The chemically directed assembly of nanoparticle clusters from superparamagnetic iron-oxide nanoparticles

Abstract

We describe the controlled synthesis of nanoparticle clusters (NCs) based on the chemically directed assembly of superparamagnetic iron-oxide nanoparticles. The NCs, with their “raspberry-like” shape, have a large effective surface area, uniform size, and contain a large fraction of the magnetic maghemite.

---

Superparamagnetic nanoparticles are being intensively investigated for a variety of biomedical applications, including bioseparations, in vitro cell manipulation, and magnetically targeted drug delivery.^1–5 Usually, these applications require a strong magnetic force (F_M) acting on the individual magnetic nanoparticle in a magnetic-field gradient (∇H) to effectively lead the nanoparticles in the desired direction. The magnetic force F_M should prevail over the thermal energy (k_BT), causing a Brownian motion, and over the hydrodynamic drag in the liquid medium. The magnetic force F_M acting on the nanoparticle is proportional to both its magnetization (M) and volume (V), i.e., F_M = μ₀MV∇H (where μ₀ is the permeability of free space), with the magnetization being mainly defined by the magnetic material. In biomedical applications, it is almost only iron oxide maghemite (γ-Fe₂O₃) that is used as the magnetic material, since it has proven to be non-toxic and was also registered by the FDA.⁶ Particle size (d) has only a minor effect on the magnetization M of nanoparticles, while the force F_M has a much larger impact, since F_M increases with d³.⁷ From the point of view of the magnetic properties, there is a clear advantage if the nanoparticles used in the specific biomedical application are smaller than approximately 15 nm, so as to be in the superparamagnetic state. This is because such superparamagnetic nanoparticles exhibit no remanent magnetization. As a result, there are no attractive magnetic dipole–dipole forces acting between the nanoparticles that will cause them to magnetically agglomerate in a suspension. However, the force F_M acting on an individual superparamagnetic nanoparticle is generally too weak to enable effective magnetic manipulation; this is due to its small volume. However, if a stable suspension of superparamagnetic nanoparticles, i.e., a ferrofluid, is exposed to a magnetic-field gradient the whole suspension is attracted to the magnet, rather than the individual nanoparticles. In contrast, larger, ferrimagnetic particles can be easily manipulated in a magnetic-field gradient; however, they tend to agglomerate due to the magnetic forces and so dispersing them homogeneously in a liquid medium is very difficult. Thus, the only way we can effectively increase the magnetic force F_M acting on a magnetic particle in a stable suspension exposed to a magnetic-field gradient, while maintaining the superparamagnetic state, is to increase the particle volume. Therefore, because of the size limitation intrinsic to superparamagnetism, the only possibility is to create an assembly of a large number of superparamagnetic nanoparticles, which is called a superparamagnetic nanoparticle cluster (NC). In addition to an ability to effectively manipulate these NCs with a magnetic-field gradient, the NCs should also have a large, effective surface area (S_E) for the subsequent conjugation of the various molecules needed for a specific application. Thus, the optimal size for a NC is a compromise between the F_M and the S_E. The minimum size for NCs of superparamagnetic iron-oxide nanoparticles that can be effectively manipulated by a magnetic-field gradient was estimated to be ∼50 nm.⁸

The ideal NCs for applications related to magnetic manipulation should be of uniform size; they should have a high, effective specific surface area; and they should contain a large fraction of the magnetic material. Additionally, it is very desirable for the NCs to be functionalized during their synthesis, i.e., grafted with a layer of functionalization molecules that provide the specific functional groups needed for additional conjugations. Besides the mentioned biomedical applications, superparamagnetic NCs are also being considered as attractive building blocks for the synthesis of advanced nanostructures, such as photonic crystals.⁹

Various strategies have been employed to prepare superparamagnetic NCs.^8–16 The most frequently used being the assembly of superparamagnetic nanoparticles in their aqueous suspensions using amphiphilic copolymers during the synthesis of the nanoparticles. The copolymers are attached to the surfaces of the iron oxide via attractive van der Waals forces or through bidentate bonding between the Fe atoms and carboxylate functionalities on the copolymers, and thus they link a certain number of the nanoparticles in the NCs.^8,10,11 The attached copolymers could be additionally cross-linked to prevent subsequent desorption.¹² The main advantage of using this strategy is that it provides functional groups (i.e., carboxyl and amine groups) on the surfaces during the formation of the NCs and there is no need for any post-synthesis functionalization. However, this strategy suffers from poor control over the size of the formed NCs and, even more importantly, the magnetic fraction is usually low because of the excess of polymers being used. Another commonly used strategy for the formation of NCs is their formation in micelles of different emulsion systems. Usually, hydrophobic nanoparticles in organic solvents are used as an oily phase in O/W emulsion systems. The size of the oily droplets can be precisely tuned to define the size of the synthesized NCs; nevertheless, these NCs need to be additionally functionalized to be applicable in biomedicine.⁹ They may have a high magnetic fraction, but they have a relatively low specific surface area due to the dense packing of the hydrophobic nanoparticles in the core of the NCs.¹³ A similar situation occurs with the strategy based on the incorporation of hydrophobic nanoparticles mixed with non-polar, biodegradable copolymers into magnetic polymeric spheres, which are interesting for magnetic delivery systems.¹⁴ The incorporation of hydrophilic, magnetic nanoparticles into porous silica spheres or the synthesis of magnetic nanoparticles inside the silica spheres during the silica particles' synthesis are more useful strategies for the preparation of larger particles containing small superparamagnetic nanoparticles.^15,16

The objective of the present study was to synthesize functionalized, superparamagnetic NCs with a controlled size, in the range between 30 and 70 nm, containing a large fraction of magnetic material. The NCs were synthesized using a strategy based on the chemically directed assembly of functionalized nanoparticles in aqueous suspensions. The nanoparticles were assembled into the NCs by the covalent bonds between the two types of surface-functional groups situated on the two types of functionalized nanoparticles. The suspensions of the carboxyl- and amine-functionalized nanoparticles were mixed to enable the covalent chemical reactions between the two types of surface-functional groups using carbodiimide chemistry, as schematically presented in Fig. 1a. The nanoparticles with the surface carboxyl-groups (C@NPs) were expected to be in the centre of the cluster and to be surrounded with a number of amine-functionalized nanoparticles (A@NPs).

Fig. 1 Idealized drawing of the preparation of the nanoparticle clusters (a), transmission electron microscopy images of the representative nanoparticle clusters (NCs) at low (b) and high (c) magnification.

First, citrate-stabilized maghemite nanoparticles, 13.7 ± 2.9 nm in size (measured using TEM), were coated with a 2 nm-thick, homogeneous, silica shell, as reported in one of our previous papers.^17,18 The XRD of the precipitated nanoparticles showed a single spinel phase, whereas the chemical analyses showed that there was less than 3% of the iron present in the oxidation state 2+, confirming that the nanoparticles were composed of maghemite (Fig. S1, ESI†).^19,20 The silica-coated nanoparticles were then functionalized, as reported by Kralj et al.^20,21 For the amine functionalization (A@NPs) the 3-(2-aminoethylamino)propylmethyldimethoxy silane was covalently grafted onto the surface of the nanoparticles. For the carboxyl functionalization the surface amine groups of the A@NPs were reacted with succinic anhydride (S2, ESI†).

The changes in the ζ-potential of the functionalized nanoparticles in their aqueous suspensions with the pH were measured using a Brookhaven Instruments Corporation, ZetaPALS (Fig. 2a). The A@NPs show an isoelectric point (IOP) at a pH of approximately 7.5 and the C@NPs at a pH of approximately 4.5. The aqueous suspensions of the functionalized nanoparticles were prepared by intensive washing with a diluted solution of NaOH, while maintaining a pH of 8.5. Thus, after the functionalization process at a high pH above 10, the nanoparticles were all the time maintained at pH values higher than the IOP, thus preventing their agglomeration. At a pH of 8.5 both types of functionalized nanoparticles display a relatively strong negative zeta-potential (approx. −30 mV for the A@NPs and approx. −50 mV for the C@NPs), ensuring their good colloidal stability. Fig. 2b shows the number-weighted distribution of the hydrodynamic diameters of the nanoparticles in the aqueous suspensions at pH values of 8.5, measured using dynamic light scattering (DLS-ALV-6010/160). The average hydrodynamic diameters of the A@NPs and the C@NPs determined with DLS measurements were 21 nm and 32 nm, respectively. The DLS hydrodynamic size is in reasonable agreement with the size determined from the TEM images,^17,18 proving that the suspensions contained well-dispersed nanoparticles.

Fig. 2 ζ-potential curves as a function of pH for the A@NPs, the C@NPs, and the NCs produced in the aqueous suspensions (a) and a normalized number-weighted distribution of the nanoparticles' hydrodynamic size in the suspensions at pH = 8.5, measured by DLS (b).

In order to assemble the two types of nanoparticles into NCs, the carboxyl groups on the surfaces of the C@NPs were activated using carbodiimide chemistry. In brief, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC, 2.5 mg) and N-hydroxysulfosuccinimide (sulfo-NHS, 2.7 mg) were added to a suspension of the C@NPs (3 mL, 12 mg) at pH 6.0 and the mixture was vigorously stirred for 30 minutes to activate the carboxyl groups on the surface of the C@NPs. Then, the suspension of activated C@NPs was washed with distilled water (10 mL, 4 °C) using ultrafiltration (membrane M_w 30 000 Da) and re-dispersed in 3 mL of diluted NaOH to set the pH at 8.5. Then, the suspension of the activated C@NPs was added drop-wise to the suspension of the ultrafiltrated A@NPs (36 mL, 144 mg) with the pH of 8.5. Finally, the reaction of the amide bonds' formation between the activated C@NPs and the A@NPs was carried out during vigorous stirring for 2 hours at room temperature. The number ratio between the A@NPs and the C@NPs in the reaction mixture was 12. The surplus of A@NPs in the reaction mixture ensured that each of the C@NPs was surrounded by as many of the A@NPs as possible. The yield of produced NCs relative to the loaded nanoparticles was ∼60%. It seems that the A@NPs on the “outer layer” of NCs could not pack very closely to each other due to the repulsive electrostatic forces of negatively-charged A@NPs at pH 8.5 and consequently remained as individuals in suspension. The produced NCs were separated from the remaining, non-reacted, individual nanoparticles using an external permanent magnet. The magnetic field at the surface of the magnet was measured with a gaussmeter to be approximately 0.5 T.

Fig. 1b and c show TEM images (Jeol JEM 2100) of the synthesized NCs. The NCs composed of 5–12 nanoparticles had relatively uniform sizes of approximately 30–70 nm (S3, Fig. S2–S5, ESI†). The average hydrodynamic diameter of the NCs was around 80 nm (Fig. 2b), which is in agreement with the average size of the NCs (∼50 nm), estimated from the TEM images (Fig. 1b and c), proving that the NCs are well dispersed in the suspension. Their “raspberry-like” shape, defined by the assembled nanoparticles, is desired for the conjugation of a high surface concentration of different molecules, usually needed for different applications in biomedicine. By considering the size of an individual silica-coated nanoparticle and their spherical shape, their specific surface area was calculated to be 127 m² g⁻¹. The A@NPs and the “raspberry-like” NCs' specific surface areas (N2-BET) were determined to be 69 ± 6 m² g⁻¹ and 58 ± 8 m² g⁻¹, respectively. The NCs' specific surface area is higher compared to the specific surface area of most commercially available superparamagnetic beads.

The concentrations of the nanoparticle suspensions were also found to be important for the effective assembly of the NCs. In the case of the 5-fold-higher concentration of both types of nanoparticles (compared to the concentrations used in the case presented above), the uncontrolled agglomeration took place during the synthesis of the NCs, whereas the 10-fold-diluted reaction mixture resulted in the formation of NCs that was too slow.

The NCs obviously formed due to the chemically directed assembly and not due to electrostatic attractions, since both types of functionalized nanoparticles display a negative surface charge (Fig. 2a) under the conditions of the assembly (pH = 8.5), where repulsive electrostatic forces are expected. Additionally, we showed that the clustering of amine- and carboxyl functionalized nanoparticles without the carbodiimide-activation of carboxyl groups (C@NPs) under the same conditions for the assembly did not result in the formation of NCs. The formed NCs showed a similar zeta-potential curve to that of the A@NPs, in accordance with their presence in the outer layer of the NCs. Thus, the NCs are readily amine-functionalized. The carboxyl-functionalized NCs were simply synthesized by using excess amounts of C@NPs.

The FTIR (FT-IR Nexus, Nicolet Smart) spectrum of the NCs (S4, ESI†) showed the presence of the characteristic band for the amide bond at 1655 cm⁻¹. This information cannot conclusively prove the presence of the formed amide bonds between both types of functionalized nanoparticles since the C@NPs alone contain amide bond in their surface carboxyl-silane molecule. The absence of the characteristic band for the carbonyl group of the –COOH at 1705 cm⁻¹ in the NCs compared to the C@NPs suggested that all surface carboxyl groups of the C@NPs reacted with the amine groups of the A@NPs. However, the direct confirmation of the formed covalent bonds between both types of functionalized nanoparticles by a specific technique is made difficult due to the “steric-inaccessibility” of the formed amide bonds by the NCs' outer layer of the nanoparticles. Moreover, the physical stability of the final NCs with respect to mechanical stress was tested by exposing the NCs to strong ultrasound pulses (SONICS, Vibra cell™ ultrasonificator, 500 W power, 10 seconds pulse, sample volume 20 mL). The NCs retained an unchanged ability to be magnetically retracted from the suspension after the ultrasound treatment, strongly suggesting that they remained structurally intact. Moreover, increasing the ionic strength (5 wt% of NaCl) and subsequent dialysis did not destroy the structure of the NCs. The rigidity of the NCs suggested that they were built by a number of amide bonds between the two types of functionalization molecules situated on the nanoparticles.

Fig. 3a shows room-temperature measurements of the magnetization as a function of the magnetic field for the maghemite nanoparticles, the silica-coated nanoparticles and the synthesized superparamagnetic NCs. All the nanoparticles show zero coercivity, which is in agreement with their superparamagnetic nature. Since the nanoparticles are coated with a silica shell, we would not expect their superparamagnetism to be significantly affected by the magnetic, inter-particle interactions, when assembled into the NCs.²² The magnetization of the 13 nm maghemite nanoparticles measured at 1 T, where it approaches the saturation value M_S, was 67.8 emu g⁻¹. The magnetization of the silica-coated nanoparticles decreased by approximately 30% (47.7 emu g⁻¹) due to the dilution of the magnetic phase by the non-magnetic, 2 nm-thick, silica shell. The saturation magnetization of the NCs was even lower, i.e., 46.4 emu g⁻¹, because of the additional presence of the silane functionalization molecules. To the best of our knowledge, the achieved fraction of magnetic material in the NCs, i.e., 70% by weight, is among the highest reported for superparamagnetic NCs or superparamagnetic beads.²³ The magnetic loading can be further increased by decreasing the thickness of the silica shell.

Fig. 3 Room-temperature measurements of the specific magnetization (M) as a function of magnetic field (H) for the maghemite nanoparticles, the silica-coated nanoparticles, and the NCs (a). The NCs can be collected by a permanent magnet after 1 h, whereas for the A@NPs no visible movement towards the magnet was observed, even after 24 h (b).

The NCs can be effectively captured from the suspension by the magnetic-field gradient, as illustrated in Fig. 3b. The suspensions containing dispersed NCs and A@NPs were placed close to the surface of a permanent magnet (B = 0.5 T). The NCs could be collected on the vial's wall within 1 hour (Fig. 3b), confirming their good responsiveness, while the A@NPs could not be magnetically separated from the stable suspension at all.

Conclusions

In summary, we have shown a versatile method for synthesising superparamagnetic nanoparticle clusters (NCs) containing a large fraction of magnetic maghemite nanoparticles. The method is based on the chemically directed assembly of the functionalized, silica-coated, maghemite nanoparticles in the suspensions. The NCs were formed by covalent bonds between the two types of surface-functionalization molecules on the nanoparticles: amine terminal groups of the amine-functionalized nanoparticles were reacted with activated carboxyl terminal groups of the carboxyl-functionalized nanoparticles. Using the same approach, other types of covalent bonds can also be applied, for example, the use of click chemistry. The NCs have a “raspberry-like” shape, which increases their effective surface area. Their size of approximately 50 nm allows them to be effectively manipulated in the suspensions using an external magnetic-field gradient. They are readily functionalized for further conjugation with the various molecules needed in a certain application. Such NCs could be a better choice for applications involving manipulation with an external magnetic field, such us magnetic-field-assisted drug delivery, bioseparations, or magnetofection, compared to individual nanoparticles. The same synthesis approach can also be used to synthesize (multifunctional) nanocomposite particles by assembling the nanoparticles of different functional materials.

Acknowledgements

The financial support by the Ministry of Higher Education, Science and Technology of the Republic of Slovenia within the Research Program P2-0089 is gratefully acknowledged. The authors also acknowledge the use of equipment in the Center of Excellence on Nanoscience and Nanotechnology – Nanocenter. The authors also thank Dr Alenka Mertelj for the DLS analyses and Dr Nace Zidar from the University of Ljubljana for the ATR-FTIR measurements.

Notes and references

1. Y. Yin and A. P. Alivisatos, Nature, 2005, 437, 664–670.
2. H. Chen, J. Yeh, L. Wang, H. Khurshid, N. Peng, A. Y. Wang and H. Mao, Nano Res., 2010, 3, 852–862.
3. S. R. Deka, A. Quarta, R. Di Corato, A. Riedinger, R. Cingolani and T. Pellegrino, Nanoscale, 2011, 3, 619–629.
4. E. A. Rozhkova, Adv. Mater., 2011, 23, H136–H150.
5. S. Tong, S. Hou, Z. Zheng, J. Zhou and G. Bao, Nano Lett., 2010, 10, 4607–4613.
6. M. Arruebo, R. Fernandez-Pacheco, M. R. Ibarra and J. Santamaria, Nano Today, 2007, 2, 22–32.
7. R. H. Kodama, J. Magn. Magn. Mater., 1999, 200, 359–372.
8. A. Ditsch, P. E. Laibinis, D. I. C. Wang and T. A. Hatton, Langmuir, 2005, 21, 6006–6018.
9. J. P. Ge, Y. X. Hu, M. Biasini, W. P. Beyermann and Y. D. Yin, Angew. Chem., Int. Ed., 2007, 46, 4342–4345.
10. H. Amiri, M. Mahmoudi and A. Lascialfari, Nanoscale, 2011, 3, 1022–1030.
11. J. F. Berret, N. Schonbeck, F. Gazeau, D. El Kharrat, O. Sandre, A. Vacher and M. Airiau, J. Am. Chem. Soc., 2006, 128, 1755–1761.
12. K. Y. Yoon, C. Kotsmar, D. R. Ingram, C. Huh, S. L. Bryant, T. E. Milner and K. P. Johnston, Langmuir, 2011, 27, 10962–10969.
13. P. Qiu, C. Jensen, N. Charity, R. Towner and C. Mao, J. Am. Chem. Soc., 2010, 132, 17724–17732.
14. P. Kocbek, S. Kralj, M. Erdani Kreft and J. Kristl, Eur. J. Pharm. Sci., 2013, 50, 130–138.
15. Y. Deng, Y. Cai, Z. Sun, J. Liu, C. Liu, J. Wei, W. Li, C. Liu, Y. Wang and D. Zhao, J. Am. Chem. Soc., 2010, 132, 8466–8473.
16. M. Feyen, C. Weidenthaler, F. Schüth and A.-H. Lu, J. Am. Chem. Soc., 2010, 132, 6791–6799.
17. S. Čampelj, D. Makovec and M. Drofenik, J. Magn. Magn. Mater., 2009, 321, 1346–1350.
18. S. Kralj, D. Makovec, S. Campelj and M. Drofenik, J. Magn. Magn. Mater., 2010, 322, 1847–1853.
19. D. Makovec, A. Košak, A. Žnidaršič and M. Drofenik, J. Magn. Magn. Mater., 2005, 289, 32–35.
20. S. Kralj, M. Drofenik and D. Makovec, J. Nanopart. Res., 2011, 13, 2829–2841.
21. S. Kralj, M. Rojnik, M. Jagodič, R. Romih, J. Kos and D. Makovec, J. Nanopart. Res., 2012, 14, 1151.
22. D. Makovec, S. Campelj, M. Bele, U. Maver, M. Zorko, M. Drofenik, J. Jamnik and M. Gaberscek, Colloids Surf., A, 2009, 334, 74–79.
23. U. Jeong, X. Teng, Y. Wang, H. Yang and Y. Xia, Adv. Mater., 2007, 19, 33–60.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra00776j

---