Use of feedback system control in optimizing chemical combinations to synthesize nanoparticles with desired characteristics

Abstract

Nanoparticles (NPs) have found extensive applications in biological diagnosis as well as new energy and chemical devices. One of the major routes to fabricate NPs is chemical synthesis under specifically controlled conditions, when surfactants are often added during or immediately after precipitation. However, interactions between multiple surfactants and reactants could be extremely complicated. Therefore, identifying the optimal formula of multiple chemicals to synthesize NPs with desired characteristics is a daunting task. In this study, we applied an experimental feedback system control (FSC) method and rapidly identified optimal chemical combinations that synthesize silver iodide (AgI) NPs with a target size of 10 nm by only testing less than 0.5% of the total possible chemical combinations. The FSC approach allows rapid optimization of combinatorial chemicals without the requirement of a complete understanding of the underlying chemical interactions. The method introduced here, with slight modification, could be easily applied for other NPs synthesis scenarios.

---

Introduction

Over the past years, NPs have generated great interest due to their specific properties and applications in anti-bacterial, drug delivery and ultrasensitive detection device, etc.^1–6 It is now recognized that the functions of NPs are usually dependent on not only their external physical and chemical conditions but also their characteristics, such as size⁷ and structure.⁸ Therefore, to synthesize NPs with desired characteristics is of vital importance for their proper functions.

In order to prepare NPs with desired properties, the nanoparticle synthesis methods have been advanced extensively in the past few decades. Currently, a large number of strategies have been applied in synthesizing NPs, such as direct chemical reaction synthesis,⁹ microemulsion synthesis,¹⁰ template-assisted synthesis,¹¹ electrochemical synthesis,¹² microwave synthesis,¹³ light-induced synthesis,¹⁴ supercritical synthesis¹⁵ and ultrasonic-assisted synthesis.¹⁶ Among them, direct chemical reaction synthesis is simple, fast, and therefore, the most commonly used approach. However, it is difficult for this approach to simultaneously control the reaction rate and size/shape of the products, which limits the application of the final products. Thus, direct chemical reaction synthesis is not suitable for large scale production of NPs with controlled sizes and shapes. In contrast, soft template synthesis is an effective method that can control the size and structure of the final product through the regulating effect of template and space limitation. Compared to direct chemical reaction synthesis, soft template approach has many advantages, such as precisely control over the size, shape and structure of the products, reacting under normal conditions, address the problem of colloidal instability, simple process suitable for producing in large scale. Soft template syntheses usually involve ordered aggregates formed by amphiphilic molecules, including the micelles, microemulsions, vesicles as well as lytropic liquid crystals (LLC). These templates can precisely control the particle shape, size and orientation through the ordered microscopic structure and the hydrophilic or oleophilic areas, respectively. Soft template method has already been applied to synthesize a large number of NPs at room temperature and under atmospheric pressure, such as Au, Zo₂, PbS, CdS, SiO₂.^17–23 Recently, Thu Huong Tran et al. reported that the sizes of YVO₄:Eu³⁺ NPs could be controlled in the range from 12 nm to 16 nm by using hexadecyl trimethyl ammonium bromide (HTAB), sodium dodecyl sulfate (SDS), dioctyl sodium sulfosuccinate (AOT).²⁴ Li et al. adopted a facile soft template-assisted water-bath method to synthesize pure phase single-crystalline β-cobalt hydroxide with distinct morphologies by using polyvinyl pyrrolidone (PVP), polyethylene glycol (PEG) and cetyltrimethyl ammonium bromide (CTAB) as soft templates.²⁵

A typical process of soft template synthesis often involve adding surface control agents (SCAs) during or after the precipitation aim to interfere with the stages of nucleation and growth to avoid particles agglomeration. However, when multiple chemicals simultaneously present in the reaction system, the potential chemical interactions could be extremely complicated. As a result, identifying the optimal formula to synthesize NPs with desired properties is not a straightforward task. Furthermore, M chemicals with N concentration levels lead to N^M total combinations, make the attempt to obtain the optimal combination be both laborious and costly.

Herein, we introduce a totally different method, FSC approach, which allows to rapidly identify the optimal combination from a large number of possibilities without the requirement of complete understanding the underlying mechanisms. Rather than testing chemical combinations based on rational design, FSC employs an engineering feedback control search algorithm, tests only a few initial chemical combinations and identifies the optimal formula through iterative optimization. The approach was originally developed for the purpose of optimizing drug combinations and has been successfully applied in various cases, including reactivating VSV virus, inhibiting Herpes Simplex Virus type 1 (HSV-1) infection and elongating human embryonic stem cell maintenance.^26–28 To the best of our knowledge, no reports have discussed the potential applications of FSC approach in synthesizing NPs.

As a demonstration of applying FSC in fabricating NPs with desired properties, AgI NPs with controlled size were produced in large scale. The AgI nanoparticle has a wide range of applications in photographic, pharmaceutical and anti-bacterial area due to its magnetic, optical and catalytic performance.^29–35 More remarkably, AgI is a well-known ionic conductor for which the high-temperature α-phase shows a superionic conductivity. That paves the way for its application in solid battery. According to the research of Kyushu University in Japan, the phase transition temperature of AgI is correlated with its particle size.³⁶ The temperature required for high ion-conductive phase transition of AgI decreases as the particle diameter shrinks. When particle size decreases to about 6.3 nm, the ultra-ion-conductive condition is available at 37 °C, which is close to the room temperature.³⁷

Our aim was to identify the optimal combination and concentration of multiple SCAs to synthesize AgI NPs with controlled size of 10 nm. 5 SCAs, including sodium citrate (Na₃C₆H₅O₇), AOT, SDS, cetrimonium bromide (cTAB), and Tween20 were employed in this study. However, the potential chemical interactions between the multiple reagents were not fully understood. With the FSC platform, we identified several effective combinations of the five SCAs, that could generate uniform and dispersive AgI NPs at sizes around 10 nm when reacting with potassium iodide (KI) and silver nitride (AnNO₃) solutions at the optimized ratios, as evidenced by various standard detections, such as dynamic light scattering (DLS), scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The method introduced here does not require any complicated tools, enabling the scale-up for industrialization purposes. The approach presented in this paper, could be potentially applied for synthesizing other NPs with desired properties by slight modification.

Results and discussion

In this study, we introduce a FSC approach, which was originally applied in biopharmaceutical study, to synthesize NPs with controlled properties through soft template method. As an example, synthesis of AgI NPs with precisely controlled size and long-term stability is demonstrated in this study. AgI NPs with controlled size of 10 nm were obtained through the reaction of KI solution and AgNO₃ solution with the presence of a group of SCAs. We firstly decided a pool of SCAs through an intensive literature search, which led to five commercially available chemicals (apart from KI solution and AgNO₃ solution).^38–42 These SCAs were reported to have influences on nanoparticle synthesis.⁴³ In addition, the synthesis of nanoparticles is effected by not only the combinations of the chemicals, but also their concentrations. Thus, four concentration levels that cover a 10–20-fold concentration range are considered for each chemical (both SCAs and reactants) in this study (Table 1). These four concentration levels are coded as “1” to “4”, with “1” representing the lowest dosage (no chemical) and “4” representing the concentrations typically found in literature.

Table 1 Concentration table of the tested chemicals (unit: M)

	Sodium citrate	AOT	SDS	cTAB	Tween20	AgNO3	KI
Level 1	0	0	0	0	0	0.01	0.01
Level 2	0.02	0.0025	0.0025	0.0005	0.0025	0.025	0.025
Level 3	0.04	0.01	0.01	0.002	0.01	0.08	0.08
Level 4	0.06	0.04	0.04	0.008	0.04	0.1	0.1

---

As shown in Fig. 1, FSC method consists of three operations in an iterative loop, generating reagent combinations, testing the particle size as readouts and linking the readouts to the combinations through a feedback search algorithm to generate new reagent combinations for the following testing generation. The key component in this method is an efficient algorithm, which is differential evolution (DE) algorithm programed by MATLAB (short for Matrix Labotary) in this study. Upon the identification of optimal surfactant combination for AgI nanoparticle synthesis, we further evaluated the products with other techniques, such as TEM, SEM and XRD. Besides, the NPs were subjected to long-term stabilization test.

Fig. 1 Illustration of experimental setup.

The success of the FSC is greatly dependents on the readouts that accurately represent the optimization purpose, in our case, being the particle sizes. The DLS was chosen as the initial readout because the measurement is rapid and straightforward. Mean particle size and standard deviation can be easily gained from the DLS measurement, as shown in Fig. S1.† However, the mean particle sizes from DLS are usually larger than the true sizes. Therefore, SEM and TEM were employed to cross-verify the effective chemical combinations that were identified through the DLS measurement. XRD was then used to assess the chemical composition of the final product.

In addition to the input and output, the search algorithm is also a key component in FSC. An efficient algorithm would guide the searching process quickly locate the optimal formula. In this study, DE algorithm, one of the well-known searching algorithms that adopt parallel searching technique, was selected based on the successful demonstrations in identifying the optimal drug combinations for inhibiting Herpes Simplex Virus type 1 infection²⁷ and human embryonic stem cell maintenance.²⁸ Process flow of DE algorithm is explained in Fig. 2. 7 chemicals with different concentration levels are studied. Firstly, the initial N (we chose 16 in this case) combinations, called trial group, are generated arbitrarily. Then, these 16 combinations are modified through mutation and crossover operation to form a test group. Finally, the trial group and test group are compared experimentally and more desired combinations carry on to the following generation to update original trial group.

Fig. 2 Demonstration of the differential evolution search algorithm.

To determine whether AgI particle sizes can be controlled by mixing KI and AgNO₃ solutions with combinatorial surfactants, we randomly selected 16 surfactant combinations by a random number generator in numerical analysis software, MATLAB, and analyzed the particle sizes. To set up the experiment, 0.01 M KI solution was pre-mixed with each surfactant combination in a 50 mL Falcon tube. The mixture was then gently injected to 0.1 M AgNO₃ bulk solution at volume ratio 1 : 1 with continuous oscillation. The control sample was purely mixture of 0.01 M KI and 0.1 M AgNO₃ without any additional surfactant. Fig. 3 shows a diverse range of effects when randomized surfactant combinations were applied, including cases outperforming the control sample, cases close to the control sample, and cases resulting even larger particle sizes compared to the control sample. Overall, the best surfactant combination was 1211414 (the combination is represented by the dosage levels in Table 1), and the worst combination was 1141242, with an average AgI product size of 70.8 nm and 17 153.4 nm, respectively.

Fig. 3 Pilot studies of 16 random surfactant combinations.

We next attempted to decrease the particle size by optimizing the combinations and concentrations of surfactants and reactants through the FSC platform. For the first generation, the same 16 random combinations in the pilot study were chosen. An additional 16 trial combinations were generated through the DE searching algorithm. For the subsequent generations, the 16 winner combinations from the adjacent prior generation and corresponding 16 trial combinations were tested. We kept the same experimental setup as in the pilot study, by pre-mixing chemical combinations with KI solution before reacting with AgNO₃ solution.

During the initial generation, or Generation 1, no combination was able to produce AgI particles with mean size smaller than 60 nm (blue line in Fig. 4(a)). The combinations in the plot are ranked from left to right, based on sizes of final products. Combinations with larger AgI particle sizes were placed to the left of those with smaller sizes. After the first optimization attempt through FSC, combinations in Generation 2 showed obvious size reduction (red line in Fig. 4(a)). Even so, the best performer in Generation 2 was still not desired, with mean particle size larger than 60 nm. The FSC optimization continued for 3 more generations. After the fifth generation (orange line in Fig. 4(a)), by which we cumulatively tested 80 out of 16384 possible combinations, we obtained a condition 1411344 that produced AgI NPs at a mean size of 19.5 nm. Compared to the non-optimal chemical combinations, the FSC optimized combination showed about 100 fold size reduction in AgI particle size (Fig. 4(b) and (c)). Similar to our earlier pilot studies in Fig. 3, a wide range of particle sizes was observed, suggesting a very unique chemical response per combination. Although, DLS is a rapid and simple readout allows FSC platform iteratively search for chemical combinations with reduced particle size. However, the true diameter of AgI particle was supposed to be smaller than the DLS readout due to a hydrated layer was outside the particle. The accurate particle size and structure was then visualized through SEM and TEM.

Fig. 4 Defining desired chemical combinations through FSC.

We cross-verified the AgI nanoparticle sizes with other standard visualization techniques, including TEM and SEM, in order to determine whether the DLS readout could faithfully guide our search for minimized NPs. The TEM images clearly shows that the NPs generated through our optimized formula 1411344 have uniform sizes ranging mainly between 5 nm to 10 nm (as showed in Fig. 5(a) and (b)), while the AgI particles generated by non-optimized chemical combinations have wide particle size distributions with typical mean sizes larger than 60 nm. As showed in Fig. 5(a) and (b), the particles obtained from optimized formula performance well in monodispersity and have a mean diameter of 10.2 ± 2.2 mm (Fig. 5(a)), smaller than DLS measurement results.

Fig. 5 TEM images of AgI NPs generate through optimal formulation.

In order to visualize the AgI particles with SEM, a layer of Pt NPs was deposited on the AgI samples. Consequently, the AgI particle diameter measured from the SEM images are supposed to be 10–20 nm larger than the actual diameters. Considering this fact, the SEM images (Fig. 6) between AgI particles generated through optimized formula 1411344 and non-optimized formula 1211244 also obtained the similar conclusions as that in TEM images. The mean diameter of particles generated by optimal formula is 26.6 ± 5.8 mm from SEM images (Fig. 6(a)).

Fig. 6 SEM images of AgI NPs, (a) and (b) NPs synthesize through optimal formulation (c) and (d) synthesize through non-optimal formulation.

Comparison the results between DLS, TEM and SEM of the same particles obtained through the optimized formula 1411344, we can conclude that the mean diameter get from DLS (19.5 mm) and SEM (26.6 mm) are larger than the TEM (10.2 mm). As mentioned before, the possible reason is that the DLS readout contains the size of hydrated layer outside the particle. As for SEM, a layer of Pt NPs is deposited on the samples results to the size increase.

For the purpose of verifying the chemical structure of the product, we tested the AgI NPs with XRD measurement. From the Fig. 7, we can easily get the conclusion that all the major peaks originating from diffraction of γ-AgI can be discerned clarified. The peaks centering at 23.7° and 39.2° are ascribed to diffractions of (111) and (220) and facets of γ-AgI crystals. Diffraction peaks from any other crystals, like Ag or I₂, cannot be observed, indicating that the investigated sample which obtain under the condition of formula 1411344 is γ-AgI NPs. Above results demonstrate that the FSC approach applied here is successful in fabricating AgI NPs with expected crystalline phase and controlled size in a very efficient manner. When studying the UV-Vis absorbance of the sample got from the same formula, a shoulder appeared at 330 nm (Fig. 8). It was observed a large blue shift of the excitonic peaks from bulk value at 450 nm due to the size quantization effects.

Fig. 7 XRD analysis of the synthesized AgI products.

Fig. 8 Absorption spectrum of AgI NPs.

To confirm whether the AgI particles generated through the optimized formula would suffer any size variation, we examined the particles for long-term size stability. We selected the optimized chemical combination 1411344 and the other two sub-optimal chemical combinations. DLS was selected as the readout in this investigation in order to have a quick evaluation of the size stability. We tracked the non-purified AgI particles generated through these conditions at room temperature for up to 6 days. Size variations are showed in Fig. 9(a). DLS results suggested the optimal formula has the best performance in maintaining the particle size, with mean diameter increase from 19.5 nm to 39.8 nm after 6 days (blue line in Fig. 9(a)). The sub-optimal combinations showed 2–3 times more increase in particle sizes (green and red line in Fig. 9(a)). As we can see from Fig. 9(a), the growth rate of NPs is great in the first 2 days. Hence, we selected three formulas with greater difference in initial size to further explore the nano-size growth pattern (Fig. 9(b)). AgI NPs obtained from the optimal formula and the other two sub-optimal chemical combinations were tracked for 2 days. DLS results suggested the third group is the worst one to control the growth, with an average AgI particle size increase from 45.1 nm to 163.9 nm after 2 days (green line in the Fig. 9(b)) and the optimal formula have the best performance in controlling the size increase (blue line in Fig. 9(b)). Basically, particle size growth rates have slowed down after 24 hours, moreover, the optimized formula has a better control of size increase. Therefore, the optimized formula was not only the optimal combination to generate AgI NPs with the smallest mean diameter, but also effective in maintaining the particles at small size for at least 6 days.

Fig. 9 Investigation of AgI NPs for size variation.

In the context of AgI nanoparticle, sizes of 10 nm or less are usually needed for the benefit of ambient transition temperature. Yet, not all NPs are desired for size of 10 nm. Some materials may have ideal properties at other particular sizes.^44,45 To demonstrate the possibility of controlling AgI particle size with altered chemical combinations, we showed a wide range of AgI sizes which can be accurately controlled with various chemical combinations and concentrations (Fig. 10). NPs sized from 10 nm to 120 nm were prepared by changing the combination and concentration of the chemicals. Each formula was repeated 3 times. The detailed composition of each combination and particle sizes in Fig. 10 is provided in ESI Table S1.†

Fig. 10 AgI NPs with a wide range of sizes can be accurately controlled with different chemical combinations.

Conclusions and discussion

We have demonstrated a scalable and reproducible FSC method to synthesize AgI NPs with desired sizes by using combinatorial chemicals. We have shown that FSC technique could quickly identify the optimal chemical combinations by testing less than 0.5% of the total possibilities. AgI NPs with controlled sizes around 10 nm can be uniformly and massively produced with the optimal formula. The reducing of several orders of magnitude workload, which required to identify the optimal formula, presents a great opportunity to utilize FSC technique for broaden applications in other synthesis scenarios.

The purification could be another factor influencing the size and stability of the final AgI produce. We tested several solutions to purify the products, including DI water, methanol and alcohol. Our results indicated that the AgI particles washed by DI water showed the most stable characteristics. The AgI NPs washed by organic solvent appeared to suffer obvious size increase as a result of particle cluster. One possible reason is the organic solvent might damage the reactant–surfactant reaction equilibrium.

In this study, we demonstrated the ability of a unique set of surfactants to synthesize AgI NPs at sizes around 10 nm when mixed with KI solution and AgNO₃ solution. In addition, this particular formula allows the AgI products maintain their small sizes for at least 6 days. Derivation of new chemical combination containing multiple chemicals will help both scientific research and industry by providing a fast and cost-effective way to generate uniform purified AgI NPs. Future study may unveil the underlying mechanism of the complex chemical reactions.

Experimental section

Materials

Potassium iodide (Cat. # AC418265000) powder, silver nitride, 0.1 N standard solution (Cat. # AC12425-0010) were purchased from Thermo Fisher Scientific Inc. Sodium citrate (Cat. # W302600), dioctyl sodium sulfosuccinate (Cat. # 323586), sodium dodecyl sulfate (Cat. # 436143), cetrimonium bromide (Cat. # H6269), and Tween20 (Cat. # P1379) were purchased from Sigma-Aldrich Co. All of the reagents and solvents were used as received without further purification. Deionized (DI) water was used throughout.

Synthesis of AgI NPs

In a typical synthesis, KI solution was pre-mixed with each surfactant combination, the mixture was stirred at room temperature for 10 min to yield a homogeneous solution. Then, AgNO₃ solution was slowly added to the mixture at volume ratio 1 : 1 with continuous oscillation. After precipitation, the product was collected by centrifugation and washed several times with DI water.

Purification

10 mL reaction solution containing AgI NPs was placed in a 15 mL Falcon tube and centrifuged at 8000 rpm for 10 min. The supernatant was aspirated and replaced with 10 mL DI water. The AgI pellet was broken down by 1 min vortex, followed by 1 min ultrasound and another 3 min vortex, till the liquid became clear again. This purification process was repeated for 6 times. After the last centrifugation and aspiration, the solid AgI pellet was dried either on a filter paper or on a copper net, depending on the following purposes. The final product was harvested and kept in a vacuum light-proof chamber.

Feedback system control

The feedback system control scheme mainly consists of three iterative parts, including generating the testing solutions, verifying the solutions through experiments and a closed-loop control algorithm linking the first two parts. In our study, the testing solutions were combinations of 5 surfactants (sodium citrate, AOT, SDS, cTAB, and Tween20) and 2 reactants (KI and AgNO₃). The experimental readout was the DLS. The control algorithm was DE algorithm. All the programming was realized by the MATLAB software.

Dynamic light scattering characterization

DLS analysis was done on a Coulter Beckman Dynamic Light Scattering Analyzer (N4 Plus). The AgI particles generated in the reaction solution was diluted 10 times with DI water. 1 mL diluted sample was transferred into a transparent cubic tube and sent to the DLS analyzer immediately after preparation.

Scanning electron microscopy visualization

SEM visual characterization of AgI NPs employed a HITACHI SU8010 Scanning Electron Microscopy. A water droplet containing purified AgI particles was placed on a surface membrane filter. After the water evaporated, a conductive layer consist of Pt NPs was deposited on the surface of the sample by an ion sputtering coater. The AgI particles (with conductive layer) were then visualized with the SEM operating at 10 kV.

Transmission electron microscopy visualization

TEM Visual characterization of AgI NPs were examined by a FEI Tecnai F20 Transmission Electron Microscopy. A water droplet containing purified AgI particles was placed on a copper net. The grid was transferred to the TEM for visualization after the water droplet was evaporated.

Conflict of interest

The authors declare no competing financial interest.

Acknowledgements

This research was supported by the project supported by National Natural Science Foundation of China (Grant No. 51375444).

References

1. G. Q. Chen, J. W. Zhang, D. X. Li, C. H. Ren, C. W. Ou, L. Wang and M. S. Chen, RSC Adv., 2014, 4, 61324–61326.
2. A. Gulzar, S. L. Gai, P. P. Yang, C. X. Li, M. B. Ansaric and J. Lin, J. Mater. Chem. B, 2015, 3, 8599–8622.
3. P. Zolotavin and P. G. Sionnest, ACS Nano, 2012, 6, 8094–8104.
4. A. Wijaya, S. B. Schaffer, I. G. Pallares and K. H. Schifferli, ACS Nano, 2009, 3, 80–86.
5. W. S. Lee, V. Sunkara, J. R. Han, Y. S. Parka and Y. K. Cho, Lab Chip, 2015, 15, 478–485.
6. H. J. Jang, Y. K. Kim, H. Huh and D. H. Min, ACS Nano, 2014, 8, 467–475.
7. C. Dhand, M. P. Prabhakaran, R. W. Beuerman, R. Lakshminarayanan, N. Dwivedid and S. Ramakrishna, RSC Adv., 2014, 4, 32673–32689.
8. Q. Yang, F. Lan, Q. Y. Yi, Y. Wu and Z. W. Gu, Nanoscale, 2015, 7, 17617–17622.
9. F. M. Abel, V. Tzitzios and G. C. Hadjipanayis, J. Magn. Magn. Mater., 2016, 400, 286–289.
10. S. Sacanna, G. H. Koenderink and A. P. Philipse, Langmuir, 2004, 20, 8398–8400.
11. C. Xu, Y. Zeng, X. H. Rui, N. Xiao, J. X. Zhu, W. Y. Zhang, J. Chen, W. L. Liu, H. T. Tan, H. H. Hng and Q. Y. Yan, ACS Nano, 2012, 6, 4713–4721.
12. N. Shpaisman, U. Givan and F. Patolsky, ACS Nano, 2010, 4, 1901–1906.
13. M. N. Nadagouda, T. F. Speth and R. S. Varma, Acc. Chem. Res., 2011, 44, 469–478.
14. T. Wang, R. Yang, S. S. Ouyang, H. B. Shi and S. Wang, RSC Adv., 2015, 5, 48992.
15. J. H. Liu and X. F. Wei, RSC Adv., 2015, 5, 7798–7802.
16. Y. Z. Zhou, G. H. Yang, H. B. Pan, C. Z. Zhu, S. F. Fu, Q. R. Shi, D. Du, X. N. Cheng, J. Yang, C. M. Waic and Y. H. Lin, J. Mater. Chem. A, 2015, 3, 8459–8465.
17. X. P. Gao, F. Lu, B. Dong, T. Zhou, W. F. Tiana and L. Q. Zheng, Chem. Commun., 2014, 50, 8783–8786.
18. S. Cheng, D. Yan, J. T. Chen, R. F. Zhuo, J. J. Feng, H. J. Li, H. T. Feng and P. X. Yan, J. Phys. Chem. C, 2009, 113, 13630–13635.
19. D. Kumara, G. Agarwal, B. Tripathi, D. Vyas and V. Kulshrestha, J. Alloys Compd., 2009, 484, 463–466.
20. X. J. Wu and D. S. Xu, Adv. Mater., 2010, 22, 1516–1520.
21. Q. S. Wu, G. X. Zhang and Y. P. Ding, J. Nanopart. Res., 2006, 8, 737–742.
22. S. Ahmad, U. Riaz, A. Kaushik and J. Alam, J. Inorg. Organomet. Polym. Mater., 2009, 19, 355–360.
23. B. Zhao, Y. L. Wang, H. J. Zhang, Z. Jiao, H. B. Wang, G. J. Ding and M. H. Wu, J. Nanosci. Nanotechnol., 2008, 8, 1–4.
24. T. H. Tran, K. A. Tran, T. H. Pham, T. V. Le and Q. M. Le, Adv. Nat. Sci.: Nanosci. Nanotechnol., 2012, 3, 035012.
25. C. Li, X. Zhang, P. Yu, H. T. Zhang, X. Z. Sun and Y. W. Ma, CrystEngComm, 2014, 16, 7478–7485.
26. P. K. Wong, F. Q. Yu, A. Shahangian, G. H. Cheng, R. Sun and C. M. Ho, PNAS, 2008, 105, 5105–5110.
27. X. T. Ding, D. J. Sanchez, A. Shahangian, I. A. Shyoukh, G. H. Cheng and C. M. Ho, Int. J. Nanomed., 2012, 7, 2281–2292.
28. H. Tsutsui, B. Valamehr, A. Hindoyan, R. Qiao, X. T. Ding, S. L. Guo, O. N. Witte, X. Liu, C. M. Ho and H. Wu, Nat. Commun., 2011, 2, 167.
29. J. S. Ghomi and M. A. Ghasemzadeh, J. Chem. Sci., 2013, 125, 1003–1008.
30. Y. Kobayashi, K. Misawa, M. Takeda, M. Kobayashi, M. Satake, Y. Kawazoe, N. Ohuchi, A. Kasuya and M. Konno, Colloids Surf., A, 2004, 251, 197–201.
31. D. A. Reddy, J. H. Choi, S. Lee, R. Ma and T. K. Kim, RSC Adv., 2015, 5, 67394–67404.
32. C. H. Liang, K. Terabe, T. Tsuruoka, M. Osada, T. Hasegawa and M. Aono, Adv. Funct. Mater., 2007, 17, 1466–1472.
33. B. Xue, T. Sun, J. k. Wu, F. Mao and W. Yang, Ultrason. Sonochem., 2015, 22, 1–6.
34. Q. Wang, M. M. Chen, N. X. Zhu, X. D. Shi, H. Jin, Y. Zhang and Y. Q. Cong, J. Colloid Interface Sci., 2015, 448, 407–416.
35. T. Yamamoto, H. Kobayashi and H. Kitagawa, Chem. Lett., 2014, 43, 1355–1356.
36. R. Makiura, T. Yonemura, T. Yamada, M. Yamauchi, R. Ikeda, H. Kitagawa, K. Kato and M. Takata, Nat. Mater., 2009, 8, 476–480.
37. S. Yamasaki, T. Yamada, H. Kobayashi and H. Kitagawa, Chem.–Asian J., 2013, 8, 73–75.
38. W. Z. Zhang, X. L. Qiao, J. G. Chen and H. S. Wang, J. Colloid Interface Sci., 2006, 302, 370–373.
39. S. G. Graña, B. Goris, T. Altantzis, C. F. Loópez, E. C. Argibay, A. G. Martínez, N. A. Barrios, N. Loópez, I. P. Santos, J. P. Juste, S. Bals, G. V. Tendeloo and L. M. L. Marzaán, J. Phys. Chem. Lett., 2013, 4, 2209–2216.
40. J. X. Wang, J. Chen, Y. K. Yu, W. J. Yu, X. R. Meng, J. S. Chen and D. Z. Li, CrystEngComm, 2015, 17, 6661–6668.
41. Z. Q. Yang, H. J. Qian, H. Y. Chen and J. N. Anker, J. Colloid Interface Sci., 2010, 352, 285–291.
42. J. Duy, L. B. Connell, W. Eck, S. D. Collins and R. L. Smith, J. Nanopart. Res., 2010, 12, 2363–2369.
43. R. G. Chaudhuri and S. Paria, J. Colloid Interface Sci., 2010, 343, 439–446.
44. S. D. Sung, D. P. Ojha, J. S. You, J. Lee, J. Kim and W. I. Lee, Nanoscale, 2015, 7, 8898–8906.
45. A. Mondal and N. R. Jana, RSC Adv., 2015, 5, 85196–85201.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra24218e

---