Largely-increased length of silver nanowires by controlled oxidative etching processes in solvothermal reaction and the application in highly transparent and conductive networks

Abstract

To increase the aspect ratio of metal nanowires and meanwhile understand its effect on the conducting behavior of transparent and conductive films is important for the optoelectronic application of these nanowires. Here a strategy is disclosed to control nanowire length and hence the aspect ratio by importing HNO₃ into reactions. With a small amount of HNO₃ added, the length of silver nanowires was increased by 50 fold, from several micrometers to hundreds of micrometers, and the average aspect ratio reached 821. Carefully designed experiments show that the improved nanowire length is due to oxidative behavior between HNO₃ and O₂, which screens nuclei in the solution and thus only a small part of them could grow into longer nanowires. Transparent & conductive films were fabricated using these long nanowires. Longer nanowires not only showed better transparency (+10%), reaching 92.1% (transmittance@550 nm) and 12.6 ohm sq⁻¹ (sheet resistance), but also presented better stability against heating, even in open air. An inert gas environment like N₂ is preferred for the heat treatment of silver nanowires. Atom re-arrangement was observed due to Rayleigh instability. Moreover, Monte Carlo based simulation is performed on networks that consisted of randomly arranged silver nanowires. Efficient ratio of nanowire length that contributes to conduction is simulated in cases of varied nanowire length, and the effect of the efficient ratio on electrical conductivity of the network is studied. At similar transparency, longer nanowires could help to form more conductive routes in the networks, which is beneficial to the conductivity of the network.

---

1. Introduction

Silver nanowire (AgNW hereafter) based films have been studied intensively due to their high transparency and film conductance^1,2 as well as sound mechanical flexibility.^3,4 They have been widely used in optoelectronic devices like organic photovoltaic cells (OPVs),^5,6 organic light emitting diodes (OLEDs)⁷ and also the recent hot topic of perovskite solar cells (PSCs).^8,9 Increasing the aspect ratio of such nanowires and hence understanding the correlation between the aspect ratio of nanowires and conductivity of the transparent & conductive films (TCFs) is important for these optoelectronic applications.

After the original work done in early 2000s by Xia and coworkers who introduced the so-called polyol method^10–12 to synthesise silver nanowires and other kinds of nanostructures,^13,14 recently many works were subjected to either decreasing nanowire width or increasing nanowire length. For example, in 2013, using high pressure polyol method, Lee et al. synthesized AgNWs of 15–22 nm in width, ∼20 μm in length and ∼1000 in aspect ratio, and fabricated TCFs with optical transmittance (T%@550 nm) of 88% and sheet resistance (R_sh) of 40 ohm sq⁻¹.¹⁵ In 2015, by tuning reacting procedure and also the concentration of Br⁻ in polyol reaction, Li et al. obtained AgNWs with diameters of 20 nm and aspect ratios up to 2000. And with further selective purification as to exclude nanoparticles, they fabricated TCFs with optoelectronic performance of 99.1% (T%) and 130 ohm sq⁻¹ (R_sh).² Very recently, very thin CuNWs with average width/length of 17.5 nm/17 μm and corresponding TCFs with 90% in T% and 34.8 ohm sq⁻¹ were also obtained by Cui and coworkers. Meanwhile, progresses have also been achieved in increasing nanowire length. In 2009, using nanowires with length of 6.5 μm (width of 85 nm), Coleman and co-workers reported TCFs with T% (@550 nm) of 85% and R_sh of 13 ohm sq⁻¹.¹⁶ Then in 2012, Lee et al. proposed successive multistep growth method, by which they pushed the nanowire length from 10.2 μm to 400–500 μm (width of ∼150 nm) by 7-steps reaction, and optoelectronic performance from 69%, 9 ohm sq⁻¹ to 90%, 19 ohm sq⁻¹ (T% and R_sh).^17,18 Other methods have also been reported to synthesis AgNWs with length up to 100–230 μm (width of 60–90 nm) and obtain optoelectronic performance of 24–109 ohm sq⁻¹ and 94–97% (R_sh and T%).^19,20 All of these methods could improve optoelectronic performance of silver nanowire based TCFs.

However, these methods need careful control over the reaction parameters like stirring speed, reacting processes and even gas background, which increased the cost. As a result, it is still a challenge to explore simplified reaction route to grow AgNWs with controlled aspect ratio. On the other hand, though increasing aspect ratio of nanowires is known to benefit the conductivity of 1D networks, and the topic has been discussed from classical percolation theory,^21–23 the detailed effect of internal structure on conductance of network is yet unclear.

Solvothermal reaction is a well established 1-pot reacting platform for nanostructure growth,^15,24–27 which have also been tried to grow AgNWs though relative short nanowires (20 μm in length) were obtained.^24,28 Here in the article, the method is developed to grow silver nanowires with largely increased length by adding HNO₃, which helps to control the oxidative etching process and hence the nanowire coarsening dynamics in the reaction. These nanowires were used to fabricate TCFs. Effect of nanowire length, annealing temperature and background gas environment on optoelectronic properties of TCFs was examined, and the conducting mechanism was explored with assistance of Monte Carlo based simulation. Topological property determined conducting behavior was revealed.

2. Experimental section

Reagents of ethylene glycol (EG, 99.0%), silver nitrate (AgNO₃, 99.8%), polyvinylpyrrolidone (PVP, average MW ∼ 40 000), potassium chloride (KCl, 99.5%), and nitric acid (HNO₃) were purchased from domestic market and used as received. AgNWs were synthesized using one step solvothermal reaction. Firstly, solution was prepared by dissolving silver nitrate (AgNO₃, 0.024 mol L⁻¹), polyvinylpyrrolidone (PVP, 0.135 mol L⁻¹), and potassium chloride (KCl, 0.02 mmol L⁻¹) in certain volume of ethylene glycol (EG) (72.5–76 mL). Small amount of HNO₃ solution (0.5–4 mL, 0.22 mol L⁻¹ in EG) was also added. The final volume of solution was fixed to 76.5 mL, or 76.5% of the autoclave capacity (100 mL). Secondly, the solution was transferred into an autoclave and heated at temperature of 120–160 °C for period of 5–14 h. At last, the obtained solution was diluted with ethanol and filtered so as to remove excess PVP and other unwanted silver nanoparticles before being stored in ethanol with desired concentration.²⁹ AgNWs TCFs were prepared by spin-coating the stock suspension on glass substrate. Before film deposition, glass substrates were ultrasonically cleaned by acetone, distilled water, ethanol each for 10 min, and then dried in oven. To fabricate films, an optimized volume of AgNWs suspension was dropped on glass substrate and spin-coated at speed of 600 rpm for 3 s and sequent 2000 rpm for 20 s in air, followed by annealing at 200 °C in air for 20 min.

Crystallographic properties of AgNWs were characterized by X-ray diffraction (D8, Bruker), while morphological properties of AgNWs were monitored by scanning electron microscopy (FE-SEM Nova NanoSEM 230) with an acceleration voltage of 10 kV. Length/width distribution was obtained by measuring the nanowires in SEM images using home-made software. UV-Visible extinction spectra were acquired by an ultraviolet-visible spectrophotometer (TU-1800) with wavelength between 350 and 1100 nm. Sheet resistance of the AgNWs films was measured using four-point probe method (SDY-4D).

3. Result and discussion

Pre-addition of small amount of HNO₃ solution in hydrothermal reaction could help to grow very long silver nanowires. As depicted in inset of Fig. 1(a), when there was no HNO₃ pre-added, only silver particles were obtained, after 0.5 mL HNO₃ solution (0.22 mol L⁻¹) was imported, nanowire length was increased from several micrometers to dozens of micrometers. Nanowire length was further increased with volume of HNO₃ solution it reached 8 mL, however, further increase in volume (e.g. 16 mL) harvest only particles in return. The effect of HNO₃ on nanowire length could be described in more details by Fig. 1(c). It is clearly reflected that nanowire length distribution could be tuned by pre-addition of small amount of HNO₃. Before going to discuss the mechanism beneath the observation, we note that, concentration and molecule weight of PVP could also affect the nanowire aspect ratio. By tuning these two parameters, AgNWs longer than 400 μm, average length/width of 138 μm/168 nm (corresponding average aspect ratio of 821) could be obtained, of which the surface morphological properties as well as length distribution were depicted by Fig. 1(d). It is also worthy to note that, for each sample, short nanowires (<25 μm) appeared in all of the depicted samples (volume of HNO₃ from 0.5 to 8 mL), but after filtration by funnel equipped with sand core (model G4, pores in size of 4–10 μm), they could all be removed.

Fig. 1 (a) SEM image of AgNWs synthesized with addition of 0.5 mL HNO₃ solution (0.22 mol L⁻¹) (inset shows samples W/O HNO₃ pre-added). (b) SEM image of AgNWs synthesized with addition of 8 mL HNO₃ solution (inset shows samples synthesized with 16 mL HNO₃ solution, the concentration is same like above). (c) Effect of HNO₃ solution volume on length distribution of AgNWs. (d) SEM image of typical long AgNWs (after filtration), a higher resolution (higher magnification) image of the long wires (inset) and the corresponding length/width distribution histogram (inset).

The increased nanowire length by HNO₃ is caused by the screening effect brought by the oxidative etching behavior between HNO₃ and O₂. To reveal more details about the mechanism, time-controlled experiments were done with or W/O HNO₃ pre-added.

As shown in Fig. 2(a), without HNO₃ pre-added, color of solution changed repeatedly. It was dark red at the first hour, but faded to colorless at 4 h. It then turned back to grey at 5 h, but became clear again at 6 h. Finally, it recovered to dark grey at 7 h, and became white grey after further reaction. While for the case with HNO₃ pre-added, the solution remained colorless for the first 5 hours, but changed gradually to slightly grey at 6 h, and became heavier at 11 h. UV-Visible extinction spectrums of these samples were characterized. Due to localized surface plasma resonance absorption (LSPR), solution color always changes with morphology of nanostructures. As shown in Fig. 2, absorption peaks lying between 350 and 600 nm were due to LSPR absorption of silver nanostructures. More specifically, peaks around 380 nm/500 nm are due to transverse/longitudinal mode of AgNWs respectively,^13,18,30 which could be well reflected by SEM examination as will be shown later. For reactions at 4 h and 6 h (W/O HNO₃ pre-added), and also that from 1 h to 5 h with HNO₃ pre-added, we actually have tried to get some useful product by high rate centrifugation (10k rpm) but failed to get anything. After consideration on the extinction data, we came to suggest that there was no obvious silver nanostructures that had formed in these reactions.

Fig. 2 UV-Visible spectra of solution synthesized from solvothermal reaction at 125 °C from 1 h to 11 h. (a) W/O HNO₃ pre-added (solution of 1 to 4 h were diluted by 2 times each, and 40, 100, 30, 100 times for 5 h, 7 h, 9 h, 11 h respectively); (b) with 0.5 mL HNO₃ (0.22 mol L⁻¹). Solutions of 7 h, 9 h, 11 h were diluted by 10, 30 and 100 times respectively.

Nanostructures in these reactions were monitored carefully by SEM. As shown in Fig. 3(a) and (b), W/O HNO₃ pre-added, both particles and short rodes (marked by yellow arrows) were obtained at the first 1 h, while only nanoparticles were harvested for both 2 h and 3 h reaction (3 h sample was not shown here). The presence of short nanorods tells why extinction peak appeared at round 500 nm. Beyond that, as marked by red arrows, etching appears in samples from 1 h and 2 h reaction W/O HNO₃ pre-added, and also 6 h reaction with HNO₃ pre-added. But for others, for example, sample from 11 h reaction W/O HNO₃ pre-added or 7 h reaction with HNO₃ pre-added, smooth surface was observed, indicating less etching had taken place. By comparison between the first 4 hours reactions (W/O HNO₃ pre-added), one could also find that the originally formed nanostructures had all been “dissolved”, regardless of either particles or rods, and moreover, the rods disappeared more quickly due to the pristine defective nature. A cyclic reaction of “growth–dissolution–growth” could be revealed in the case of W/O HNO₃. Such behavior was similar to that described in literatures,^31–33 where two kinds of oxidants, or Cl⁻/O₂ and oxidative HNO₃, were suggested to dissolve the originally formed silver nanoparticles.

Fig. 3 Typical SEM images of nanostructures from reactions W/O HNO₃ pre-added: (a) 1 h, (b) 2 h, inset of (d) 11 h, and (c) SEM images of that with from 6 h reaction with HNO₃ pre-added (0.5 mL, 0.22 mol L⁻¹).

The cyclic reaction is caused by the reduction–oxidation between Ag/Ag⁺ in the solution. There existed two possible oxidization mechanisms in the reaction. The first one comes from the oxidative nature of HNO₃ as previously discussed in literature.^13,34 With reduction reaction going on, HNO₃ appears as byproduct and accumulates, as described by eqn (1). Due to the oxidative nature of HNO₃, oxidation on originally reduced Ag atoms could easily happen, as described by following eqn (2).^31,34 And since there was certain amount of O₂ in the autoclave, HNO₃ could be regenerated from NO due to eqn (3):

2CH₃CHO + 2AgNO₃ → CH₃COCOCH₃ + 2Ag + 2HNO₃ (1)

3Ag + 4HNO₃ → 3AgNO₃ + NO + 2H₂O (2)

4NO + 3O₂ + 2H₂O → 4HNO₃ (3)

These three equations seem to be reasonable in understanding the cyclic reaction depicted in Fig. 2 and 3. However, considering eqn (1) and (2), only one quarter of Ag⁺ could be reduced to Ag; as eqn (3) is involved, even less Ag⁺ could be reduced. This deviates greatly from the experiment: when there was no HNO₃ pre-added, almost all Ag⁺ could be reduced (Fig. S1†). In order to clarify this question, it is needed to know the exact role played by HNO₃ in these reactions, oxidative, acidic or the both, which was verified by another group of “indicative experiments”. As indicated in Fig. S2 and Table SI,† gas byproduct was judged by litmus solution. It changed from blue to red when meeting with acidic gas. After monitoring the color evolution in four orthogonal reactions, we come to show that, HNO₃ only present oxidative nature in case of high temperature and high concentration. For the reaction at relative low temperature of 125 °C ([AgNO₃] = 0.024 mol L⁻¹), HNO₃ only presented acidic nature. As a result, we come to the second possible oxidative etching mechanism which is from the synergy between O₂ and acidic HNO₃: the originally reduced Ag atoms is oxidized to Ag₂O by O₂ at first, and then dissolved by acidic HNO₃, as described by eqn (4) and (5):

4Ag + O₂ → 2Ag₂O (4)

Ag₂O + 2HNO₃ → 2AgNO₃ + H₂O (5)

Combination between eqn (1), (4) and (5), the etching behavior as well as the cyclic reaction could also be well explained. Moreover, this could help to solve the awkward question met above by the first possible mechanism. Actually, concentration of O₂ in the autoclave is limited, it was 2.7 mmol L⁻¹, which is 11% of [Ag] (0.024 mol L⁻¹) in the system. After consumption of O₂ by few cycles, silver nanostructures could grow freely. When there was few amount of HNO₃ pre-added, the consumption was accelerated, this is why no obvious nanostructure could exist at the first 5 hours (as shown in Fig. 2).

The newly proposed oxidative etching behavior could be simulated in Fig. 4. Based on such mechanism, nucleus could be screened thus only limited part of them could grow into longer nanowires. It should be acknowledged that, HNO₃ showed oxidative nature to some extent. In experiment without addition of PVP, solution color changed little after reaction, and nothing was left at the bottom of centrifuge tube even though very high rate centrifugation (10k rpm/30 min) was used. As a result, less silver ions had been reduced. The appearance of silver nanostructures thus implies that PVP polymer chains have provided protection against oxidative etching. As a result, in the relatively low temperature reaction as described here, oxidative etching behavior was mainly exerted by O₂ and acidic HNO₃. Too high concentration is again not preferred due to slowed reacting rate brought by Le Chatelier's principle. Such effect could be well demonstrated by monitoring the remained Ag⁺ in the reaction. As shown in Fig. S1,† though reacted at same temperature and also same period, Ag⁺ has not been reacted completely in reaction with small amount of HNO₃ pre-added.

Fig. 4 Schematic of confined seeds nucleating process brought by HNO₃, enhanced oxidative etching behavior and the application in growing very long AgNWs.

The screen effect is very useful to synthesis very long AgNWs. Besides the solvothermal reaction, pre-addition of small amount of HNO₃ could also help to achieve nanowires longer than 170 μm in one step polyol method (Fig. S3†). And in the time-controlled experiment at similar reacting period at 11 h, about 265 μm-long AgNWs were achieved with pre-addition of HNO₃, which was 50 times longer than that W/O HNO₃ pre-added, as depicted by Fig. S4.† The HNO₃ assisted solvothermal reaction is different from the high pressure polyol method,¹⁵ which was high pressure reaction (100–150 psi, equaling to 6.8–10.2 atm) created by N₂ along with high temperature heating (170 °C). It is easy to follow, especially for mass production.

The increased nanowire length is beneficial to improvement in optoelectronic performance of TCFs. As shown in Fig. 5(a), for TCFs made of longer nanowires (according to Fig. 1(d), average length of 138 μm) with transmittance decreased from 94.2% to 83.1%, R_sh decreased from 40.6 to 4.5 Ω sq⁻¹. T% & R_sh of 92.1% (@550 nm), 12.6 ohm sq⁻¹, and 89.5%, 9.2 ohm sq⁻¹ were achieved. Such performance is slightly better than that reported in literature, at similar transparency (around 90%), the achieved R_sh was nearly 50% of that reported.¹⁷ Moreover, at similar R_sh, TCFs fabricated from longer nanowires were more transparent (>10%) than that using shorter ones (average length/width of 55.75 μm/237 nm), which is similar to that reported before.¹⁶ The performance is also comparable to the commercial ITO, but AgNWs TCFs showed much wider transmittance window, especially in two areas 350–600 nm and 900–1100 nm, which is appealing for optoelectronic applications.³⁵ It is noted that, besides the improved nanowire length, thermal annealing also contributes to the improved film conductance. As could seen in Fig. S5,† welding was realized between neighboring AgNWs at intersecting area after annealing at 200 °C for 20 min, which is beneficial to the film conductance. However, longer annealing time causes broken nanowires (Fig. S5†), so does higher annealing temperature. In following sections, to better illustrate the effect of higher annealing temperature as well as nanowire length on conductance of AgNWs TCFs, stability-test experiment and simulation were performed.

Fig. 5 (a) UV-Visible spectra and corresponding sheet resistance (R_sh) of TC films made of long AgNWs and ITO (inset shows T%@550 nm and average T% between 300 and 1100 nm). (b) Relationship between T% and R_sh of TC films consisting of long AgNWs (average length/width = 138 μm/168 nm) and shorter ones (average length/width = 55.75 μm/237 nm).

In order to show the stability against heating, TCFs made of nanowires with two kinds of length distribution (short/long ones, SNWs/LNWs) were heated at open air and N₂ respectively. As shown in Fig. 6(a), when treated in open air, nanowires were easily broken into short parts, which were due to oxidization by O₂;³⁶ While in N₂, the nanowires turned to be made of several thicker rods separated by much thinner ones, but less of them were broken, even though they were heated at 360 °C for 0.5 h. The re-arrangement of silver atoms along nanowire length could be ascribed to the so-called Rayleigh instability.³⁷ Nanowire length was collected and shown in Fig. 6(b), after heating, both SNWs/LNWs were cut down to be around 25 μm, which is one quarter (LNWs) or half (SNWs) of the starting length. But for long nanowires, there are more nanowires with length between 50 and 120 μm. The evolvement of morphology has drawn much effect on conductance of TCFs. As shown in Fig. 6(c), for TCFs treated in open air, sheet resistance increases quickly with temperature. For that made of short nanowires and high transparency (S88%), it even turned to be insulate. While for that treated in N₂, film conductance changed slightly (Fig. 6(d)). Meanwhile, transparency also changed slightly, this once again shows that silver atoms have been just re-arranged along the nanowire length according to Rayleigh instability. In both of these two cases, LNWs performed better than SNWs in outputting relatively higher conductance. The stability test also implies that thermal annealing in protected gas environment is preferred for silver nanowire based TCFs.

Fig. 6 (a) SEM images of AgNWs heated in N₂ at 360 °C for 0.5 h and that in air at 300 °C for 0.5 h (inset). (b) Length distribution of silver nanowires obtained before and after annealing (300 °C denotes heating temperature). (c) Effect of annealing temperature on sheet resistance of TCFs made of silver nanowires with different original length distribution and different transmittance. “S/L” represents “short/long” nanowires, while the following data (like 89.9%) marks the transmittance (inset shows width distribution). (d) Sheet resistance of TCFs before and after heating in N₂ (T% is depicted on top of each R_sh).

From above discussion one could find that nanowire length (hence the aspect ratio) plays an important role in the conducting behavior of TCFs. In order to give more details about the effect, simulations were done based on Monte Carlo and Kirchhoff's circuit law.^38,39 For each point, ten times of separated simulation were done, and the average (along with standard error) was used. Seeing from Fig. 7(a)–(c) one can find that, when nanowires are relative short (aspect ratio is small), for example 10 μm in average (aspect ratio of 40 in average), no conductive routes could be formed, thus the film is insulate. Then as it is 15 μm, conductive routes start to form stably, implying that the network is conductive.²³ At this very point, 51.4% of total nanowire length becomes effective in form conduction routes, and the ratio increases further with nanowire length, which is reflected from Fig. 7(d). Besides the efficient ratio, film conductance is also improved by increment of the length, which relates closely to the efficient ratio. As shown in Fig. 7(e), sheet resistance of TCFs decreases quickly with increase of efficient ratio. The relationship between the two could be well understood by Kirchhoff's circuit law. As efficient ratio increases, more conductive routes are formed between the left/right boundaries of network, which cuts down the resistance between these two counter boundaries (as depicted in inset of Fig. 7(e)). It is also worthy noting that, only resistance from nanowire itself has been considered in the calculation, and the corresponding R_sh is 1.74 ohm sq⁻¹, which is much smaller than the experimental value of 12.87 ohm sq⁻¹. The gap between the two is from the contact resistance between neighboring nanowires.^1,3,38,40,41

Fig. 7 Networks generated by Monte Carlo using silver nanowire with average length equaling to (a) 10 μm, (b) 15 μm, (c) 60 μm (routes painted by blue represent efficient part in conduction); average width and area fraction was set at ∼250 nm and 11.5% respectively. (d) Effect of average length on efficient ratio of nanowires and sheet resistance of TCFs (inset shows length distribution). (e) Relationship between efficient ratio and sheet resistance of TCFs (inset shows equivalent circuit).

At last, the relative larger width came along with the ultra-long length was also caused by the oxidative etching processes, because nanowires larger in size were more stable when growing in corrosive background. But nanowire width could be further decreased through replacing Cl⁻ by Br⁻,² employing longer chain PVP polymers,¹⁵ as well as creating less oxidative etching background. For example, by replacing Cl⁻ by Br⁻, we have recently synthesized silver nanowires with width from 30–90 nm, of which the detailed mechanism is under study.

4. Conclusion

Finally, very long AgNWs had been successfully synthesized by controlled oxidative etching process in one step solvothermal reaction. A new kind of oxidative etching behavior was evidenced from the residential O₂ and acidic HNO₃. The disclosed synthesis route is suitable for mass production of very long AgNWs, and could also find interest in growth of other kinds of 1D metal structure. The improved nanowire length plays important role in achieving better conductance of TCFs, which is due to the increased conductive routes in the network.

Acknowledgements

This work was supported by National Natural Science Foundation of China (No. 61306080, No. 11334014), Natural Science Foundation of Hunan Province (No. 2015JJ3143), Scientific and Technological Project of Hunan Provincial Development and Reform Commission, and partially supported by the Project of Innovation-driven Plan in Central South University (2015CXS036) and the Fundamental Research Funds for the Central Universities of Central South University (2016zzts233).

References

1. E. C. Garnett, W. Cai, J. J. Cha, F. Mahmood, S. T. Connor, M. G. Christoforo, Y. Cui, M. D. McGehee and M. L. Brongersma, Nat. Mater., 2012, 11, 241–249.
2. B. Li, S. Ye, I. E. Stewart, S. Alvarez and B. J. Wiley, Nano Lett., 2015, 15, 6722–6726.
3. X. Xia, B. Yang, X. Zhang and C. Zhou, Mater. Res. Express, 2015, 2, 075009.
4. L. Hu, H. S. Kim, J.-Y. Lee, P. Peumans and Y. Cui, ACS Nano, 2010, 4, 2955–2963.
5. H. Lu, D. Zhang, J. Cheng, J. Liu, J. Mao and W. C. H. Choy, Adv. Funct. Mater., 2015, 25, 4211–4218.
6. K. H. Ok, J. Kim, S.-R. Park, Y. Kim, C.-J. Lee, S.-J. Hong, M.-G. Kwak, N. Kim, C. J. Han and J.-W. Kim, Sci. Rep., 2015, 5, 9464.
7. Z. Fang, H. Zhu, Y. Yuan, D. Ha, S. Zhu, C. Preston, Q. Chen, Y. Li, X. Han, S. Lee, G. Chen, T. Li, J. Munday, J. Huang and L. Hu, Nano Lett., 2014, 14, 765–773.
8. X. Dai, Y. Zhang, H. Shen, Q. Luo, X. Zhao, J. Li and H. Lin, ACS Appl. Mater. Interfaces, 2016, 8, 4523–4531.
9. A. Kim, H. Lee, H. C. Kwon, H. S. Jung, N. G. Park, S. Jeong and J. Moon, Nanoscale, 2016, 8, 6308–6316.
10. F. Fievet, J. P. Lagier, B. Blin, B. Beaudoin and M. Figlarz, Solid State Ionics, 1989, 32–33, 198–205.
11. F. Fievet, J. P. Lagier and M. Figlarz, MRS Bull., 1989, 14, 29–34.
12. P. Y. Silvert and K. Tekaia-Elhsissen, Solid State Ionics, 1995, 82, 53–60.
13. Y. Sun, Y. Yin, B. T. Mayers, T. Herricks and Y. Xia, Chem. Mater., 2002, 14, 4736–4745.
14. Y. Sun, B. Gates, B. Mayers and Y. Xia, Nano Lett., 2002, 2, 165–168.
15. E. J. Lee, M. H. Chang, Y. S. Kim and J. Y. Kim, Appl. Phys. Lett., 2013, 1, 042118.
16. S. De, T. M. Higgins, P. E. Lyons, E. M. Doherty, P. N. Nirmalraj, W. J. Blau, J. J. Boland and J. N. Coleman, ACS Nano, 2009, 3, 1767–1774.
17. J. H. Lee, P. Lee, D. Lee, S. S. Lee and S. H. Ko, Cryst. Growth Des., 2012, 12, 5598–5605.
18. J. Lee, P. Lee, H. Lee, D. Lee, S. S. Lee and S. H. Ko, Nanoscale, 2012, 4, 6408–6414.
19. T. Araki, J. T. Jiu, M. Nogi, H. Koga, S. Nagao, T. Sugahara and K. Suganuma, Nano Res., 2014, 7, 236–245.
20. J. Jiu, T. Araki, J. Wang, M. Nogi, T. Sugahara, S. Nagao, H. Koga, K. Suganuma, E. Nakazawa, M. Hara, H. Uchida and K. Shinozaki, J. Mater. Chem. A, 2014, 2, 6326–6330.
21. I. Balberg, N. Binenbaum and C. H. Anderson, Phys. Rev. Lett., 1983, 51, 1605–1608.
22. J. Li and S. L. Zhang, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2009, 80, 040104.
23. J. Li and S. L. Zhang, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2010, 81, 021120.
24. J. Jiu, T. Sugahara, M. Nogi and K. Suganuma, J. Nanopart. Res., 2013, 15, 1588.
25. C. Zhou, J. Ouyang and B. Yang, Mater. Res. Bull., 2013, 48, 4351–4356.
26. C. H. Zhou, X.-Z. Zhao, B. C. Yang, D. Zhang, Z. Y. Li and K. C. Zhou, J. Colloid Interface Sci., 2012, 374, 9–17.
27. C. Zhou, S. Xu, Y. Yang, B. Yang, H. Hu, Z. Quan, S. Bobby, B. Chen, Q. d. Tai, Z. h. Sun and X. Zhao, Electrochim. Acta, 2011, 56, 4308–4314.
28. Z. Yang, H. Qian, H. Chen and J. N. Anker, J. Colloid Interface Sci., 2010, 352, 285–291.
29. T. Kim, A. Canlier, G. H. Kim, J. Choi, M. Park and S. M. Han, ACS Appl. Mater. Interfaces, 2012, 5, 788–794.
30. B. J. Wiley, Y. Chen, J. M. McLellan, Y. Xiong, Z.-Y. Li, D. Ginger and Y. Xia, Nano Lett., 2007, 7, 1032–1036.
31. S. H. Im, Y. T. Lee, B. Wiley and Y. N. Xia, Angew. Chem., Int. Ed., 2005, 44, 2154–2157.
32. B. Wiley, T. Herricks, Y. G. Sun and Y. N. Xia, Nano Lett., 2004, 4, 1733–1739.
33. Y. Zheng, J. Zeng, A. Ruditskiy, M. Liu and Y. Xia, Chem. Mater., 2014, 26, 22–33.
34. M. J. Hu, J. F. Gao, Y. C. Dong, S. L. Yang and R. K. Y. Li, RSC Adv., 2012, 2, 2055–2060.
35. C.-C. Chen, L. Dou, R. Zhu, C.-H. Chung, T.-B. Song, Y. B. Zheng, S. Hawks, G. Li, P. S. Weiss and Y. Yang, ACS Nano, 2012, 6, 7185–7190.
36. J. Jiu, J. Wang, T. Sugahara, S. Nagao, M. Nogi, H. Koga and K. Suganuma, IEEE, International Conference on Nanotechnology, 2015, 1390–1393.
37. J. Y. Huang, Y. C. Lo, J. J. Niu, A. Kushima, X. Qian, L. Zhong, S. X. Mao and J. Li, Nat. Nanotechnol., 2013, 8, 277–281.
38. R. M. Mutiso, M. C. Sherrott, A. R. Rathmell, B. J. Wiley and K. I. Winey, ACS Nano, 2013, 7, 7654–7663.
39. R. M. Mutiso and K. I. Winey, Phys. Rev. E, 2013, 88, 032134.
40. R. Zhu, C.-H. Chung, K. C. Cha, W. Yang, Y. B. Zheng, H. Zhou, T. B. Song, C. C. Chen, P. S. Weiss, G. Li and Y. Yang, ACS Nano, 2011, 5, 9877–9882.
41. T. B. Song, Y. Chen, C. H. Chung, Y. Yang, B. Bob, H.-S. Duan, G. Li, K. N. Tu and Y. Huang, ACS Nano, 2014, 8, 2804–2811.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra20147d

---