Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spider-web

Nano Energy (]]]]) ], ]]]–]]]

1

Available online at www.sciencedirect.com

3 5 journal homepage: www.elsevier.com/locate/nanoenergy

7 9

RAPID COMMUNICATION

11

Q2

Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spider-web

Q1

Enzheng Shia,1, Hongbian Lib,1, Wenjing Xua, Shiting Wua, Jinquan Weic, Ying Fangb,n, Anyuan Caoa,n

13 15 17 19 21 23 25 27 29

a

Department of Materials Science and Engineering, College of Engineering, Peking University, Beijing 100871, PR China b National Center for Nanoscience and Technology, 11 Beiyitiao Street, Zhongguancun, Beijing 100190, PR China c School of Materials Science and Engineering, Tsinghua University, Beijing 100084, PR China

31 Received 3 April 2015; received in revised form 30 July 2015; accepted 24 August 2015

33 35 37 39 41 43 45 47 49

KEYWORDS

Abstract

Graphene; Carbon nanotube; Silicon; Solar cell; Photovoltaics

Graphene–Si solar cells have the potential to become low-cost, high-efficiency photovoltaics owing to simplified manufacturing process and compatible integration of two-dimensional graphene with Si wafers. Here, we present an effective way to improve the solar cell performance by introducing carbon nanotubes (CNT) into graphene, a structure called CNTembroidered graphene (CeG) film, and achieved a power conversion efficiency of 15.2% under standard illumination conditions, which is the highest efficiency for graphene-involved Si solar cell reported so far. The bi-continuous structure consisting of an interconnected CNT spiderweb uniformly embedded in graphene film is the material-related factor for improvement, while the co-existence of both graphene–Si and CNT–Si junctions is distinct from previous solar cells based on only one of these junctions. The stability of our CeG–Si solar cells in air storage or chemical doping has been studied, in which the influence of oxide growth and a linear relationship between the Schottky barrier height and open-circuit voltage have been experimentally observed. & 2015 Published by Elsevier Ltd.

51 53

63

55 n

57

Corresponding authors. E-mail addresses: [email protected] (Y. Fang), [email protected] (A. Cao). 1 These authors contributed equally to this work.

59 61

http://dx.doi.org/10.1016/j.nanoen.2015.08.018 2211-2855/& 2015 Published by Elsevier Ltd.

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018

65 67 69 71

2

E. Shi et al.

1

Introduction

3

As a typically two-dimensional (2D) sheet of sp2-hybridized carbon atoms, graphene has motivated tremendous interests in photovoltaics owing to its ultrathin geometry, high mechanical strength, good conductivity and high transparency [1–4]. Generally, graphene acts as the transparent conducting electrode to collect carriers generated by other semiconductors like semiconducting polymer [5,6], dyes [7], quantum-dot [8], and mono-crystalline Si [9]. Among those, due to the excellent light absorption of Si and relatively simple device fabrication process, graphene–Si (G–Si) solar cell has become one alternative to achieve high efficiency as well as low cost [9–15]. In the past four years, with many efforts such as the improvement in graphene quality, effective modulation of the electrical property of grapheme [10–12], and fabrication of antireflection microstructure [12,13] or addition of antireflection layer [14], graphene– Si solar cells have experienced a rapid progress in power conversion efficiency from initially about 1.5% to a current level of 14.5%. In graphene–Si solar cells, the most widely used graphene was synthesized by a chemical vapor deposition (CVD) method. During fabrication, CVD-grown graphene on Cu must be transferred to Si substrate. To date, the most commonly investigated approach is to use a support layer of poly (methyl methacrylate) (PMMA) [16,17], a thermal release tape [18,19], an elastomer stamp [19,20] or a self-releasing layer [21,22]. However, undesired polymer residues were frequently induced in the transfer process. Furthermore, the electrical conductivity and light transmittance are important factors underlying the energy conversion process. Although single-layer graphene has a high theoretical transmittance (97.7%), it also has a relatively high sheet resistance (usually on the order of hundreds to a few thousand Ω) which limits the solar cell efficiency. Therefore, it is a critical point to improve the graphene conductance without substantially decreasing the transparency. Recently, we have fabricated a carbon nanotube (CNT) embroidered graphene (CeG) film by direct CVD growth [23], in which a CNT spider-web was intimately embroidered into a continuous graphene sheet. The addition of CNTs not only improved the conductance of graphene sheet, but also led to coexistence of CNT–Si and G–Si junctions in the device. As a result, the power conversion efficiency (PCE) of the CeG–Si solar cell has been improved to 15.2%, higher than previous cells using only graphene. In addition, the stability and related mechanism for CeG–Si solar cell stored in air or under chemical doping have been analyzed in detail.

5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51

Results and discussion

53 55 57 59 61

The synthesis of CeG film was based on a low-pressure CVD method using a single-walled nanotube covered Cu foil substrate [23]. After CVD, the CeG film was released in freestanding form by dissolving Cu, and transferred without polymer support onto other substrates such as silicon wafers or transmission electron microscopy (TEM) grids. From the top view (Figure 1a) and bottom view (Figure 1b) of the suspended CeG film on TEM grid, we can see that onedimensional (1D) CNTs are uniformly distributed in

graphene, forming a bi-continuous 1D–2D hybrid structure. As the morphology of CNTs and graphene is similar when viewed from either the top or bottom side, this structure is called a CNT-embroidered graphene film as reported in our recent work [23]. Embedding such a CNT network actually does not disturb the continuity of CVD-grown graphene, but leads to a seamless combination of CNT skeletons with graphene although small gaps at the CNT–graphene interface were occasionally observed (Figure 1a and b). This perfect integration indicates that the high quality of CVDgrown graphene has been maintained after CNT integration. Although CNTs have been embroidered into graphene, it is not sure whether the high-temperature CVD process has degraded the original quality of CNTs. To analyze this, Raman spectra were recorded on a CNT spider-web in pristine form and after high-temperature annealing on a Cu foil (similar process to CVD growth of graphene), respectively, and statistical data were collected on about 20 positions for each sample (Figure 1c). For the pristine CNT film, the negligible D-band and the typical G-band indicate a well-crystallized nanotube structure with low defects. After annealing, only a few positions show increased D-band, which may be induced by Cu etching at elevated temperature. Most of CNTs exhibit similar Raman spectra compared with that of the pristine form, indicating high stability of CNTs. The ratio between D-band to G-band was less than 0.08 for all spider-webs subjected to hightemperature annealing, which is important for maintaining the structural integrity in the resulting CeG film. Embroidering a CNT spider-web in graphene caused slight decrease of optical transparency, but simultaneously improved the electrical conductance. For the optimal thickness of CeG film in high-performance solar cells, the light transmittance generally ranged from 80% to 90%. The as-synthesized CeG film demonstrated a light transmittance (T%) of 85.6% (at 550 nm wavelength), which was 5.8% lower than CNT spider-web (91.4%). A graphene film grown by the same process but without the CNT spider-web was predominantly mono-layer with a T% of 97.3%. Based on the difference in T% between CeG film and CNT spider-web, the mean layer of graphene in CeG (consisting of graphene domains with different layers) can be roughly estimated to be 5.8/2.3=2.5. Meanwhile, the sheet resistance of graphene, CNT spider-web, CeG film was measured to assess their respective conductance, which was more than 1000 Ω/□ for the graphene film, and 270 Ω/□ for the CNT spider-web. Notably, the sheet resistance of the CeG film was decreased to 170 Ω/□, indicating that the introduction of CNTs largely improved the conductance of graphene by more than 4-fold. Moreover, despite of the slightly increased defects in some CNTs (Figure 1c), CeG film demonstrated a better conductance than the pristine CNT spider-web. Therefore, CNTs and graphene have been combined into a highly transparent and conductive CeG film, a promising material in photovoltaic application. The fabrication of CeG–Si solar cells involves 3 steps. First, a floating CeG film released from Cu foil was transferred to a SiO2(300 nm)/n-type Si substrate (Figure 2a) from the surface of deionized water, which was followed by high-purity N2 blowing to remove the residue water between CeG and Si. Then the SiO2 layer was etched away by HF vapor, leaving the CeG film in direct contact to n-Si. After that, TiO2 colloids were spin-coated onto the CeG–Si surface as an anti-

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018

63 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93 95 97 99 101 103 105 107 109 111 113 115 117 119 121 123

Improvement of graphene–Si solar cells

3

1

63

3

65

5

67

7

69

9

71

11

73

13

75

15

77

17

79

19

81

21

83

23

85

25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61

Figure 1 Characterization of CeG film. (a) SEM image of suspended CeG film from the top-view side. (b) SEM image of suspended CeG film from the bottom-view side. (c) Raman spectra of 23 positions in one pristine CNT spider-web, and 20 positions of the CNT spider-web after high-temperature annealing. (d) Light transmission of CNT spider-web, mono-layer graphene and CeG film. Inset is a photograph of CeG film transferred onto quartz glass.

reflection layer to inhibit light reflection from Si. Crosssectional SEM image reveals that the CNT spider-web and graphene were sandwiched between the TiO2 coating and Si substrate (Figure 2b). The TiO2 coating was very uniform in an optimal thickness (60–70 nm) and compactly adhering to the beneath solar cell. Herein, CeG film serves as the transparent electrode for light transmission and charge collection, and Si was in charge of photon absorption. When either side of the CeG film was placed on Si, both CNTs and graphene made direct contact to Si, resulting in co-existence of CNT–Si and graphene–Si junctions, both are effective for charge separation in the CeG–Si solar cell (Figure 2c). The formation of CNT–Si or graphene–Si junctions depends on the relative position of CNT bundles and graphene sheets at the boundaries defined by the porous CNT network. During the CVD process, the graphene growing on the Cu surface may pass through the CNTs from the top side (if CNTs have sunk into melting Cu) or underneath (if CNTs are suspended above Cu); in the former case a CNT–Si junction will form during solar cell fabrication. Since the CNT spider-web is highly porous, we think that the graphene–Si junctions are dominating in the solar cell. Furthermore, the 2D graphene has a conformal contact to planar Si with nearly 100% surface coverage, and the increased junction area is a favorable factor for increasing the collecting efficiency of minorities (holes). Meanwhile, the highly conductive CNT spider-web can increase the transport efficiency of holes to outside electrode, thereby avoiding the loss of charge carriers caused by the large resistance of graphene. Therefore, the unique embroidering structure of CeG film made it an optimized candidate when combined with planar n-type Si to fabricate solar cells. The synergy

between CNT–Si and graphene–Si junctions led to highly efficient carrier collection and transportation. The photovoltaic properties were characterized under air mass 1.5 global (AM 1.5G) illumination at an intensity of 100 mW/cm2 (Figure 2d) The initial CeG–Si solar cell exhibited a short-circuit current density (JSC) of 23.5 mA/ cm2, an open-circuit voltage (VOC) of 0.387 V and a fill factor (FF) of 44.2%, which yielded a power conversion efficiency (PCE) of 4.0%. After the TiO2 layer was coated, the JSC reached 31.9 mA/cm2, increasing by 36% owing to the anti-reflection effect. And then, to further optimize the solar cell, chemical agents including concentrated HNO3 and H2O2 vapors were used to dope the cell, further improving VOC from 0.458 V to 0.618 V, and FF from 46.5% to 75.5%, respectively. Finally, the optimized CeG–Si solar cell showed a PCE of 15.2%, which was the highest reported efficiency for graphene-involved Si solar cells. Compared with the pure graphene–Si solar cell (with a best PCE of 14.5% [14]), embroidering CNTs into graphene is an effective method to improve photovoltaic performance. Also, this value is comparable with the best reported CNT–Si solar cell (15.1%) [24]. The stability of solar cells is important for practical applications. For a pristine CeG–Si solar cell without TiO2 coating and chemical doping, there appeared to be a gradual change in photovoltaic performance when the cell was stored in air (Figure 3a). Over an 11-day storage period, the J–V curves shifted toward larger Voc as the arrow indicated. And the evolution of the key parameters for this cell, including JSC, VOC, FF and PCE, as a function of storage time has been presented (Figure 3b). Initially, the fresh CeG–Si cell without native oxide at Si surface exhibited a low VOC (0.287 V) as

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018

87 89 91 93 95 97 99 101 103 105 107 109 111 113 115 117 119 121 123

4

E. Shi et al.

1

63

3

65

5

67

7

69

9

71

11

73

13

75

15

77

17

79

19

81

21

83

23

85

25

87

27

89

29 31 33

Figure 2 CeG–Si solar cells. (a) Schematic illustration of the fabrication of CeG–Si solar cell. (b) SEM image of the cross section of TiO2-coated CeG–Si solar cell. (c) Schematic image of the cross section of CeG–Si solar cell before and after adding the antireflection layer. The junction surrounded by the black dashed box is graphene–Si junction, while the junction surrounded by the red dashed box is CNT–Si junction. (d) J–V curves of CeG–Si solar cells at initial state, with TiO2 coating and by subsequent chemical doping under 100 mW/cm2 illumination. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

35 37 39 41 43 45 47 49 51 53 55 57 59 61

91 93 95 97

well as a low FF (24.7%). During the storage in air, the CeG–Si solar cell demonstrated a growing VOC from 0.287 V to 0.498 V. In addition, the FF also increased at the early stage to 46.9% (till 94-h storage). This was caused by the growth of a thin native oxide layer at Si surface, transforming the CeG– Si solar cell from metal–semiconductor (M–S) to metal–oxide– semiconductor (M–I–S) structure [25–27]. Insertion of a thin insulating layer between metal and semiconductor could boost the VOC [26,28,29]. However, for a given M–I–S diode structure, there is an optimum value of oxide thickness. With the thickening of oxide layer between CeG film and Si, the FF began to decrease possibly because the oxide layer had grown beyond the optimum thickness, reducing minority (hole) tunneling probability. During the storage, the JSC kept relatively stable. To systematically study the evolution of CeG–Si solar cells, some parameters have been extracted from the dark characteristics of CeG–Si solar cell under different storage times (from the fresh state to storage for 266 h in air), including the parasitic resistances (series resistance RS and shunt resistance RSh), diode ideality factor n and Schottky barrier height ΦSB. By plotting dV/d(lnJ) as a function of J (Figure 3d), the product of RS and the effective area of this solar cell Aeff, RS n Aeff, was obtained from the slope of the curve [30], while the reciprocal of the slope at negative bias voltage of dark I–V curve yielded RSh [31]. Besides, lnJ as a function of V within the forward bias voltage (0–1 V) yielded n and ΦSB

based on the formula below [10]. ln J ¼ lnðAnn T 2 Þ 

e e ΦSB þ V kB T nkB T

where Ann is Richardson constant (112 A/(cm2 K2)), kB is the Boltzmann constant (1.38n10  23 J/K), e is the electronic charge (1.6n10  19 C), T is the absolute temperature ( 298 K here). From the evolution of these parameters, it is found that n is small and ranges from 1.4 to 2.0 during the storage period. The low value and stability of n indicated restricted carrier recombination at the interface. And the gradually increased ΦSB directly resulted in the improvement of the VOC because of the linear correspondence between them [10], which was attributed to the growing native oxide at Si surface. In addition, both of RS and RSh increased with storage time in air. The increase in RSh will effectively lower the current leakage, whereas increased RS indicated more loss in carrier transportation. Generally, at the early storage stage (from the pristine state to 94 h), the FF kept increasing, indicating that the increase in RSh dominated the evolution, whereas later on (from 94 h to 266 h), the FF began to drop, indicating that the increase in RS dominated. Overall, with the storage in air, a native oxide layer was formed between Si and CeG film, which effectively modulated the photovoltaic performance of CeG–Si solar cell. Besides, chemical doping was usually used to improve the performance of graphene–Si solar cells, including HNO3,

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018

99 101 103 105 107 109 111 113 115 117 119 121 123

Improvement of graphene–Si solar cells

5

1

63

3

65

5

67

7

69

9

71

11

73

13

75

15

77

17

79

19

81

21

83

23

85

25

87

27

89

29

91

31

93

33

95

35 37 39 41

Figure 3 Evolution of CeG–Si solar cell stored in air. (a) J–V curves under 92.8 mW/cm2 illumination of CeG–Si cell with different storage times in air. (b) Calculated photovoltaic parameters (from the J–V curves in a)—JSC, VOC, FF and PCE—as a function of storage time in air. (c) Dark characteristics of the solar cell in a. (d) dV/d(lnJ)–J curve used to calculate RS. (e) lnJ–V curves used to calculate Q6 ΦSB and n. (f) Evolution of RS, RSh, n, ΦSB with storage time. The black arrows in (a), (c) and (e) indicate the evolution of J–V curves with storage time. The curves in black, red, green, blue, cyan, magenta colors in (a), (c), (d) and (e) correspond to the cell under 0, 26, 46, 73, 94, and 266 h storage, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

43 45 47 49 51 53 55 57 59 61

97 99 101 103 105

SOCl2 and AuCl3 [10,12,14]. As shown above, chemically doped CeG–Si solar cell exhibited a PCE of 15.2% (Figure 2d). The enhancement by chemical doing mainly arises from the improvement of FF and VOC. Previous studies have revealed that the chemical doping could increase the work function and conductivity of grapheme [32,33]. In addition, oxidative chemicals could also create a thin oxide layer at Si surface even for a short duration time, which transforms the M–S solar cell into M–I–S solar cell. Herein, we further discussed the function of HNO3 doping to the same CeG–Si cell discussed in Figure 3. After the storage in air for 266 h, the cell was doped by concentrated HNO3 vapor for 30 s. Then the VOC and FF immediately increased from 0.498 V to 0.588 V and 41.7% to 75.3%, respectively, both of which led to an enhanced PCE of 10.4%. This value was larger than that of the best graphene–Si (without surface nanostructures or antireflection layer) solar cell reported before [34]. Compared with the pristine CeG film, 30s-doping by concentrated HNO3 vapor led to a pronounced increase in

conductivity by 57% (Figure 4b). Meanwhile, the G-band in Raman spectra of the CeG film shifted from 1588 to 1591 cm  1 after HNO3 doping (Figure 4c), indicating the p-type doping of CeG film by HNO3 [35].However, with the desorption of HNO3 molecules on CeG film, the conductivity increase by doping gradually reduced to 40% after 17-min exposure to air (Figure 4b). The real-time photovoltaic performance had been recorded, exhibiting a shrinking tendency of J–V curves with continuous drop of FF (Figure 4a), thus demonstrating a degrading performance with increasing exposure time in air after doping. To observe the evolution of solar cell more clearly, JSC, VOC, FF and PCE have been plotted as a function of time after exposure (Figure 4d), respectively. It is noted that PCE was monotonically decreasing with the desorption of HNO3, as a result of the decreasing FF, decreasing VOC, and rather stable JSC. In addition, by analyzing the dark J–V curves (Figure 4e), the RS decreased upon doping and gradually increased with the desorption of HNO3, which is consistent with the change of the conductivity of CeG film after doping

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018

107 109 111 113 115 117 119 121 123

6

E. Shi et al.

1

63

3

65

5

67

7

69

9

71

11

73

13

75

15

77

17

79

19

81

21

83

23

85

25

87

27

89

29

91

31

93

33

95

35

97

37 39 41 43 45

Figure 4 Evolution of CeG–Si solar cell after HNO3 doping. (a) J–V curves under 92.8 mW/cm2 illumination of CeG–Si cell after HNO3 doping. (b) The change of conductance of CeG film upon and after doping. (c) The Raman shift of G-band after doping. (d) The calculated photovoltaic parameters (from the J–V curves in (a))—JSC, VOC, FF and PCE—as a function of time after doping. (e) Dark characteristics of the solar cell upon and after doping, corresponding to (a). The inset shows the evolution of RS, RSh, n, and ΦSB with time after doping. (f) VOC plotted as a function of ΦSB. All (VOC, ΦSB) dots are collected from the cell under different states in Figure 3 (in black dashed circle) and in this figure (in red dashed circle). The black arrows in (a) and (e) indicate the evolution of J–V curve with time after doping. And the curves in black, red, green, blue, cyan, and magenta colors in (a) and (e) correspond to the states of 0, 2, 3, 7, 10, and 17 min after doping, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

47 49 51 53 55 57 59 61

99 101 103 105 107 109

(Figure 4b). And the chemical p-doping increased the ΦSB from 0.855 V to 0.904 V. For this CeG–Si solar cell, we have studied the stability of CeG–Si solar cells with the growth of native oxide on Si surface in air and by chemical doping. In these two states, by plotting all the VOC values with the corresponding ΦSB together, we have experimentally confirmed the approximate linear relationship between VOC and ΦSB (Figure 4f). This revealed that increasing the ΦSB by chemical doping or natively growing an oxide layer is a feasible approach to tailor solar cell characteristics such as VOC. Meanwhile, during the preparation of this article, we noticed the publication of high-efficiency graphene Si solar cell with an efficiency of 15.6% [36], which was achieved by modulating the thickness of interfacial oxide at Si surface in combination with chemical doping and TiO2 antireflection

layer. This highlights that there is still room for improvement in the efficiency of graphene–Si solar cells.

111 113

Conclusion

115

In summary, we have fabricated CeG–Si solar cells with a PCE of 15.2%, which is the highest efficiency for graphene-involved Si solar cells. The synergy between highly conductive CNTs and ultra-smooth graphene contributes to the improved performance. Besides, from the systematic study on the storage in air and chemical doping of CeG–Si solar cells, we have revealed how the photovoltaic performance was effectively modulated by growing an insulation layer at Si surface and p-doping of CeG

117

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018

119 121 123

Improvement of graphene–Si solar cells film. It is possible to optimize graphene-involved Si solar cells by structural engineering of graphene with CNTs.

1

7

Structural characterizations and electrical measurements

3

65

5 Q4

Methods

7

Preparation of CNT spider-webs 9 11 13 15 17 19 21 23 25 27

CNT films were grown by atmospheric CVD (APCVD) with ferrocene/xylene/sulfur as precursors. The CVD furnace was heated to 1160 1C under 20 sccm Ar gas flow (99.999% purity). And then a solution of ferrocene (0.045 g/mL) and sulfur (0.001 g/mL) dissolved in xylene was injected into the upstream side at a rate of 5 mL/min. At the same time, the carrier gas was switched to a mixture of Ar and H2 (v/ v= 0.85:0.15) flowing at a rate of 1500 sccm. The reaction time at 1160 1C was controlled from 5 to 60 min to grow CNT films with different thicknesses. CNT films were floating to the downstream end and deposited on the inner wall of quartz tube (or collected by a nickel foil inserted in the downstream of the furnace). Lastly, a CNT film was peeled off from the inner wall of quartz tube or nickel foil and soaked into ethanol where individual SWNTs form bundles and the CNT film collapses into a spider-web. The CNT spider-web was then transferred onto water surface to expand into a floating film.

The structures of graphene, CNT films, CeG films, and monolithic G-CeG films were characterized by an optical microscope (Olympus BX51M) and scanning electron microscope (Hitachi S4800). Light transmission spectra of CNT spider-web, graphene and CeG film were recorded by an Agilent Cary 5000 UV–vis–NIR spectrophotometer. Raman spectra were recorded with a micro-Raman spectrometer (Renishaw inVia plus). Solar cell characteristics were tested by a source meter (Keithley 2635A) and a solar simulator (Newport Thermo Oriel 91195A-1000) under AM 1.5G condition, calibrated by a standard Si solar cell (91,150 V).

Synthesis of CeG films

33

The CeG films were synthesized by a low pressure CVD (LPCVD) method. A CNT spider-web was firstly transferred onto a 25 μm thick copper foil (99.8% purity) before it was loaded into a quartz tube inside a horizontal Lindberg/Blue furnace (Thermo Scientific). The system was evacuated to 40 mTorr, and then heated to 1050 1C under 7 sccm H2 gas flow (99.999% purity). After annealing in H2 for 30 min, CH4 (99.999% purity) at a flow rate of 10 sccm was introduced into the system to grow graphene. After exposure to CH4 for 30 min, the furnace was cooled to room temperature. The as-prepared CeG films can be transferred to arbitrary substrates with a polymer-free method. By simply etching the Cu foil in a FeCl3 (0.5 M) solution for 2 h, a floating CeG film was obtained. After rinse in water for three times, the CeG film was transferred onto an arbitrary substrate.

39 41 43 45 47 49 51

High-temperature annealing of CNTs

53

The annealing process is similar to the synthesis of CeG films. A CNT spider-web was firstly transferred onto a 25 μm thick copper foil (99.8% purity) before it was loaded into a quartz tube inside a horizontal Lindberg/Blue furnace (Thermo Scientific). The system was evacuated to 40 mTorr, and then heated to 1050 1C under 7 sccm H2 gas flow (99.999% purity). After annealing in H2 for 60 min, the furnace was cooled to room temperature. The annealed CNT spider-web would be transferred to SiO2/Si substrate for Raman characterization.

55 57 59 61

69 71 73 75

Acknowledgments A. C. acknowledges funding support from the National Natural Q5 Science Foundation of China (NSFC 51325202). Y. F. acknowledges NSFC (21322302 and 21173055). J. W. acknowledges NSFC (51172122). H. L. acknowledges NSFC (51202042).

79 81 83 85

Appendix A.

Supporting information 87

Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.nanoen. 2015.08.018.

89 91

31

37

67

77

29

35

63

References

93 95

[1] C. Lee, X.D. Wei, J.W. Kysar, J. Hone, Science 321 (2008) 385–388. [2] A.K. Geim, K.S. Novoselov, Nat. Mater. 6 (2007) 183–191. [3] K.S. Novoselov, V.I. Fal'ko, L. Colombo, P.R. Gellert, M.G. Schwab, K. Kim, Nature 490 (2012) 192–200. [4] A.S. Mayorov, R.V. Gorbachev, S.V. Morozov, L. Britnell, R. Jalil, L.A. Ponomarenko, P. Blake, K.S. Novoselov, K. Watanabe, T. Taniguchi, A.K. Geim, Nano Lett. 11 (2011) 2396–2399. [5] J.B. Wu, H.A. Becerril, Z.N. Bao, Z.F. Liu, Y.S. Chen, P. Peumans, Appl. Phys. Lett. 92 (2008) 263302. [6] L.G. De Arco, Y. Zhang, C.W. Schlenker, K. Ryu, M.E. Thompson, C.W. Zhou, ACS Nano 4 (2010) 2865–2873. [7] X. Wang, L.J. Zhi, K. Mullen, Nano Lett. 8 (2008) 323–327. [8] C.X. Guo, H.B. Yang, Z.M. Sheng, Z.S. Lu, Q.L. Song, C.M. Li, Angew. Chem. Int. Ed. 49 (2010) 3014–3017. [9] X.M. Li, H.W. Zhu, K.L. Wang, A.Y. Cao, J.Q. Wei, C.Y. Li, Y. Jia, Z. Li, X. Li, D.H. Wu, Adv. Mater.22 (2010) 2743–2748. [10] X. Miao, S. Tongay, M.K. Petterson, K. Berke, A.G. Rinzler, B.R. Appleton, A.F. Hebard, Nano Lett.12 (2012) 2745–2750. [11] X. Li, L. Fan, Z. Li, K. Wang, M. Zhong, J. Wei, D. Wu, H. Zhu, Adv. Energy Mater. 2 (2012) 425–429. [12] . Lin, X. Li, D. Xie, T. Feng, Y. Chen, R. Song, H. Tian, T. Ren, M. Zhong, K. Wang, H. Zhu, Energy Environ. Sci. 6 (2013) 108–115. [13] C. Xie, X.J. Zhang, K.Q. Ruan, Z.B. Shao, S.S. Dhaliwal, L. Wang, Q. Zhang, X.W. Zhang, J.S. Jie, J. Mater. Chem. A 1 (2013) 15348–15354. [14] E.Z. Shi, H.B. Li, L. Yang, L.H. Zhang, Z. Li, P.X. Li, Y.Y. Shang, S.T. Wu, X.M. Li, J.Q. Wei, K.L. Wang, H.W. Zhu, D.H. Wu, Y. Fang, A.Y. Cao, Nano Lett. 13 (2013) 1776–1781. [15] F. Bonaccorso, L. Colombo, G. Yu, V.T. Stoller, A.C. Ferrari, R.S. Ruoff, V. Pellegrini, Science 347 (2015) 1246501. [16] X.S. Li, Y.W. Zhu, W.W. Cai, M. Borysiak, B.Y. Han, D. Chen, R.D. Piner, L. Colombo, R.S. Ruoff, Nano Lett. 9 (2009) 4359–4363.

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018

97 99 101 103 105 107 109 111 113 115 117 119 121 123

8 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

E. Shi et al.

[17] X.L. Liang, B.A. Sperling, I. Calizo, G.J. Cheng, C.A. Hacker, Q. Zhang, Y. Obeng, K. Yan, H.L. Peng, Q.L. Li, X.X. Zhu, H. Yuan, A.R.H. Walker, Z.F. Liu, L.M. Peng, C.A. Richter, ACS Nano 5 (2011) 9144–9153. [18] Y. Lee, S. Bae, H. Jang, S. Jang, S.E. Zhu, S.H. Sim, Y.I. Song, B.H. Hong, J.H. Ahn, Nano Lett.(2010) 490–493. [19] J. Kang, S. Hwang, J.H. Kim, M.H. Kim, J. Ryu, S.J. Seo, B.H. Hong, M.K. Kim, J.B. Choi, ACS Nano 6 (2012) 5360–5365. [20] S.J. Kang, B. Kim, K.S. Kim, Y. Zhao, Z.Y. Chen, G.H. Lee, J. Hone, P. Kim, C. Nuckolls, Adv. Mater. 2011 (23) (2011) 3531–3535. [21] X.D. Chen, Z.B. Liu, C.Y. Zheng, F. Xing, X.Q. Yan, Y.S. Chen, J.G. Tian, Carbon 56 (2013) 271–278. [22] J. Song, F.Y. Kam, R.Q. Png, W.L. Seah, J.M. Zhuo, G.K. Lim, P.K.H. Ho, L.L. Chua, Nat. Nanotechnol. 8 (2013) 356–362. [23] E. Shi, H. Li, L. Yang, J. Hou, Y. Li, L. Li, A. Cao, Y. Fang, Adv. Mater. 27 (2015) 682–688. [24] E. Shi, L.H. Zhang, Z. Li, P. Li, Y. Shang, Y. Jia, J. Wei, K. Wang, H. Zhu, D. Wu, S. Zhang, A. Cao, Sci. Rep. 2 (2012) 884. [25] Y. Jia, A.Y. Cao, F.Y. Kang, P.X. Li, X.C. Gui, L.H. Zhang, E.Z. Shi, J.Q. Wei, K.L. Wang, H.W. Zhu, D.H. Wu, Phys. Chem. Chem. Phys. 14 (2012) 8391–8396. [26] D.L. Pulfrey, IEEE Trans. Electron Devices 25 (1978) 1308–1317. [27] M. Morita, T. Ohmi, E. Hasegawa, M. Kawakami, M. Ohwada, J. Appl. Phys. 68 (1990) 1272–1281. [28] J.P. Ponpon, P. Siffert, J. Appl. Phys. 47 (1976) 3248–3251. [29] Q. Zhang, X. Wang, D. Li, Z. Zhang, Adv. Funct. Mater.(2014) 7613–7618. http://dx.doi.org/10.1002/adfm.201402099. [30] S.K. Cheung, N.W. Cheung, Appl. Phys. Lett. 49 (1986) 85–87. [31] J.H. Werner, Appl. Phys. A 47 (1988) 291–300. [32] Y.M. Shi, K.K. Kim, A. Reina, M. Hofmann, L.J. Li, J. Kong, ACS Nano 4 (2010) 2689–2694. [33] K.K. Kim, A. Reina, Y.M. Shi, H. Park, L.J. Li, Y.H. Lee, J. Kong, Nanotechnology 21 (2010) 285205. [34] X. Li, D. Xie, H. Park, M. Zhu, T.H. Zeng, K. Wang, J. Wei, D. Wu, J. Kong, H. Zhu, Nanoscale 5 (2013) 1945–1948. [35] W. Zhou, J. Vavro, N.M. Nemes, J.E. Fischer, F. Borondics, K. Kamaras, D.B. Tanner, Phys. Rev. B 71 (2005) 205423. [36] Y. Song, X. Li, C. Mackin, X. Zhang, W. Fang, T. Palacios, H. Zhu, J. Kong, Nano Lett. 15 (2015) 2104–2110.

37 39 41 43 45 47 49 51 53 55 57 59

Enzheng Shi received his B.S. degree from School of Materials Science and Engineering, Shandong University and Ph. D. degree from Department of Materials Science and Engineering, Peking University. His research interests mainly focus on the synthesis of CNTs, graphene, and CNT/graphene hybrid films; the assembly of one-dimensional nanomaterials; the fabrication of CNT and graphene-involved solar cells; flexible allcarbon transistors and thermoelectric devices. Hongbian Li received her Ph.D. in 2008 in School of Chemistry and Chemical Engineering from Nanjing University. Then she worked as a postdoc researcher from 2009 to 2011 in Department of Materials Science and Engineering, Peking University. Currently, she is an associate professor in National Center for Nanoscience and Technology (NCNST). Her research interests focus on: (1) controllable synthesis of graphene-based film and their application in energy and sensors. (2) Fabrication of three dimensional carbon macrostructures and their application in environmental adsorption and filtration.

Wenjing Xu received her B.S. degree in School of Physics and Engineering from Zhengzhou University, China in 2013. She is currently a Ph. D. student, majoring in Material Science and Engineering in College of Engineering in Peking University. Her research interest focuses on fabrication and characterization of carbon-based thin-film solar-cell for efficient energyconversion.

63

Shiting Wu received her B. S. degree in Chemistry Materials from Central South University, China in 2012. She is a current Ph. D. candidate in Material Science and Engineering at college of engineering in Peking University, China. Her research interest focuses on the assembly of 1-dimensional nanomaterials (such as Te nanowires and carbon nanotubes) and 2-dimensional nanosheets by a blown bubble method. The assembled arrays are used in electronic devices (such as photo detectors and gas sensors) and solar cells.

73

Jinquan Wei received his B.S.and Ph.D. degrees from Department of Mechanical Engineering at Tsinghua University in 1999 and 2004, respectively. After two years of postdoctoral training at Tsinghua, he took a faculty position at Department of Mechanical Engineering at Tsinghua University, China. In 2013, he transferred to School of Materials Science at the same university. His research areas include nanomaterials and nanocomposites, solar cells and devices for energy storage.

85

65 67 69 71

75 77 79 81 83

87 89 91 93 95

Ying Fang received her B. S. degree from University of Science and Technology of China. And then she got her Master and Ph. D. degrees from Harvard University. Currently, she is a professor in National Center for Nanoscience and Technology (NCNST). Her research interests mainly focus on cardiac/brain activity mapping, nanoelectronics, bioelectronics and flexible electronics. Anyuan Cao received his PhD degree in Mechanical Engineering from Tsinghua University. He has spent 3years in Rensselaer Polytechnic Institute as a postdoc researcher, and 3 years in the University of Hawaii at Manoaas as an assistant professor. He is currently a professor in the Department of Materials Science and Engineering, College of Engineering, Peking University. His research areas include controlled synthesis of macroscopic structures based on carbon nanotubes and graphene, self-assembly, nanocomposites, nanoelectronics, energy, and environmental applications. He has published over 100 peer-reviewed journal papers.

61

97 99 101 103 105 107 109 111 113 115 117 119 121 123

Please cite this article as: E. Shi, et al., Improvement of graphene–Si solar cells by embroidering graphene with a carbon nanotube spiderweb, Nano Energy (2015), http://dx.doi.org/10.1016/j.nanoen.2015.08.018