Evidence for a very slow disaggregation of lignosulfonates

Lignosulfonates

All our experiments were performed with NaLS to avoid electrostatic bridging between molecules due to divalent ions like Ca²⁺ (Collins et al. 1977). Most of the studies were performed with an ultrafiltrated softwood (spruce) LS. The product was ion exchanged from calcium to the sodium form after ultrafiltration, and spray dried. Two experiments were done with modified samples described under reduction of the carbonyls.

The analytical data of the samples are presented in Table 1, and were determined by the methods described by Myrvold (2013a). All chemicals were from Merck (Darmstadt, Germany). The molecular weight (MW) determination by size exclusion chromatography (SEC) was described in Fredheim et al. (2002) and Fredheim et al. (2003). Column: Jordi Glucose DVB 10⁴ Å pore size (Jordi, Mansfield, MA, USA) in combination with a TSK gel PWXL pre-column. The UV detector was set to 280 nm. The eluent was made from 3200 g water, 323.6 g DMSO, with 242.9 g Na₂ HPO₄·7H₂ O and 3.2 g sodium dodecylsulfate. The pH was adjusted to 10.5 with NaOH. 0.2% solutions of LS were filtered through a 0.22 μm Millipore filter before injection. All measurements were done in duplicate. The standard deviation of the ratio “height of peak I (at low RT) divided by the height of peak II (at high RT)” is 0.5%. Two broad MW LS standards of M_w 33500 and M_w 6000 served for calibration based on the original light scattering work of Fredheim et al. (2002, 2003).

Instrument for viscosity measurements

An automatic Schott Geräte AVS 310 (Schott, Mainz, Germany) with a 532 01/0a capillary was used.

Temperature

A temperature of 25.00±0.02°C was used (Schott CT 52 water bath, Schott, Mainz, Germany). Each value presented is an average of at least five individual measurements. 1.1% solutions were made in water adjusted to pH 12.7 with NaOH. The solutions were filtered through a 0.45 μm Millipore filter before filling into the viscosimeter. The run-out time for each sample was measured repeatedly over several weeks.

Reduction of carbonyls

One gram of LS was dissolved in 5% NaOH solution. 0.5 g NaBH₄ was added. The mixture was stirred for 97 h, poured into 2 l water, and HCl (7.5%) was added, until a pH of 6 was reached, to destroy the excess NaBH₄. The water was removed in a rotary evaporator and the product was further dried at 105°C overnight. The reduction was done with the standard sample and with a sample ultrafiltrated with a much higher cut-off of 100 kDa.

Determination of carbonyl groups

A total of 400 mg LS was dissolved in 1 l 0.1 M NaOH. Two 100 ml samples were withdrawn. To one, 31 mg NaBH₄ was added and stirred for 71 h. The other was left undisturbed. Both samples were diluted to 800 ml. The UV spectra were recorded on a Perkin Elmer Lambda 25 UV-Vis spectrophotometer (Perkin Elmer, Waltham, MA, USA) with the untreated sample in the reference beam.

The difference in absorbance was 0.03940 at 356 nm, 0.03582 at 342 nm and 0.02346 at 310 nm. Following the calculations of Adler and Marton (1959) this corresponds to 0.032 mg g^-1 4-hydroxy benzophenone structures, 0.010 mg g^-1 etherified benzophenone structures and 0.035 mg g^-1 cinnamic aldehyde structures.

Sample preparations

The pH adjustments were done by adding 50% NaOH and water to a 40% LS sample, to obtain the desired pH for the most concentrated samples (40%). This sample was diluted with NaOH solutions to the same pH to achieve the lower concentrations. The samples were stored in glass bottles for the desired time. The heated samples were stored in thick walled borosilicate glass bottles in a heating cabinet for the desired time. The samples were stored in similar bottles at room temperature (r.t.) until the chromatograms were recorded.

Ultra high MW of NaBH₄ reduced sample

The SEC experiments were run directly after diluting and filtrating the sample in the eluent. In total, 40 injections in 30-min intervals (the time needed for one chromatogram) were made (Figures 1 and 2a).

Figure 1

SEC of NaBH₄ reduced lignosulfonates with ultrahigh molecular weights. The samples were repeatedly injected (from top to bottom): 0, 30, 60, 90, 120, 150, 180, 210, 540, and 1170 min after the first injection. Peak I is for the aggregated molecules, while peak II represents the non-aggregated well-dissolved molecules.

Figure 2

Various data of lignosulfonates as a function of disaggregation time. a) Sample shown in Figure 1. Degree of aggregation: 100×height peak I/height peak II in Figure 1. b) Data of 1% LS sample at pH 12.7 (filled circles) or water (open circles). The sample was stored at 10% and then diluted to 1% at different times. c) Run-out time (measured in the capillary viscosimeter) of a 1.1% LS sample at pH 12.7 (filled circles), the NaBH₄ reduced sample (open circles), and a fully disaggregated sample (a 5% solution heated to 105°C for 16 h), which was dried in a rotary evaporator before (red crosses). The samples were left in the capillary viscosimeter and measured several times a day over several weeks.

It is seen from Figure 1 that peak I at the r.t., around 14.5 min, representing high MW LS, diminishes rapidly. The chromatograms are scaled so that the maximum height of peak II, with retention times (RT) around 18.0 min, is constant in all chromatograms. The shape of this peak is unchanged over the 20-h experimentation period and it is supposed to represent non aggregated single molecules, while the rapidly changing first peak represents the aggregated polymer molecules. A similar effect has been described for hydrolysis of lignin (Guerra et al. 2008). The SEC column will separate M_W up to ∼400 kDa (Fredheim et al. 2002). The RT of the aggregate peak corresponds to a M_W of ∼500 kDa and it is probably an exclusion peak and does not represent a true M_W. The area of the largest peak I corresponds to 12% of the total area of the chromatogram. There is no reason to assume any differences in the chromophores (detected at 280 nm) due to the aggregation. Thus no more than 12% of the monomers take part in the aggregate formation of this sample. The peak II for the single molecules at RT around 18 min retains its shape when the aggregate peaks dissolve. There is thus no change to the individual macromolecules during the disaggregation. The molecules in the aggregates must be similar to the individual molecules, e.g., there is no preference for high or low molecular weight. The height of the well-resolved aggregate peak I in Figure 1 is 40% of the maximum peak (Figure 2).

The radius of the LS molecules scales as the MW to the 0.49 power (Myrvold 2007a,b). The factor of 49.5 higher calculated MW for the aggregate peak thus corresponds to 49.5^0.49=6.8 times larger radius than that for the single molecules. This is in reasonable agreement with the relative sizes of single molecules and aggregates described in the literature (Parfenova 2006; Qiu et al. 2010; Yan et al. 2010). Due to the necessary filtration of all the samples through a 0.22 μm Millipore filter to remove possible insolubles, the maximum size of any aggregates detected will be ∼220 nm. Thus the two different aggregate peaks observed by Qiu et al. (2010) cannot be seen in our experiment.

The M_w is strongly dependent on the abundance of higher M_w species, and in the case of the sample depicted in Figure 1 the calculated M_w depends strongly on the size of the aggregate peak. The calculated M_w for the samples in Figures 1 and 2 decreases from 69 600 to 15 600 Da. The aggregation phenomenon will be described in the following as degree of aggregation (%) calculated as: 100×intensity of peak I divided by intensity of peak II (Figure 1).

Concentration dependence

The formation of aggregates from single macromolecules is a reversible process with an equilibrium between the two stages. To observe the concentration dependent equilibrium, several LS solutions in two different solvents in various concentrations were prepared. For the MW determination, the samples were diluted to the same concentration. The composition of the solvent during the MW determination will thus vary slightly depending on the initial concentration. To avoid this complication, one series of samples with different LS concentrations was made in the eluent used for the MW determination and then stored for different times. The results are presented in Figure 3. There are no qualitative differences between the samples stored in NaOH (pH 12.7) or between those stored in the eluent. For the samples in the eluent, there is a weak concentration dependence for a short time. The higher the concentration, the higher the aggregate peak. Increasing the storage time, the aggregates at r.t. increase at higher concentrations (>10%). For the lower concentrations (<10%) the aggregates decrease. At 10%, there is hardly any change and this seems to be the critical aggregation concentration (CAC). For the sample in these experiments, the reduced viscosity is ∼0.1 dl g^-1. This is interpreted to be an over-lap concentration of ∼10%. Below this concentration, there is space for the macromolecules to move apart and the aggregation process is stopped. Above 10% concentration, the molecules are packed so densely that there is little room to keep a distance or separate. Aggregates might also collide and form even larger aggregates. Our results agree well with the 10–19% CAC found by Rana et al. (2002).

Figure 3

Plots of degree of aggregation vs. storage concentration. LS stored < 24 h at pH 12.7 (open circles), 14 days at pH 12.7 (filled circles), 7 days at pH 12.7 including 16 h at 105°C (open triangles) the same sample stored an additional 30 days (filled triangles), 16 days in the eluent (open squares), 61 days in the eluent (gray squares) and 99 days in the eluent (filled squares).

Solutions with different concentrations of NaOH at pH 12.7 were made. As seen from Figures 3 and 4, this pH is high enough to speed up disaggregation. For the samples in alkaline solution at r.t., the same trends are visible as for the samples stored in the eluent. The cross-over between aggregation and disaggregation at ∼10% is also the same. The DMSO added to the eluent is included to reduce the formation of aggregates (Fredheim et al. 2002). This could possibly explain the lower aggregation level in this solvent mixture compared to alkaline water.

Figure 4

Plots of degree of aggregation vs. heating time. The 5% LS samples were stored at pH 12.7 at different temperatures, or at 105°C at different pH.

For the heated samples, 1 week or 1 month storage at r.t. was included before the actual MW measurements. For the higher LS concentrations, there is no re-aggregation once the aggregates have been broken apart. This agrees well with the observation of Qian et al. (2014) that the aggregates are formed at temperatures above 38°C, when the charges of the molecules are lost. When storing at r.t., the molecules remain highly charged and cannot form aggregates. If a solution is cooling down, when the molecules are still uncharged and when the thermal movements are slowed down, little time is left to form new aggregates. This is the reason why no measurable amounts of aggregates are formed in such solutions.

Time dependence

There are early reports on slow disaggregation for LS (Benko, 1961a,b, 1964), but this effect was not considered in recent investigations. Two 10% solutions were prepared, one in pure water and one in NaOH at pH 12.7. At this concentration, no time dependent aggregation can be seen (Figure 3). A sample of 2 ml was withdrawn and diluted to 1% with the same solvent. This was repeated daily for three weeks and the MWs of all the samples were determined in the same run. Thus the samples were stored for different times as 1% solutions. The results are presented in Figure 2b.

In many cases it has been found that the specific viscosity of 1% solutions is higher 1 h after dissolution than after equilibrium is reached (Benko 1964). The specific viscosity is given by η_sp=(t–t₀)/t₀ for dilute solutions where η_p is the specific viscosity, t₀ is the run-out time for the solvent from the capillary and t is the run-out time for the solution (Young 1981). The run-out times for the solvent and the concentrations are constant, thus the only factor that changes over time is the difference of run-out times for the solution (t–t₀), and the latter is shown in Figure 2c.

The samples disaggregate very slowly at r.t. This is seen both from the chromatograms (Figure 2b), and from the viscosity data (Figure 2c). The disaggregation is faster at pH 12.7 than in pure water. This is unsurprising as a result of the larger repulsive forces due to the dissociation of the phenolic hydroxyl groups.

It is found, as expected, that the run-out time decreases over time as the aggregates are broken apart. High pH will cause a dissociation of the phenolic hydroxyl groups. Thus, a stronger polyelectrolyte expansion is expected at higher pHs. However, at pHs above 11, the concentration of ions in the solvent is so large that it effectively screens the charges along the LS chain and there will be no polyelectrolyte expansion. This screening does not seem to lower the CAC.

Benko (1961a,b) noticed that aggregates formed during spray-drying of the LS solution. In this process, small droplets of LS are subjected to high temperature gases and the water rapidly evaporates. The LS molecules in the concentrated solution, in close proximity to one another, have many contact points. The dissolved sample – which disaggregated by heating at high pH and then dried in the rotary evaporator – showed no time-dependent viscosity increment, except for the first 2 h. After 1–2 h, the run-out time was constant within the experimental error. The NaBH₄ reduced and rotary evaporator dried sample also shows a rapid viscosity decrease over the first 4–5 h. After that, there was only a slow disaggregation for the next 700 h.

The laboratory prepared samples were dried in a rotary evaporator. This is a slow process. The water is gradually removed, and the system goes through intermediate steps in highly concentrated solutions. However, as there is still much water present, the number of contact points are limited and a more open and loose structure is formed. This is evidenced by a shorter retention time in SEC indicating a larger volume of the aggregate peak, while the UV absorbance of the aggregate peak is the same (or maybe less) than for the spray dried sample indicating the presence of the same number of molecules in the larger particle. The open and less connected structure is also evidenced by the much faster disaggregation. The disaggregation shown in Figure 2a is one or two orders of magnitude faster than that in 2b. There are fewer connections that must be broken and thus a faster disaggregation is observed. For the NaBH₄ reduced sample, this is partially off-set by the increased amount of hydroxyls, and also the larger MWs. For this sample, the aggregates dissolved in 10–15 h at r.t. The extremely slow rate of disaggregation has not been taken into account in most of the previous studies on the aggregation phenomenon.

Temperature and pH dependence

Benko (1961a,b) found a decreasing MW with increasing pH. A temperature dependence has been found for the disaggregation of hydrolysis lignin (Contreras et al. 2008), which formed fewer aggregates at higher temperatures. LS aggregates were only formed above a certain threshold temperature (Qian et al. 2014).

Solutions with 5% LS at pH 12.7 were prepared and heated at four different temperatures for durations between 1 and 24 h. Samples at different pHs were heated to 105°C for different times. Figure 4 shows the effects. Expectedly, higher temperature leads to a faster disaggregation under heating, the equilibrium between aggregated and not aggregated molecules should be shifted towards the latter because the thermal motion will favor a break-up of the aggregates. The loss of charge for LS in a neutral solution has been investigated by Mafe et al. (1995). The strong pH effect seen in Figure 4 clearly shows that the charge is not lost for the phenolate group at high pH. The disaggregation seems to follow a first order rate law, with a linear decrease in the aggregation peak on a logarithmic time scale. For the samples with 5% LS at pH 12.7, the half-lives and disaggregation rates were calculated (Table 2).

Reaction kinetics can be described by the Arrhenius equation (McMurray and Fay 2008), ln(k)=(E/R)(1/T)+ln(A), where k is the reaction rate, A is the temperature independent frequency factor, T is the absolute temperature, E is the energy of disaggregation and R is the gas constant. Based on the data in Table 2, E∼79 kJ mol^-1 can be calculated. The M_n of the fully disaggregated LS molecule is 2300. This corresponds to about 10 monomers per molecule. The net adhesive energy is thus about 8 kJ mol^-1 per monomer. If the calculations are based on the most probable molecular weight (6000), 27 monomers per molecule and a net attractive energy 3 kJ mol^-1 monomer is obtained.

An interaction energy of 3–8 kJ mol^-1 monomer is found. Clearly, not all monomers in one polymer chain will take part in bonding to other polymer chains. In solution, the LS molecules are folded into a more or less globular shape (Myrvold 2007a,b). Only the monomers on the outside of folded LS molecules will participate in the interactions. The aggregates have fairly open structures (Qiu et al. 2010). For a 3D aggregate, the average polymer chain must interact with more than two others. With two contact points and two monomers per contact, four of the monomers are taking part in intermolecular binding. The energy per binding will thus be 10/4 higher than the calculated 8 kJ mol^-1 monomers. This gives ∼20 kJ mol^-1 binding and this value is higher than the binding energy of 7.5–11.7 kJ mol^-1 for π-π bonds in benzene (Sinnokrot and Sherrill 2006). Substituents like -OH or -CH₃ reduce the π-π interactions by 0–2 kJ mol^-1. The value found agrees well with the strength of hydrogen bonds, which is typically 10–40 kJ mol^-1 (McMurray and Fay, 2008). It should also be noted that the net force is calculated from data at pH 12.7, where the repulsive forces are highest due to the deprotonation of the phenolic groups. Thus 20 kJ mol^-1 per intermolecular bond is an underestimate of the situation in neutral solutions. As Figures 2b and 4 clearly show, the disaggregation is much slower in a neutral solution, e.g., the net attractive forces are larger. The observations are summarized schematically in Figure 5.

Figure 5

Summary of the observations.

Disaggregation of molecules from spray drying is a very slow process. For a molecule to leave the aggregate, all the contact points must be broken. As long as at least one contact point remains, the participating molecules will stay in close proximity, promoting the re-aggregation. Thus, aggregates with many contact points are difficult to separate. For the NaBH₄ treated samples, there is a 10% increase in aliphatic hydroxyls. Therefore, this sample disaggregated much more slowly than the other samples dried by rotary evaporation.

Many studies reported lower CACs than found in this work. The differences are partly sample specific but most of the very low values are probably due to a retention of the aggregates formed at higher LS concentrations during the experiments, which had insufficient time for complete disaggregation over weeks or months. Yean et al. (1964) heated a sample for 30 min at 120°C or 6 h at 95°C, without finding any disaggregation. However, 30 min is too short to effectuate any disaggregation. Kontturi et al. (1992) looked at the effective charge number and diffusion coefficients and ruled out LS aggregation. However, the highest concentration in this work was 0.1%, which is too low for aggregate formation. The very low aggregation concentrations found by fluorescence (Yan et al. 2010; Qiu et al. 2010; Deng et al. 2012; Li and Ouyang 2012) might be due to the definition of intramolecular interactions as “intramolecular aggregates.”