of SiNWs in HIT solar cells brings an additional challenge of passivation. Indeed, while diffusion-fabricated p-n junctions are formed inside the SiNW bulk, for HIT devices the pn junction is formed at the surface of the SiNW which may present many surface defects. Today, only few works have been published on HIT solar cells based on SiNWs 21–24. So far, the best efficiency reported is 10.04% for SiNWs upon addition of a thin insulating layer of Al2O3 to passivate the SiNWs surface 22. The atomic layer deposition (ALD) technique, characterized by its slow deposition rate, was used for the Al2O3 and contact deposition. Another interesting work claimed an efficiency of 12.2 % for HIT solar cells based on silicon microwires (SiMWs)


25


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 34

with radii ranging from 1.5 µm to 50 µm fabricated by DRIE. Nevertheless, microwires do not boost the light trapping inside the cells as much as SiNWs can. So far, most of the studies published on SiNW based HIT solar cells are based on thick crystalline silicon wafers, while the final objective is to apply this type of architecture to Si thin films. Indeed few works have been published on SiNWs based thin film solar cells

14,23,26

. In particular, Jia et al.26 have reported

recently an efficiency of 10% for solar cells based on SiNWs etched on poly-crystalline thin film of 8 µm. In this work we present a comprehensive study of HIT solar cells based on SiNWs using exclusively low-cost processes. The SiNWs are fabricated by a low-cost top-down approach – the standard metal assisted chemical etching (MACE) process

27–29

and the p-n junction is

formed by a low temperature PECVD process at 200 °C. Different a-Si:H deposition conditions were studied in order to optimize the performances of the devices. The effect of SiNWs length is also investigated. The devices are characterized by electron beam induced current (EBIC) which provides the current generation mapping at nanoscale. The effective carrier lifetime inside the devices is estimated using time resolved microwave conductivity (TRMC), to understand the electrical activity of the SiNWs. Finally the fabrication procedure is transferred to epitaxial c-Si thin films and the performance of such devices is discussed.


Page 5 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(a)

(b)

500 nm Figure 1: (a) Schematic of the structure of the HIT solar cell where the n-type cSi wafer has been nanostructured to improve light trapping. (b) SEM crosssection of a fabricated device.

2. Experimental 2.1

Fabrication of SiNW arrays by metal assisted chemical etching SiNW arrays were fabricated by the MACE, which is a method of anisotropic etching

relying on a local oxidation and reduction reactions in presence of noble metals [26-28]. MACE process enables the fabrication of dense SiNW arrays over large areas. This type of arrays is considered as random or disordered as there is no direct control over the diameter and arrangement on the C-Si wafer. The SiNWs diameters typically lie in the range of 20 − 200 nm, whereas their length is controlled by the etching time. Note that such small diameters are not suitable for the diffusion process generally used to form p-n junction.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

To start the fabrication, single-crystal N-type [100] wafers with resistivity 1-3 Ω-cm and a thickness of 550 µm, were cleaned in acetone, isopropyl alcohol (IPA) and deionized (DI) water by an ultrasonic bath for 10 min each at room temperature. The etching bath consisted of a solution of 4.6 M HF (Sigma Aldrich) and 0.02 M AgNO3 (Sigma Aldrich, Purity = 99.9999%) at room temperature. The length of the NWs was tuned by the etching time. Typically, NWs with lengths of 500 nm, 2 µm, 5-6 µm were obtained for etching times of 5 min, 12 min and 25 min respectively. The samples were then thoroughly rinsed for 5 min in DI water. The silver dentrites, produced by the etching process, were subsequently removed by subjecting the samples to a 15 minutes 65% nitric acid bath (Sigma Aldrich). The samples were thoroughly rinsed again in DI water and dried under nitrogen gas. 2.2

Fabrication of HIT devices by PECVD and finalization of devices The SiNW samples previously prepared by MACE were transferred to a RF PECVD

reactor (ARCAM) 30 after a prior removal of the native oxide layer by a 5% HF bath. The plasma deposition process was performed at 200 °C. Fig. 1 (a) and (b) show the schematic architecture of our device and its scanning electron microscopy (SEM) image respectively. The structure is as follows: indium tin oxide (ITO)/(p) a-Si:H/(i) a-Si:H/(n) c-Si/(i) a-Si:H/(n++) a-Si:H/ITO/Ag. First, an additional very thin layer of intrinsic amorphous silicon carbide (i) a-SiC:H was deposited on the top of the nanowires from the dissociation of a gas mixture of SiH4:CH4 (25:50) for 30 seconds with a low plasma power of 1 watt. This layer prevents the epitaxial growth of silicon which has a detrimental effect on the surface passivation and carrier lifetime. A thin intrinsic hydrogenated amorphous silicon layer (i) a-Si:H was then deposited using pure SiH4 (50 sccm, 1 watt). Finally, the window layer (p) a-Si:H was deposited from a gas mixture of


Page 6 of 34

Page 7 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


SiH4:trimethylboron (TMB) in argon. The gas ratio was varied during deposition, initially 50:5, then 50:10 and finally 50:20 to achieve a doping gradient. Different doping gradients were obtained by varying the deposition time. For the backside, a thin a-SiC:H layer (30s) followed by an (i) a-Si:H (1 min) layer were deposited. Following that, an (n++) a-Si:H layer was deposited from a gas mixture of SiH4:PH3 with ratio of 50:1 and 50:2 for 1 and 3 minutes, respectively, on the back of the samples, providing both surface passivation and ohmic contact to the ITO electrode. Finally top and bottom ITO contacts were deposited by sputtering through a shadow mask while the bottom Ag was deposited by thermal evaporation. The current density-voltage (J-V) characteristics of cells were measured under 100 mW/cm2 illumination (AM 1.5G) with a commercial solar simulator (Oriel AAA). The intensity of the light source was calibrated using a crystalline Si reference cell. Carrier lifetime measurements were carried out by using the time resolved microwave conductivity (TRMC) technique 31. The pulse light source used in our experiments is a Nd:YAG laser operating at 532 nm with 3 ns full width at half maximum (FWHM) pulses. The microwave part of the set up consists of a gun diode (Quinstar QTM 2720, emitting 100mW at 27GHz) generating the microwaves, a circulator to direct the generated and reflected microwaves and a Schottky diode used to detect the reflected microwave. The three components are connected via 3D waveguides. Electron beam induce current (EBIC)

32

measurements were carried out at room

temperature in a Hitachi SU8000 SEM system controlled by a Gatan DigiScan system. Micro-



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

manipulators, directly connected to Stanford Research System SR570 current amplifier, were used for contacting the devices. EBIC measurements were performed under zero bias. 3. Results and discussion 3.1 Effect of the thickness of hydrogenated amorphous silicon The main difficulty faced in SiNW based solar cells is the proper passivation of NW surface by a-Si:H. Indeed, the small size and the high aspect ratio of NWs make achieving conformal coverage extremely challenging. Considering this aspect, the SiNW cells are much more complex to fabricate than the reference planar cells. Besides, the deposition conditions have to be optimized for SiNW based solar cells. To optimize the passivation scheme for SiNWs, we have tested 6 different deposition conditions for the two front layers (i) a-Si:H and (p) a-Si:H. Disordered SiNWs with a height of 500 nm are used in the experiments. Six different samples are obtained, they are labeled as sample A, B, C, D, E and F, and are summarized in Table 1. All the samples studied have the same back contact, and only the front deposition conditions are different. The area of each cell is 0.126 cm2. The nominal thickness corresponds to the overall thickness of a-Si:H (Da-Si:H) measured on flat reference thin films deposited under the same conditions, i.e. Da-Si:H corresponds to the total thickness of both (i) a-Si:H and (p) a-Si:H layers. The thicknesses have been deduced from spectroscopic ellipsometry measurements and modeling using Tauc Lorentz dispersion law 33,34.


Page 8 of 34

Page 9 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 1: Summary of various deposition conditions with details of the deposition time of every layer. Samples Total deposition time Deposition time (p) a-Si:H + Deposition time (i) a-Si:H Overall thickness measured on planar Si (Da-Si:H)

A

B

C

D

E

F

4 min

7 min

8 min

13 min

17 min

21 min

14 nm

28 nm

32 nm

48 nm

64 nm

80 nm

The J(V) and EQE curves of the different type of cells are displayed in Fig. 2 (a) and Fig. 2 (b) respectively. The averaged values and the standard deviation of the corresponding solar parameters are summarized in Table 2. J(V) curve of sample C presents the smallest series resistance compared to samples with other thicknesses. A record efficiency of 12.9% was achieved for the best cell from sample C. (a)

(b)

Figure 2: Dependence of (a) (J,V) and (b) EQEcurves on the thickness of the amorphous layer (see Table 1).



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 34

Table 2: The average value and the standard deviation of Voc , Jsc , FF and η of core-shell heterojunction solar cells. A B C D E F Sample Voc Jsc

0.33 ± 0.01

Efficiency

0.53 ± 0.01

77.33 ± 5.0

81.05 ± 5.2

14.21 ± 2.5

28.71 ± 0.44

4.015 ± 0.31

10.73 ± 0.81

41.44 ± 1.9

FF

0.48 ± 0.01

29.27 ±1.1

12.43 ± 0.33

0.52 ± 0.005

30.17 ± 0.50

61.91 ± 2.0

9.66 ± 0.35

0.53 ± 0.004

25.03 ± 0.11

66.41 ± 1.3

8.31 ± 0.13

0.52 ± 0.005

23.71 ±0.51

64.43 ± 0.42 7.64 ± 0.21

Fig. 3 summarizes effects of the amorphous layer thickness Da-Si:H on the photovoltaic parameters (Jsc, FF, Voc and η) of the fabricated devices. It can be seen that Da-Si:H has a drastic effect on the performances of solar cells. If Da-Si:H is too thin (series A, 14 nm), the samples are characterized by Voc, Jsc, FF and η that are lowest from all sample types. It indicates a poor passivation by a-Si:H most likely due to an insufficient a-Si:H covering at the bottom of the NWs. When Da-Si:H is increased to 32 nm, the open-circuit voltage reaches 0.53 V as seen in Fig. 3 (a). There are two clearly identifiable trends: For 14 nm < Da-Si:H < 32 nm, the Voc increases until it reaches a plateau value which holds for larger thicknesses. Moreover both Jsc and FF achieve a good performance at the optimal a-Si:H thickness of 32 nm. Fig. 3 (b) illustrates the effect of thickness on Jsc. The poor coverage induced by too thin layers of a-Si:H does not provide enough surface passivation. When Da-Si:H is increased from 28 to 48 nm, Jsc increases before it drops drastically for Da-Si:H above 48 nm. We can relate these observations to the fact that thinner a-Si:H may not be able to provide surface passivation for the arrays of SiNWs, while thicker a-Si:H will lead to optical loss by absorption in the a-Si:H layer. As shown in Fig. 3 (c), the FF shows a maximum for a thickness of about 32 nm with an excellent value above 80%. On the other hand, lower FF values for thicknesses over 45 nm can be attributed to non-conformal ITO coating due to a large a-Si:H bump on the top of NWs. As a result, the best efficiency is reported for Da-Si:H = 32 nm, as illustrated in Fig. 3 (d).


Page 11 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(a)

(b)

(c)

(d)

Figure 3: The average value and the standard deviation of Voc (a), Jsc (b), FF (c) and η (d) as a function of Da-Si:H.

The SEM pictures of the samples D and F after a-Si:H PECVD deposition and before ITO sputtering are displayed in Figs. 4(a) and (b). Fig. 4 (c) shows sample F after ITO deposition. SEM pictures show that the coating of a-Si:H is becoming thicker with increasing deposition time. However this deposition is not uniform as shown by the bump shape of the heteroemitter coating where maximum Da-Si:H is observed at the top of the SiNWs. Da-Si:H is then reducing along the SiNW length from the top to the bottom , which might be explained by the challenging deposition of a-Si:H in high aspect ratio structure. The discrepancy in a-Si:H thickness between top and bottom of wires is becoming larger with increasing deposition time. Differences in thicknesses between the top and the bottom of SiNWs reaches up to 50% for sample F while



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 34

this difference is less than 10% for sample A. Fig. 4(c) shows that the ITO coating is not conformal as well and large bumps are formed at the top of the SiNWs. For sample F, bumps are very large and touch each other. For some SiNWs, the ITO is not reaching the bottom of the SiNWs, confirming our previous hypothesis. (a)

(b)

(c)

Figure 4: SEM pictures of SiNW solar cells after a-Si:H PECVD deposition of sample D (a) and sample F (b) before ITO deposition and sample F (c) after ITO deposition.

The stronger absorption loss at the front part of the cells with thicker a-Si:H film is confirmed by EQE measurements as shown in Fig. 2 (b). The blue response is strongly reduced with increasing Da-Si:H, indicating that a significant portion of the light is absorbed by a-Si:H near the front part of the cells but does not contribute to the collected current. However, for wavelength (λ) between 800 nm and 1000 nm, we observe higher EQE response for cells from samples B, C, D and E compared to cells from sample A with thinnest a-Si:H layer. The higher performance in the long wavelength regime is explained by the reduced surface recombination due to a proper surface passivation in combination with the light trapping capability of SiNWs. As for sample F, we observe an overall weaker EQE response as the thicker a-Si:H has resulted in substantial optical loss. Moreover, we suspect that large bumps observed for thicker a-Si:H deposition are making the conformal deposition of ITO difficult (see Fig. 4(c)) and may therefore play a significant role in the reduced EQE response measured. For λ > 1000 nm, all the EQE curves are nearly identical. Note that sample A displays a good EQE but poor J(V)


Page 13 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


characteristics. This can be attributed to the extremely thin a-Si:H passivation layer for this sample. In this case the passivation of the SiNWs surface is strongly deteriorated and therefore we obtain very poor J(V) characteristics under AM1.5 illumination conditions. Yet, the EQE shows a good collection as it is performed at much lower light intensity with a monochromatic light. In fact, sample A is a typical example stressing the importance of optimizing a-Si:H thickness - too thin might be good for optics, but not necessarily for electric properties.

The quality of the passivation provided by the different thicknesses of a-Si:H was then investigated by the time resolved microwave conductivity (TRMC) technique. We first review the main features and capabilities of the TRMC set up. The principle of the non-invasive and contactless TRMC technique is to measure the change in the microwave power reflected by a sample

35

. The laser pulse (λexc= 532 nm) generates an excess of free carriers () which

modifies the conductivity of the sample and therefore its refractive index at microwaves frequency. The relative change in the microwave power - the so-called microwave photoconductivity or TRMC signal ΔP(t) /P - is proportional to the variation of the sample’s conductivity Δσ(t) for small perturbations. Therefore we can write =

∆()

= ∆() = ∑ ()

(1)

where () is the number of excess charge carriers k at time t, is the mobility of the carrier k, e is the electron charge and A is a constant. TRMC enables us to monitor the kinetic decay of excess charge carriers caused by the recombination or trapping and can be used to evaluate the effective carrier lifetime and assess the surface passivation in a non-destructive way

35–40

. The measurement depth corresponds to the penetration depth of the green



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 34

excitation laser, which is in the order of one micrometer for crystalline silicon. Carriers are therefore photogenerated in the crystalline silicon wafer and thus we measure the effective lifetime in the wafer. For FZ material this amounts to characterizing the surface passivation. TRMC signals can usually be fitted by an exponential decay using the following expression:

=

(2)

with τ! being the effective lifetime. We have studied the effect of Da-Si:H deposited on the NWs on the TRMC signal in Fig. 5. With increasing Da-Si:H, the intensity of the signal is increasing and the decay much slower . The best effective lifetime is achieved for Da-Si:H = 32 nm (sample C), the same thickness as for the best Voc, confirming the proper passivation by a-Si:H for sample C. For thicker a-Si:H layers (Da-Si:H = 64 nm and Da-Si:H = 80 nm), the signal is very weak and we were not able to fit the data. The variations observed in the decay time and shape of the curves of the different TRMC signal implied different passivation quality and/or junction properties. A low TRMC signal might result from two main phenomena: the electron-hole recombination and carrier transfer from amorphous to crystalline layer. Devices with Da-Si:H = 14 nm , Da-Si:H = 28 nm and Da-Si:H = 32 nm present a principal decay mode which kinetics can be expressed by an exponential function. Note that we can observe a fast initial response. The fast initial decay could be related to the carriers which have diffused in the a-Si:H layer, where the lifetime is much shorter than in the c-Si wafer. For solar cells with Da-Si:H = 48 nm, 3 decay modes can be identified: a fast initial decay followed by a slower intermediate decay and the principal decay mode. In this experiment, the pulsed laser light entered the device through the front side of the cells, that is, through the SiNWs. We have also flipped the samples to have the


Page 15 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


laser light penetrating through the backside of the cells, which has a planar geometry. Generally, we observe a signal with a lower intensity and a slightly higher lifetime in the order of 10 µs. The lower intensity in the case of laser illumination of the planar backside is explained by the higher reflectivity of the planar backside which reduces the absorption in the wafer. We have observed a slightly better effective lifetime on the backside because of reduced recombination. However, irrespective of the side measured, the effective lifetime is very small whereas for planar cells the signal is in the order of ms independently of the illumination side. As all solar cells, including the planar ones, have the same backside, we can conclude that the recombination at the NW surface is the limiting factor.

Figure 5: TRMC signal as a function of time for HIT solar cells with different a-Si:H thicknesses. The inset shows the signal in logarithmic scale.

3.2 Effect of the SiNW length The effect of SiNW length on the performance of SiNW based HIT devices was further investigated. This work was motivated by the total reflectance spectra measured for different lengths of SiNWs 41 using a UV/Vis/NIR PE LAMBDA 750 spectrophotometer with an integrating



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 34

sphere as illustrated in Fig. 6(a). This figure shows that SiNWs provide very low reflectance from the visible to near infrared range and that the reflectance reduces with NW length.

(a)

(b)

High recombination rate with increasing SiNWs length

Better light trapping for longer NWs

Figure 6: (a) Reflectance spectra of SiNWs with different length. (b)Effect of the SiNWs length on the EQE of the cells.

Therefore, we have fabricated SiNW based HIT solar cells with SiNWs of different length. Figure 6(b) and Table 3 shows respectively the EQE curves and the current density-voltage J(V) characteristics of HIT solar cells with SiNWs length (L) of 800 nm, 1.5 µm and 10 µm, measured under 100 mW/cm2 illumination (AM 1.5G). The Voc for all the cells are lower than the highest value obtained for reference planar cell of 0.58 V. We should note that the reference planar cells were co-deposited in the same chamber as SiNW samples and naturally the deposition


Page 17 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


conditions are not optimal for them. From Table 3, the Voc and FF are observed to decrease with L, attributed to the fact that it is harder to properly passivate the long wires. Indeed, it is difficult to achieve conformal coating in high aspect ratio structure with the PECVD technique as it involves energetic bombardment. Therefore the coating of a surface depends on its slope. This leads to depositing a thicker layer in step corners and a thinner one in bottom corners. The ALD technique could be used to obtain a more uniform layer but it would be at the detriment of the deposition rate. This difficulty to passivate long SiNWs is corroborated by the EQE measurements shown in Fig. 6 (b), where the blue response weakens with increasing L, indicating that more recombination occurs at the front part of the cells where the NWs are located. Indeed longer SiNWs result in larger surface area, which means the presence of more defects and correspondingly higher recombination rate. Interestingly the overall EQE response for 700 nm < λ < 1000 nm increases for SiNWs with intermediate lengths of 800 nm and 1.5 µm. Such improvement is explained by the enhanced light trapping capability of longer SiNWs in combination with a proper surface passivation by a-Si:H. Higher EQE response in the long wavelength regime for 1.5 µm SiNWs shows that the improvement in light trapping in longer SiNWs overcomes the losses due to higher recombination. This increase in EQE is nevertheless not observed for very long SiNWs (L= 10 µm) despite the fact that longer SiNWs have better light harvesting properties as confirmed in Fig. 6(a). Therefore the reduction in EQE for very long SiNWs must be explained by the fact that enhanced light trapping offered by very long SiNWs cannot counterbalance the very high recombination rate due to an insufficient surface passivation. It is also possible that such very high aspect ratio structure prevents the formation of the junction at the bottom of the NWs which is also supported by the low Voc of the cells



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 34

with 10 µm long NWs (see Table 3). Likewise ITO may not reach the bottom of the wires and therefore the quality of the contacts might be degraded 42. In Fig. 6 (b), no significant difference in EQE is observed for λ > 1000 nm since the longer wavelength light is absorbed deep in the substrate and hence is not much affected by the SiNWs length. Table 3: Dependence of the photovoltaic parameters Jsc, FF, Voc and η on the length of SiNWs. Length of Voc (V) Jsc FF (%) η (%) 2 SiNWs (mA/cm )

planar 0.8 µm 1.5 μm 10 μm

0.58 0.49 0.37 0.28

29.4 28.9 26.59 8.38

60.2 69.2 42.33 36.62

10.25 9.69 4.16 0. 85

The decrease of efficiency of HIT SiNW based solar cells with increasing SiNW length (L) can be correlated with the TRMC signal measured on samples with different L. Indeed, for longer SiNWs the signal is weaker and the decay is much faster, indicating a decreasing carrier lifetime with increasing L as illustrated in Fig. 7. We were able to fit the TRMC signals with an exponential decay with good accuracy (χ 2> 0.998 with χ

2

being the fitting error of the

experimental curve by the theoretical model). To fit the experimental decay, we used equation (2). We found effective lifetimes of 12 µs, 11 µs and 6.9 µs for SiNW length of 300 nm, 500 nm and 1.5 µm, respectively. For 10 µm long SiNWs, we were unable to fit the data as the signal was too weak. For comparison, we have obtained a lifetime of the order of a millisecond for the reference planar cells. Similar to the case of SiNW HIT cells with different a-Si:H thickness, we therefore conclude that for all NW lengths the performance is limited by surface recombination.


Page 19 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


300 nm SiNWs 500 nm SiNWs 1.5 µm SiNWs 10 µm SiNWs

Figure 7: TRMC signal as a function of time for samples with different NW lengths. The inset shows the signal in logarithmic scale.

3.3 Electron Beam Induced Current (EBIC) microscopy As SiNW based HIT devices show reduced performances compared to their reference counterparts, a legitimate question arises: are SiNWs really contributing to the generated current, i.e. are SiNWs electrically active. To answer this question, SiNW based solar cells were characterized by the electron beam induced current (EBIC) technique 32. EBIC has been widely used to characterize nanowires and provides a unique tool that gives insight into the electrical parameters of nanowire devices

43–49

. This technique uses the interaction between a SEM

focused beam of electrons and a device to generate electron-hole pairs, which are separated by the built-in electric fields and then collected by probes connected to ammeter and measured as a current. The electron beam focused in a less than 10 nm2 spot scans the sample surface while building the SEM image and measuring the induced current at each point giving the corresponding EBIC map. Note that the volume of excitation is much bigger, even for low acceleration voltages.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 34

The influence of the electron beam energy was first investigated. Fig. 8 shows EBIC mappings of the cross-section of a HIT solar cell based on sample C, at different acceleration voltages of 3 keV Fig. 8 (a), 5 keV Fig. 8 (b) and 10 keV Fig. 8 (c). We have used the software Casino

50

to estimate the excitation volume for different acceleration voltages. Simulation

shows that the excitation volume is a sphere with its center close to the surface, sphere radius increases with the electron beam energy from 60 nm at 3 keV to 300 nm at 5 keV and finally to 1 µm at 10 keV. The dimensions of the excitation volume, calculated from modeling, are displayed in the inset of Fig. 8. At low acceleration voltage, the signal from SiNWs is small and the EBIC mapping is slightly distorted because of charging effects that cause the image to drift as shown in Fig. 8 (a). The EBIC signal increases at higher acceleration voltage. The maximal EBIC signal is generated close to the SiNWs/substrate interface, more precisely the signal maximum is located 100 nm down into the Si wafer from the interface as extracted from the EBIC profile in Figure 9.

a)

b)

200 nm

320 nm 100 nm

0.30x10-7 A

500 nm

c)

1.1 µm 240 nm

1.20x10-7 A 1.2x10-7 A

500 nm

2.42x10-7A

900 nm

1.3x10-7 A

500 nm

3.7x10-7 A

Figure 8: EBIC mapping of the HIT NW cross-section under various accelerating voltages (a) 3 keV, (b) 5 keV and (c) 10 keV.


Page 21 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Zone 1

Zone 2

Figure 9: EBIC current profile at 10 keV as a function of position of the beam on the sample C with the corresponding EBIC mapping in the inset. The red arrow in the inset indicates the corresponding signal position on the sample. A cross-section schematic of the structure is displayed at the bottom.

Fig. 9 shows a typical EBIC signal profile measured on the cross-section of a HIT solar cell based on disordered SiNWs (sample C) when the electron beam scans the sample from the top of the wires down into the substrate. The red arrow in the inset indicates the corresponding position on the sample where the EBIC profile has been extracted. The EBIC signal is the highest when the electron beam impinges the c-Si wafer at about 100 nm from the NW/c-Si interface, and it decays as the excitation moves further away from the interface, towards the nanowires or the bulk wafer. The signal decay away from the junction into the wafer reflects the minority carrier diffusion process, i.e. the decrease of the probability for carriers generated deep in the wafer to be collected by the junction. The signal decay on the NW side is due to the lower conversion efficiency of the NWs compared to the wafer as well as to the increased ITO thickness toward the NW top. The e-beam current Iebic and the diffusion length L are expected to follow the relationship given by Ioannou and Dimitriadis 51:



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

"#$%& =

' !()*

+, . -

Page 22 of 34

(3)

0 / 21

where A is a constant, L is the diffusion length and x is the distance between the junction and the electron beam. By plotting 0

= (3 1 "#$%& )

(4)

as a function of x, we obtain a straight line and the diffusion length L can be extracted from the slope equal to 1/L. The data from zone 2, i.e. the silicon substrate, in Fig. 9 was plotted and compared with Eq. (4) in Fig. 10. A good agreement between the fitted theoretical and experimental results is observed, with χ2> 0.995. From this adjustment, we obtained an effective diffusion length of 3 µm.

Figure 10: Equation (4) as a function of x, the distance from the junction, obtained from experimental data of zone 2 in Fig. 9.

SiNWs are covered with ITO as we need to contact them to collect the current in the EBIC set up. However, even a thin ITO layer reflects and absorbs a non-negligible amount of impinging electrons and therefore has an impact on the EBIC measurements that we need to consider when analyzing the results. The effect of ITO on the EBIC signal is shown in Figs. 11(a),


Page 23 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


11(b) and 11 (c). Fig 11 (d) displays the EBIC mapping of a solar cell that has underwent a special treatment and will be detailed later on in the paper. Fig. 11 shows a high magnification SEM image of a 500 nm long NWs (a) and the corresponding EBIC mapping image at 10 keV (b). The SiNW is partially covered by a 200 nm thick layer of ITO which acts as an electron barrier layer and reduces the signal intensity generated by the SiNWs since a darker contrast is observed in the zone covered by ITO. This is not the case for the bare bulk underlying wafer as we can observe in Fig. 11 (b). The reduction of the signal due to the presence of the ITO layer is clearly observed in Fig 11(c), where the wafer and the bottom of the NW, which is not covered with ITO have similar intensities while a decrease in the signal is observed on the portion of the NW covered by ITO. To confirm these observations, we have estimated the volume of excitation at 10 keV for a planar silicon substrate, a planar silicon substrates covered by 100 nm and 200 nm of ITO by the Casino software. As shown in Fig. 12, the excitation volume increases when reducing ITO thickness. The penetration depth inside a bare silicon substrate is about 900 nm (Fig. 12 (a)), which is reduced to 550 nm (Fig. 12 (b)) and 350 nm (Fig. 12 (c)) in the presence of 100 nm and 200 nm thick ITO film respectively, including ITO thickness. As mentioned earlier, the ITO coating is not uniform over the wires, as shown in Fig. 1(b) and Fig. 4(c). Indeed, the ITO coating appears as a bump on the top of SiNWs whose thickness reduces along the length. This is due to the difficulty in coating conformally high aspect ratio SiNWs by sputtered ITO. Therefore the effect of ITO on the measured data might be more prominent at the top of the NWs where the ITO thickness is larger. In conclusion the EBIC current generated by the SiNWs is underestimated because of their ITO coating.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

a)

Page 24 of 34

b) ITO with

ITO

Darker contrast

100 nm

100 nm

Bare Wafer

Bare Wafer

c)

d)

1 µm

Figure 11: (a) Cross-section SEM image of the bottom of a SiNW partially covered by ITO and (b) corresponding EBIC mapping with the red arrow indicating the corresponding signal position on the sample and (c) the corresponding EBIC current profile at 10 keV as a function of the position of the beam on the sample. (d) Cross-section EBIC mapping of samples with “dead” and electrically active SiNWs. The inset shows the EBIC top-view EBIC mapping confirming that only few nanowires are active.

a)

c) 200 nm ITO

b) 100 nm ITO

SILICON 550 µm

SILICON 550 µm 550 nm

900 nm

1.1 µm

880 nm

SILICON 550 µm

350 nm

600 nm

Figure 12: Modeling of the volume of excitation produced by a 10 keV electron beam impinging on a c-Si substrate with (a) no ITO, (b) 100 nm of ITO and (c) 200 nm of ITO.


Page 25 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


a)

b)

500 nm

Figure 13: EBIC current profiles (a) and cross-section EBIC current mapping (b). The 4 different arrows indicate the localization of the different EBIC profiles measured on the sample.

A noteworthy observation is the confirmation of the electrical activity of SiNWs, assessed by the EBIC current profile, although SiNWs are generating less current than the underlying bulk wafer. For each of our standard solar cells, the EBIC signal is uniform over the entire sample as we obtained similar EBIC profile for every SiNW we tested, as the one shown in Figure 9. For comparison an EBIC mapping of a solar cell that has underwent a special thermal oxidation treatment in order to damage the SiNWs is displayed in Fig. 11(d). This solar cell is presenting “dead” SiNWs (e.g not electrically active with dark contrast) and active SiNWs (bright contrast) while for our standard SiNWs solar cells, we observe uniform bright contrast for all the SiNWs. The nanoscale resolution provided by the EBIC measurements allows us to further analyze the carrier collection along the radial direction. This is illustrated in Fig. 13 showing the EBIC current profile of a long single nanowire of about 2 µm length (Fig. 13 (a)) with the corresponding EBIC mapping at 10 keV (Fig. 13 (b)). We can observe in Fig. 13 (a) that the signal is decreasing in the axial direction towards the NW tops, which means that the NW top part is



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 34

contributing less to the current than its bottom part which can be attributed to a reduction in efficiency. However, the ITO layer being thicker at the top of the wires and thinner at the bottom, the electron beam attenuation in the ITO layer might also have an impact on the EBIC signal intensity. Overall, we observe that SiNWs can generate and separate carriers but less effectively than the bulk underlying wafer. Moreover, the EBIC signal is very uniform for SiNWs over the entire sample and therefore every SiNW is contributing equally to the total generated current. 3.4 SiNWs cells based on epi-layer As mentioned in the introduction, the final objective of incorporating SiNWs into HIT devices is to use thinner Si wafers. With this objective in mind, we have fabricated HIT devices based on SiNWs prepared on a 5 µm thick commercial epitaxial silicon layer. The characteristics of the epitaxial thin film are as follows: thickness of 5 µm (+/- 0.5 µm), n-type doping (2-4 ohm.cm) and grown on a highly doped n-type silicon wafer (0.001-0.005 ohm.cm). With a doping concentration in the order of 1x1020 cm-3, the minority carrier diffusion length in the highly doped n-type substrate is less than 0.3 µm 52. Therefore the n++ substrate is acting as a conducting contact and its contribution to the photocurrent can be neglected. Hence, we can use this sample to simulate a silicon thin film with an effective absorbing thickness of 5 um. This assumption is confirmed by the cross-section EBIC picture of the solar cell device as shown in Fig. 14 (a), where we can clearly see that the highly doped wafer does not significantly contribute to the current and that only the epitaxial film is electrically active. Fig. 14 (b) shows the corresponding SEM pictures. These experiments provide a proof of


Page 27 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


concept to establish the feasibility of using thin film technology for HIT solar cells based on SiNWs. The SiNWs used for the purpose of this experiment have a length of 500 nm. Planar cells based on the 5 µm Si epitaxial films were also fabricated in parallel to serve as a reference.

a)

0.11x10-7 A

0.33x10-7 A

b)

5 µm

Figure 14: (a) EBIC mapping of SiNW solar cells based on 5 µm epitaxial film and (b) corresponding SEM picture.

Fig. 15 (a) and Fig. 15 (b) show respectively the J(V) and EQE curves of reference planar and SiNW based HIT devices fabricated on a 5 µm Si absorbing layer under 100 mW/cm2 illumination (AM 1.5G). The average photovoltaic parameters of Jsc, Voc, FF, and PCE are summarized in Table 4. The efficiency of the planar cells is higher than its SiNWs counterpart due to a sharp drop in the Voc, as well as a significant drop in the Jsc, despite the gain due to light trapping in the near-IR, mainly induced by the high recombination rate at the SiNWs surface in the front part of the cell, as confirmed by the EQE measurements shown in Fig. 15 (b). Interestingly, the FF is higher for SiNWs cells. The PCE of our cells are limited by their low Jsc which we can attribute to the insufficient light absorption for the 5 µm thin film. Moreover



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 34

there is almost no reflection between the epitaxial film and the supporting bulk wafer, which limits light trapping capabilities within the epitaxial film. To improve the performances of such devices, a thicker epitaxial film can be used. We can also consider transferring this device to a substrate which is either textured or is acting as a back reflector in order to promote light absorption within the thin film. (a)

(b)

Figure 15: EQE (a) and (I,V) (b) curves of reference planar cell and SiNWs based solar cells on 5 µm Si epitaxial thin film.


Page 29 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 4: Photovoltaic parameters planar cells and SiNWs based solar epitaxial layer. Solar cells Voc (V) Jsc 2 (mA/cm ) Reference 0.63 18.4 planar cell 500 nm 0.512 12.4 SiNWs cell

for reference cells on 5 µm FF (%)

η (%)

70.2

8.06

74.8

4.74

4. Conclusion We have investigated HIT devices based on random SiNW arrays fabricated by wet etching. The simplicity of the fabrication process exposed here coupled with the excellent FF (more than 80%) of the fabricated device proves the potential of using SiNWs for cheap and highly efficient HIT solar cells on low-cost thin films. The process optimization is essential in order to get a proper surface passivation. If the amorphous layer is too thin, it is impossible to cover conformally the extra surface induced by SiNWs. However, if too thick, the absorption in the amorphous layer reduces the intensity of the current generated by the devices because of strong absorption losses. By optimizing the deposition conditions, a remarkable efficiency of 12.9% was achieved for such kind of devices. The effect of the length of SiNWs was investigated. Only short SiNWs (< 1 µm) are suitable for PV applications as the performances are degrading for too long SiNWs. Indeed, long SiNWs are more difficult to passivate with aSi:H. The passivation problem results in a larger amount of surface defects responsible for a shorter carrier lifetime and subsequent degradation of the performances. The current generated by the SiNWs was mapped using EBIC microscopy. We have observed that SiNWs are electrically active but generate less current that the underlying bulk wafer due to strong carrier



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 34

recombination at the nanowires surface. The effect is becoming more ubiquitous with increasing length. Finally, we have proven that HIT cells based on SiNWs can be successfully transferred to the thin film technology. An efficiency of 4.74% was obtained for a silicon thin film with an effective absorbing thickness of 5 µm incorporating SiNWs.

5. Acknowledgments This work was partly supported by the Merlion project no. 2.02.11 sponsored by the French embassy in Singapore. The authors also acknowledge the financial support provided by Nanyang Technological University in Singapore. We also acknowledge the funding support from the Singapore Ministry of Education Academic Research Fund Tier 2, Grant No: MOE2012-T2-1104. Finally we would like also to thank the support from the ANR project NATHISOL (PROGELEC 2012) and ANR PLATOFIL (2014). The financial support from RTRA for the purchase of EBIC equipment is acknowledged. The authors also acknowledge F. Bayle for his assistance with EBIC set-up. 6. References (1) (2) (3)

(4) (5) (6)

Panasonic Global Press Release. http://news.panasonic.com (accessed Oct 29, 2014). Saga, T. Advances in Crystalline Silicon Solar Cell Technology for Industrial Mass Production. NPG Asia Mater. 2010, 2, 96–102. Yoon, J.; Li, L.; Semichaevsky, A. V.; Ryu, J. H.; Johnson, H. T.; Nuzzo, R. G.; Rogers, J. A. Flexible Concentrator Photovoltaics Based on Microscale Silicon Solar Cells Embedded in Luminescent Waveguides. Nat. Commun. 2011, 2, 343. Garnett, E.; Yang, P. Light Trapping in Silicon Nanowire Solar Cells. Nano Lett. 2010, 10, 1082– 1087. Basu Mallick, S.; Sergeant, N. P.; Agrawal, M.; Lee, J.-Y.; Peumans, P. Coherent Light Trapping in Thin-Film Photovoltaics. MRS Bull. 2011, 36, 453–460. Teo, B. K.; Sun, X. H. Silicon-Based Low-Dimensional Nanomaterials and Nanodevices. Chem. Rev. 2007, 107, 1454–1532.


Page 31 of 34

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(7) (8) (9) (10) (11)

(12)

(13) (14)

(15) (16)

(17) (18)

(19) (20) (21) (22) (23)

(24) (25) (26)

Kobayashi, Y.; Adachi, S. Properties of Si Nanowires Synthesized by Galvanic Cell Reaction. Jpn. J. Appl. Phys. 2010, 49, 075002. Bhattacharya, S.; Banerjee, D.; Adu, K. W.; Samui, S.; Bhattacharyya, S. Confinement in Silicon Nanowires: Optical Properties. Appl. Phys. Lett. 2004, 85, 2008. Peng, K.-Q.; Lee, S.-T. Silicon Nanowires for Photovoltaic Solar Energy Conversion. Adv. Mater. 2011, 23, 198–215. Garnett, E. C.; Brongersma, M. L.; Cui, Y.; McGehee, M. D. Nanowire Solar Cells. Annu. Rev. Mater. Res. 2011, 41, 269–295. Misra, S.; Yu, L.; Foldyna, M.; Roca i Cabarrocas, P. High Efficiency and Stable Hydrogenated Amorphous Silicon Radial Junction Solar Cells Built on VLS-Grown Silicon Nanowires. Sol. Energy Mater. Sol. Cells 2013, 118, 90–95. Misra, S.; Yu, L.; Foldyna, M.; Roca i Cabarrocas, P. New Approaches to Improve the Performance of Thin-Film Radial Junction Solar Cells Built Over Silicon Nanowire Arrays. IEEE J. Photovolt. 2015, 5, 40–45. Peng, K.; Xu, Y.; Wu, Y.; Yan, Y.; Lee, S.-T.; Zhu, J. Aligned Single-Crystalline Si Nanowire Arrays for Photovoltaic Applications. Small 2005, 1, 1062–1067. Sivakov, V.; Andrä, G.; Gawlik, A.; Berger, A.; Plentz, J.; Falk, F.; Christiansen, S. H. Silicon Nanowire-Based Solar Cells on Glass: Synthesis, Optical Properties, and Cell Parameters. Nano Lett. 2009, 9, 1549–1554. Fang, H.; Li, X.; Song, S.; Xu, Y.; Zhu, J. Fabrication of Slantingly-Aligned Silicon Nanowire Arrays for Solar Cell Applications. Nanotechnology 2008, 19, 255703. Peng, K.; Wu, Y.; Fang, H.; Zhong, X.; Xu, Y.; Zhu, J. Uniform, Axial-Orientation Alignment of OneDimensional Single-Crystal Silicon Nanostructure Arrays. Angew. Chem. Int. Ed. 2005, 44, 2737– 2742. Lin, X. X.; Hua, X.; Huang, Z. G.; Shen, W. Z. Realization of High Performance Silicon Nanowire Based Solar Cells with Large Size. Nanotechnology 2013, 24, 235402. Kumar, D.; Srivastava, S. K.; Singh, P. K.; Husain, M.; Kumar, V. Fabrication of Silicon Nanowire Arrays Based Solar Cell with Improved Performance. Sol. Energy Mater. Sol. Cells 2011, 95, 215– 218. Huang, B.-R.; Yang, Y.-K.; Lin, T.-C.; Yang, W.-L. A Simple and Low-Cost Technique for Silicon Nanowire Arrays Based Solar Cells. Sol. Energy Mater. Sol. Cells 2012, 98, 357–362. Lu, Y.; Lal, A. High-Efficiency Ordered Silicon Nano-Conical-Frustum Array Solar Cells by SelfPowered Parallel Electron Lithography. Nano Lett. 2010, 10, 4651–4656. Jia, G.; Steglich, M.; Sill, I.; Falk, F. Core–shell Heterojunction Solar Cells on Silicon Nanowire Arrays. Sol. Energy Mater. Sol. Cells 2012, 96, 226–230. Jia, G.; Eisenhawer, B.; Dellith, J.; Falk, F.; Thøgersen, A.; Ulyashin, A. Multiple Core–Shell Silicon Nanowire-Based Heterojunction Solar Cells. J. Phys. Chem. C 2013, 117, 1091–1096. Jia, G.; Gawlik, A.; Bergmann, J.; Eisenhawer, B.; Schonherr, S.; Andra, G.; Falk, F. Silicon Nanowire Solar Cells With Radial P-N Heterojunction on Crystalline Silicon Thin Films: Light Trapping Properties. IEEE J. Photovolt. 2014, 4, 28–32. Li, K.; Wang, X.; Lu, P.; Ding, J.; Yuan, N. The Influence of Passivation and Photovoltaic Properties of α-Si:H Coverage on Silicon Nanowire Array Solar Cells. Nanoscale Res. Lett. 2013, 8, 396. Gharghi, M.; Fathi, E.; Kante, B.; Sivoththaman, S.; Zhang, X. Heterojunction Silicon Microwire Solar Cells. Nano Lett. 2012, 12, 6278–6282. Jia, G.; Andrä, G.; Gawlik, A.; Schönherr, S.; Plentz, J.; Eisenhawer, B.; Pliewischkies, T.; Dellith, A.; Falk, F. Nanotechnology Enhanced Solar Cells Prepared on Laser-Crystallized Polycrystalline Thin Films (

Core–Shell Heterojunction Solar Cells Based on Disordered Silicon

Recommend Documents