Formation Mechanisms of Iron Oxide Nanoparticles in Different

Feb 24, 2012 - Institute for Particle Technology, TU Braunschweig, Volkmaroder ... TU Braunschweig, Mendelssohnstrasse 3, 38106 Braunschweig, Germany...
0 downloads 0 Views 379KB Size
Article pubs.acs.org/crystal

Formation Mechanisms of Iron Oxide Nanoparticles in Different Nonaqueous Media Ilka-Marina Grabs,† Christian Bradtmöller,† Dirk Menzel,‡ and Georg Garnweitner*,† †

Institute for Particle Technology, TU Braunschweig, Volkmaroder Strasse 5, 38104 Braunschweig, Germany Institute of Condensed Matter Physics, TU Braunschweig, Mendelssohnstrasse 3, 38106 Braunschweig, Germany



ABSTRACT: The formation of iron oxide nanoparticles in a solvothermal synthesis with two different nonaqueous solvents, benzyl alcohol (BA) and triethylene glycol (TEG), was studied. Additionally, a scale-up of the synthesis from lab scale autoclaves (45 mL volume) to a 1.5 L reactor was performed. The differences in both reaction vessels as well as for both solvents were investigated regarding the particle size and crystallinity, the magnetic properties, and the stability of the produced particles. A two-step mechanism was identified, with a fast particle formation step and a distinct, slower crystallization step resulting in a strong increase in magnetization. Strong differences in the particle formation between the two reaction media were identified. The stability of the produced particles against agglomeration was screened by a qualitative comparison of sedimentation time in different polar and nonpolar solvents.



INTRODUCTION Iron oxide nanoparticles, mostly magnetite (Fe3O4) and maghemite (Fe2O3), are highly attractive for diverse applications in medical, biological or even ecological fields.1−4 The single-step synthesis of highly crystalline, monodisperse nanoparticles in large quantities thereby is important to tap the full potential of these materials.5 Iron oxide nanoparticles today are mainly produced in aqueous media via coprecipitation of ferrous (Fe(II)) and ferric (Fe(III)) salt solutions.6−10 This process involves high particle formation rates and therefore, particle size and size distribution can hardly be controlled.11 To avoid the limitations of these reactions, different alternative strategies have been developed, such as hydrothermal treatment.12 Several promising nonaqueous methods to metal oxide nanoparticles were also established that lead to more uniform particles than the classical coprecipitation methods.13 In principle, these strategies can be subdivided into hot-injection approaches, where the precursors are injected into a hot reaction mixture, and conventional reaction strategies where a reaction mixture is prepared at room temperature and then heated in a closed or open reaction vessel. For example, the injection of Fe(cup)3 as precursor into hot trioctylamine at 300 °C yielded uniform maghemite particles around 6−7 nm in size.14 Hyeon et al. presented a more complex approach, involving the injection of Fe(CO)5 as precursor into a mixture of octyl ether and oleic acid at 100 °C to form metallic iron nanoparticles. In a next step, these were oxidized by addition of (CH3)3NO to obtain monodisperse maghemite nanocrystals in a size range of 4−16 nm.15 Other precursors that inherently contain the stabilizing oleate groups can be used to control the particle formation and to prevent agglomeration, as has been shown by Park et al.16 The precursor iron(III) acetylacetonate, Fe(acac)3, has been applied in different reaction mixtures, for example, in a mixture of diphenyl ether, 1,2hexadecanediol, oleic acid and oleylamine,17 or 2-pyrrolidone.18 On © 2012 American Chemical Society

the other hand, it has also been shown that the reaction of Fe(acac)3 in pure benzylamine or benzyl alcohol as solvents yielded highly crystalline iron oxide nanoparticles, avoiding the usage of surfactants like oleylamine or oleic acid. In these systems, the reaction medium concurrently also served as surface-binding ligand.19,20 Such solvent-controlled approaches generally are simple and robust, and possess the advantage of leading only to low amounts of organic impurities in the products.21 These systems appear to be highly suitable to study the nanoparticle formation in detail because of their simplicity. However, we have reported earlier that due to organic side reactions occurring during the synthesis, a number of additional side products may be produced in the reaction mixture, influencing the formation mechanisms which cannot be assumed to proceed like a simple decomposition of a precursor in the medium.22 Recently, the formation of mixed magnetite−maghemite nanocrystals by decomposition of iron oleate has been studied by Kwon et al. in more detail.23 By measuring the magnetic moment of the reaction mixture versus time, the nanocrystal formation kinetics could be determined. A mechanism analogous to the formation of nanocrystals in hotinjection methods was identified.23 This implies the intermediate formation and subsequent aggregation of monomers in the crystallization process.24 It is however not clear yet whether these results are generally applicable to the nonaqueous synthesis of nanoparticles in other media, and if solvent-controlled methods can be explained by a similar model. Additionally, for the biomedical application of nanoparticles it is necessary to obtain water dispersible products, because most biological media are nearly neutral aqueous solutions. Received: November 25, 2011 Revised: January 17, 2012 Published: February 24, 2012 1469

dx.doi.org/10.1021/cg201563h | Cryst. Growth Des. 2012, 12, 1469−1475

Crystal Growth & Design

Article

out on a JEOL JEM-2100 instrument at 100 kV. The dispersions in ethanol (diluted to about 0.09 wt % solid content) were placed on a 400-mesh Formvar-coated copper grid (Plano) and left to dry under ambient conditions; no additional contrasting was applied. The magnetic properties of the nanoparticles were determined using a superconducting quantum interference device (SQUID) magnetometer (Quantum Design MPMS-5S). The samples were sealed in a polyvinyl chloride container, and the magnetization was measured at room temperature as function of the applied field. Subsequently, the diamagnetic moment of the container was subtracted from the data.

The approaches presented above are however carried out in hydrophobic media, resulting in nanoparticles mostly being well-dispersible in organic media but incompatible with water. Therefore, Cai et al. presented the use of triethylene glycol (TEG) as reaction medium instead of, for example, benzyl alcohol (BA) for the synthesis of iron oxide nanoparticles from Fe(acac)3.25,26 Triethylene glycol thereby acts not only as a solvent but also as a stabilizer and reducing agent. Maity et al. studied the thermal decomposition of Fe(acac)3 in TEG and predicted a good suitability of the synthesis for scale-up.27 The mechanisms of particle formation in TEG are however not known yet and it remains unclear whether the concepts developed for hydrophobic organic media can be applied also for the hydrophilic TEG systems. Here, we present a comparison of benzyl alcohol and triethylene glycol as solvents for the nonaqueous solvothermal synthesis of iron oxide nanoparticles. A scale-up of the synthesis from lab scale (45 mL) to 1.5 L scale was performed for both systems. To study the formation mechanisms in detail, the concentration decrease of the precursor as well as the heating rates for both scales were measured, followed by kinetic studies of particle growth and crystallization, as well as the magnetic properties. Moreover, the stability of dispersions of the produced particles in different solvents was screened to investigate the differences in particle properties for the two reaction media and their possible a priori prediction based on solvent properties.





RESULTS AND DISCUSSION A great deal of research has been done on iron oxide nanoparticles, possibly making them the best-studied magnetic nanomaterial. The formation process of the magnetic nanoparticles has however not become clear yet for many reaction systems, preventing the rational synthesis of desired sizes and shapes. In the nonaqueous synthesis, the molecular reaction mechanism of the precursor iron acetylacetonate (Fe(acac)3) has been identified in the solvent benzyl alcohol (BA),22 but the actual particle formation process from these molecular building units remained unexplained. Therefore, the objective of this report is to clarify the growth and magnetic behavior of the nanoparticles during the early stages of the reaction for two nonaqueous solvents, benzyl alcohol and triethylene glycol (TEG). In analogy to the investigations of the nonhydrolytic particle formation mechanism proposed and discussed by Kwon et al.23 and following our earlier investigations,22 we assume a general two-step particle formation mechanism: The first step is a solvolysis of the precursor, involving ligand exchange as well as organic reactions, resulting in hydroxyl groups coordinated to the iron center. The following step is a condensation initiated by the hydroxyl groups to form iron− oxygen-iron clusters and ultimately, the iron oxide nanocrystals. It is not clear whether the intermediate species exist in the form of a single-center complex or some hydroxyl groups condensate immediately to substructures of the later iron oxide crystals. Therefore, it is still not clarified which structure can be defined as monomeric species for the crystallization process. To study the formation mechanism in detail, it would be necessary to determine the concentration of each species during the synthesis; it is however not possible to measure the concentration of the intermediate substructures nor to determine their precise chemical composition by conventional spectroscopic methods. On the other hand, the concentration of the precursor Fe(acac)3 can be traced by UV/vis spectroscopy because of its deep red color (the absorption at λ = 430 nm was used for the calculation). To determine the reaction kinetics of the precursor, samples were taken during the first hours of the reaction of Fe(acac)3 in BA and TEG at 200 °C in the 1.5 L reaction system. The samples were centrifuged to remove the produced particles, and the supernatants subsequently analyzed by UV/vis spectroscopy. In Figure 1, the calculated precursor concentrations are shown for both solvents as a function of the reaction time. It can be seen that the decrease of Fe(acac)3 concentration is slightly slower in TEG than in BA, but in both cases approaches zero within the first two hours. Therefore, we conclude that within that time period Fe(acac)3 is fully transformed into the intermediate species as shown in Scheme 1. Because the concentration of the intermediates cannot be measured, it would be useful to determine the solid content and hence the concentration of the condensed product. The solid

EXPERIMENTAL SECTION

Synthesis of Magnetic Nanoparticles. Ethyl acetate, benzyl alcohol (BA, 97%), triethylene glycol (TEG, 97%) and iron(III) acetylacetonate (Fe(acac)3, 97%) (all obtained from Sigma-Aldrich) were used without further purification. In a typical procedure for the autoclave reactions, 2.8 mmol Fe(acac)3 and 20 mL of the respective solvent were placed into a 45 mL Teflon cup inside a steel autoclave (Parr Instruments). The autoclaves were closed and put in a preheated oven at 200 °C for the desired reaction time ranging between a few minutes and 48 h. For the scale-up experiments, a stainless steel 1.5 L double wall reactor (Polyclave type 3/1, Büchi Glas Uster) was used, being heated via an external thermostat (Huber Tango HT). The reactor was filled with a mixture of 50 g Fe(acac)3 and 1 L of solvent. The reactor was equipped with a mechanical propeller mixer being operated at 50 rpm, a carbon slide ring seal to ensure a closed reaction system developing autogenous pressure during the reaction, and a sampling system to enable the withdrawal of samples at different reaction times with pressure compensation. Characterization. The precipitates obtained from the synthesis in BA were washed with ethanol twice and then dried at 75 °C or redispersed in chloroform with addition of oleic acid (Sigma Aldrich) as stabilizer for the measurement of the particle size. The stable dispersion samples prepared with TEG as solvent were used as received or precipitated with ethyl acetate and dried at 75 °C. All samples were measured by dynamic light scattering (DLS) to determine the particle size distribution (Zetasizer NS, Malvern Instruments). The dried samples were measured by powder X-ray diffractometry (XRD, Cu Kα radiation in reflection mode, Si wafer, X’Pert PRO MPD instrument, PANalytical). Supernatants from the reaction mixture for the BA samples and the TEG samples after removal of the particles by precipitation were analyzed by UV/vis spectroscopy (Beckman Coulter) to determine the concentrations of the precursor. First, a calibration curve was determined from solutions at different concentrations of Fe(acac)3 that showed a linear dependence on the Fe(acac)3 concentration, without any influence of the respective solvent. The absorption at 430 nm was used for the calculation of the Fe(acac)3 concentration in the supernatants. Transmission electron microscopy (TEM) measurements were carried 1470

dx.doi.org/10.1021/cg201563h | Cryst. Growth Des. 2012, 12, 1469−1475

Crystal Growth & Design

Article

The solvent is heated much slower inside the autoclaves than in the reactor due to the required heat conduction through the wall of the inner Teflon cup that acts as thermal insulation. The double-wall reactor, on the other hand, is quickly heated by the external thermostat using circulating silicone oil. The reaction temperature of 200 °C is reached within 30 min in the reactor, but only within 5 h in the autoclaves. This has to be considered for all following observations. The synthesis was performed with both solvents, benzyl alcohol and triethylene glycol, to compare the two systems at both scales. For the synthesis at larger scale in the reactor, it is possible to withdraw samples at different times during the reaction, but for the closed autoclaves, separate experiments for each reaction time had to be done. Each sample was measured by dynamic light scattering to investigate the formation and growth of the nanoparticles. The TEG samples could be measured directly in the as-received stable dispersions, whereas the BA samples needed to be dispersed in chloroform with oleic acid as stabilizer to get stable dispersions, in order to be able to determine the primary particle size. The results for the benzyl alcohol-based synthesis in the reactor and the autoclave as a function of reaction time are presented in Figure 3, top.

Figure 1. Concentration of Fe(acac)3 as a function of reaction time in benzyl alcohol (BA) and triethylene glycol (TEG) at 200 °C as determined by UV/vis spectroscopy.

Scheme 1. Proposed Two-Step Reaction Mechanism of the Precursor Fe(acac)3

content however is difficult to be quantified for the benzyl alcohol system because the particles strongly sediment in the medium and therefore, are not uniformly distributed in the reaction volume when taking a sample, which results in a high error. Therefore, a comparison between the BA system and the TEG system is not meaningful in this respect. Instead, the kinetics of particle formation were determined via the particle size as analyzed by DLS. The following investigations were all performed in lab scale autoclaves (45 mL) as well as in a 1.5 L reactor to study the differences between both scales. For a detailed comparison of the results, it is first necessary to measure the temperature profile of the reaction mixture during the first hours (heating-up profile) for both reaction systems. The solvent TEG was used for this purpose; however analogous results can be expected for BA. The cold autoclaves were filled with the solvent, closed and put in a preheated oven at 200 °C. A thermocouple sensor was inserted through the opening in the steel mantle and a little hole drilled in the cover of the Teflon cup for temperature detection. The reactor, on the other hand, was filled with TEG and heated up to 200 °C via the external thermostat, as for a standard reaction. This results in strongly different heating rates for the solvent in both vessels. The measured temperatures as a function of time can be seen in Figure 2.

Figure 3. Particle size as a function of reaction time for the autoclave and the reactor samples for the benzyl alcohol system (top) and triethylene glycol system (bottom) as determined by DLS.

For benzyl alcohol, it can be seen that the final particle size is nearly the same in both reaction vessels (around 14−15 nm). This indicates that the synthesis indeed can be successfully scaled up. For the particles synthesized in the autoclaves, particle growth can be observed in the range of 0−8 h. On the other hand, the particles in the reactor show nearly the same particle size already for the first sample taken only 40 min after

Figure 2. Heating rates for the reactor and the autoclave systems to a set target temperature of 200 °C. 1471

dx.doi.org/10.1021/cg201563h | Cryst. Growth Des. 2012, 12, 1469−1475

Crystal Growth & Design

Article

the reaction mixture reached the target temperature of 200 °C. Here, the particle size remains practically constant throughout the reaction. Therefore, the growth observed for the autoclave samples is attributed to the slower heating rates as discussed before and does not constitute a growth step that would be related to the synthesis. A similar trend is observed for the TEG synthesis in Figure 3, bottom. Because the temperature in the reactor reaches 200 °C within 30 min, again no significant growth step is observed within the first hours. The measured particle sizes for the TEG system are with ∼8 nm nearly the same for both reaction vessels up to a reaction time of about 15 h, apart from also a small growth step in the first few hours for the autoclave samples due to the slower heating rates. However, after about 15 h a steady increase in particle size up to ∼30 nm is measured for the reactor samples. For autoclave samples at 48 h, a particle size of ∼28 nm was measured, in agreement to the size observed for the reactor system. It seems here that although the solvent acts as a stabilizer, aggregation and ripening or even growth processes from potentially remaining intermediate species take place. Ripening effects would lead to a broader primary particle size distribution observable in DLS measurements or TEM images. Aggregation or agglomeration, on the other hand, would lead to an increase of particle size in the DLS measurements with constant primary size in the TEM pictures, while growth processes would result in bigger particles detected by both analyses. Barker et al. observed ripening effects (Ostwald and Smoluchowski-type) in the solvothermal synthesis of magnetite particles with Fe(acac)3 and trioctylamine as solvent at 260 °C.28 In our case, the volume-weighted particle size distribution curves obtained by DLS (as depicted in Figure 4 for selected examples), are significantly shifted but

Figure 5. TEM image of the particles synthesized using TEG in the reactor system after 48 h of reaction.

reported by Barker et al. are attributed to the lower reaction temperature (200 vs 260 °C) or may even be ascribed to the different reaction medium. Aside from the particle size distribution, the magnetic properties of the nanoparticles are also interesting and may serve to investigate their formation process. Therefore, we measured the magnetization as function of the applied field B at room temperature for different reactor samples in BA and TEG withdrawn from the reaction mixture after 0.25 and 4 h of reaction time, assuming no changes in the nature of the samples after quenching to room temperature. The magnetization does not show any remanence nor coercivity and saturates at fields of approximately 1−2 T, indicating superparamagnetic behavior of the particles (Figure 6). The magnetization M can be fitted using the Brillouin function BJ as M(B) = MS × BJ(B) with BJ (B) =

⎛ 2J + 1 ⎛ μ B⎞ μ B⎞ 2J + 1 1 coth⎜ ·gJ · B ⎟ − coth⎜gJ · B ⎟ ⎝ 2 ⎝ 2kT ⎠ 2J kT ⎠ 2J

in which MS is the saturation magnetization, J is the total angular momentum, and gJ the Landé factor. Assuming a magnetic moment of 4 μB per formula unit and a quenched orbital moment, that is, gJ = 2, the average size of the Fe3O4 particles was calculated from the total magnetic moment μ = gJ × μB × J for the different samples (Table 1). Thereby, a spherical particle shape is assumed. With a lattice parameter of a = 8.396 Å of a cubic unit cell containing 8 formula units,29 the particle volume, V, and diameter, d, can then be derived using V = 1/32 × a3 × J and d = 2 × (3/(4 × π) × V)1/3. The shown values are in reasonable agreement with the results obtained by DLS when taking into account that the size of the magnetic core is detected here while DLS yields the hydrodynamic particle size that includes an organic shell around the particles, thus being slightly larger. Furthermore, the development of the saturation magnetization over time was traced for the two reaction systems, with a focus on the early stages of reaction within the first 4 h (Figure 7). The saturation magnetization tends to increase with the reaction time and reaches 53.3 and 44.7 Am2/kg for the BA and TEG samples after 4 h, respectively, which is significantly lower as compared to bulk Fe3O4 (MS,bulk = 92 Am2/kg).30 It is however well-known for magnetite nanoparticles that the saturation magnetization declines with decreasing particle size, showing values of ∼50 Am2/kg for particles of ∼10 nm.31,32 The trend that the BA samples are slightly larger in size than the TEG samples is also reflected by the larger MS for reaction

Figure 4. Volume-weighted particle size distribution curves obtained from DLS measurements of BA and TEG samples after different reaction times in the reactor system.

only slightly broadened during the second stage of the reaction. Additionally, we performed TEM measurements of a TEG sample after 48 h (Figure 5), showing a uniform size and shape of the formed nanoparticles. Here, a primary particle size of ∼8 nm is detected that is in good agreement with the DLS measurements of the initial sample, proving that hardly any particle growth occurs during the reaction. The shift of the particle size distribution can therefore be attributed to aggregation or agglomeration processes taking place during the second phase of the reaction, and ripening or growth can be neglected. The difference between these results and those 1472

dx.doi.org/10.1021/cg201563h | Cryst. Growth Des. 2012, 12, 1469−1475

Crystal Growth & Design

Article

clear step is observed, with a sharp increase of magnetization at about 2 h of reaction to a plateau value of about 50 Am2/kg. For the TEG system, the saturation magnetization increases to a final value of ∼70 Am2/kg for long reaction times of 40 h. To correlate these results with the particle crystallinity, powder X-ray diffractometry (XRD) was performed with samples withdrawn from the two systems at different reaction times (Figure 8). The observed reflections correspond to the

Figure 6. Hysteresis loops for the nanoparticles obtained from the BA system (top) and the TEG system (bottom) from the reactor after 0.25 and 4 h of reaction.

Table 1. Magnetic Moment and Diameter per Fe3O4 Particle for Both Solvents solvent, reaction time (h) BA, 0.25 BA, 4 TEG, 0.25 TEG, 4

magnetic moment μ per Fe3O4 particle (μB) 11255 25584 16104 5103

particle diameter (nm) 7 10 8 6

particle diameter (by DLS) (nm) 10 13 7 8

Figure 8. XRD results of as-prepared samples from the reactor for both solvents at different reaction times, with the reflections of magnetite from the database (JCPDS 19-629) indicated underneath.

magnetite structure (Fe3O4, JCPDS 19-629); it however cannot be excluded that also the maghemite phase (γ-Fe2O3) is present because of the high signal-to-noise ratio in the patterns due to the X-ray fluorescence of Fe, as the patterns differ only by some low-intensity reflections of the maghemite structure.19 The intensities and therefore the crystallinity of the samples are very different for both solvents, especially the changes in the course of the reaction. For the benzyl alcohol system, the particles are somewhat crystalline already in the first hours, but crystallinity does not increase significantly from 0.25 to 1.5 h. Only after 4 h, a pronounced increase in crystallinity can be observed. Due to the strong sedimentation of the product particles in the BA system, the quantity of nanoparticles obtained from the samples is generally rather low. The rather sharp signals marked with asterisks are attributed to carbon (JCPDS 46-943) impurities stemming from the carbon slide ring seal, which showed strong wear for these experiments and thus was replaced by better quality material. For the TEG system, it is obvious that the first sample is amorphous but afterward a continuous increase of the crystallinity can be distinguished. Comparing the crystallinity with the magnetic behavior, it can be seen that there is a sharp step-type increase in both properties for the benzyl alcohol

Figure 7. Saturation magnetization measured for both reactor systems as a function of the reaction time.

times >2 h. Interestingly, whereas the magnetization in the TEG system gradually increases over time, in the BA system a 1473

dx.doi.org/10.1021/cg201563h | Cryst. Growth Des. 2012, 12, 1469−1475

Crystal Growth & Design

Article

of Rohrschneider’s polarity index, the stability of dispersions of the nanoparticles in diverse solvents can be roughly predicted by comparing the polarity of the reaction medium and the desired solvent, but a more detailed elucidation of the surface chemistry of the obtained particles needs to be performed to allow a precise explanation of compatibility and stability.

system and a slower, continuous increase for the triethylene glycol system. The results indicate that the magnetization behavior is determined by the crystallinity of the particles. The crystallinity itself is strongly influenced by the reaction medium and therefore also by the initial particle formation. Additionally, the particle formation and the crystallization need to be seen as two distinct processes, with different dependence on the reaction medium. Notably, this is quite different from the synthesis of ZnO in BA reported recently, where an instantaneous formation of crystalline nanoparticles with a continuous increase in crystal size was observed.33 Apart from the formation mechanism, the stability of dispersions of the resulting nanoparticles against agglomeration was also observed to be very different for both systems. Whereas the nanoparticles from the BA system do not form stable dispersions in hydrophilic solvents, in particular water, the nanoparticles synthesized in TEG are well dispersible in water. To just give a small overview of the stability of dispersions of both nanoparticle types in different solvents, a quick screening is presented (Table 2). Samples were taken from both reaction



CONCLUSIONS We have studied the differences in the solvothermal synthesis of magnetite nanoparticles in benzyl alcohol (BA) and triethylene glycol (TEG). The experiments were performed in lab scale autoclaves (45 mL) and a 1.5 L reactor. Our results regarding particle size, crystallinity and magnetization indicate that the synthesis can be scaled up for both solvents. For the BA system, the nanoparticles show strong agglomeration and sedimentation in the reaction system, causing problems for the withdrawal of samples from the steel reactor vessel. The particle growth was traced by DLS measurements, which could be performed directly for the TEG system but required an additional stabilization step for the BA system due to agglomeration. While for BA, the particles showed a constant size of about 14 nm throughout the reaction, in the TEG system somewhat smaller particles of ∼8 nm in size were observed. In the TEG system, additionally agglomeration processes of the particles were observed after a reaction time of approximately 12 h, whereas the agglomeration in the BA system could not be traced due to sedimentation. Magnetization measurements revealed a superparamagnetic behavior of the particles and allowed to calculate the particle sizes for the different systems, which were in reasonable agreement with the results from DLS. Samples taken at early stages of the synthesis only exhibited a low magnetization. For the BA system, a sharp increase in the magnetization was observed during the first hours, while for TEG a slow rise of the magnetization was detected. These results could be correlated to the crystallinity of the samples, proving that the nanoparticles formed initially possess low crystallinity and that the reaction medium has a strong influence not only on the formation, but also the subsequent crystallization of the particles. Finally, the stability of the particles in different solvents was tested, and also in this case, the results showed a strong dependence on the properties of the used reaction medium. Although the particles did not form stable dispersions in all solvents that are miscible with the reaction medium, the polarity of the respective solvent, as expressed by Rohrschneider’s index of polarity, could be used to generally allow a prediction of the stability of particle dispersions, based on its similarity to the reaction medium.

Table 2. Stability of Dispersions of the Nanoparticles in Different Solvents Judged by Sedimentation Time (- -, 5 min) and Sorted by Rohrschneider’s Polarity Parameter (Showa Denko Europe GmbH) water methanol ethanol chloroform THF isopropanol dichloromethane toluene n-hexane a

polarity index

BA

TEG

10.2 9.2 4.3 4.1 4.0 3.9 3.1 2.4 0.1

- -a -++ -+ ++ + - -a

++ ++ + ++ -+ - -a - -a

Pure solvents not miscible.

systems after 10 h, centrifuged and washed twice. Then the nanoparticles were dispersed in the respective solvent simply by shaking for 5 min. The different solvents were sorted by Rohrschneider′s polarity parameter, a typical parameter used to characterize the polarity of stationary phases in chromatography.34 The strong differences in stability show that the reaction medium used for the synthesis has a great influence on the particle surface, which is in general more compatible with organic solvents when BA is used and more compatible with polar solvents when TEG is used. It is interesting that in several cases, the nanoparticles produced in TEG or BA are not stable in solvents that are miscible with the particular reaction solvent, for example, THF for BA particles or isopropanol for TEG particles. This indicates that not only the solvent but the whole reaction system and therefore also possible side products may have an influence on the particle surface and hence the stability. Although side products such as 4-phenyl-2-butanone as identified for the BA system are expected to be efficiently removed in the washing process due to their weaker binding to the particle surface, the synthesis is known to result in a large number of organic byproducts,22 and the two reaction systems also can be expected to result in a different concentration of hydroxyl groups on the particle surface, rendering the influence of the reaction medium rather complex. On the basis



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: + 49 (531) 3919615. Fax: + 49 (531) 391-9633. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was financially supported by the Deutsche Forschungsgemeinschaft (DFG) within the Collaborative Research Center (SFB) 578, project C7, at TU Braunschweig, as well as project GA 1492/3-1. The authors thank Mrs. Peggy Knospe from the Institute of Particle Technology, TU Clausthal, for the TEM measurements, and Mrs. Ursula Jahn, Institute of Pharmaceutical Technology, TU Braunschweig, for the XRD measurements. 1474

dx.doi.org/10.1021/cg201563h | Cryst. Growth Des. 2012, 12, 1469−1475

Crystal Growth & Design



Article

REFERENCES

(1) Bergemann, C.; Müller-Schulte, D.; Oster, J.; à Brassard, L.; Lübbe, A. S. J. Magn. Magn. Mater. 1999, 194, 45. (2) Nunez, L.; Kaminski, M. D. J. Magn. Magn. Mater. 1999, 194, 102. (3) Berry, C. C.; Curtis, A. S. G. J. Phys. D 2003, 36, R198. (4) Tsang, S. C.; Caps, V.; Paraskevas, I.; Chadwick, D.; Thompsett, D. Angew. Chem., Int. Ed. 2004, 43, 5645. (5) Cademartiri, L.; Ozin, G. A. Phil. Trans. R. Soc. A 2010, 368, 4229. (6) Kang, Y. S.; Risbud, S.; Rabolt, J. F.; Stroeve, P. Chem. Mater. 1996, 8, 2209. (7) Bee, A.; Massart, R.; Neveu, S. J. Magn. Magn. Mater. 1995, 149, 6. (8) Ishikawa, T.; Kataoka, S.; Kandori, K. J. Mater. Sci. 1993, 28, 2693. (9) Kandori, K.; Kawashima, Y.; Ishikawa, T. J. Colloid Interface Sci. 1992, 152, 284. (10) Willis, A. L.; Turro, N. J.; O′Brien, S. Chem. Mater. 2005, 17, 5970. (11) Vatta, L. L.; Sanderson, R. D.; Koch, K. R. J. Magn. Magn. Mater. 2007, 311, 114. (12) Daou, T. J.; Pourroy, G.; Bégin-Colin, S.; Grenèche, J. M.; Ulhaq-Bouillet, C.; Legaré, P.; Bernhardt, P.; Leuvrey, C.; Rogez, G. Chem. Mater. 2006, 18, 4399. (13) Garnweitner, G.; Niederberger, M. J. Mater. Chem. 2008, 18, 1171. (14) Rockenberger, J.; Scher, E. C.; Alivisatos, A. P. J. Am. Chem. Soc. 1999, 121, 11595. (15) Hyeon, T.; Lee, S. S.; Park, J.; Chung, Y.; Na, H. B. J. Am. Chem. Soc. 2001, 123, 12798. (16) Park, J.; An, K.; Hwang, Y.; Park, J.-G.; Noh, H.-J.; Kim, J.-Y.; Park, J.-H.; Hwang, N.-M.; Hyeon, T. Nat. Mater. 2004, 3, 891. (17) Sun, S.; Zeng, H. J. Am. Chem. Soc. 2002, 124, 8204. (18) Li, Z.; Chen, H.; Bao, H.; Gao, M. Chem. Mater. 2004, 16, 1391. (19) Pinna, N.; Grancharov, S.; Beato, P.; Bonville, P.; Antonietti, M.; Niederberger, M. Chem. Mater. 2005, 17, 3044. (20) Pinna, N.; Garnweitner, G.; Antonietti, M.; Niederberger, M. J. Am. Chem. Soc. 2005, 127, 5608. (21) Pinna, N.; Niederberger, M. Angew. Chem., Int. Ed. 2008, 47, 5292. (22) Niederberger, M.; Garnweitner, G. Chem.Eur. J. 2006, 12, 7282. (23) Kwon, S. G.; Piao, Y.; Park, J.; Angappane, S.; Jo, Y.; Hwang, N.-M.; Park, J.-G.; Hyeon, T. J. Am. Chem. Soc. 2007, 129, 12571. (24) de Mello Donegá, C.; Liljeroth, P.; Vanmaekelbergh, D. Small 2005, 1, 1152. (25) Wan, J.; Cai, W.; Meng, X.; Liu, E. Chem. Commun. 2007, 47, 5004. (26) Cai, W.; Wan, J. J. Colloid Interface Sci. 2007, 305, 366. (27) Maity, D.; Kale, S. N.; Kaul-Ghanekar, R.; Xue, J.-M.; Ding, J. J. Magn. Magn. Mater. 2009, 321, 3093. (28) Barker, A. J.; Cage, B.; Russek, S.; Stoldt, C. R. J. Appl. Phys. 2005, 98, 063528. (29) Hu, P.; Zhang, S.; Wang, H.; Pan, D.; Tian, J.; Tang, Z.; Volinsky, A. A. J. Alloys Compd. 2011, 509, 2316. (30) Cullity, B. D. Introduction to Magnetic Materials; Addison Wesley: New York, 1972. (31) Lin, C.-R.; Chu, Y.-M.; Wang, S.-C. Mater. Lett. 2006, 60, 447. (32) Demortière, A.; Panissod, P.; Pichon, B. P.; Pourroy, G.; Guillon, D.; Donnio, B.; Bégin-Colin, S. Nanoscale 2011, 3, 225. (33) Bilecka, I.; Elser, P.; Niederberger, M. ACS Nano 2009, 3, 467. (34) Bertold, A. Anal. Chem. 1995, 67, 849.

1475

dx.doi.org/10.1021/cg201563h | Cryst. Growth Des. 2012, 12, 1469−1475