The effect of cationic and anionic blocks on temperature-induced micelle formation†
aiNANO Interdisciplinary Nanoscience Center and Department of Chemistry, Aarhus University, Gustav Wieds Vej 14, Aarhus C, DK-8000, Denmark, bFaculty of Engineering, Østfold University College, N-1757 Halden, Norway, cDepartment of Pharmacy, School of Pharmacy, University of Oslo, PO Box 1068, Blindern, Oslo, N-0316, Norway, and dDepartment of Chemistry, University of Oslo, PO Box 1033, Blindern, Oslo, N-0316, Norway
*Correspondence e-mail: email@example.com
The influence of electrostatic interactions on micelle formation has been investigated in two series of charged temperature-sensitive triblock copolymers using small-angle X-ray scattering [Behrens, Lopez, Kjøniksen, Zhu, Nyström & Pedersen (2011). Langmuir, 28, 1105–1114; Behrens, Kjøniksen, Zhu, Nyström & Pedersen (2011). Macromolecules, 45, 246–255]. The results of these studies are compared to further elucidate the effect of an anionic and a cationic block in the triblock copolymer species. The two series of block copolymers have common water-soluble and temperature-sensitive blocks, methoxypoly(ethylene glycol) and poly(N-isopropylacrylamide), respectively, whereas the charged block differs in the two series: one has a cationic block, poly[(3-acrylamidopropyl) trimethyl ammonium chloride], and the other copolymer has an anionic block, poly(4-styrenesulfonic acid sodium). From the small-angle X-ray scattering study, performed in a temperature range between 293 and 343 K, all copolymers can be described as molecularly dispersed copolymers at ambient temperature, displaying inter-chain repulsion in aqueous solution, and these can be screened by addition of 30 mM NaCl. Both copolymer series display a transition temperature at elevated temperature, where the molecularly dispersed copolymers self-assemble into larger structures, which can be described as either spherical or cylindrical micelles. These structures are affected by the polymer composition, both with respect to the length of the temperature-sensitive block and the specific character of the charged block.
Understanding and controlling polymeric self-organization is of vital interest for the use of such types of systems in practical applications, as this process can be utilized to form specific structures with high versatility. Many commercial products involve self-assembly of polymers and surfactants, and common examples are cosmetics and detergent formulations. The structures formed are, to a large extent, governed by the chemical composition of the polymer. The most common self-assembled structure is a micelle, in which the hydrophobic part of the block copolymer forms the core and the hydrophilic part forms the outer shell in aqueous media. These structures can be formed by having a responsive block in the block copolymer driving micelle formation in response to a change in external parameters. Systems with temperature-responsive blocks have been studied extensively with a focus on use in, for example, drug delivery (Chuang et al., 2009; Muthukumar et al., 1997; Discher & Eisenberg, 2002; Liu et al., 2001).
Another class of copolymers that have been intensively investigated are cationic block copolymers and their interaction with DNA for gene therapy (Wang et al., 2006; Choi et al., 2000; Toncheva et al., 1998). Here the copolymer can have either a hydrophobic or a temperature-sensitive block in addition to the charged block. It is well known that cationic polymers interact strongly with DNA (Holmberg et al., 2002). However, to understand more about the effect of a charged block on copolymer self-assembly, investigation of triblock copolymers with charged and hydrophilic blocks in the micelle corona can be of interest. This will give information on the effects of introducing charges into a hydrophobic/hydrophilic system. In this context, comparison of the effect of an anionic or a cationic block in the copolymer is of interest. This can help to elucidate which effect originates from simply the charge of the polymer block and which from the specific hydrophilic or hydrophobic character of the charged block.
The work presented here originates from two individual studies where the solution behavior has been investigated for an anionic and a cationic temperature-sensitive triblock copolymer series (Behrens, Kjøniksen et al., 2011; Behrens, Lopez et al., 2011), respectively. Both studies have been performed using small-angle X-ray scattering (SAXS). The anionic and cationic polymers only differ in structure with respect to the charged block. Therefore, a comparative study is presented to investigate the effects of the two oppositely charged blocks on micelle formation with varying thermoresponsive block size.
The polymers are composed of a methoxypoly(ethylene glycol) (MPEG) block of 45 repeating units located at one end of the copolymer and a middle block of poly(N-isopropylacrylamide) (PNIPAAM) with varying block length (given in Table 1). The charged block is located at the opposite end in relation to the MPEG block and it differs between the two series of copolymers. One type of copolymer has a poly(styrenesulfonic acid sodium) [PSSS(−)] block and the other copolymer series a poly[(3-acrylamidopropyl)trimethyl ammonium chloride] [PN(+)] block. The two copolymers investigated are designated MPEG-PNIPAAM-PSSS(−) (anionic) and MPEG-PNIPAAM-PN(+) (cationic), respectively (Fig. 1).
The individual blocks making up the thermoresponsive triblock copolymers have specific characteristics governing their solution behavior. Polyethyleneglycol (PEG) is hydrophilic until its transition temperature is reached, which is above 373 K in water (Sommer et al., 2004; Holmberg et al., 2002). The solvent quality is easily varied by changing the temperature, owing to changes in the polymer configuration and breaking of hydrogen bonds between water molecules and the oxygen in PEG. These hydrogen bonds can also be disrupted by the introduction of ions into the solution (Ataman, 1987; Ataman & Boucher, 1982). MPEG is expected to behave in a similar way to PEG in aqueous solution, as the methoxy group is small compared to the overall size of the PEG block. Concerning the PNIPAAM block, this exhibits a coil–globule transition in water at approximately 305 K (Stieger et al., 2004; Schild, 1992). This does, however, depend on both the molecular weight and composition of the block copolymer that it is part of, as well as the concentration of the polymer (Pamies et al., 2009). The thermoresponsive behavior of PNIPAAM can induce micelle formation with a reversible transition for the current block copolymers.
Considering the charged blocks, both the PSSS(−) and the PN(+) block are soluble in salt-free aqueous solution; however, addition of salt will significantly affect characteristics of the blocks. The PSSS(−) block will become more hydrophobic when counter-ions are associated, owing to the aromatic character of the block. The PN(+) block is also expected to decrease its solubility in water upon addition of salt. However, whether the block becomes completely hydrophobic upon counter-ion association is unclear. Furthermore, it is likely that the two blocks will interact differently with water as cationic charges normally form strong bonds with water and hence effectively disrupt the hydrogen bonds in water, whereas the anionic charges normally form weaker bonds with water and the hydrogen bonds in water are not significantly affected (Grossfield, 2005; Nucci & Vanderkooi, 2008; Zhang & Cremer, 2006; Mountain & Thirumalai, 2004).
In this article, the focus will mainly be on the effect of electrostatics on micelle formation observed in the anionic and cationic triblock copolymers. Specific details on data modeling of the two systems will not be presented in the present paper as these can be found in the original papers concerning each system individually (Behrens, Kjøniksen et al., 2011; Behrens, Lopez et al., 2011).
The polymers were synthesized by atom transfer radical polymerization. For further information on the MPEG-PNIPAAM-PSSS(−) polymers, see Kjøniksen et al. (2011), and for the MPEG-PNIPAAM-PN(+) polymers, see Behrens, Lopez et al. (2011). The polymers were prepared in aqueous solution without salt and with 30 mM NaCl and the desired weight fractions were obtained by dissolving appropriate amounts of the copolymers in the two solvents.
All small-angle X-ray scattering experiments were performed on a modified NanoSTAR camera from Bruker AXS (Pedersen, 2004). The experiments were performed on samples of 1 wt% and with a temperature variation from 293 to 343 K in steps of 5 K. For further information on the individual studies see Behrens, Kjøniksen et al. (2011) for the MPEG-PNIPAAM-PSSS(−) polymers and Behrens, Lopez et al. (2011) for the MPEG-PNIPAAM-PN(+) polymers.
The scattering data for both polymer series are temperature dependent. Examples of the scattering data are shown in Fig. 1 for the copolymers with the medium size PNIPAAM block. For the molecularly dispersed copolymers in aqueous solution, a correlation peak is observed, and this feature is ascribed to inter-chain interactions resulting from the charged character of the polymers. At elevated temperatures, the scattering intensity at low q increases. This originates from the self-assembly of the polymers mediated by the decreased solvent quality of water mainly towards the PNIPAAM block. This increase in scattering intensity masks the chain–chain correlation peak. Furthermore, addition of 30 mM NaCl is sufficient to screen most of the electrostatic interactions between the molecularly dispersed copolymers, as the correlation peak is no longer clearly seen at this salinity (data not shown).
The critical micelle temperature (CMT) can be estimated from the scattering data (Table 1). Neither of the copolymers with the short PNIPAAM block self-assembles in the temperature range investigated. This is believed to be a consequence of the size of the PNIPAAM block in relation to the size of the hydrophilic and charged blocks. These are in good solvent conditions in the entire temperature range investigated. The attraction between the PNIPAAM blocks in the individual copolymer is not strong enough to overcome the repulsion originating from the charged blocks and the solvation of the MPEG chains. The copolymers with the medium and the large size PNIPAAM blocks self-assemble above their respective CMTs. In the anionic triblock copolymer series, the polymers with the medium size PNIPAAM block self-assemble at higher temperatures than those with the large PNIPAAM block. In contrast, in the cationic triblock copolymer series variation in CMT is not observed with increasing PNIPAAM block length. This is ascribed to the different structure of the charged blocks. It is believed that the anionic block [PSSS(−)] forms weaker bonds with water than the cationic block [PN(+)], a phenomenon that has been discussed previously (Grossfield, 2005). Moreover, the charged block in the cationic copolymer can contract in response to changes in temperature, whereas the anionic block is less likely to contract because of the steric hindrance originating from the stiff character of the phenyl ring.
Therefore, one could speculate that the CMT of the cationic block is affected by the tendency of the charged block to contract and it is also likely that it interacts more strongly with water. By balancing these two effects, a CMT independence of PNIPAAM block length may be obtained. On the other hand, the anionic block cannot contract and therefore the CMT is governed by the length of the PNIPAAM block, as observed for the homopolymer of PNIPAAM (Pamies et al., 2009). Only a slight effect of the added salt is seen on the CMT in both polymer series, suggesting that the micelle formation is predominately governed by the length of the PNIPAAM block.
The scattering originating from the molecularly dispersed copolymers is described by a Gaussian chain model including an effective hard-sphere structure factor. In this model the center of the interaction potential is placed at one end of the polymer to account for the repulsion between the polyelectrolyte part of the copolymers. For further details on the modeling of the scattering data the reader is referred to Behrens, Kjøniksen et al. (2011) and Behrens, Lopez et al. (2011). One of the parameters obtained from the modeling is the radius of gyration, Rg, of the polymer chain. From Fig. 2(a), it is clear that this depends on the size of the copolymer species, as expected, because the various blocks are all in relatively good solvent conditions. A difference between the two copolymer series is seen, where the MPEG-PNIPAAM-PSSS(−) copolymers have gyration radii larger than those of the MPEG-PNIPAAM-PN(+) polymers. This is ascribed to the bulky character of the PSSS(−) block, as a consequence of the phenyl ring, which makes the PSSS(−) block less flexible. Hence, it is believed that the PSSS(−) block is more stretched than the PN(+) block.
Turning to the copolymers in the presence of salt (Fig. 2b), it is clear that the cationic copolymer does not change Rg upon addition of salt, whereas Rg of the anionic polymer increases slightly as salt is added. The behavior of the anionic copolymer could arise from the association of counter-ions with the charged block, which cannot contract because of the bulky character of the phenyl ring when the charges are screened. Conversely, the cationic block is less bulky and can contract when the charges are screened, because of its greater flexibility. However, the radius of gyration for the cationic copolymer does not decrease, which is likely to be a consequence of the additional volume originating from the associated ions.
Another characteristic parameter obtained is the effective hard-sphere radius, which can be associated with the range of the electrostatic interactions or with the inter-polymer distance if this is less than the range of the interactions. The effective hard-sphere radius increases with polymer size, as expected, when the range of the electrostatic interaction is larger than the inter-polymer distance, and it is constant up to the transition temperature (Fig. 2c). This indicates that the polymers do not contract until they reach the CMT and that this is a relatively sharp transition. Comparing the two polymer series, one sees that the MPEG-PNIPAAM-PSSS(−) polymers have larger effective hard-sphere radii. This is associated with the difference in number density of the two copolymer series at a concentration of 1 wt%. The molecular mass of the MPEG-PNIPAAM-PSSS(−) copolymer is higher than that of the MPEG-PNIPAAM-PN(+) copolymer. This results in fewer polymer chains in solution for the anionic copolymer than for the cationic copolymer at the same concentration, thus giving a lower number density, which in turn allows the inter-polymer distance to be larger.
Above the CMT, the polymers self-assemble to form micelle structures of either spherical or cylindrical geometry. The micelle morphology is dependent on the PNIPAAM block size. The polymers with the medium size PNIPAAM block form spherical micelles and those with the large block form cylindrical micelles. This is due to the increase in the size of the PNIPAAM block, which changes the volume ratio between the temperature-sensitive and hydrophilic blocks. Polymers composed of PEG and PNIPAAM have previously been shown to form spherical micelles in aqueous solution (Topp et al., 1997).
Different behaviors are observed for the different copolymer series, without and with salt. Focusing on the spherical micelles, growth occurs as a function of temperature in both the salt-free and salt-containing solution for the anionic copolymer. In contrast, the spherical micelles formed by the cationic copolymer shrink as the temperature is increased (Fig. 3). It should be noted here that the outer radius, Rout, of the micelle shown in Fig. 3, is the distance where the scattering length density of the micelle is equal to that of the solvent. The growth observed for the anionic copolymer at elevated temperatures is associated with decreased solvent quality for MPEG and PSSS(−). The decrease in micelle radius observed for the cationic polymer is also a consequence of the decreased solvent quality. The difference between the anionic and cationic copolymers is thought to arise from the difference in the charged block, as the PSSS(−) block is bulky, because of the presence of phenyl rings, and therefore has limited freedom to reorganize. In contrast, the PN(+) block does not have steric restrictions that counteract the decrease in solvent quality towards the hydrophilic blocks with increased temperature, and hence can pack closer as a result of charge screening upon addition of salt. Furthermore, one can speculate that the solvent quality for the cationic block does not decrease as much as for the anionic block owing to the character of the PSSS(−) block with its phenyl ring. This may influence the anionic polymer, thereby favoring interaction with itself over water and thus inducing growth. Finally, if the charged block and the MPEG block are both present in the corona of the micelle, the interaction between these may also play a role in the different behaviors of the anionic or cationic block copolymer micelles. It is assumed that only the MPEG and charged blocks are located in the corona of the micelles, as the experiments performed in these studies did not allow for explicit determination of the positions of the individual blocks, because of the similar scattering length densities of the individual blocks. It is expected that only weak or no interactions at all will be found between the cationic block and MPEG, as there are not expected to be any electrostatic interactions. Conversely, the anionic block is expected to interact with the MPEG block, as observed for the interaction between anionic surfactants and PEG (Schwuger, 1972), where a partial transfer of positive charge from the head group of the anionic surfactant to the oxygen of MPEG can take place and thus enhance the attraction between the MPEG block and the anionic surfactant. These two factors, coupled with the fact that the polymer cannot contract completely owing to steric hindrance, will induce micelle growth and not shrinkage.
One sees that for the cylindrical micelles the cross-section radius of the micelles formed by the anionic polymer is slightly larger in the aqueous salt solution compared to the salt-free aqueous solution at 323 K. However, the micelles in water grow with temperature, whereas the size of those in the salt solution is virtually independent of temperature, giving similar sized micelles at 343 K. For the cationic copolymer the micelle cross-section radius decreases with temperature in both solvents and the smallest micelles are present in the salt solution (Fig. 3). The difference in solution behavior for the two copolymers is ascribed to the different nature of the charged blocks, as in the case of the spherical micelles. However, the constant cross-section radius observed for the anionic polymer could arise from a balance between decreased solvent quality of water to MPEG and the charge screening of the PSSS(−) block by the added salt. This may allow the block to contract slightly in spite of the steric restrictions, which is only possible because more volume is available on the micelle surface of the cylindrical micelle than on the spherical micelle.
The temperature-induced changes in the self-assembly behavior of the charged triblock copolymers are depicted in Figs. 4 and 5 in the absence and presence of salt, respectively. In aqueous solution at ambient temperature the polymers are present as singly dissolved chains. Electrostatic repulsion is observed regardless of the polymer size and the nature of the charged block (anionic or cationic). Raising the temperature does not induce any changes to the polymers with the small PNIPAAM block length and the repulsive inter-particle interactions persist. The polymers with the medium and large PNIPAAM block size both exhibit a CMT, and spherical or cylindrical micelles are formed, respectively. This behavior is independent of the nature of the charged block. Increasing the temperature further to 343 K induces changes in the micelles, whereas the molecularly dispersed copolymers do not show any temperature-induced changes. The changes observed for the micelles are dependent on the block composition of the copolymers. The micelles formed by the polymers with the anionic block [PSSS(−)] increase in radius for the spherical micelles and in cross-section radius for the cylindrical micelles. In contrast, the micelles formed from the polymers with the cationic block [PN(+)] contract. This is ascribed to the difference in the character of the two charged blocks used, as previously described.
The charged triblock copolymers also show molecularly dispersed chain behavior in the presence of 30 mM NaCl at ambient temperature. However, significant signatures of inter-particle interactions are not observed, because of charge screening. It is thus found that 30 mM salt is sufficient to screen most of the charge interactions in both polymer systems. Increasing the temperature does not induce micelle formation in the polymers with the short PNIPAAM block in this solvent either, and the polymers are present as molecularly dispersed chains in the entire temperature range investigated. As in the aqueous solution without salt, increasing the temperature induces micelle formation for the polymers with the medium and larger PNIPAAM block size. Again the length of the PNIPAAM block governs the micelle morphology. The polymers with the medium PNIPAAM block size form spherical micelles and the polymers with the large PNIPAAM block size form cylindrical micelles. This is observed for both polymer species, as in the salt-free aqueous solution. Increasing the temperature further induces changes to the micelles. Both the spherical and cylindrical micelles formed by the polymers with the cationic block [PN(+)] decrease in radius and cross-section radius, respectively, as in pure water. However, the micelles formed by the polymers with the anionic block [PSSS(−)] do not follow the same trend as the micelles formed in the salt-free aqueous solution. The spherical micelles grow in radius with increasing temperature as in the salt-free aqueous solution, whereas the cylindrical micelles do not exhibit any temperature dependence as they do in the aqueous solution without salt. This can be explained by a balance between the reduced solvent quality for the MPEG and the PSSS(−) blocks, which induce growth and a slight contraction of the PSSS(−) block when counter-ions are present and thus screen the interactions between the charged repeat units, allowing the block to contract slightly. This is most likely not observed in the spherical micelles because the volume available to the PSSS(−) block is not sufficient to facilitate contraction of the block.
The complexity of the charged triblock copolymers resulted in aggregation behavior dependent on a variety of parameters: polymer composition, temperature and ionic strength of the solution. These competing effects made it difficult to completely rationalize the solution behavior of the systems. In spite of this several trends could be identified. At low temperature, prior to micelle formation, both copolymer series dissolve as molecularly dispersed copolymers in solution. In aqueous solution without salt, repulsive interparticle interactions are present, and the electrostatic interactions giving rise to this repulsion are screened by addition of salt. Therefore, in the aqueous salt solution no signatures of inter-particle repulsion are observed between the molecularly dispersed chains.
The self-assembly behavior is dependent on the nature of the charged block in the copolymer. The CMT of the triblock copolymer with varying PNIPAAM block length is temperature dependent, when the charged block is anionic, and temperature independent for the copolymer with the cationic block. This effect is believed to originate from the difference in polymer–water interactions of the two systems.
Self-assembly is not observed for the polymers with the short PNIPAAM block length in either system. Thus, the hydrophobicity of the PNIPAAM block is not sufficient to drive micelle formation. The polymers with the medium and long PNIPAAM block length self-assemble and the structure of the micelles varies with the PNIPAAM block length. The polymers with the medium PNIPAAM block length form spherical micelles, whereas the polymers with the long PNIPAAM block length form cylindrical micelles. Variations induced by an increase in temperature within a specific geometry are found to be dependent on the nature of the charged block, as these differ in geometry (or how much they can contract with respect to steric hindrance) as well as the different solvent quality of water towards the different polymer blocks.
†This article will form part of a virtual special issue of the journal, presenting some highlights of the 15th International Small-Angle Scattering Conference (SAS2012). This special issue will be available in early 2014.
‡Present Address: Division of Physical Chemistry, Department of Chemistry, Lund University, Gettingsvægen 60, Box 124, Lund, SE-221 00, Sweden.
MAB and JSP acknowledge support from the Danish Council for Independent Research for Natural Sciences. MAB gratefully acknowledges support from the Villum Kann Rasmussen Foundation. JSP and BN acknowledge support from the People Program (Marie Curie Actions) of the European Union's Seventh Framework Programme FP7/2007–2013/ under REA grant agreement No. 290251.
Ataman, M. (1987). Colloid Polym. Sci. 265, 19–25. CrossRef CAS Web of Science
Ataman, M. & Boucher, E. A. (1982). J. Polym. Sci. Part B Polym. Phys. 20, 1585–1592. CrossRef CAS Web of Science
Behrens, M. A., Kjøniksen, A.-L., Zhu, K., Nyström, B. & Pedersen, J. S. (2011). Macromolecules, 45, 246–255. Web of Science CrossRef
Behrens, M. A., Lopez, M., Kjøniksen, A.-L., Zhu, K., Nyström, B. & Pedersen, J. S. (2011). Langmuir, 28, 1105–1114. Web of Science CrossRef PubMed
Choi, J. S., Joo, D. K., Kim, C. H., Kim, K. & Park, J. S. (2000). J. Am. Chem. Soc. 122, 474–480. Web of Science CrossRef CAS
Chuang, C., Don, T. & Chiu, W. (2009). J. Polym. Sci. Part A Polym. Chem. 47, 5126–5136. Web of Science CrossRef CAS
Discher, D. E. & Eisenberg, A. (2002). Science, 297, 967–973. Web of Science CrossRef PubMed CAS
Grossfield, A. (2005). J. Chem. Phys. 122, 024506. Web of Science CrossRef PubMed
Holmberg, K., Jönsson, B., Kronberg, B. & Lindman, B. (2002). In Surfactants & Polymers in Aqueous Solution, ch. 4, pp. 97–118, and ch. 13, pp. 277–316. Chichester: Wiley.
Kjøniksen, A. L., Zhu, K., Behrens, M. A., Pedersen, J. S. & Nyström, B. (2011). J. Phys. Chem. B, 115, 2125–2139. Web of Science PubMed
Liu, J., Zhang, Q., Remsen, E. E. & Wooley, K. L. (2001). Biomacromolecules, 2, 362–368. Web of Science CrossRef PubMed CAS
Mountain, R. D. & Thirumalai, D. (2004). J. Phys. Chem. B, 108, 19711–19716. Web of Science CrossRef CAS
Muthukumar, M., Ober, C. K. & Thomas, E. L. (1997). Science, 277, 1225–1232. CrossRef CAS Web of Science
Nucci, N. V. & Vanderkooi, J. M. (2008). J. Mol. Liq. 143, 160–170. Web of Science CrossRef PubMed CAS
Pamies, R., Zhu, K., Kjøniksen, A. & Nyström, B. (2009). Polym. Bull. 62, 487–502. Web of Science CrossRef CAS
Pedersen, J. S. (2004). J. Appl. Cryst. 37, 369–380. Web of Science CrossRef CAS IUCr Journals
Schild, H. (1992). Prog. Polym. Sci. 17, 163–249. CrossRef CAS Web of Science
Schwuger, M. J. (1972). J. Colloid Interface Sci. 43, 491–498. CrossRef Web of Science
Sommer, C., Pedersen, J. S. & Stein, P. C. (2004). J. Phys. Chem. B, 108, 6242–6249. Web of Science CrossRef PubMed CAS
Stieger, M., Richtering, W., Pedersen, J. S. & Lindner, P. (2004). J. Chem. Phys. 120, 6197–6206. Web of Science CrossRef PubMed CAS
Toncheva, V., Wolfert, M. A., Dash, P. R., Oupicky, D., Ulbrich, K., Seymour, L. W. & Schacht, E. H. (1998). Biochim. Biophys. Acta, 1380, 354–368. Web of Science CrossRef CAS PubMed
Topp, M. D. C., Dijkstra, P. J., Talsma, H. & Feijen, J. (1997). Macromolecules, 30, 8518–8520. Web of Science CrossRef CAS
Wang, Y., Gao, S., Ye, W. H., Yoon, H. S. & Yang, Y. Y. (2006). Nat. Mater. 5, 791–796. Web of Science CrossRef PubMed CAS
Zhang, Y. & Cremer, P. S. (2006). Curr. Opin. Chem. Biol. 10, 658–663. Web of Science CrossRef PubMed CAS
© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.