Introduction

Chitosan represents a collective name for a group of polysaccharides consisting of glucosamine and N-acetylglucosamine or glucosamine only. Thus, the chemical structure of chitosan includes both cationic and hydrophobic sites. Chitosans differing in molecular weights are usually categorized into high molecular weight (HMW) chitosan, low molecular weight (LMW), and oligochitosan. Although the boundaries between these groups are rather uncertain, it has been suggested to restrict the term oligochitosan to a polymer with less than 100 saccharide units [1]. It is well known that physicochemical and biological properties of chitosan depend on its molecular weight (MW) and degree of acetylation [2, 3].

The presence of charged groups in polymer chains and their mobile counterions leads to additional nonlocal interactions of the chains and increases the diversity of the properties of polyelectrolyte solutions. In solution, aggregating groups of polyelectrolytes can form both intra- and intermolecular reversible junctions. As a result, associating polyelectrolytes can form either finite size clusters (aggregates) or infinite clusters (physical gels) as it has been predicted theoretically [4–6]. These predictions have been proven experimentally by many authors who revealed intermolecular aggregates in chitosan solutions below its pKa value. The aggregation phenomenon occurring in solutions of a HMW chitosan has been demonstrated by dynamic and static light scattering (DLS and SLS) techniques [7–12]. The formation of hydrophobic domains accompanying an aggregation of HMW chitosan also has been evidenced by fluorescence spectroscopy [10, 12, 13]. Some authors have shown the presence of intermolecular aggregates in dilute solutions of HMW chitosan by means of SEC coupled with a mass-sensitive light scattering detector [14–16]. TEM and AFM studies have visualized aggregates of HMW chitosan and its derivatives [9, 12, 17–19].

Thorough knowledge of self-aggregation effects occurring in chitosan solutions is highly significant not only for accurate chitosan characterization, but also for practical applications of physical hydrogels and chitosan-based nanomaterials in medicine and technology. The understanding is also important for biomedical applications of chitosan since this phenomenon can affect its biological properties.

It is known that oligochitosan has several advantages over HMW chitosan in some biomedical and industrial applications [20]. In particular, due to a lower viscosity and higher compatibility with many additives, the application of oligochitosan has no undesirable impact on the physicochemical properties of consumer goods. Additionally, in terms of biological action and biomedical applications, oligochitosan and its derivatives have also been shown to have some advantages compared to HMW and LMW chitosans [1, 18, 21–24]. The enhanced oral absorption of oligochitosans opens a perspective of using oligochitosan as an alternative/additive for traditional antibiotics [25].

Most of the structural studies of chitosan solutions being performed for HMW and LMW chitosans [7, 9–11], the associative behavior of oligochitosan has been studied obscurely so far [18]. Indeed, in order to thoroughly understand the mode of oligochitosan bactericidal activity its molecular heterogeneity and self-assembling properties should be considered also.

The present paper describes the self-assembly effects occuring in oligochitosans varying in molecular weight in aqueous media. The influence of different procedures of solution preparation and pre-treatment on molecular structure of oligochitosan diluted solutions are examined by light scattering, HP SEC and microscopy methods, and the data about structural features of oligochitosan solutions, which are indispensable for preparation of injectable chitosan compositions, are discussed.

Materials and methods

Synthesis of oligochitosan samples

All oligochitosan hydrochloride samples used were prepared in accordance with the procedure described in [18]. Degree of acetylation (DA, mol. %) was determined by 1H-NMR method [26].

Sample solutions preparation

Dilute solutions were prepared as follows: oligochitosan hydrochloride sample was dissolved either in 0.3MAcONa/AcOH (pH 5.4) buffer or in 0.1 M acetic acid (sample concentrations was c < 1/[η]). To study the aggregation as a function of pH 3 M aqueous tris(hydroxymethyl)aminomethan (TRIS) solution was added drop wise for pH adjustment. Before measurement, the solution was purified by filtration through one of the following porous membranes: Filter A – 1.2 μm cellulose ester membrane (Millipore), Filter B - 0.45 μm cellulose ester (Millipore), Filter C – 0.2 μm regenerated cellulose (Whatman), or Filter D – 0.22 μm hydrophilic polyvinylidene fluoride (PVDF) membrane (Millipore).

Viscometry

Measurements were performed at 25 °C with an Ubbelohde capillary viscometer. The intrinsic viscosity [η] was calculated by extrapolating reduced viscosity values to zero concentration.

High performance size exclusion chromatography (HP SEC)

Agilent 1200 Series Chromatography system equipped with an isocratic pump, refractive index detector, and PL-OH mixed column was applied for HP-SEC. 0.2M CH3COOH/0.15M CH3COONH4 buffer solution (pH 4.5) filtered through a 0.22 μm Durapore membrane and degassed was used as the eluent at flow rate 1 ml/min. The column and detector temperatures were kept at 25 °C, injection volume was 50 μl, and sample concentration was 0.001 g/ml. A series of monodispersed pullulans (Fluka) with M p from 1.08 to 710 kg · mol−1 were used as calibration standards. Molecular characteristics of chitosan samples are shown in Table 1.

Table 1 HP SEC and viscosity characterization of chitosan macromolecules measured in 0.3 M AcONa/0.1 M AcOH buffer (pH 5.4)
Full size table

Light scattering

SLS and DLS measurements were performed using a PhotoCor Complex spectrometer (PhotoCor Instruments, Russia) equipped with a He–Ne laser (λ = 633 nm, 10 mW) as the light source and a pseudo cross-correlation system of photon counting that allowed to avoid the distortions of the correlogram due to afterpulsing of the photo multiplier at small lag times and to measure Rh values down to approximately 1 nm.

The real-time correlator was employed in the logarithmic configuration. Measurements were performed in dilute solutions at 25ºС within the range of scattering angles of 30–140o. Distributions over decay time and hydrodynamic radius were obtained by means of a nonlinear regularized inverse Laplace transformation method (CONTIN). Apparent self-diffusion coefficients D were determined for each diffusive mode from angular dependence of the reciprocal relaxation time τ in accordance with the relation D = 1/τq 2, where q = (4πn/λ)sin(θ/2) is wave vector magnitude. The corresponding hydrodynamic radii R h were calculated from Stokes-Einstein equation: R h  = kT/6πηD, where k is Boltzmann’s constant, η is the solvent viscosity. When the distribution of scattered intensity over decay time A(τ ) contained two distinct diffusive modes A fast(τ) and A slow(τ), diffusion coefficients for each mode were determined at fixed concentration from the slope of the linear dependence 1/τ vs q 2, and from the beginning slope of this plot if it was nonlinear (second-order polynomial 1/τ = ax + bx 2, where x = q 2). Radii of gyration of aggregates R g agg were estimated from the plots of the reciprocal excess intensity normalized on the relative contribution of the slow mode [I agg(q) = I(q)A slow(q)] vs q 2 using Zimm relation. Apparent molar weights M w app were calculated according to Zimm relation either at fixed concentration or by double extrapolation to q = 0 and c = 0. Apparent molar masses of aggregates M w agg were calculated from SLS and DLS by following relations:

$$ M_w^{app }=yM_w^{uni }+\left( {1-y} \right)M_w^{agg }, $$
$$ {{\left. {\frac{{{A_{slow }}}}{{{A_{fast }}}}} \right|}_{q=0 }}=\frac{{\left( {1-y} \right)M_w^{agg }}}{{yM_w^{uni }}}, $$

where M w (SEC) values were used as M w uni, second virial coefficients of macromolecules and aggregates being ignored, A slow and A fast are relative contributions of two modes into LS intensity.

Dn/dc in various solvents (AcOH, AcOH/TRIS, and AcONa/AcOH buffers) were measured on Wyatt Optilab 903 refractometer at λ = 633 nm. Solutions of chitosan hydrochloride sample 5 (38.6 kg · mol−1) were dialysed against the solvent using Spectrapore membrane with 3.5 kg · mol−1 cutoff. The values dn/dc = 0.163 ± 0.003 and 0.154 ml/g were found in pH range 2.8÷5.4 for solutions in AcOH/TRIS and + AcONa/AcOH buffers, respectively.

Transmission electron microscopy

Transmission electron microscopy was performed with a LEO912 AB OMEGA electron microscope. The samples were prepared using 0.001–0.005 g/ml solutions of oligochitosan in 0.1 M acetic acid. A droplet of sample solution was deposited on a Formvar-coated copper grid and dried for 1 min, and then the excess of the solution was blotted off. The staining solution (1 % solution of uranyl acetate) was added, blotted off, and dried in the air.

Atomic force microscopy

2.5-μl drops of dilute solutions of samples in 0.1 M acetic acid were deposited onto freshly cleaved mica and dried in air. The sample morphology was examined with a scanning probe microscope FemtoScan produced by Advanced Technologies Center, Russia. AFM images were obtained in contact mode using CSG10 cantilevers at room temperature in air. The images were processed and analyzed by FemtoScan Online software.

Results and discussion

Equilibrium theory of multisticker polyelectrolytes predicts the formation of optimal size clusters (aggregates) in dilute solutions [6]. The stabilization of optimum size clusters of spherical form is a result of competition between the attraction of associating groups and the repulsion of the chains caused by their charges as well as by translational entropy of counterions. The sizes of clusters are determined by the distances between the stickers, charged units, and average polymer volume fraction. As a result, the size of cluster is expected to be independent of the chain length. In this theory, authors consider a salt-free system where subchains between the associating groups are assumed to be flexible and swollen with respect to their Gaussian size due to the osmotic pressure of counterions and repulsive forces between the charged units.

Another approach has been used for theoretical consideration of association of rigid/semirigid charged macromolecules with relatively high attraction energies [27], and a tendency to form filaments, where chains were aligned in parallel, has been predicted. As a consequence, the authors have predicted high kinetic hindrances for structural rearrangements in such systems.

Molecular dynamics simulations of macromolecules with amphiphilic monomer units, where hydrophobic fragments form the backbone of the chain, and hydrophilic fragments constitute side groups attached to the backbone, seems to be rather valuable for predicting conformation transitions and aggregation peculiarities of a wide variety of natural polymers including polysaccharides. So, it has been found that those macromolecules with amphiphilic monomer units were particularly structured in semi dilute solutions [28]. During deterioration in solvent quality, flexible amphiphilic macromolecules formed soluble globules, whereas rigid-chain macromolecules form fibril-like aggregates. Additionally, the structure of concentrated solutions depends on the method of solution preparation. For rigid amphiphilic macromolecules capable of ordering into a local helical structure, it has been shown that the compaction of a macromolecule during deterioration of solvent quality leads to the stabilization of the secondary helical structures [29].

Simple chemically, oligochitosan macromolecules have a rather complicated 3D structure. Their specific structural features include semi rigid conformation, changeable hydrophilic-hydrophobic balance of glucosamine monomer unit depending on protonation-deprotonation process in addition to the presence of hydrophobic acetylglucosamine units, the ability to form helical conformations stabilized with hydrogen bonds between the neighboring units, and the ability to form intermolecular hydrogen bonds. The self-aggregation phenomenon for oligochitosans was firstly observed in our previous publications [18, 30] where it was shown that hydrodynamic radii of aggregates R h agg are virtually independent of pH in AcOH/TRIS solution below pH critical values despite the continuous variation in degrees of ionization. At critical pH values macro-phase separation into gel phase and supernatant starts, and the co-existence of two aggregation modes (primary aggregates and their clusters) was observed by DLS.

DLS-SLS studies

In various investigations of dilute solutions of HMW and LMW chitosans different procedures of solution filtration have been used. Filtering membranes differing in membrane material and porosity: 0.05–0.2 μm Sartorius nitrocellulose membranes [7, 31], 0.45–0.22 μm Millex-LCR hydrophilic PTFE membranes [11], and 0.45–0.22 μm membranes of unknown type [32, 33] have been used for the purposes of molecular characterization or self-aggregation study. As a result, either true molecular solutions or aggregated systems were observed. Moreover, no changes or negligible decrease of chitosan concentration after filtration were experimentally determined in these works. In addition, no solution aging effects were observed [7, 31]. Also when 0.45 μm Millex-LCR hydrophilic PTFE membrane was used for chitosan solution filtration the aging effect wasn’t observed while molecularly dispersed solutions (prepared by filtration through a 0.22 Millex-LCR membrane) were subjected to self-aggregation in time [11].

Thus, studying behaviour of oligochitosan solutions and self-aggregation phenomenon in details, we should, in particular, answer the following questions: a) does sample filtration method affect the associative properties of oligochitosans in their solutions?; b) do the aggregates exist in the solution in the state of a dynamic equilibrium with individual polymer chains?

In order to examine the influence of purification procedure on DLS-SLS results dilute solutions of oligochitosans were filtered through membrane filters of different porosities (filters A-D). Using CONTIN analysis of autocorrelation functions (examples are shown in Fig. 1a), two diffusive modes ascribed to two populations of species, namely, single macromolecules and aggregates, were observed in both 0.1M acetic acid and 0.3 M AcONa/0.1M AcOH (pH 5.4) buffer solutions. Table 2 demonstrates that aggregate sizes are highly dependent on filter porosity and type in both solvents. In general, the larger filter porosity is, the higher measured value of R h agg is. The intensity of scattered light and, as a consequence, M w app values sharply decreased when a finer filter was used (filters A, B, and C). Figure 1 illustrates the variation of distribution over hydrodynamic radii R h agg (Fig. 1a) and angular dependences of reciprocal relaxation times (Fig. 1b) of slow mode (i.e. D agg) of sample 2 as the result of filtration through cellulose filters of different porosity. These figures show that filtrations through fine cellulose filters (either 0.45 or 0.2 μm) provide linear 1/τ vs q 2 dependences with different slopes resulting in different diffusion coefficients and hydrodynamic radii. On the contrary, solutions purified by the coarse filter A are characterized by nonlinear 1/τ vs q 2 plots (i.e. with the angular dependences of the apparent diffusion coefficients). The following causes for this nonlinearity are possible: firstly, internal diffusion motions in large (qR g ≥1) fractal aggregates, secondly, superposition of translational and rotational diffusions of large anisometric particles, and thirdly, particle heterogeneities both in sizes and forms.

Fig. 1
figure 1

Distributions over hydrodynamic radii of sample 2 (7.7 kg · mol−1) in AcOH (a) depending on purification procedure: filters A (1), B (2), C (3); and angular dependences of reciprocal relaxation times of slow mode (1/τ vs q 2 plots) for the same solutions (b)

Full size image
Table 2 Effect of solution purification protocol on chitosan aggregation in 0.1 M AcOH and 0.3 M AcONa/AcOH (pH 5.4) buffer
Full size table

It should be noted that the solvents taken for LS studies of chitosans essentially differ in their ionic strengths. In accordance with Debye-Huckel theory, the low ionic strength of acetic acid solution (I = 0.0013 M) is insufficient to screen electrostatic interactions in the system in contrast to 0.3 M AcONa/0.1 M AcOH buffer. It is well known that conventional relations usually applied to uncharged polymers cannot be used for analyzing LS data of polyelectrolytes dissolved in a solvent of extremely low ionic strength [32–37]. Due to strong intermolecular repulsions between polyions, the intensity of light scattered by a dilute salt-free solution decreases with dilution and becomes as low as that of the solvent. Therefore, realistic dimensions of polycationic oligochitosan macromolecules cannot be measured by SLS and DLS in salt-free systems. Relations applied to deriving the distribution function over relaxation times in standard DLS software are correct for a weakly interacting system only. For this reason, R h values of aggregates only, but not apparent values R h uni of macromolecules and relative contributions from the fast and slow modes, were determined for samples dissolved in 0.1 M acetic acid. Sizes of unimers R h uni as well as relative contributions of both modes to the correlation functions were estimated in solvents of higher ionic strength (0.3 M AcONa/0.1 M AcOH and 0.1 M AcOH/TRIS buffers at pH ≥ 5). These relative contributions Afast and A slow were used to calculate radii of gyration of aggregates and apparent molar masses of aggregates M w agg.

The ratio of two differently defined equivalent sphere radii, R g /R h, (shape factor), can be used as a qualitative criterion for analyzing aggregates structures if they are sufficiently uniform in size [38–41]. Magnitudes of aggregate shape factors determined in 0.3 M AcONa/ 0.1 M AcOH solutions filtered through finest filters C and D (Table 2) were rather near to 1 or slightly greater, and were independent of chain length. Similar values were found for 0.1 M AcOH/TRIS solutions of various chitosan samples. These values were in agreement with theoretical predictions for hollow spheres (shape factor =1), slightly anisometric prolate ellipsoids, “soft spheres” and dendrimers (0.977), as well as “regular stars” with many arms (1.07 – 1.225) [40]. Thus, it can be supposed that the aggregates present in filtered chitosan and oligochitosan solutions have approximately spherical shapes.

SLS-DLS results are affected not only by the nominal porosity of filter, but also by the filter’s material. Indeed, different results were obtained for the sample solutions filtered through cellulose and PVDF membranes with the mean porosity 0.2 μm (filters C) and 0.22 μm (filter D). Consequently, aggregates were observed after filtration through filter C, whereas in the latter case aggregates were not observed in most solutions with the exception of chitosan samples 4 and 5 having relatively longer polymer chains (21.5 and 38.6 kg · mol−1, respectively) in 0.3 MAcONa/0.1 M AcOH buffer. The following reasons of aggregate disappearance are possible: a) particles were separated (filtered off) by the filter D (0.22 μm PVDF membrane); b) aggregates were disrupted when passing through the membrane pores. It should be mentioned that no solution aging effect was observed in solutions filtered through filter D during the storage time of a month.

In order to elucidate the question whether the scattered light drop after the filtration was the result of the loss of chitosan or it was due to disruption of aggregates and their conversion to individual macromolecules, oligochitosan 2 (7.7 kg · mol−1) solutions purified by different filters (A, B, and D) were examined by HP SEC. The SEC analysis showed that the peak areas were independent of the solution purification (the standard deviation was 2.5 %) and calculated M w and polydispersity indexes were also independent of the purification method (standard deviations were 3.0 and 1.0 %, respectively) for all solutions. These results show that the changes in LS data were caused by disruption of aggregates. In addition, the weight of chitosan was determined after the solution out-coming from filter D was lyophilized, and the complete balance of in- and out-coming chitosan was confirmed.

To find out whether chitosan sample (7.7 kg · mol−1) solutions in 0.1 M acetic acid or 0.3 MAcONa/0.1 M AcOH buffer filtered through the 0.22 μm PVDF membrane D can “remember” their unimolecular organizations, these solutions were freeze-dried in dust-free flasks and then redissolved in deionized water. Afterwards, the solutions were studied by SLS-DLS both without additional filtration and after the filtration through a 0.22 μm PVDF membrane (filter D). SLS-DLS studies showed that redissolution of purified samples resulted in the formation of a aggregated system, and the hydrodynamic radii of aggregates were almost independent of sample concentration in the range 0.003 – 0.01 g/ml. High R h agg (180–270 nm) and M w app (3,000–10,000 kg · mol−1) values indicated the presence of large aggregates with high aggregation numbers. On the other hand, additional filtration through filter D led to disruption of aggregates as it was described above.

Taking into account the above shown influence of solution filtration procedure on aggregation phenomena, it was reasonable to compare the aggregates in their sizes and apparent molar masses (M w app) depending on the chain length when the solutions were purified identically. Figure 2 shows growth of R h agg in two series of solutions filtered through either 0.45 μm or 0.2 μm cellulose filter. It represents also the variation of the mean R h agg values in 0.1 M AcOH/TRIS solutions at pH 5.0 ± 0.5, i.e. below critical points of phase separation, as a function of chain lengths (prior to measurements all solutions were filtered through membrane C). The presented dependences are non-monotonic and approximately parallel one to the others. In particular, the M w range up to 11.5 kg · mol−1 (which can be attributed to oligomeric macromolecules) is characterized by virtually constant sizes and masses of aggregates whereas above this range the increase of aggregate sizes and masses is observed with the increase of chain lengths. It is interesting that the sizes of oligochitosan aggregates are significantly lower than the membrane pore sizes, therefore these constant values cannot be limited by membrane pores. Values of apparent molecular weights of aggregates M w agg demonstrated the similar non-monotonic variation (not shown). It is clear that if M w agg remains virtually constant with the increase of M w uni , apparent aggregation numbers decrease in this sample series. These results prove that short chain chitosan samples having Mw below 12 kg · mol−1 form aggregates of constant sizes and can be selected into a series of oligochitosans (less that 100 monomer units) whose aggregation behavior is somewhat different of that occurring in solutions of higher molecular weight chitosan.

Fig. 2
figure 2

Variations of the hydrodynamic radius of aggregates as functions of molecular weight of chitosan sample 2 in: 0.1 M AcOH 1(1, 2) and 0.1 M AcOH/TRIS buffer (3), depending on purification procedure: filter B (1) and filter C (2, 3)

Full size image

HP SEC and viscometry characterization of oligochitosan samples

The studied series of short chain chitosans of low DA prepared by acidic depolymerization were characterized by HP SEC and viscometry (Table 1). All samples demonstrated unimodal chromatograms and possessed rather low polydispersities. HP SEC results (calculated average MWs and peak areas) were independent of the filtration procedures. In other words, the results remained the same whether the sample solution taken for HP SEC analysis contained either large or small aggregates (filters A-C) or the aggregates were destroyed by filtration through a membrane filter D (section DLS-SLS studies). The intrinsic viscosity of oligochitosan sample 5 (38.6 kg · mol−1) measured in 0.3 M AcONa/0.1 M AcOH buffer (pH 5.4) also was found independent of the filtration protocol. Therefore, it was supposed that only unimolecular species of oligochitosan were observed in HP SEC and viscometry experiments without an essential impact on aggregation phenomenon.

The values of viscometric radii R η are shown in Table 1 for comparison with R h uni determined by DLS. It has to be taken into account that hydrodynamic radii R η of oligochitosan samples shown in Table 1 were estimated using the Stokes equation and M w (SEC) values (determined relatively pullulans) and therefore they must be considered as a first approximation. As it was discussed above, R h uni values deduced from DLS fast mode were in reasonable agreement with R η , whereas average M w app values were usually much higher than M w (SEC) shown in Table 1.

The comparison of viscometric behavior of chitosan samples in two solvents differing in ionic strength (0.1 M AcOH and 0.3 M AcONa/0.1 M AcOH buffer) has shown the differences characteristic for polyelectrolyte solutions: primary and tertiary electroviscous effects were observed for all samples in salt-free system (the example is shown in Fig. 3). Indexes of Mark-Houwink equation calculated from experimental data varied from a = 0.81 ± 0.06 in buffer to 1.03 ± 0.02 in the salt-free acetic acid solution indicating a tendency to enhance the chain rigidity. This discovery together with the conclusion inferred from HP SEC allowed supposing that large supramolecular species deaggregated during chromatography and capillary viscometry. It seems that the common feature of these techniques is the action of a shear stress. Actually, variation of the applied shear rate by a factor of 10 is characteristic for two methods: ~ 1,000 s−l for capillary viscometry and ~ 10,000 s−l for SEC. The phenomenon of shear deaggregation during capillary viscometry experiments also has been observed in the study of micellar solutions of block-copolymers [42]. Besides, it is well known that HP SEC coupled with mass-sensitive LS detector, where light scattering is performed in a capillary shear field, is the most reliable method for molecular characterization of chitosan [8, 29, 43]. Therefore, the influence of shear stress can be a cause of deaggregation in oligochitosan solutions during HP SEC and viscometry experiments.

Fig. 3
figure 3

Viscosity plots for oligochitosan sample 4 (21.5 kg · mol−1) in 0.1 M AcOH (1) and 0.3 M AcONA/0.1MAcOH (pH 5.4) buffer (2)

Full size image

TEM and AFM images

In order to elucidate the morphology of aggregates, they were visualized by TEM and AFM methods.

Figure 4 shows several images of chitosan samples whose solutions were purified using different filters. Negatively stained spherical particles were found characteristic of all samples after solution filtration through fine cellulose membranes (B and C). The example is shown on Fig. 4b. As it was shown by DLS, sample solutions filtered through PVDF membrane (D) didn’t contain aggregates, and the sample TEM images were not representative to demonstrate well-defined aggregates. Small spherical particles (Fig. 4a) were observed for sample 5 only. One should note that these TEM micrographs were normally made after the drying of solutions on the support films so that the particles shown on Fig. 4a were formed in the course of the sample drying procedure. On the other hand, a wide morphological diversity of aggregates was observed by TEM for solutions filtered through a coarse membrane (A). Actually, Fig. 4c and d show the images of sample 2 containing a collection of both fibril-like structures and more or less spherical and rod-like particles. It should be reminded here that the lyophilization of oligochitosan solutions filtered through membrane D and the redissolution of the sample resulted in similar morphological diversity of aggregates.

Fig. 4
figure 4

TEM micrographs of chitosan samples 5(a) and 2 (b, c, d) in 0.1 M AcOH depending on purification procedure: filters D (a), C (b), and A (c, d)

Full size image

Sample 5 was additionally inspected by AFM (Fig. 5). The figure corroborates the shown above variations of morphology caused by different sample filtrations. The solution of chitosan sample 5 purified by filtration through the filter D (0.22 μm PVDF) resulted in a rather smooth plane on mica surface (Fig. 5a) whereas the filtration through the filter A (pore size 1.2 μm) produced essentially not uniformed spherulite -like patterns of much larger sizes (Fig. 5b).

Fig. 5
figure 5

AFM images of sample 5 depending on purification procedure: filter D (a) and filter A (b)

Full size image

Thus, the microscopy data are in a good agreement with the conclusions made after DLS-SLS studies: size, form, and heterogeneity of aggregates depend on the sample preparation method. TEM images of spherical particles of oligochitosan, whose solutions were purified by filtration through the finest filters, match with the conclusions about the form and uniformity of particles made on the basis of SLS-DSL studies.

To summarize the results, the following tendencies should be pointed out: a) the effect of sample purification method on the observed aggregation phenomenon must be critically considered since the results of the structural study of chitosan solutions are essentially affected by the purification procedure; b) porous membranes apparently play an important role in the structural rearrangements occurring in chitosan solution during filtration with the mechanism of this rearrangement not yet clear and requiring further investigations; c) roughly spherical aggregates seem to be the most characteristic for purified solutions of oligochitosan and rather short chain oligochitosans as well while longer oligochitosan chains seem to be somewhat less liable to deaggregation during filtration; d) AFM and TEM data show that spherical particles and their conglomerates are the main, but not unique, structural units: fibril-like particles and also bundles of filaments are observed as well; e) the preservation of the variety of structures may be connected with a relatively high potential barrier for changing the structural morphology of aggregates. In particular, the enhanced kinetic stability of intermolecular junctions inside the aggregates can be caused by a cooperativity of hydrogen bonding in junction domains where worm-like macromolecules are placed parallel and aligned in the order similar to crystalline domens existing in chitin. As a result, the equilibrium state of oligochitosan solution is not easily accessible so that certain structures formed under different conditions (for example, in the course of the sample lyophilization) can preserve their identity until they are destroyed by an additional treatment.

Conclusion

In this paper, we have described the self-aggregation phenomenon and the impact of the improved filtration method of oligochitosan sample solutions varying in molar mass on solution homogenity that can be helpful for preparation of injectable chitosan compositions. Actually, light scattering and microscopy studies show that dilute solutions of oligochitosans are partially associated at the entire pH range below the critical points of phase separation in solvents of various ionic strengths. The observed structural diversity of the solutions: from highly associated systems, containing either fibril-like particles or spherical aggregates relatively homogeneous in their sizes, up to the totally deaggregated systems, – depends on the purification procedure, specifically, on the pore size of the membrane filter and its material. Redissolution of the sample prepared by lyophilization of totally deaggregated solution results in a partial recovery of supramolecular structure of the solution. It is significant that HP SEC and capillary viscometry studies do not reflect the supramolecular structure of solutions, and their results are independent of solution pre-treatment. We can infer that the initial supramolecular structures present in oligochitosan solutions are disrupted under shear force action in the course of HP SEC and viscometric measurements. Under filtration through a porous membrane, interactions of macromolecules and morphological units with pore surfaces can play an essential role in the structural rearrangements in addition to the shear stress. We suppose therefore that non-equilibrium states of solutions are generally observed in practice, and theoretical and modeling approaches reflect various characteristic features of structural organization of such complex objects as chitosan and oligochitosan.