Influence of Temperature and Lignin Concentration on Formation of Colloidal Lignin Particles in Solvent-Shifting Precipitation

Abstract

Colloidal lignin particles offer a promising route towards material applications of lignin. While many parameters influencing the formation of these particles in solvent-shifting precipitation have been studied, only a small amount of research on the influence of temperature has been conducted so far, despite it being a major influence parameter in the precipitation of colloidal lignin particles. Temperature influences various other relevant properties, such as viscosity, density, and lignin solubility. This makes investigation of both temperature and lignin concentration in combination interesting. The present work investigates the precipitation at different temperatures and initial lignin concentrations, revealing that an increased mixing temperature results in smaller particle sizes, while the yield is slightly lowered. This effect was strongest at the highest lignin concentration, lowering the hydrodynamic diameter of the particles from 205 to 168 nm. Decreasing the lignin concentration resulted in significantly smaller particles (from 205 to 121 nm at 20 °C mixing temperature) but almost no change in particle yield (between 81.2 and 84.6% at 20 °C mixing temperature). This opens up possibilities for the process control and optimization of lignin precipitation.

1. Introduction

Lignin is an abundantly available biopolymer, which is currently mostly used for energy production and thus underutilized as a material. Several authors have stressed the importance of lignin valorization into high added-value products for the sustainability of biorefineries [1,2,3]. Therefore, evaluating the material uses of lignin becomes more relevant, especially considering the multiple properties and functionalities that lignin has (biodegradability, biocompatibility, UV-resistance, and low toxicity) [4]. In recent years, investigations of colloidal lignin particles (CLPs) for material applications have shown some promising results [4,5,6,7,8]. Due to the high specific surface area, lignin nanomaterials have shown improved qualities compared to bulk materials [9], which makes them interesting for applications such as sunscreens [8], food packaging [5], and emulsifiers [7], among others [4].

Organosolv pretreatment is a well-known method for the extraction of lignin suitable for high-value applications, using organic solvents such as aqueous ethanol, organic acids, or acetone [10,11]. There are several methods to produce CLPs from the extracted lignin (ideally directly from the liquor), such as pH shifting, solvent shifting, or polymerization [12]. Among these methods, solvent shifting (lignin precipitation by mixing lignin solution with an antisolvent) has been intensively investigated and has shown promising results [12] but has the downside of high solvent consumption and often low precipitation yields [13]. This leads to a low overall process efficiency, since solvents need to be recovered downstream, and substantial amounts of lignin stay in solution after precipitation. Thus, methods to improve the efficiency of CLP production need to be investigated.

In previous works [13], the influence of the antisolvent, ratio, and flowrate in the mixer were investigated, which indicated that the flow regime in the mixer has a strong impact on particle formation. The flow regime in a tube can be characterized by the Reynolds number, which is influenced by fluid viscosity and density, both temperature dependent properties. Additionally, the degree of local supersaturation is known to influence particle formation [14], which is influenced by the temperature-dependent solubility of the precipitated compound. In summary, the temperature influences several properties that are critical for the formation of colloidal lignin particles. This makes predicting the influence of temperature on particle formation difficult and requires experimental research, which has not been investigated much, despite being a major process parameter [15].

Varying the precipitation temperature also opens up interesting possibilities from a process perspective. On a laboratory scale, lignin precipitations are usually conducted at ambient temperature [13,16,17,18]. This is not necessarily representative of precipitation in a biorefinery process, since temperatures in the range of 150–210 °C are commonly applied in organosolv pretreatment [19,20]. Hence, cooling the liquor to ambient temperature after pretreatment implicates an energy demand that should be fulfilled only if necessary, which is currently a matter of uncertainty. If higher liquor temperatures still result in colloidal particles, this could be a step towards the optimization of a biorefinery process producing CLPs. Additionally, since lignin solubility increases with the temperature [21], the process efficiency could be further improved if the lignin concentration can be increased at higher temperatures. This would reduce the amount of solvent needed per the number of CLPs produced.

To summarize, the influence of the temperature in the solvent-shifting precipitation of lignin has not been sufficiently investigated so far, despite being a major process parameter with high relevance in scaled-up processes. Since the temperature influences lignin solubility, which in turn impacts process efficiency, it makes sense to investigate temperature and lignin concentration simultaneously. In this work, we therefore varied the temperature of the lignin solution and the mixer, as well as the initial lignin concentration in the precipitation of a commercial organosolv-lignin solution using a T-mixer. The resulting suspensions were characterized by particle diameter and yield to gain knowledge on the impacts on product quality and process efficiency. Selected samples were analyzed regarding their molecular weight distribution.

2. Materials and Methods

2.1. Materials

The lignin used in this work was a commercial organosolv-lignin from an annual plant supplied by ChemicalPoint (Oberhaching, Germany), which had a lignin content of 93.7 wt%, ash content of 1.6 wt%, and carbohydrate content of 0.2 wt%. For the solutions, 99.9 wt% undenatured ethanol (Chem-Lab, Zedelgem, Belgium), and ultrapure water (produced with a Sartorius arium pro system at 18 MΩ/cm²) were used.

2.2. Preparation of Lignin Solutions

Commercial organosolv-lignin and 60 wt% aqueous ethanol where mixed in a ratio of 15 g dry lignin per 1 L of aqueous ethanol and stirred at 20 °C for 24 h. After that, the liquids and solids were separated by filtration using cellulose nitrate filters (Whatman, Maidstone, Great Britain) with a pore size of 0.1 µm. The particle-free filtrate was used as lignin stock solution in the precipitation experiments. The concentration of this stock solution was 12 g/kg, which was determined by drying a sample of the stock solution in a drying oven at 105 °C until it reached a constant weight.

2.3. Precipitation

For the precipitation experiments, the lignin concentration of the solution was set to three different concentrations by volumetric addition of 60 wt% aqueous ethanol. The concentration levels used were that of the lignin stock solution (12 g/kg), a dilution to 75% (9 g/kg), and a dilution to 50% (6 g/kg) of that concentration. The lignin was precipitated in a T-mixer with ultrapure water as the antisolvent, as described by Beisl et al. [22] and depicted in Figure 1. The volumetric flow in the mixer was set to 112.5 mL/min and the volumetric ratio of extract to antisolvent was kept at 1:5 for all experiments. The temperature of the lignin solution, the antisolvent, and the mixer was controlled with water baths. Antisolvent and lignin solution were overheated to compensate for heat losses during syringe filling and pumping. The temperature was checked at the mixer inlet before the precipitation, in the bath, and in the tempered beaker where the suspension was collected. The temperature was varied over three different values, 20, 40, and 60 °C. Three precipitations were carried out for each experimental condition; the presented results are averages and standard deviations of the three repetitions. Two sets of precipitation experiments were conducted. In the first set, the temperature of lignin solution, antisolvent, and mixer was always set to the same value to facilitate a better understanding of the temperature’s influence. In the second set, only the lignin solution’s temperature was varied, while the other temperatures were always set to 20 °C, to simulate conditions closer to an industrial process.

2.4. Analytics

2.4.1. Particle Size

The hydrodynamic diameter of the lignin particles was determined with an Anton Paar Litesizer 500 (Graz, Austria). The suspensions were diluted 1:150 with ultrapure water before the measurements. The refractive index of the particles was set to 1.53 and the absorbance to 0.1. For each precipitation, the average of three measurements was calculated.

2.4.2. Yield

The particle yield of the precipitations was determined by filtering (filtrate) the suspensions through a hydrophilic polyethersulfone membrane with a 30 kDa cutoff (supplied by Nadir^®) and comparing the dry matter content of the filtrate to that of the suspensions, as shown in Equation (1). The dry matter content was determined by drying the samples in a drying oven at 105 °C until they reached a constant weight.

Particle Yield = (DMS − DMF)/DMS × 100%,

(1)

where DMS is the dry matter of the suspension and DMF is the dry matter of the particle-free filtrate. Preliminary tests showed that this filtration method leads to similar but more consistent results compared to the ultracentrifugation used in previous works [13], with lower operational and material expenditure.

2.4.3. Molecular Weight Distribution

The molecular weight distribution was determined through high-performance size exclusion chromatography (HP–SEC), with 10 mM NaOH as eluent with three TSK-Gel columns in series at 40 °C (PW5000, PW4000, PW3000; TOSOH Bioscience, Darmstadt, Germany) using an Agilent 1200 HPLC system (flow rate: 1 mL/min, DAD detection at 280 nm). The pH of liquid samples was adjusted to that of the eluent with NaOH for analysis. Polystyrene sulfonate reference standards (PSS GmbH, Mainz, Germany) with molar mass peak Maxima at 78,400 Da, 33,500 Da, 15,800 Da, 6430 Da, 1670 Da, 891 Da, and 208 Da were used for calibration.

3. Results and Discussion

3.1. First Experimental Plan: Variation of Mixing Temperature

Both the lignin concentration and the temperature were varied to three different levels (6 to 12 g/kg and 20 to 60 °C, respectively). For the first set of experiments, the temperature for the lignin solution and the antisolvent was set to the same temperature for each precipitation. As can be seen in Figure 2, increasing the mixing temperature in the lignin precipitation led to smaller particle sizes, while an increased lignin concentration increased the particle size. This means that the increase in particle size at higher lignin concentrations can be at least partially compensated for by increasing the mixing temperature. In the temperature and concentration ranges used in this work, the influence of the initial lignin concentration was stronger than that of the mixing temperature. Increasing the lignin concentration is advantageous for process efficiency, since less solvent and antisolvent are needed per the number of CLPs produced. However, this also leads to larger particles, which can be partially compensated for by increasing the mixing temperature, as the results indicate. Significantly higher lignin concentrations than those used in this study have been reported in the literature (Goldmann et al. [23] achieved a kraft lignin concentration of 235.89 g/L in 60 wt% aqueous ethanol). However, the temperature cannot be increased much higher than the maximum temperature applied in this work without potentially significantly altering the lignin molecular structure. Thus, the results indicate that the possibility to compensate for higher concentrations by higher mixing temperatures is limited. Additionally, the influence of the temperature is stronger for higher lignin concentrations, while the particle diameter stays similar for all temperatures at the lowest concentration. This indicates that the particle size converges towards a lower limit with increasing precipitation temperatures.

The increase in particle size with the increasing concentration of lignin is a well-documented phenomenon [24,25] and can be explained by a larger number of solubilized molecules available for particle growth after primary nucleation [26]; however, contrasting results have also been reported [27]. So far, little research has been conducted on the influence of temperature on lignin precipitation [15]. According to Elnashaie et al. [14], an elevated temperature tends to lead to larger particles due to it hindering primary nucleation. However, in the case examined in this work, other effects seem to have had a stronger influence, since the particle size decreased with the mixing temperature.

In this work, the particle size was only determined by dynamic light scattering (DLS), which cannot differentiate between single particles and aggregates. However, pictures obtained by scatter electron microscopy (SEM) published in previous works [13,22] showed that the particle diameters as determined by DLS are in the same order of magnitude as those determined by SEM.

A previous work [13] showed that increasing the volumetric flow in the T-mixer up to a certain point leads to decreasing particle sizes. This was explained by improved mixing at higher flowrates, which can be expressed as higher Reynolds (Re) numbers in the mixer. Therefore, a possible explanation for the decreasing particle sizes with increasing mixing temperature could be the increase in the Reynolds number in the mixer from 396 to 1017 due to the decrease in the viscosity with the temperature (from 1.6 to 0.6*10³ Pas). The correlation of particle size with Re in the mixer and the viscosity of the mixture are shown in Figure 3. The density and dynamic viscosity of the suspension were approximated using literature data from Belda et al. [28]. Due to the proportional increase in the Re with the temperature in the mixer, the correlation of particle size and Re is similar to that of particle size and temperature (Figure 3a). On the other hand, Figure 3b shows that the particle size linearly increased with the viscosity. These correlations suggest that the particle size is primarily determined by the mixing quality, which improves with an increased mixing temperature due to the decrease in viscosity. These results are in agreement with results from other works correlating smaller particle sizes with improved mixing quality [29,30].

Another relevant factor for the particle formation is the lignin supersaturation in the mixture, the driving force for precipitation. The solubility of lignin increases with increasing temperature [21], which should lower the driving force for precipitation and thus lead to higher particle diameters due to the lower precipitation speed [14]. While the lower supersaturation might still be a factor, the results indicate that the improved mixing quality outweighs its influence and leads to smaller particle diameters.

Since the experiments were carried out with ethanol–water mixtures, it is not certain whether the results can be transferred to other solvents commonly used for solvent-shifting precipitation, such as acetone [25] or tetrahydrofuran (THF) [31]. However, the increase in particle size with increasing concentration is well-established for precipitation [24,25]. As stated earlier, information from the literature on the influence of temperature on lignin precipitation is limited. Based on the explanation that the particle size decreases due to improved mixing with decreasing viscosity, it should be possible to also transfer the results to other solvents, since the same principles should apply. This could be investigated in future works to confirm or refute the explanation given in the present work.

The particle yield of the precipitations ranged from 76.7 to 85.8% (Figure 4), which can be considered high compared to the results of previous works [13]. When comparing the results from different conditions, there was a slight trend to lower yields at higher temperatures, and higher yields at higher initial lignin concentrations. The latter result matches findings by Xiong et al. [24], who precipitated enzymatic hydrolysis lignin from THF at initial lignin concentrations ranging from 0.5 to 2 mg/mL; however, the changes in the yield were small in the present work, especially considering the deviations of the experiments at the same conditions. Since the yield was determined gravimetrically, the influence of random errors was higher at lower dry matter concentrations, which explains the tendentially higher deviations for yield and filtrate dry matter at lower lignin concentrations (e.g., Figure 4).

The decreased yield at higher temperatures could be explained by the increasing solubility of lignin, leading more lignin to stay in solution. The yield was determined at ambient temperature, which would suggest that nonprecipitated lignin stays in solution even after cooling down.

Since the precipitation is assumed to be solubility-driven, it is noteworthy that there was only a slight decrease in the yields with decreasing initial lignin concentrations (Figure 4b). If the solubility limit of lignin at a certain ethanol content is assumed to be constant, the concentration of solubilized lignin after the precipitation should be constant. This would result in a direct correlation between initial lignin concentration and particle yield. For the particle yield determination, the lignin particles were removed using membrane filtration (30 kDa cutoff), and the dry matter contents of the particle-free filtrates were calculated. These dry matter contents are plotted in Figure 5, showing that the dry matter content of the filtrates from the two lower lignin concentrations was significantly lower than that of the highest concentration. This indicates that the concentration of solubilized lignin after precipitation depends on the initial lignin concentration, meaning that the amount of solubilized lignin is higher when more lignin is present.

This could be explained by the polydispersity of lignin and the different solubility limits of lignins with different molecular structures and weights, which was also found by Buranov et al. [32]. Previous studies have also shown the fractionation of lignin by molecular weight in solvent-shifting precipitation, meaning that predominately high-molecular-weight lignin precipitates while low-molecular-weight lignin predominately stays in solution [33,34]. Figure 6 shows the molecular weight distributions of lignin in filtrates from experiments at different initial lignin concentrations. The non-normalized distributions (Figure 6a) show a significant increase in lignin concentration over the whole molecular weight spectrum from the lowest initial lignin concentration. Interestingly, the distribution of the filtrates from two higher initial concentrations are very similar, despite a difference in dry matter content; the main difference is an increase in the peak at 400 Da, which was found in earlier works to be influenced by p-hydroxycinnamic acids such as ferulic acid and p-coumaric acid, both monomeric and connected to lignin fragments [35]. This suggests that the difference in residual solubilized lignin between these samples is mostly influenced by this lignin fraction.

The area-normalized distributions (Figure 6b) show that the filtrate from the lowest initial concentration has the highest ratio of lignin at 400 Da, while the ratio of the highest molecular weight fraction is very low. The distributions of the two higher initial concentrations are very similar, the main difference being a higher ratio of the 400 Da fraction for the highest initial lignin concentration.

The results suggest that the higher-molecular-weight fractions soluble at 10 wt% ethanol (ethanol content after precipitation) are completely dissolved at 6 g/kg initial lignin concentration, but hit a solubility limit at 9 g/kg. The lignin fraction at 400 Da exhibits better solubility. The increase of this fraction from 9 to 12 g/kg suggests that the concentration of this fraction is limited by the concentration in the solution at 9 g/kg, not by solubility. Generally, the results demonstrate the high complexity of lignin solubility. Therefore, the solubility of different lignin fractions will be further investigated in future works.

From a process perspective, the observation that the decrease in yield was less than proportional with the decrease in lignin concentration has interesting implications, since it would allow the tailoring of the particle size by adjusting the lignin concentration of the solution with only a minor impact on the yield. For example, when the lignin concentration of the solution is reduced to 50% at 20 °C mixing temperature, the particle size is reduced from 205 ± 4 to 121 ± 2 nm (reduction by 41%), while the yield does not change significantly. However, the merits of this depend on the requirements and value of the final product, since a higher solvent consumption lowers process efficiency.

In the context of a biorefinery process producing CLPs by precipitation directly from the organosolv extract [13], the results of the first set of experiments lead to the conclusion that cooling the extract to ambient temperature after lignin extraction is not necessary. Higher mixing temperatures still result in colloidal particles and even lower particle diameters compared to ambient mixing temperature, while the yield does not change significantly.

3.2. Second Experimental Plan—Variation of Lignin Solution Temperature

While it may be advantageous to use warm extract from a process perspective, it is disadvantageous to use warm antisolvent, since this would result in increased energy demand for the precipitation process. While warming antisolvent and lignin solution to the same temperatures for the precipitations was necessary to facilitate a better understanding of the influence of the temperature, in an industrial process, different temperatures for the solution and antisolvent are more likely. Thus, a second experimental plan was conducted, in which the temperatures of the antisolvent and the mixer were kept at 20 °C, while the temperature of the lignin solution was varied.

Figure 7 shows that there are only small differences in the particle diameter for different solution temperatures. The ratio of extract to antisolvent was kept at 1:5 for all experiments, so the mixing temperature was influenced more by the temperature of the antisolvent and the mixer than that of the lignin solution. This meant there were only small temperature changes in the temperature of the suspension after mixing in this experimental plan. It is noteworthy that the experiments with the solution warmed to 60 °C consistently resulted in the smallest average particle sizes. As in the first set of experiments, the influence of temperature was most visible at the highest lignin concentration. This supports the findings from the first experimental plan, since a higher solution temperature results in higher temperatures in the mixer. These results suggest that the temperature and properties of the mixture are relevant for the precipitation, rather than the temperature and properties of the lignin solution or antisolvent. This could mean that the size of the final particles is decided after the solution and antisolvent are mixed.

Similar to the particle sizes, the particle yield of the second experimental plan showed no strong dependency on the lignin solution temperature (Figure 8a). As in the first experimental plan, the correlation between the initial lignin concentration and the particle yield was less than proportional (Figure 8b). The results from the second experimental plan indicate that elevated temperatures of the lignin solution alone have a negligible influence on both the particle size and yield. In a biorefinery process, this would mean that it is not necessary to cool the liquor to ambient temperature after biomass pretreatment.

4. Conclusions

In the present work, the influence of the temperature and lignin concentration on the precipitation of CLPs was investigated. The particle size increased with an increasing concentration of lignin in the solution and decreased with increasing mixing temperatures. The particle yield was slightly lowered by increasing the mixing temperature, while lower initial lignin concentrations resulted in lower yields. The correlation between the particle yield and lignin concentration was underproportional, which was explained by the polydispersity of lignin and the different solubilities of different lignin fractions. In a second set of experiments, lignin solutions warmed to different temperatures were precipitated with antisolvent at ambient temperature, which had only a minor influence on the particle size and no significant influence on the yield. The lignin concentration showed the same influence on the particle size as in the first set of experiments.

From a process perspective, the results suggest that the precipitation of lignin particles directly from warm extract is not disadvantageous with respect to particle size and yield, but may in fact be advantageous for the formation of CLPs with smaller diameters. This finding could help to increase the efficiency of a biorefinery producing CLPs by removing the necessity to cool the liquor to ambient temperature after lignin extraction.

The changes in the molecular weights of lignin still dissolved after precipitation demonstrate the complexity of lignin solubility even in a relatively simple experimental plan and stress the necessity of further investigating this topic. The solubility and interaction of different lignin fractions will be investigated in future works, as well as downstreaming and valorization methods for the nonprecipitated lignin.