Green microfluidic synthesis of monodisperse silver nanoparticles via genetic algorithm optimization

Abstract

A scalable and green procedure for the microfluidic flow synthesis of monodisperse silver nanoparticles is reported. Beetroot extract is used both as a reducing and growth-regulating agent. A multi-objective genetic algorithm was used to automate the optimization of the reaction and reduce sample polydispersity observed in previous reports. The proposed methodology ensures high-quality nanoparticles in a rapidly manner and devoid of human skill or intuition, essential for method standardization and implementation.

---

Metallic nanoparticles (NPs) have been used for the development of numerous technological applications due to their unique properties compared to bulk materials.¹ However, the production of high-quality NPs in a controlled and reproducible manner is hard to achieve and is acknowledged to be the biggest obstacle for exploitation of many nanoscale phenomena. Hence, it is desirable to derive green and scalable synthetic methods to produce metal nanocrystals with controlled morphology and tuned properties.²

Top-down engineering of NPs from bulk materials is often based on the combination of lithography, micromachining methods, and etching.³ This approach can be challenging, costly and although reproducible, hard to scale-up. On the other hand, the chemical growth of NPs on an atom-by-atom basis, namely bottom-up strategy, is a scalable and cost-effective method based on standard synthetic chemistry lab techniques.⁴ Besides the large number of complex chemical strategies for producing near defect-free and monodispersed NPs using bottom-up synthesis, often by bulk methods, their application often requires a combination of skill, intuition, and extensive experimentation.⁵ In this sense, microfluidic flow systems are ideal for NPs production because they allow rapid and controlled thermal and mass transfer by continuous manipulation and processing of sub-microliter volumes of solution.⁶ This method has many advantages over batch reactions because it allows precise and automated control over reaction conditions, and provide both high scalability and ability to combine multiple chemical processes into a single integrated device.⁷ Microfluidic synthesis approach has been applied to accelerate nanoparticle preparation for a multitude of applications, including clinical translation,⁸ biosensing,⁹ magnetic resonance imaging,¹⁰ nanofluids,¹¹ and fuel applications,¹² and to diversify the type of materials produce, leading to the synthesis of for example polymeric nanoparticles,¹³ hybrid nanoparticles with controlled lipid layer,¹⁴ and functionalized metallic nanostructures.¹⁵

In regard to green technology, natural products are seen as benign alternatives to reactive organometallic reductants, such as NaBH₄ and N₂H₄, and growth-limiting (stabilizing) agents.¹⁶ Betalains are versatile antioxidant alkaloids found in plants and fungi¹⁷ that have been used in the development of biotechnological applications.¹⁸ Beetroots contain elevated sugar and betalain content¹⁹ and, thus, red beet juice was exploited for bottom-up synthesis of noble metal NPs leading to broad size and shape distributions.²⁰

Herein, we present a green and scalable method for the preparation of monodispersed silver NPs based on the reaction between betanin-enriched beetroot extract, silver nitrate and sodium hydroxide in an automated microfluidic flow system (Fig. 1), controlled by a Genetic Algorithm (GA). The GA is a method for solving optimization problems based on natural selection. At each cycle, the genetic algorithm selects the best individuals (condition ratios) from the current population (array of different condition ratios) to be parents and uses them to produce the children for the next generation (array of different condition ratios to be tested). Over successive generations (cycles), the population evolves (converge) for an optimal solution. This approach is decoupled from human skill and/or intuition and eliminates the use and generation of hazardous substances.

Fig. 1 Schematic representation of the automatic reactor for the Ag NPs production.

For the synthesis of silver NPS we used betanin (Bn, beetroot extract diluted with dextrin, Sigma-Aldrich), sodium hydroxide (NaOH, 99.5%, VWR), and silver nitrate (AgNO₃, ≥99%, Sigma-Aldrich) were used without any further purification. All solutions were prepared using ultra-pure water (18.2 MΩ cm at 25 °C) from Ångström laboratory, Uppsala University. The samples are labelled based on the basis of volume ratio and reaction temperature. However, in order to keep information about the total flow-rate between different ratios conditions, the volume ratio was normalized in respect to the minimum flow-rate used in the experiments. By using 1 : 1 : 1 or 2 : 2 : 2 means that the total flow-rate in the first case it is 3 and in the second case the total flow-rate is 6. This information it is important since the resident time inside the chip is different, even though the ratio between betatin : NaOH : AgNO₃ still the same.

The microfluidic reactor system was purchased as a ASIA unit from Syris Ltd. It contains two pumps containing valves and pressure sensors and two 50 μL/100 μL syringes, a climate controller (temperature range −50 to 150 °C), a pressure module, two chip headers, and two chips as depicted in Fig. 1 (micromixer and microfluidic reactor). Liquids are transported in PTFE tubing. It should be mentioned that at anytime one can flow up to four different chemical compositions, which can be increased by adding more syringe pumps. This is easily achieved since it is a completely modular reactor, where things can be added or removed at any time.

UV-Vis spectra were acquired online with a setup equipped with a deuterium–tungsten–halogen light source (DH-2000-BAL, Ocean Optics) and a USB4000 spectrometer (Ocean Optics). The solution was flowing continuously in a Z-cell with an optical path of 10 mm. The size a distribution index of the particles was determined online using dynamic light scattering (DLS) in a flow-cuvette 3 mm (Malvern NanoS). The same equipments were used to measure quality of the samples prepared in batch mode. An aliquot of the final solution of each batch was measured using a plastic cuvette (o.p. 10 mm).

Atomic Force Microscopy (AFM) was employed for particle imaging using a Nanosurf easyScan 2 microscope (Nanosurf AG, Switzerland). AFM images were taken with silicon cantilevers with force constant 0.188 N m⁻¹, tip height 14–16 μm, contact mode. Diluted samples were prepared by drop deposition on a mica substrate. Data processing was done with Gwyddion program. Around 700 height measurements on individual particles were performed for particle size analysis.

Silver nanoparticles were synthesized using an automated system consisting of a microfluidic flux reactor (ASIA, Syrrus), online detectors (UV-Vis and DLS), and decision-making controlling program (GA). The microfluidic system has four pumps, a micromixer chip for solution mixing,²¹ and a thermostatted glass microreactor, which were subsequently connected to the analyzer setups flow cell according to Fig. 1. Reaction was monitored online by UV-Vis spectrophotometry and DLS and reaction conditions were analysed and modified automatically to new operation conditions by using a multi-objective genetic algorithm (GA) implemented in software developed in the Matlab environment.²² Briefly, the process starts with the specification of NPs optical properties requirement, absorption maximum wavelength, and the GA population structure to be used in the controlling software. Betanin (3 μmol L⁻¹ – estimated using UV-Vis measurements and a molar attenuation coefficient (ε) of 6.5 × 10⁴ L mol⁻¹ cm⁻¹, channel A′) and AgNO₃ aqueous solutions (1.0 mmol L⁻¹, channel B′) were pumped into the micromixer (chip-1, for details see S1a†) at room temperature. The output of the micromixer is then mixed with a solution of NaOH in water (75 mmol L⁻¹, pH = 12.9) inside the thermostated glass microreactor (chip-2, for details see S1b†) at 353 K (80 °C). All pumps and detectors are controlled by the developed software, employing GA as optimization tool in the decision software.

The resulting silver NPs suspension is monitored online by UV-Vis absorption spectroscopy and DLS to deliver real-time feedback to the optimization tool (Fig. 1). The GA modifies experimental conditions until an appropriate set of flow-rate conditions resulting in NPs with the desired optical properties and size distribution is found. The system is automatic washed with aqueous solutions of NaOH (75 mmol L⁻¹) and HNO₃ (4.0 mol L⁻¹) between runs of different conditions.

Batch synthesis was performed using the same work solutions used in microfluidic synthesis. A solution of betanin (1 mL of 3 μmol L⁻¹) and AgNO₃ (0.5 or 1 mL of 1.0 mmol L⁻¹) was heated up (353 K or 373 K) for 3 min under magnetic stirring. Next, a solution of NaOH (1 mL of 75 mmol L⁻¹) was added to the reaction flask and the resulting suspension was kept under stirring for three more minutes at the same temperature. The suspension was let cool down and analyzed by UV-Vis absorption spectroscopy and DLS.

The batch reaction between betanin (Bn), AgNO₃ and NaOH was used as a control system for the study of the microfluidic flow reaction. UV-Vis and DLS were used to evaluate the quality of the produced particles. The UV-Vis spectra for samples prepared in batch mode have a rather narrow absorption band centered at 404 nm originated from the intense plasmon resonance of small size Ag NPs (Fig. 2A).²³

Fig. 2 Spectroscopic data of Ag NPs prepared in batch mode. (A) Normalized UV-Vis spectra and (B) DLS, particles size analysis from intensity profiles.

Decreasing the NaOH concentration in half did not affect the shape of the spectra; however, no reaction was observed in the absence of base. Fig. 2B shows the DLS of the samples prepared in batch. DLS signal analysis by either intensity or volume (Fig. S2 in the ESI†) show broad range of particles size distribution,^20b,24 i.e. batch reaction produce polydisperse Ag NPs. The broad size distribution was confirmed by AFM imaging (Fig. 3). Some aggregation of particles is observed within the sample, which can be due to both sample preparation and initial Ag NPs inhomogeneity. However, on closer inspection at higher magnification (Fig. 3B) it is evident that the sample is generally polydispersed. Analysis of AFM micrographs by Gwyddion showed particle sizes varying between 10 nm and 180 nm, which agree with the DLS data (Fig. 2B).

Fig. 3 AFM micrographs of Ag NPs prepared in batch mode (1 : 1 : 1@353 K) for (A) low and (B) high magnification.

The size distribution and absorption profile of the Ag NPs was controlled by using an automated microfluidic flow method. A multi-oriented genetic algorithm was used to optimize solution flux and ratio between Bn, AgNO₃ and NaOH to fine tune the size and reduce the polydispersity of the resulting Ag NPs. The second derivative absorption spectra of the resulting Ag NPs are given in Fig. 4. Second derivative spectroscopy makes data analysis simpler by increasing band resolution and eliminating scattering and baseline effects.²⁵

Fig. 4 Spectroscopic data of Ag NPs prepared in the microfluidic reactor at 353 K. (A) Second derivative absorption spectra (inverse, for clarity), and (B) DLS, particles size analysis from intensity profiles.

Comparison of absorption profiles before (1 : 1 : 1 and 1.5 : 2.7 : 2.1) and after reaction optimization in the microfluidic setup (3.4 : 2.5 : 3.3, 3.0 : 2.6 : 3.8 and 3.8 : 2.7 : 3.2) shows that initial conditions result in Ag NPs with broad bands or bimodal absorption whereas optimized conditions lead to narrow absorption bands, which is consistent with an improvement in monodispersion (Fig. 4A). This result is further corroborated by DLS analysis, which found an average size of the optimized Ag NPs between 40–45 nm (Fig. 4B and S3 in the ESI†). The optimization process took roughly a week and it ran more than 160 different condition ratios given by GA. This level of optimization would be difficult to achieve using batch reactions within the same amount of time.

Samples of Ag NPs produced under optimized conditions where collected at the exit of the DLS instrument and submitted to AFM imaging without further purifications. Micrographs in Fig. 5 show that the sample of Ag NPs is homogeneous with some minor aggregation resulting from sample preparation.²⁶ Particles seemed to be quasi-spherical in shape and analysis of the AFM micrographs by Gwyddion of more than 700 counts showed a particle average size of 41 nm (Fig. S4 in the ESI†), which is in agreement with the size found using DLS analysis of the same sample (44 nm).

Fig. 5 AFM micrographs of Ag NPs prepared in the microfluidic reactor from a ratio between Bn : NaOH : AgNO₃ of 3.4 : 2.5 : 3.3@353 K. (A) Low and (B) high magnification.

While full chemometric analysis is required to establish correlations between particle sizes, distribution and homogeneity, analysis of data reveals the following general trends: (i) increase of betanin AgNO₃ ratio increases particle size and (ii) NaOH content is important for the reaction to proceed, but seems to increase size distribution when is equal or higher than betanin and AgNO₃.

The commercial sample of betanin used in this work is a beetroot extract diluted with dextrin. HPLC analysis reveals that betanin and its epimer isobetanin are the main betalains of the sample, in agreement with our previous findings (Fig. S6 in the ESI†).¹¹ Alkaline hydrolysis of betalains yields betalamic acid and the corresponding amino acid, e.g., betanin is converted into betalamic acid and 5-O-glycosylated ciclo-DOPA.²⁷ We were unable to find residual betalamic acid when NaOH was added to betanin in the presence of AgNO₃, suggesting that either betalamic acid is not formed or it is completely consumed to reduce silver cations. The role of betalamic acid in nucleation and growth of metal NPs is under study and will be reported in due time.

In conclusion, monodispersed Ag NPs were synthesized from beetroot extract on a scalable microfluidic setup, paving the way for homogeneous nanomaterials synthesis with green credentials. By combining the microfluidic reactors and an optimization algorithm, it was possible to improve output quality of the NPs produced and standardization of synthesis procedure. This was accomplished in a slither of time and devoid of human skill or intuition.

Acknowledgements

The authors would like to thank Uppsala University, ÅForsk (grant no. 15-455), STINT (grant no. IB2015-6474), São Paulo Research Foundation – FAPESP (grant 2014/14866-2) and Swedish Research Council (grant no. 2015-03764) for the financial support. We would also like to thank Syrris, in particular Paul Oakley and Ulf Olsson for all the help with microfluidic setup.

Notes and references

1. (a) S. Guo and E. Wang, Nano Today, 2011, 6, 240; (b) T. K. Sau and A. L. Rogach, Adv. Mater., 2010, 22, 1781; (c) B. Lim, M. Jiang, P. H. Camargo, E. C. Cho, J. Tao, X. Lu, Y. Zhu and Y. Xia, Science, 2009, 324, 1302; (d) M. Schrinner, M. Ballauff, Y. Talmon, Y. Kauffmann, J. Thun, M. Moller and J. Breu, Science, 2009, 323, 617.
2. C. M. Donegá, S. G. Hickey, S. F. Wuister, D. Vanmaekelbergh and A. Meijerink, J. Phys. Chem. B, 2003, 107, 489–496.
3. D. Xia, D. Li, Z. Ku, Y. Luo and S. R. Brueck, Langmuir, 2007, 23, 5377.
4. J. Park, K. An, Y. Hwang, J. G. Park, H. J. Noh, J. Y. Kim, J. H. Park, N. M. Hwang and T. Hyeon, Nat. Mater., 2004, 3, 891.
5. C. D. Dushkin, S. Saita, K. Yoshie and Y. Yamaguchi, Adv. Colloid Interface Sci., 2000, 88, 37.
6. X. Z. Lin, A. D. Terepka and H. Yang, Nano Lett., 2004, 4, 2227.
7. A. J. DeMello, Nature, 2006, 442, 394.
8. P. M. Valencia, O. C. Farokhzad, R. Karnik and R. Langer, Nat. Nanotechnol., 2012, 7, 623–629.
9. J. Boken, S. K. Soni and D. Kumar, Crit. Rev. Anal. Chem., 2016, 46, 538–561.
10. J. Wang, K. Zhao, X. Shen, W. Zhang, S. Zhang, Y. Song, X. Zhang, R. Rong and X. Wang, J. Mater. Chem. C, 2015, 3, 12418–12429.
11. Y. Zhang, W. Jiang and L. Wang, Microfluid. Nanofluid., 2010, 9, 727–735.
12. F. S. Majedi, M. M. Hasani-Sadrabadi, S. H. Emami, M. Taghipoor, E. Dashtimoghadam, A. Bertsch, H. Moaddel and P. Renaud, Chem. Commun., 2012, 48, 7744–7746.
13. J.-M. Lim, N. Bertrand, P. M. Valencia, M. Rhee, R. Langer, S. Jon, O. C. Farokhzad and R. Karnik, Nanomedicine, 2014, 10, 401–409.
14. L. Zhang, Q. Feng, J. Wang, S. Zhang, B. Ding, Y. Wei, M. Dong, J.-Y. Ryu, T.-Y. Yoon, X. Shi, J. Sun and X. Jiang, ACS Nano, 2015, 9, 9912–9921.
15. L. Uson, V. Sebastian, M. Arruebo and J. Santamaria, Chem. Eng. J., 2016, 285, 286–292.
16. (a) J. Virkutyte and R. S. Varma, Chem. Sci, 2011, 2, 837; (b) M. N. Nadagouda and R. S. Varma, Green Chem., 2008, 10, 859; (c) B. Baruwati and R. S. Varma, ChemSusChem, 2009, 2, 1041.
17. (a) K. M. Herbach, F. C. Stintzing and R. Carle, J. Food Sci., 2006, 71, R41; (b) D. Strack, T. Vogt and W. Schliemann, Phytochemistry, 2003, 62, 247.
18. (a) L. C. P. Gonçalves, S. M. Da Silva, P. C. DeRose, R. A. Ando and E. L. Bastos, PLoS One, 2013, 8, e73701; (b) L. C. P. Gonçalves, R. R. Tonelli, P. Bagnaresi, R. A. Mortara, A. G. Ferreira and E. L. Bastos, PLoS One, 2013, 8, e53873; (c) G. Calogero, A. Bartolotta, G. Di Marco, A. Di Carlo and F. Bonaccorso, Chem. Soc. Rev., 2015, 44, 3244.
19. L. C. P. Gonçalves, M. A. D. Trassi, N. B. Lopes, F. A. Dorr, M. T. dos Santos, W. J. Baader, V. X. Oliveira and E. L. Bastos, Food Chem., 2012, 131, 231.
20. (a) L. Castro, M. L. Blazquez, F. Gonzalez, J. A. Munoz and A. Ballester, Chem. Eng. J., 2010, 164, 92; (b) J. H. Kou and R. S. Varma, RSC Adv., 2012, 2, 10283.
21. B. K. H. Yen, N. E. Stott, K. F. Jensen and M. G. Bawendi, Adv. Mat., 2003, 15, 1858.
22. J. Sá, Chem. Mater. Eng., 2014, 2, 166.
23. E. Tomaszewska, K. Soliwoda, K. Kadziola, B. Tkacz-Szczesna, G. Celichowski, M. Cichomski, W. Szmaja and J. Grobelny, J. Nanomater., 2013, 2013, 60.
24. M. Baalousha and J. R. Lead, Environ. Sci. Technol., 2012, 46, 6134.
25. L. C. P. Gonçalves, B. M. Di Genova, F. A. Dorr, E. Pinto and E. L. Bastos, J. Food Eng., 2013, 118, 49.
26. A. Alahmad, M. Eleoui, A. Falah and I. Alghoraibi, Physical Sciences Research International, 2013, 1, 89.
27. F. H. Bartoloni, L. C. P. Gonçalves, A. C. B. Rodrigues, F. A. Dorr, E. Pinto and E. L. Bastos, Monatsh. Chem., 2013, 144, 567.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental section and recyclability of catalyst. See DOI: 10.1039/c6ra20877k

---