Theoretical study on two-dimensional MoS 2 piezoelectric nanogenerators

Transcription

1 Nano Research DOI /s Nano Res 1 Theoretical study on two-dimensional MoS 2 piezoelectric nanogenerators Yongli Zhou 1,, Wei Liu 1, (*), Xin Huang 1,, Aihua Zhang 1, Yan Zhang 2, and Zhong Lin Wang 1,3 (*) Nano Res., Just Accepted Manuscript DOI: /s on November. 26, 2015 Tsinghua University Press 2015 Just Accepted This is a Just Accepted manuscript, which has been examined by the peer-review process and has been accepted for publication. A Just Accepted manuscript is published online shortly after its acceptance, which is prior to technical editing and formatting and author proofing. Tsinghua University Press (TUP) provides Just Accepted as an optional and free service which allows authors to make their results available to the research community as soon as possible after acceptance. After a manuscript has been technically edited and formatted, it will be removed from the Just Accepted Web site and published as an ASAP article. Please note that technical editing may introduce minor changes to the manuscript text and/or graphics which may affect the content, and all legal disclaimers that apply to the journal pertain. In no event shall TUP be held responsible for errors or consequences arising from the use of any information contained in these Just Accepted manuscripts. To cite this manuscript please use its Digital Object Identifier (DOI ), which is identical for all formats of publication.

2 64 Nano Res. Theoretical study on two-dimensional MoS 2 piezoelectric nanogenerators Yongli Zhou 1,, Wei Liu 1,,*, Xin Huang 1,, Aihua Zhang 1, Yan Zhang 2, and Zhong Lin Wang 1,3,* 1 Beijing Institute of Nanoenergy and Nanosystems, Chinese Academy of Sciences, Beijing , China 2 Institute of Theoretical Physics, and Key Laboratory for Magnetism and Magnetic Materials of MOE, Lanzhou University, Lanzhou , China 3 School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA The present theoretical study investigates both the static and dynamic output of the two-dimensional MoS 2 nanogenerator. Due to the symmetry restriction, nanogenerator consisting of odd number of MoS 2 layers is capable of giving obvious piezoelectric output voltage and current, which decreases as the increasing of the layer number. Zhong Lin Wang, Nano Research

3 Nano Research DOI (automatically inserted by the publisher) Research Article Theoretical study on two-dimensional MoS 2 piezoelectric nanogenerators Yongli Zhou 1,, Wei Liu 1, (*), Xin Huang 1,, Aihua Zhang 1, Yan Zhang 2, and Zhong Lin Wang 1,3 (*) 1 Beijing Institute of Nanoenergy and Nanosystems, Chinese Academy of Sciences, Beijing , China 2 Institute of Theoretical Physics, and Key Laboratory for Magnetism and Magnetic Materials of MOE, Lanzhou University, Lanzhou , China 3 School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA These authors contributed equally to this work Received: day month year Revised: day month year Accepted: day month year (automatically inserted by the publisher) Tsinghua University Press and Springer-Verlag Berlin Heidelberg 2014 KEYWORDS piezoelectric nanogenerator, MoS2, two-dimensional, mechanical-electrical energy conversion, high-frequency application ABSTRACT Recent experiment has demonstrated the nanogenerators fabricated using two-dimensional MoS2 flakes, which may find potential applications in electromechanical sensing, wearable technology, pervasive computing and implanted devices. In the present study, we theoretically examine the effect of the number of atomic layers in MoS2 flakes on the nanogenerators output. Under a square-wave applied strain, MoS2 flakes with an even number of atomic layers do not give piezoelectric output, which is due to the presence of projected inversion symmetry. On the other hand, for MoS2 flakes with an odd number of layers, owing to the lack of inversion symmetry, the piezoelectric output voltage and current are obvious and decreases as the increasing of the layer number. Furthermore, since MoS2 flakes are only few-atoms thick, the capacitance of the MoS2 nanogenerators is at least an order of magnitude smaller than that of nanowire/film based nanogenerators, which enables MoS2 nanogenerators for high-frequency applications. Our results explain the experimental observations and provide guidance on optimizing and designing the two-dimensional nanogenerators.

4 2 Nano Res. 1 Introduction Mechanical energy, such as air flow/vibration, hydraulic pressure, and body movements, widely exist in ambient environment and human daily life, which can be utilized to supplement the world s energy demands. Piezoelectric semiconductors, such as wurtzite-structured ZnO [1-3], GaN [4-6], and InN [7, 8], have coupled piezoelectric and semiconductor properties and can be used to fabricate the piezoelectric nanogenerators (NGs), which convert the mechanical energy into electricity [9]. Under an external applied strain, the piezoelectric semiconductor nanowires/films generate a piezoelectric potential, and strain-induced piezoelectric charges (piezocharges) at nanowire/film edges can drive the flow of electrons in an external circuit. When the applied strain is released, electrons flow back in the opposite direction. Up to date, NGs with different structures have been developed, such as the thin-film based nanogenerators [10], vertical/lateral nanowire arrays integrated nanogenerators [11], and two-dimensional (2D) woven nanogenerators [12], towards the goal for industrial productions and scalable applications. Recently, piezoelectricity and piezotronic effect have been observed in 2D atomic-thin MoS2 flakes for the first time [13]. Compared with the early-fabricated nanowires/films based NGs, the MoS2 NGs have the advantage in withstanding the enormous strain (up to 17% [14]). And MoS2 based power cell can be integrated with graphene and other 2D functional units or devices to construct an atomic-thin self-powered nanosystem that can operate without external experiments that the piezoelectric output of the MoS2 NG is sensitive to the number of atomic layers in the MoS2 flakes [13]: for the even-layer flakes, almost no detectable output can be seen; on the other hand, for the odd-layer flakes, the piezoelectric output can be observed but decreases as increasing of the layer number. The above observation requires further theoretical investigation. In this report, we theoretically examine the output performance of the MoS2 NG including up to 7 atomic layers of MoS2. The even-layer NGs have almost no output, which can be attributed to the presence of projected inversion symmetry in the MoS2 flake. On the contrary, due to the lack of inversion symmetry, the odd-layer NGs give obvious signal, which decreases upon the increase of the layer number because of the increasing capacitance. Furthermore, since the thickness of the MoS2 NGs lies in the dimension of nanometer, their capacitance is at least an order of magnitude smaller than the nanowire/film based NGs, which enables the MoS2 NGs to have potential applications as high-frequency functional devices. This study provides an understanding on the performance of the 2D nanogenerators and guidance for future device design. 2 Model and method Figure 1(a) gives a typical structure of the MoS2 NG. In this model, MoS2 flake consisting of three atomic layers is sandwiched between the left-hand and the right-hand electrodes. Each MoS2 layer is parallel to the x-y plane. MoS2 layers are numbered from the bottom to the top along the z-axis. For the MoS2 layer with an odd number, its ( 1010 ) zigzag edge (terminated by S atoms) contacts with the left electrode and (10 10 ) Address correspondence to Wei Liu, wliu@binn.cas.cn; Zhong Lin Wang, zlwang@gatech.edu bias by harvesting energy from the ambient environment. However, it is found in the zigzag edge (terminated by Mo atoms) contacts with the right electrode. On the other hand, the Nano Research

5 Nano Res. 3 MoS2 layer with an even number has an inverse orientation to the adjacent layers [15]: its ( 1010 ) edge contacts with the right electrode and (10 10 ) edge contacts with the left electrode. REx is an external load resistor, which contacts with the two electrodes. The MoS2 NG with other number of layers can be modeled similarly using the above method. According to the symmetry analysis, single-layer MoS2 has only one non-zero independent piezoelectric constant e11 [13]. Thus in the present study, the external strains are applied along the x-axis, which lead to the piezoelectric polarization along the same direction. In order to reveal the intrinsic properties of the MoS2 NG, we adopt a simple resistor-capacitor (RC) circuit to describe the electrical characteristics of the nanogenerator. According to the previous experimental and theoretical analyses [13, 16], the MoS2 NG consists of three parts: RIn is the NG internal resistance, CNG is the NG capacitance, and V is the voltage source, which utilizes the piezopotential to drive the electrons flow in the circuit. The equivalent circuit of the MoS2 NG with an external resistor is given in Fig. 1(b). Figure 1 shows the model of the MoS 2 NG adopted in the present study. (a) A schematic illustration of a three-layer MoS 2 NG in connection with an external load resistor R Ex. (b) The equivalent circuit of the three-layer MoS 2 NG. The static electrical characteristics of the MoS2 NG, including the open-circuit voltage V, the surface piezocharge Q at the zigzag edge under an external applied strain, and the NG capacitance CNG, are numerically simulated by the COMSOL software package following the method in the previous study [16]. The open-circuit voltage equals to the piezoelectric potential difference between the two electrodes. And the surface piezocharges at the two zigzag edges are calculated as the negative of the artificially added charges that cancel the piezopotential and result in a zero potential difference between the two electrodes. In most of the simulations, the MoS2 flake has a length (along x-axis) of 50 nm, a width (along y-axis) of 50 nm, and a thickness of n 0.65 nm, where n is the number of atomic layers in the flake and 0.65 nm is the thickness of a single-layer MoS2 [13]. To explore the dependence of the single-layer NG output on its in-plane dimension, MoS2 flakes with different lengths and widths, from 10 to 300 nm, are also simulated (Refer to Section 3.1). Compare with the MoS2 flakes used in the previous experiment which have an in-plane size of 5 µm 5 µm [13], the flakes used in the present study are quite small, because of difficulty of the convergence in COMSOL grogram when simulating the slab materials with a length/thickness ratio larger than 100. As a matter of fact, the single-layer MoS2 flake with an in-plane size of 5 nm 5 nm can be fabricated experimentally [17], thus we believe the MoS2 NGs simulated in the present study are feasible for the future experiments. For the single-layer MoS2, the adopted relative dielectric constants is er=3.3, the Poisson ratio v=0.34, the elastic constants C1=200 GPa, C12=49 GPa, and the piezoelectric constant e11=0.45 Cm - 2 [18, 19]. In order to examine the dynamical properties of the MoS2 NG in the circuit, a square-wave applied strain (refer to Fig. 3(a) in Section 3.2) is adopted following the previous experiment [13]. The time-dependent short-circuit current and output current in the external circuit are calculated by the following two steps: (1) first the time-dependent voltage V(t) of the voltage source (refer to Fig. 1(b)) is obtained as the open-circuit voltage of the NG under the real time applied strain, and (2) then the time-dependent currents I(t) is calculated using the V(t) by the PSpice program. In the circuit simulation, the adopted capacitance of the MoS2 NG is numerically calculated by the COMSOL program as mentioned in the above. And the adopted NG internal resistance RIn is 800 MW, which is obtained as the open-circuit voltage divided by the maximum short-circuit current in the previous experiment [13]. 3 Results and discussion Nano Research

6 4 Nano Res. 3.1 Static behavior of the MoS2 NG First we focus on the static properties of a single-layer MoS2 NG. Since the MoS2 NG is a 2D atomic-thin structure, the previous used parallel-capacitor model is not suitable for describing the current system. And the analysis of the NG open circuit V, piezocharge Q and the capacitance CNG should be based on the relation dv = dq / C. Figure 2(a) gives the linear dependence of the open-circuit voltage and surface piezocharge density at the MoS2-right electrode interface (MoS edge), which is obtained as surface piezocharge Q divided by the MoS2 edge area, on the applied strain. Under a tensile strain, positive piezocharges are created at the MoS2-right electrode interface and equivalent negative piezocharges are created at the MoS2-left electrode interface, thus the single-layer MoS2 NG has a positive open-circuit voltage (the electrical potential at the right electrode is higher than the one at the left electrode). The above simulation results are in consistent with the previous studies [13, 20]. Figure 2(b) shows the dependence of the open-circuit voltage and surface piezocharge density on the MoS2 flake length for a single-layer NG under an applied tensile strain of 0.5%. As the flake length becomes longer, the open-circuit voltage of the NG increases until its saturation at about 200 nm, while the surface piezocharge density keeps almost the same value. As a matter of fact, we also find that both the open-circuit voltage and surface piezocharge density does not depend on the flake width. The saturation of the NG open-circuit voltage deviates from the previous theoretical results on the nanowire/film based NGs, in which the open-circuit voltage is linearly dependent on the nanowire length or film thickness [10, 21], indicating the previously adopted parallel-capacitor model is not appropriate for describing the static behavior of the 2D NGs and a numerical simulation for accurate capacitance is necessary. The surface piezocharge density, on the other hand, is an intrinsic property of the MoS2, which is determined by the applied strain and does not depend on the flake length. Nano Research

7 Figure 2 gives the static behaviors of the MoS2 NGs. (a) The open-circuit voltage and surface piezocharge density of a single-layer MoS2 NG are linearly dependent on the applied tensile strain. (b) The open-circuit voltage and surface piezocharge density of a single-layer MoS2 NG versus MoS2 flake length under a 0.5% applied tensile strain. (c) The influence of the MoS2 layer number n on the NG open-circuit voltage and surface piezocharge under a 0.5% tensile strain. And (d) the dependence of the odd-layer NG capacitance on the flake length. Since the the metal electrode. At the NG interface, however, single-layer NG saturates at length 200 nm and different types of defects may appear, such as does not depend on the flake width, we can vacancies or dislocations, which may break the estimate an bonding between the MoS2 and metal electrode experimentally used 5 µm 5 µm NG as about thus results in the screening of the piezoelectric 0.35 V, which is much larger than the observed output by the localized metallic states. To value of 15 mv under a 0.5% strain [13]. The large enhance the piezoelectric output of the MoS2 NG difference in the future experiment, it is helpful to optimize the open-circuit open-circuit between the voltage voltage of of calculation and experimental results may due to the three facts: (1) the NG structure to minimize the electrode in the experiment, large capacitance of the NG capacitance and to eliminate the free carriers and may rise due to the two electrode plates, thus the edge metallic states in the MoS2 surface. Next we discuss the variation of the NG piezoelectric output on the MoS2 layer number. Adopting an external tensile strain of 0.5%, from Fig. 2(c) it can be found that the open-circuit voltage and surface piezocharge density observe the similar behavior dependent on the layer number: for the NGs with an even number of MoS2 decreasing the open-circuit voltage, (2) the strain-induced piezocharges may be partially screened by free carriers in the MoS2 [13], and (3) according to the previous studies [17, 20], localized metallic states exist at the edge of MoS2 flake which can be eliminated by contacting the

8 6 Nano Res. layers, due to the existence of the inversion symmetry, the open-circuit voltage and surface piezocharge density are almost zero, indicating no piezoelectric output from these NGs; on the other hand, for the NG with an odd number of atomic layers, the inversion symmetry is broken due to the unpaired single-layer MoS2, resulting in obvious open-circuit voltage and surface piezocharge. The total surface piezocharge Q in an odd-layer NG does not change by increasing the layer number in the MoS2 flake, since all the piezocharges are contributed by the unpaired single-layer MoS2. Then the surface piezocharge density is inversely proportional to the layer number as a consequence of the enlargement of the edge surface area in NGs with more MoS2 layers. And the NG capacitance becomes larger as the increasing of the layer number as shown in Fig. 2(d), which leading to the decrease of the open-circuit voltage. Figure 2(d) also gives the length dependence of the NG capacitance, from which the saturation of the capacitance from 200 nm can be found. It is the saturation of the capacitance of the NG results in the saturation of the open-circuit voltage shown in Fig. 2(b). From Fig. 2(d), we can estimate the capacitance of a 50 nm 50 nm single-layer MoS2 flake as Fa, which close to the experimentally measured value [22]. As a comparison, the capacitance of the ZnO nanowire fabricated experimentally, which typically has a diameter of 2µm and a length of 200 µm [2, 23], is at least one order larger than the MoS2 NG. The smaller capacitance of the MoS2 NG is a result of its atom-level thickness, and is an advantage of the MoS2 NG in harvesting the high frequency mechanical signals. 3.2 Dynamical output of the MoS2 NG under a square-wave applied strain After discussing the static behavior of the MoS2 NG, the dynamical output of the MoS2 NG is investigated in the present section. Figure 3(a) shows the square-wave applied strain used in the simulation. To emphasize the ability of the MoS2 NG in harvesting high-frequency mechanical signals, we adopt a periodical strain with a frequency of 0.5 GHz, namely, the periodicity of the strain is only 2 ns. Figure 3(b) gives the short-circuit currents (setting the REx in Fig. 1(b) equals to zero) of the NGs with different MoS2 layers. Both Figs. 3(a) and 3(b) demonstrate the two cycles of the energy harvesting and conversion from the mechanical to the electrical domain by the MoS2 NG. For the first circle, at time t = 4 ns in Fig. 3(a), a 0.5% tensile strain is suddenly applied on the NG. Effective piezocharges are induced at the zigzag edges of the Figure 3 shows piezoelectric outputs of MoS 2 NGs without external load. (a) A square-wave external strain applied on and released from the NG. And (b) corresponding short-circuit currents of the odd-layer MoS 2 NGs under the applied strain. Both figures give two cycles of the energy harvesting and conversion from the mechanical to the electrical domain by the MoS 2 NGs. MoS2 flake, driving the electron flow from the left electrode to the right electrode in the external circuit and giving rise to the current peak around the same time. As the time passes, the electrons Nano Research

9 Nano Res. 7 accumulate at the interfacial region between the right electrode and MoS2; the effect of the piezocharges is balanced by the accumulated electrons and the current gradually decreases to zero. When the tensile strain is released at t = 5 ns, the piezocharges vanish immediately and the electrons previously accumulated in the interfacial region flow back from the right electrode to the left electrode through the external circuit to return the system to the original state, leading to the negative current flow during the time from 5 ns to 6 ns. Since the equivalent circuit of the MoS2 NG is a RC-circuit, the behavior of short-circuit current that decreases from the peak to the zero can be understood by the capacitor discharging. Within each half cycle (e.g. t = 4 ~ 5 ns in Fig. 3(b)), the short-circuit current Isq(t) follows: 0 0 -( t-t0 ) R C In NG I ( t) = I ( t ) e. In the equation, t0 t is the sq time when the tensile strain is applied on or released from the MoS2 NG (refer to t = 4, 5, 6, and 7 ns in Fig. 3(b)) and I0(t0) = V(t0) / RIn is the maximum current at the time t = t0. The above equation suggests that the time-dependent short-circuit current follows an exponential decay after the strain is applied on or released from the NG, which is inconsistent to the experimental results [13]. The RC time constant of the circuit, which is the time required for the current to fall to 1/e, is equal to RInCNG. Providing the same internal resistance RIn in the present study, the RC constants are larger for those NGs with more MoS2 layers, which resulting in the slower decay rate of the short-circuit current in Fig. 3(b). It is noteworthy that in the present study the output time scale of the MoS2 NG is at the nano-second, which is drastically smaller than the time scale in the previous experiment [13]. As mentioned in the previous section, the high-frequency application of the MoS2 NG is due to its small capacitance, which is a consequence of the atomic-level thickness of the MoS2 flake. On the other hand, in the MoS2 NG experiment, the large time scale of the NG output is originated from the large capacitance in the circuit, which may due to the parallel connection of the capacitance from the electrode plates. By optimizing the NG circuit and improving the fabrication techniques, we believe the high-frequency output of the MoS2 NG can be realized in the future experiments. Nano Research

10 8 Nano Res. Figure 4 gives the piezoelectric outputs of odd-layer MoS 2 NGs under a square-wave external strain with an external load resistor. (a) The output peak current depends on the load resistance. (b) The output peak voltage across the resistor versus the external resistance. (c) The output peak power on the load resistor as a function of the external resistance. And (d) the energy conversion efficiency of the NG versus the layer number. Under the same square-wave applied strain with a tensile strain equals to 0.5% (refer to Fig. 3(a)), the dynamical response of the MoS2 NGs with an external resistor REx is given in Fig. 4. As the external resistance REx increases, the output peak current (which is the maximum of the time-dependent current in the circuit, refer to t = 4 ns in Fig. 3(b)) decreases notably when REx exceeds 10 MW, as shown in Fig. 4(a). The output peak voltage across the resistor becomes obvious near REx = 10 MW and saturates near 100 GW as shown in Fig. 4(b). Figure 4(c) gives the output peak power, which is the multiple of the peak voltage and peak current. For the odd-layer NG in the present study, the output peak power reaches the maximum value at REx near 800 MW and decreases as the increasing of the layer number. Figure 4(d) shows the energy conversion efficiency of the odd-layer NGs, which is obtained as the electrical output energy of the NG divided by the mechanical deformation energy stored in the MoS2 flake after being strained [13]. As the layer number increases, the energy conversion efficiency suffers an obvious decrease, which can be attributed to the higher mechanical deformation energy, larger NG capacitance, and lower open-circuit voltage for the NG consists of more MoS2 layers. The above results indicate that single-layer NG provides largest piezoelectric output and highest energy conversion efficiency under the square-wave applied strain, and should be considered as a preferential candidate in the fabrication and application of the 2D NGs. 3.3 Output of the MoS2 NGs under high-frequency applied strains

11 Nano Res. 9 Figure 5 shows the outputs of the MoS 2 NGs under the high-frequency mechanical signals. (a) Short-circuit current of the single-layer MoS 2 NG under a 5 GHz square-wave applied strain. (b) Energy conversion efficiency of the single-layer MoS 2 NG versus the frequency of the square-wave applied strain. Last but not least, the outputs of the MoS2 NGs under high-frequency mechanical signals are investigated in this section. Figure 5(a) gives the short-circuit current of a single-layer MoS2 NG under a 5 GHz square-wave applied strain. Compare with the case under the 0.5 GHz applied strain shown in Fig. 3(b), under a 5 GHz strain the peak current hardly relaxes in a half periodicity (0.1 ns) before the next peak current appears, which must give a negative influence on the energy conversion efficiency. In Fig. 5(b) the energy conversion efficiency of the single-layer MoS2 NG under different frequency of the square-wave applied strain is given. In the low-frequency region, the conversion efficiency is almost a constant value. In contrast, in the high-frequency region, the conversion efficiency suffers an exponential decrease as the frequency increases, which agrees with the results in a previous study [24]. The decrease of the conversion efficiency is due to the lack of the discharge time of the NG capacitor, as is shown in Fig. 5(a). According to Fig. 5(b), the single-layer MoS2 NG is capable of giving satisfactory output until the signal exceeds 1 GHz, which is much higher than the frequency limit of the ZnO nanowire NG that lying in the MHz region. We expect the advantage of the MoS2 NG working under high-frequency mechanical stimuli can be realized in the future experiments and device applications. As indicated in the introduction section, the MoS2-based NGs have high tolerance to the strain amplitude. Thus in the present study, we also simulate the output characteristics of the MoS2 NGs under the larger strain amplitude (up to 15%; other parameters such as the MoS2 flake size and NG internal resistance are same to the small-strain case), the results of which are given in Figs. S1, S2, S3, and S4 as supporting information. Compared to the results of small-strain case shown in corresponding Figs. 2, 3, 4, and 5, the MoS2 NGs under the larger strain give much higher output voltages and currents, which show qualitatively similar dependence on the strain/flake length/layer number as the small-strain case. On the other hand, the energy conversion efficiency hardly changes as the external mechanical strain becomes larger, indicating the stable of energy conversion by the MoS2 NGs under the enormous applied strain. These results reveal the advantages of the MoS2 NGs under the large-strain condition, which serve as guidance for the application of next generation 2D piezoelectric nanogenerators. 4 Conclusions We have investigated both the static and dynamical properties of the MoS2 based 2D nanogenerators under the external applied strains. Due to the symmetrical restriction, only odd-layer MoS2 NGs

12 10 Nano Res. give piezoelectric output. And both the NG output current and energy conversion efficiency decrease as the increasing of the MoS2 layer number. Furthermore, due to its atomic-level thickness, the capacitance of the MoS2 NGs is at least an order of magnitudes smaller than the capacitance of the nanowire/film based NGs, which enables the MoS2 NGs for promising applications under high-frequency mechanical signals. The present study provides not only an understanding on the mechanism of the 2D nanogenerators, but also guidance for future design and optimization of 2D piezoelectric nanogenerators. Acknowledgements This work was supported by the Thousands Talents Program for Pioneer Researcher and his Innovation Team, China, and the National Natural Science Foundation of China (Grant No ). Electronic Supplementary Material: Supplementary material (Fig. S1 gives the static behaviors of the MoS2 NGs in case of large external applied strain; Fig. S2 gives the piezoelectric outputs of MoS2 NGs without external load under a 15% external applied strain; Fig. S3 gives the piezoelectric outputs of odd-layer MoS2 NGs under a square-wave external strain (15%) with an external load resistor; and Fig. S4 shows the piezoelectric outputs of the MoS2 NGs under the high-frequency mechanical signal) is available in the online version of this article at (automatically inserted by the publisher). References [1] Wang, Z. L.; Song, J., Piezoelectric Nanogenerators Based on Zinc Oxide Nanowire Arrays. Science 2006, 312, [2] Yang, R.; Qin, Y.; Dai, L.; Wang, Z. L., Power generation with laterally packaged piezoelectric fine wires. Nat. Nanotechnol. 2009, 4, [3] Kim, D. Y.; Lee, S.; Lin, Z.-H.; Choi, K. H.; Doo, S. G.; Chang, H.; Leem, J.-Y.; Wang, Z. L.; Kim, S.-O., High temperature processed ZnO nanorods using flexible and transparent mica substrates for dye-sensitized solar cells and piezoelectric nanogenerators. Nano Energy 2014, 9, [4] Lin, L.; Lai, C.-H.; Hu, Y.; Zhang, Y.; Wang, X.; Xu, C.; Snyder, R. L.; Chen, L.-J.; Wang, Z. L., High output nanogenerator based on assembly of GaN nanowires. Nanotechnology 2011, 22, [5] Huang, C.-T.; Song, J.; Lee, W.-F.; Ding, Y.; Gao, Z.; Hao, Y.; Chen, L.-J.; Wang, Z. L., GaN Nanowire Arrays for High-Output Nanogenerators. J. Am. Chem. Soc. 2010, 132, [6] Wang, C.-H.; Liao, W.-S.; Lin, Z.-H.; Ku, N.-J.; Li, Y.-C.; Chen, Y.-C.; Wang, Z.-L.; Liu, C.-P., Optimization of the Output Efficiency of GaN Nanowire Piezoelectric Nanogenerators by Tuning the Free Carrier Concentration. Adv. Energy Mater. 2014, 4, [7] Wang, X.; Song, J.; Zhang, F.; He, C.; Hu, Z.; Wang, Z., Electricity Generation based on One-Dimensional Group-III Nitride Nanomaterials. Adv. Mater. 2010, 22, [8] Huang, C.-T.; Song, J.; Tsai, C.-M.; Lee, W.-F.; Lien, D.-H.; Gao, Z.; Hao, Y.; Chen, L.-J.; Wang, Z. L., Single-InN-Nanowire Nanogenerator with Upto 1 V Output Voltage. Adv. Mater. 2010, 22, [9] Hu, Y.; Wang, Z. L., Recent progress in piezoelectric nanogenerators as a sustainable power source in self-powered systems and active sensors. Nano Energy 2015, 14, [10] Lee, K. Y.; Kumar, B.; Seo, J.-S.; Kim, K.-H.; Sohn, J. I.; Cha, S. N.; Choi, D.; Wang, Z. L.; Kim, S.-W., P-Type Polymer-Hybridized High-Performance Piezoelectric Nanogenerators. Nano Lett. 2012, 12, [11] Wang, Z. L., Self-Powered Nanosensors and Nanosystems. Adv. Mater. 2012, 24, [12] Bai, S.; Zhang, L.; Xu, Q.; Zheng, Y.; Qin, Y.; Wang, Z. L., Two dimensional woven nanogenerator. Nano Energy 2013, 2, [13] Wu, W.; Wang, L.; Li, Y.; Zhang, F.; Lin, L.; Niu, S.; Chenet, D.; Zhang, X.; Hao, Y.; Heinz, T. F.; Hone, J.; Wang, Z. L., Piezoelectricity of single-atomic-layer MoS2 for energy conversion and piezotronics. Nature 2014, 514, [14] Bertolazzi, S.; Brivio, J.; Kis, A., Stretching and Breaking of Ultrathin MoS2. ACS Nano 2011, 5, [15] Duerloo, K.-A. N.; Ong, M. T.; Reed, E. J., Intrinsic Piezoelectricity in Two-Dimensional Materials. J. Phys. Chem. Lett. 2012, 3, [16] Huang, X.; Li, L.; Zhang, Y., Modeling the open circuit output voltage of piezoelectric nanogenerator. Sci. China Tech. Sci. 2013, 56, [17] Bollinger, M. V.; Lauritsen, J. V.; Jacobsen, K. W.; Nørskov, J. K.; Helveg, S.; Besenbacher, F., One-Dimensional Metallic Edge States in MoS2. Phys. Rev. Lett. 2001, 87, [18] Zhu, H.; Wang, Y.; Xiao, J.; Liu, M.; Xiong, S.; Wong, Z. J.; Ye, Z.; Ye, Y.; Yin, X.; Zhang, X., Observation of piezoelectricity in free-standing monolayer MoS2. Nat. Nanotechnol. 2015, 10, [19] RadisavljevicB; RadenovicA; BrivioJ; GiacomettiV; KisA, Single-layer MoS2 transistors. Nat. Nanotechnol. 2011, 6, [20] Liu, W.; Zhang, A.; Zhang, Y.; Wang, Z. L., Density functional studies on edge-contacted single-layer MoS2 piezotronic transistors. Appl. Phys. Lett. 2015, 107, [21] Hu, Y.; Zhang, Y.; Xu, C.; Zhu, G.; Wang, Z. L., High-Output Nanogenerator by Rational Unipolar Nano Research

13 Nano Res. 11 Assembly of Conical Nanowires and Its Application for Driving a Small Liquid Crystal Display. Nano Lett. 2010, 10, [22] Lu, C.-P.; Li, G.; Mao, J.; Wang, L.-M.; Andrei, E. Y., Bandgap, Mid-Gap States, and Gating Effects in MoS2. Nano Lett. 2014, 14, [23] Zhou, J.; Gu, Y.; Fei, P.; Mai, W.; Gao, Y.; Yang, R.; Bao, G.; Wang, Z. L., Flexible Piezotronic Strain Sensor. Nano Lett. 2008, 8, [24] Sun, C.; Shi, J.; Wang, X., Fundamental study of mechanical energy harvesting using piezoelectric nanostructures. J. Appl. Phys. 2010, 108, Nano Research

14 Nano Res. Electronic Supplementary Material Theoretical study on two-dimensional MoS 2 piezoelectric nanogenerators Yongli Zhou 1,, Wei Liu 1, (*), Xin Huang 1,, Aihua Zhang 1, Yan Zhang 2, and Zhong Lin Wang 1,3 (*) 1 Beijing Institute of Nanoenergy and Nanosystems, Chinese Academy of Sciences, Beijing , China 2 Institute of Theoretical Physics, and Key Laboratory for Magnetism and Magnetic Materials of MOE, Lanzhou University, Lanzhou , China 3 School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA These authors contributed equally to this work Supporting information to DOI /s12274-****-****-* (automatically inserted by the publisher) Address correspondence to Wei Liu, wliu@binn.cas.cn; Zhong Lin Wang, zlwang@gatech.edu Nano Research

15 Figure S1 gives the static behaviors of the MoS2 NGs in case of large external applied strain. (a) The open-circuit voltage and surface piezocharge density of a single-layer MoS2 NG under the applied tensile strain up to 15%. (b) The open-circuit voltage and surface piezocharge density of a single-layer MoS2 NG versus MoS2 flake length under a 15% applied tensile strain. (c) The influence of the MoS2 layer number n on the NG open-circuit voltage and surface piezocharge under a 15% tensile strain. And (d) the dependence of the odd-layer NG capacitance on the flake length. 96

16 Figure S2 gives the piezoelectric outputs of MoS 2 NGs without external load under a 15% external applied strain. (a) A square-wave external strain (15%) applied on and released from the NG. And (b) corresponding short-circuit currents of the odd-layer MoS 2 NGs under the applied strain. Both figures give two cycles of the energy harvesting and conversion from the mechanical to the electrical domain by the MoS 2 NGs. 97

17 Figure S3 shows the piezoelectric outputs of odd-layer MoS 2 NGs under a square-wave external strain (15%) with an external load resistor. (a) The output peak current depends on the load resistance. (b) The output peak voltage across the resistor versus the external resistance. (c) The output peak power on the load resistor as a function of the external resistance. And (d) the energy conversion efficiency of the NG versus the layer number. 98

18 Figure S4 shows the outputs of the MoS 2 NGs under a high-frequency 15% mechanical applied strain. (a) Short-circuit current of the single-layer MoS 2 NG under a 5 GHz square-wave applied tensile strain (15%). (b) Energy conversion efficiency of the single-layer MoS 2 NG versus the frequency of the square-wave applied strain. 99

19 Silver Nanowires with Semiconducting Ligands for Low Temperature Transparent Conductors Brion Bob, 1 Ariella Machness, 1 Tze-Bin Song, 1 Huanping Zhou, 1 Choong-Heui Chung, 2 and Yang Yang 1, * 1 Department of Materials Science and Engineering and California NanoSystems Institute, University of California Los Angeles, Los Angeles, CA (USA) 2 Department of Materials Science and Engineering, Hanbat National University, Daejeon , Korea Abstract Metal nanowire networks represent a promising candidate for the rapid fabrication of transparent electrodes with high transmission and low sheet resistance values at very low deposition temperatures. A commonly encountered obstacle in the formation of conductive nanowire electrodes is establishing high quality electronic contact between nanowires in order to facilitate long range current transport through the network. A new system of nanowire ligand removal and replacement with a semiconducting sol-gel tin oxide matrix has enabled the fabrication of high performance transparent electrodes at dramatically reduced temperatures with minimal need for post-deposition treatments of any kind. Keywords: Silver Nanowires, Sol-Gel, Transparent Electrodes, Nanocomposites 100

20 1. Introduction. Silver nanowires (AgNWs) are long, thin, and possess conductivity values on the same order of magnitude as bulk silver (Ag) [1]. Networks of overlapping nanowires allow light to easily pass through the many gaps and spaces between nanowires, while transporting current through the metallic conduction pathways offered by the wires themselves. The high aspect ratios achievable for solution-grown AgNWs has allowed for the fabrication of transparent conductors with very promising sheet resistance and transmission values, often approaching or even surpassing the performance of vacuum-processed materials such as indium tin oxide (ITO) [2-6]. Significant electrical resistance within the metallic nanowire network is encountered only when current is required to pass between nanowires, often forcing it to pass through layers of stabilizing ligands and insulating materials that are typically used to assist with the synthesis and suspension of the nanowires [7, 8]. The resistance introduced by the insulating junctions between nanowires can be reduced through various physical and chemical means, including burning off ligands and partially melting the wires via thermal annealing [9, 10], depositing additional materials on top of the nanowire network [11-14], applying mechanical forces to enhance network morphology [15-17], or using various other post-treatments to improve the contact between adjacent wires [18-21]. Any attempt to remove insulating materials the network must be weighed against the risk of damaging the wires or blocking transmitted light, and so many such treatments must be reined in from their full effectiveness to avoid endangering the performance of the completed electrode. We report here a process for forming inks with dramatically enhanced electrical contact between AgNWs through the use of a semiconducting ligand system consisting of tin oxide (SnO 2 ) nanoparticles. The polyvinylpyrrolidone (PVP) ligands introduced during AgNW synthesis in order to encourage one-dimensional growth are stripped from the wire surface using ammonium ions, and are replaced with substantially more conductive SnO 2, which then fills the space between wires and enhances the contact geometry in the vicinity of wire/wire junctions. The resulting transparent electrodes are highly conductive immediately upon drying, and can be effectively processed in air at virtually any temperature below 300 C. The capacity for producing high performance transparent electrodes at room temperature may be useful in the fabrication of devices that are damaged upon significant heating or upon the application of harsh chemical or mechanical post-treatments. 2. Results and Discussion 2.1. Ink Formulation and Characterization Dispersed AgNWs synthesized using copper chloride seeds represent a particularly challenging material system for promoting wire/wire junction formation, and often require thermal annealing at temperatures near or above 200 C to induce long range electrical conductivity within the deposited network [22, 23]. The difficulties that these wires present regarding junction formation is potentially due to their relatively large diameters compared to nanowires synthesized using other seeding materials, which has the capacity to enhance the thermal stability of individual wires according to the Gibbs-Thomson effect. We have chosen these wires as a demonstration of pre-deposition semiconducting ligand substitution in order to best illustrate the contrast between treated and untreated wires. Completed nanocomposite inks are formed by mixing AgNWs with SnO 2 nanoparticles in the presence of a compound capable of stripping the ligands from the AgNW surface. In this work, we have found that ammonia or ammonium salts act as effective stripping agents that are able to remove the PVP layer from the AgNW surface and allow for a new stabilizing matrix to take its place. Figure 1 shows a schematic of the process, starting from the precursors used in nanowire and nanoparticle synthesis and ending with the deposition of a completed film. The SnO 2 nanoparticle solution naturally contains enough ammonium ions from its own synthesis to effectively peel the insulating ligands from the AgNWs and allow the nanoparticles to replace them as a stabilizing agent. If not enough SnO 2 nanoparticles are used in the mixture, then the wires will rapidly agglomerate and settle to the bottom as large clusters. Large amounts of SnO 2 in the mixture gradually begin to increase the sheet resistance of the nanowire network upon deposition, but greatly enhance the uniformity, durability, and wetting properties of the resulting films. We have found that AgNW:SnO 2 weight ratios ranging between 2:1 and 1:1 produce well dispersed inks that are still highly conductive when deposited as films. The nanowires were synthesized using a polyol method that has been adapted from the recipe described by Lee et al. [22, 23] Silver nitrate dissolved in ethylene glycol via ultrasonication was used as a precursor in the presence of copper chloride and PVP to provide seeds and produce anisotropic morphologies in the reaction products. Synthetic details can be found in the experimental section. Distinct from previous recipes, we have found that repeating the synthesis two times without cooling down the reaction mixture generally produces significantly longer nanowires than a single reaction step. The lengths of nanowires produced using this method fall over a wide range from 15 to 65 microns, with diameters between 125 and 250 nm. This range of diameters is common for wires grown using copper chloride seeds, although the double reaction produces a number of wires with roughly twice their usual diameter. The morphology of the as-deposited AgNWs as determined via SEM is shown in Figure 2(a), higher magnification images are also provided in Figures 2(c) and 2(d). 101

21 The SnO 2 nanoparticles were synthesized using a sol-gel method typical for multivalent metal oxide gelation reactions. A large excess of deionized water was added to SnCl 4 5H 2 O dissolved in ethylene glycol along with tetramethylammonium chloride and ammonium acetate to act as surfactants. The reaction was then allowed to progress for at least one hour at near reflux conditions, after which the resulting nanoparticle dispersion can be collected, washed, and dispersed in a polar solvent of choice. The material properties of SnO 2 nanoparticles formed using a similar synthesis method have been reported previously [24], although the present recipe uses excess water to ensure that the hydrolysis reaction proceeds nearly to completion. After mixing with SnO 2 nanoparticles, films deposited from AgNW/SnO 2 composite inks show a largely continuous nanoparticle layer on the substrate surface with some nanowires partially buried and some sitting more or less on top of the film. Representative scanning electron microscopy (SEM) images of nanocomposite films are shown in Figure 2(b). Regardless of their position relative to the SnO 2 film, all nanowires show a distinct shell on their outer surface that gives them a soft and slightly rough appearance, as is visible in the higher magnification images shown in Figure 2(e) and 2(f). The SnO 2 nanoparticles do a particularly good job coating the regions near and around junctions between wires, and frequently appear in the SEM images as bulges wrapped around the wire/wire contact points. The precise morphology of the SnO 2 shell that effectively surrounded each AgNW was analyzed in more detail using transmission electron microscopy (TEM) imaging. Figures 3(a) to 3(c) show individual nanowires in the presence of different ligand systems: as-synthesized PVP in Figure 3(a), inactive SnO 2 in Figure 3(b), and SnO 2 activated with trace amounts of ammonium ions in Figure 3(c). The as-synthesized nanowires show sharp edges, and few surface features. In the presence of inactive SnO 2, which is formed by repeatedly washing the SnO 2 nanoparticles in ethanol until all traces of ammonium ions are removed, the nanowires coexist with somewhat randomly distributed nanoparticles that deposit all over the surface of the TEM grid. When AgNWs are mixed with activated SnO 2, a thick and continuous SnO 2 shell is formed along the nanowire surface. In when sufficiently dilute SnO 2 solutions are used to form the nanocomposite ink, nearly all of the nanoparticles are consumed during shell formation and effectively no nanoparticles are left to randomly populate the rest of the image. As the AgNWs acquire their metal oxide coatings in solution, the properties of the mixture change dramatically. Freshly synthesized AgNWs coated with residual PVP ligands slowly settle to the bottom of their vial or flask over a time period of several hours to one day, forming a dense layer at the bottom. The AgNWs with SnO 2 shells do not settle to the bottom, but remain partially suspended even after many weeks at concentrations that are dependent on the amount of SnO 2 present in the solution. A comparison of the settling behavior of various AgNW and SnO 2 mixtures after 24 hours is shown in Figures 3(d) and 3(e). The ratios 8:4, 8:16, and 8:8 indicate the concentrations of AgNWs and SnO 2 (in mg/ml) present in each solution. The 8:8 uncoupled solution, in which the PVP is not removed from the AgNW surface with ammonia, produces a situation in which the nanowires and nanoparticles do not interact with one another, and instead the nanowires settle as in the isolated nanowire solution while the nanoparticles remain well-dispersed as in the solution of pure SnO 2. The mixtures of nanowires and nanoparticles in which trace amounts of ammonia are present do not settle to the bottom, but instead concentrate themselves until repulsion between the semiconducting SnO 2 clusters is able to prevent further settling. Our current explanation for the settling behavior of the wire/particle mixtures is that the PVP coating on the surface of the as-synthesized wires is sufficient to prevent interaction with the nanoparticle solution. The addition of ammonia into the solution quickly strips off the PVP surface coating and allowing the nanoparticles to coordinate directly with the nanowire surface. This explanation is in agreement with the effects of ammonia has on a solution of pure AgNWs, which rapidly begin to agglomerate into clusters and sink to the bottom as soon as any significant quantity of ammonia is added to the ink. We attribute the stripping ability of ammonia in these mixtures to the strong dative interactions that occur via the lone pair on the nitrogen atom interacting with the partially filled d-orbitals of the Ag atoms on the nanowire surface. These interactions are evidently strong enough to displace the existing coordination of the five-membered rings and carbonyl groups contained in the original PVP ligands and allow the ammonia to attach directly to the nanowire surface. Since ammonia is one of the original surfactants used to stabilize the surface of the SnO 2 nanoparticles, we consider it reasonable that ammonia coordination on the nanowire surface would provide an appropriate environment for the nanoparticles to adhere to the AgNWs. 102

22 Scanning Energy Dispersive X-ray (EDX) Spectroscopy was also conducted on nanoparticle-coated AgNWs in order to image the presence of Sn and Ag in the nanowire and shell layer. The line scan results are shown in Figure 3(f), having been normalized to better compare the widths of the two signals. The visible broadening of the Sn lineshape compared to that of Ag is indicative of a Sn layer along the outside of the wire. The increasing strength of the Sn signal toward the center of the AgNW is likely due to the enhanced interaction between the TEM s electron beam and the dense AgNW, which then improves the signal originating from the SnO 2 shell as well. It is also possible that there is some intermixing between the Ag and Sn x-ray signals, but we consider this to be less likely as the distance between their characteristic peaks should be larger than the detection system s energy resolution Network Deposition and Device Applications For the deposition of transparent conducting films, a weight ratio of 2:1 of AgNWs to SnO 2 nanoparticles was chosen in order to obtain a balance between the dispersibility of the nanowires, the uniformity of coated films, and the sheet resistance of the resulting conductive networks. Nanocomposite films were deposited on glass by blade coating from an ethanolic solution using a scotch tape spacer, with deposited networks then being allowed to dry naturally in air over several minutes. The as-dried nanocomposite films are highly conductive, and require only minimal thermal treatment to dry and harden the film. Without the use of activated SnO 2 ligands, deposited nanowire networks are highly insulating, and become conductive only after annealing at above 200 C. The sheet resistance values of representative films are shown in Figure 4(a). The capability to form transparent conductive networks in a single deposition step that remain useful over a wide range of processing temperatures provides a high degree of versatility for designing thin film device fabrication procedures. Figure 5(a) shows the sheet resistance and transmission of a number of nanocomposite films deposited from inks containing different nanowire concentrations. The deposited films show excellent conductivity at transmission values up to 85%, and then rapidly increase in sheet resistance as the network begins to reach its connectivity limit. The optimum performance of these networks at low to moderate transmission values is a consequence of the relatively large nanowire diameters, which scatter a noticeable amount of light even when the conditions required for current percolation are just barely met. Nonetheless, the sheet resistance and transmission of the completed nanocomposite networks place them within an acceptable range for applications in a variety of optoelectronic devices. Figure 5(b) shows the wavelength dependent transmission spectra of several nanowire networks, which transmit light well out into the infrared region. The presence of high transmission values out to wavelengths well above 1300 nm, where ITO or other conductive oxide layers would typically begin to show parasitic absorption, is due to the use of semiconducting SnO 2 ligands, which is complimentary to the broad spectrum transmission of the silver nanowire network itself. Avoiding the use of highly doped nanoparticles has the potential to provide optical advantages, but can create difficulties when attempting to make electrical contact to neighboring device layers. In order to investigate their functionality in thin film devices, we have incorporated AgNW/SnO 2 nanocomposite films as electrodes in amorphous silicon (a-si) solar cells. Two contact structures were used during fabrication: one with the nanocomposite film directly in contact with the p-i-n absorber structure and one with a 10 nm Al:ZnO (AZO) layer present to assist in forming Ohmic contact with the device. The I-V characteristics of the resulting devices are shown in Figure 6(a). The thin AZO contact layers typically show sheet resistance values greater than 2.5 kω/, and so cannot be responsible for long range lateral current transport within the electrode structure. However, their presence is clearly beneficial in improving contact between the nanocomposite electrode and the absorber material, as the SnO 2 matrix material is evidently not conductive enough to form a high quality contact with the p-type side of the a-si stack. We hope that future modifications to the AgNW/SnO 2 composite, or perhaps the use of islands of high conductivity material such as a discontinuous layer of doped nanoparticles will allow for the deposition of completed electrode stacks that provide both rapid fabrication and good performance. Figure 6(b) contains the top view image of a completed device. The enhanced viscosity of the nanowire/sol-gel composite inks allows for films to be blade coated onto substrates with a variety of surface properties without reductions in network uniformity. In contrast with traditional back electrodes deposited in vacuum environments, the nanocomposite can be blade coated into place in a single pass under atmospheric conditions and dried within moments. We anticipate that the use of sol-gel mixtures to enhance wetting and dispersibility may prove useful in the formulation of other varieties of semiconducting and metallic inks for deposition onto a variety of substrate structures. 3. Conclusions In summary, we have successfully exchanged the insulating ligands that normally surround as-synthesized AgNWs with shells of substantially more conductive SnO 2 nanoparticles. The exchange of one set of ligands for the other is mediated by 103

23 the presence of ammonia during the mixing process, which appears to be necessary for the effective removal of the PVP ligands that initially cover the nanowire surface. The resulting nanowire/nanoparticle mixtures allow for the deposition of nanocomposite films that require no annealing or other post-treatments to function as high quality transparent conductors with transmission and sheet resistance values of 85% and 10 Ω/, respectively. Networks formed in this manner can be deposited quickly and easily in open air, and have been demonstrated as an effective n-type electrode in a-si solar cells when a thin interfacial layer is deposited first to ensure good electronic contact with the rest of the device. The ligand management strategy described here could potentially be useful in any number of material systems that presently suffer from highly insulating materials that reside on the surface of otherwise high performance nano and microstructures. 4. Experimental Details Tin oxide nanoparticle synthesis. Tin chloride pentahydrate was dissolved in ethylene glycol by stirring for several hours at a concentration of 10 grams per 80 ml to serve as a stock solution. In a typical synthesis reaction, 10 ml of the SnCl 4 5H 2 O stock solution is added to a 100 ml flask and stirred at room temperature. Still at room temperature, 250 mg ammonium acetate and 500 mg ammonium acetate were added in powder form to regulate the solution ph and to serve as coordinating agents for the growing oxide nanoparticles. 30 ml of water was then added, and the flask was heated to 90 C for 1 to 2 hours in an oil bath, during which the solution took on a cloudy white color. The gelled nanoparticles were then washed twice in ethanol in order to keep trace amounts of ammonia present in the solution. Additional washing cycles would deactivate the SnO 2, and then require the addition of ammonia to coordinate with as-synthesized AgNWs. Silver nanowire synthesis. Copper(ii) chloride dihydrate was first dissolved in ethylene glycol at 1 mg/ml to serve as a stock solution for nanowire seed formation. 20 ml of ethylene glycol was then added into a 100 ml flask, along with 200 µl of copper chloride solution. the mixture was then heated to 150 C while stirring at 325 rpm, and.35g of PVP (MW 55,000) was added. In a small separate flask,.25 grams of silver nitrate was dissolved in 10 ml ethylene glycol by sonicating for approximately 2 minutes, similar to the method described here. 22 The silver nitrate solution was then injected into the larger flask over approximately 15 minutes, and the reaction was allowed to progress for 2 hours. After the reaction had reached completion, the various steps were repeated without cooling down. 200 µl of copper chloride solution and.35g PVP were added in a similar manner to the first reaction cycle, and another.25g silver nitrate were dissolved via ultrasonics and injected over 15 minutes. The second reaction cycle was allowed to progress for another 2 hours, before the flask was cooled and the reaction products were collected and washed three times in ethanol. Nanocomposite ink formation. After the synthesis of the two types of nanostructures is complete, 104

24 the double washed SnO 2 nanoparticles and triple-washed nanowires can be combined at a variety of weight ratios to form the completed nanocomposite ink. The dispersibility of the mixture is improved when more SnO 2 is used, although the sheet resistance of the final networks will begin to increase if they contain excessive SnO 2. AgNW agglomeration during mixing is most easily avoided if the SnO 2 and AgNW solutions are first diluted to the range of 10 to 20 mg/ml in ethanol, with the SnO 2 solution being added first to an empty vial and the AgNW solution added afterwards. The dilute mixture was then be allowed to settle overnight, and the excess solvent removed to concentrate the wires to a concentration that is appropriate for blade coating. Film and electrode deposition. The completed nanocomposite ink was deposited onto any desired substrates using a razor blade and scotch tape spacer. The majority of the substrates used in this study were Corning soda lime glass, but the combined inks also deposited well on silicon, SiO 2, and any other substrates tested. Electrode deposition onto a-si substrates was accomplished by masking off the desired cell area with tape, and then depositing over the entire region. The p-i-n a-si stacks and 10 nm AZO contact layers were deposited using PECVD and sputtering, respectively. ACKNOWLEDGMENTS The authors would like to acknowledge the use of the Electron Imaging Center for Nanomachines (EICN) located in the California NanoSystems Institute at UCLA. REFERENCES [1] Sun, Y.; Gates, B.; Mayers, B.; Xia, Y., Crystalline silver nanowires by soft solution processing. Nano Lett. 2002, 2, [2] Kim, T.; Kim, Y. W.; Lee, H. S.; Kim, H.; Yang, W. S.; Suh, K. S., Uniformly interconnected silver-nanowire networks for transparent film heaters. Adv. Funct. Mater. 2013, 23, [3] Hu, L.; Wu, H.; Cui, Y., Metal nanogrids, nanowires, and nanofibers for transparent electrodes. MRS Bull. 2011, 36,

25 [4] van de Groep, J.; Spinelli, P.; Polman, A., Transparent conducting silver nanowire networks. Nano Lett. 2012, 12, [5] Yang, L.; Zhang, T.; Zhou, H.; Price, S. C.; Wiley, B. J.; You, W., Solution-processed flexible polymer solar cells with silver nanowire electrodes. ACS Appl. Mater. Interfaces 2011, 3, [6] Scardaci, V.; Coull, R.; Lyons, P. E.; Rickard, D.; Coleman, J. N., Spray deposition of highly transparent, low-resistance networks of silver nanowires over large areas. Small 2011, 7, [7] Wiley, B.; Sun, Y.; Xia, Y., Synthesis of silver nanostructures with controlled shapes and properties. Acc. Chem. Res. 2007, 40, [8] Korte, K. E.; Skrabalak, S. E.; Xia, Y., Rapid synthesis of silver nanowires through a cucl- or cucl2-mediated polyol process. J. Mater. Chem. 2008, 18, [9] Anuj, R. M.; Akshay, K.; Chongwu, Z., Large scale, highly conductive and patterned transparent films of silver nanowires on arbitrary substrates and their application in touch screens. Nanotechnology 2011, 22, [10] Lee, J.-Y.; Connor, S. T.; Cui, Y.; Peumans, P., Solution-processed metal nanowire mesh transparent electrodes. Nano Lett. 2008, 8, [11] Zhu, R.; Chung, C.-H.; Cha, K. C.; Yang, W.; Zheng, Y. B.; Zhou, H.; Song, T.-B.; Chen, C.-C.; Weiss, P. S.; Li, G.; Yang, Y., Fused silver nanowires with metal oxide nanoparticles and organic polymers for highly transparent conductors. ACS Nano 2011, 5, [12] Chung, C.-H.; Song, T.-B.; Bob, B.; Zhu, R.; Duan, H.-S.; Yang, Y., Silver nanowire composite window layers for fully solution-deposited thin-film photovoltaic devices. Adv. Mater. 2012, 24,

26 [13] Kim, A.; Won, Y.; Woo, K.; Kim, C.-H.; Moon, J., Highly transparent low resistance zno/ag nanowire/zno composite electrode for thin film solar cells. ACS Nano 2013, 7, [14] Ajuria, J.; Ugarte, I.; Cambarau, W.; Etxebarria, I.; Tena-Zaera, R. n.; Pacios, R., Insights on the working principles of flexible and efficient ito-free organic solar cells based on solution processed ag nanowire electrodes. Sol. Energy Mater. Sol. Cells 2012, 102, [15] Tokuno, T.; Nogi, M.; Karakawa, M.; Jiu, J.; Nge, T.; Aso, Y.; Suganuma, K., Fabrication of silver nanowire transparent electrodes at room temperature. Nano Res. 2011, 4, [16] Lim, J.-W.; Cho, D.-Y.; Jihoon, K.; Na, S.-I.; Kim, H.-K., Simple brush-painting of flexible and transparent ag nanowire network electrodes as an alternative ito anode for cost-efficient flexible organic solar cells. Sol. Energy Mater. Sol. Cells 2012, 107, [17] De, S.; Higgins, T. M.; Lyons, P. E.; Doherty, E. M.; Nirmalraj, P. N.; Blau, W. J.; Boland, J. J.; Coleman, J. N., Silver nanowire networks as flexible, transparent, conducting films: Extremely high dc to optical conductivity ratios. ACS Nano 2009, 3, [18] Hu, L.; Kim, H. S.; Lee, J.-Y.; Peumans, P.; Cui, Y., Scalable coating and properties of transparent, flexible, silver nanowire electrodes. ACS Nano 2010, 4, [19] Garnett, E. C.; Cai, W.; Cha, J. J.; Mahmood, F.; Connor, S. T.; Greyson Christoforo, M.; Cui, Y.; McGehee, M. D.; Brongersma, M. L., Self-limited plasmonic welding of silver nanowire junctions. Nat. Mater. 2012, 11,

27 [20] Yu, Z.; Zhang, Q.; Li, L.; Chen, Q.; Niu, X.; Liu, J.; Pei, Q., Highly flexible silver nanowire electrodes for shape-memory polymer light-emitting diodes. Adv. Mater. 2011, 23, [21] Song, T.-B.; Chen, Y.; Chung, C.-H.; Yang, Y.; Bob, B.; Duan, H.-S.; Li, G.; Tu, K.-N.; Huang, Y., Nanoscale joule heating and electromigration enhanced ripening of silver nanowire contacts. ACS Nano 2014, 8, [22] Lee, P.; Lee, J.; Lee, H.; Yeo, J.; Hong, S.; Nam, K. H.; Lee, D.; Lee, S. S.; Ko, S. H., Highly stretchable and highly conductive metal electrode by very long metal nanowire percolation network. Adv. Mater. 2012, 24, [23] Lee, J. H.; Lee, P.; Lee, D.; Lee, S. S.; Ko, S. H., Large-scale synthesis and characterization of very long silver nanowires via successive multistep growth. Cryst. Growth Des. 2012, 12, [24] Bob, B.; Song, T.-B.; Chen, C.-C.; Xu, Z.; Yang, Y., Nanoscale dispersions of gelled Sno2: Material properties and device applications. Chem. Mater. 2013, 25,

28 Figure 1. Process flow diagram showing the synthesis of AgNWs and SnO 2 nanoparticles followed by stirring in the presence of ammonium salts to create the final nanocomposite ink. Transparent conducting films were produced by blade coating the completed inks onto the desired substrate. 109

29 Figure 2. (a,c,d) SEM images of as-synthesized AgNWs at various magnifications. (b,e,f) SEM images of nanocomposite films, showing the tendency of the SnO 2 nanoparticles to coat the entire outer surface of the AgNWs, increasing their apparent diameter and giving them a soft appearance. 110

30 Figure 3. Schematic diagrams and TEM images of (a) a single untreated AgNW, (b) an AgNW in the presence of uncoupled SnO 2 (all ammonium ions removed), and (c) an AgNW with a coordinating SnO 2 shell. Scale bars in images (a), (b), and (c) are 300 nm, 400 nm, and 600 nm, respectively. (d,e) Optical images of AgNW and SnO 2 nanoparticle dispersions mixed in varying amounts (d) before and (e) after settling for 24 hours. The numbers associated with each solution represent the AgNW:SnO 2 concentrations in mg/ml. The uncoupled solution contains AgNWs and non-coordinating SnO 2 nanoparticles, and shows settling behavior similar to the pure AgNW and pure SnO 2 solutions. (f) Normalized Ag and Sn EDX signal mapped across the diameter of a single nanowire, with the inset showing the scanning path across an isolated wire. 111

31 Figure 4. Sheet resistance versus temperature for films deposited using (red) AgNWs that have been washed three times in ethanol and (blue) mixtures of AgNW and SnO 2 with weight ratio of 2:1. The annealing time at each temperature value was approximately 10 minutes. The large sheet resistance values of the bare AgNWs when annealed below 200 C is typical for nanowires fabricated using copper chloride seeds, which clearly illustrate the impact of SnO 2 coordination at low treatment temperatures. 112

32 Figure 5. (a) Sheet resistance and transmission data for samples deposited from solutions of varying nanostructure concentration. Each of these samples were fabricated starting from the same nanocomposite ink, which was then diluted to a range of concentrations while maintaining the same AgNW to SnO 2 weight ratio. (b) Transmission spectra of several transparent conducting networks chosen from the plot in plot (a). 113