*2.5. Compression Testing*

Uniaxial compression testing and evaluation of the respective stress–strain relationships confirmed the typical compression behavior of lightweight cellulose II aerogels as discussed elsewhere [32]. Besides ductility, low rigidity as expressed by Young's modulus (*E* ≤ 14.5 MPa), and absence of sample buckling (zero Poisson ratio), the stress-strain curves exhibit a pronounced plateau region (10–50% strain). In this region, compression energy is dissipated by gradual collapsing of the cellular structure. Beyond that plateau, strain hardening sets in. Recently, it has been shown for aerogels from nanofibrillated 2,3-dicarboxyl cellulose that strain hardening can coincide with pore size harmonization in favor of mesopores giving access towards superinsulating aerogels [9]. Interestingly, both Young's modulus and density-normalized specific Young's modulus of the CL aerogel were higher than that of comparable CL aerogels obtained using NMMO·H2O or [EMIm][OAc]/DMSO as cellulose solvent (Table 3; [32]). Phosphorylation, at least at the envisaged low degrees of substitution, had no clear impact on the anyway low values of Young's modulus *E*, yield stress (σy) and yield strain (εy). While *E*ρ and εy slightly increased with phosphorylation for the aerogels derived from cotton linters, it was the reverse for the hwPHK samples.

**Table 3.** Young's modulus (*E*), specific modulus (*E*ρ), yield strength (σy), and yield stress (εy) as obtained from uniaxial compression testing of the prepared aerogels.


a data taken from [15], b not determined.

Based on the apparent density (ρA) of the aerogels and assuming a skeletal cellulose density of ρS = 1.56 g cm<sup>−</sup>3, porosity (Π) values ranging between 95.46 and 96.99% were calculated using the following equation: Π (%) = 1 – ρA / ρS (Table 4). At a first glance, these values combined with the information of the SEM pictures (Figure 3) could be perceived as indicative for the presence of very low specific surfaces only in all prepared aerogels. However, nitrogen sorption experiments at 77 K provided a different picture. Evaluation of the isotherms in the low relative pressure range (p/p<sup>0</sup> = 0.05–0.2) of the adsorption branches using the Brunauer–Emmett–Teller (BET) approach revealed a considerable monolayer nitrogen adsorption. It corresponds to about 350–370 m<sup>2</sup> g<sup>−</sup><sup>1</sup> which is almost as high as reported for anisotropic mesoporous aerogels of narrow size distribution [9]. These relatively high values are indicative of the presence of a well-developed nanoporous substructure not visible in the micrographs of Figure 3. Somewhat lower specific surface values were calculated for both types of phosphorylated cellulose aerogels (Table 4). However, these results require careful treatment since gas sorption in aerogels of multiscale porosity depends on many factors. It also includes the C factor of the BET equation. Its significance is low since the C factor—in the strict sense—is a measure of interaction between a non-porous surface and adsorbent molecules [48]. Assuming largely similar morphology, however, the significantly lower C values obtained for the phosphorylated samples can be interpreted as a considerable drop in interaction due to the introduction of the polar phosphate moieties.


**Table 4.** Calculated porosity, results of nitrogen sorption experiments and pore characteristics derived from thermoporosimetry. Numbers in brackets indicate the 95% confidence interval (*n* = 3).

All sorption isotherms confirmed the presence of a considerable volume fraction of mesopores as strongly evident from their IUPAC type IV shape [49]. The latter is characterized by slow monolayer adsorption at relatively low relative pressure (p/p<sup>0</sup> ≤ 60 kPa), but higher ad- and desorption rates beyond that (0.07 ≤ p/p<sup>0</sup> ≤ 0.1 MPa).

Due to the fragility of lightweight cellulosic aerogels, pore size distributions representing the true porosity are methodologically difficult to obtain, in particular for aerogels of multiscale pore size distribution. While mercury intrusion is not applicable (the high specific density causes pore collapsing), other methods are limited to a certain range of pore size. Evaluation of data points taken from the desorption branches of the nitrogen sorption isotherms using the Barrett–Joyner–Halenda model (BJH) suggested the presence of mesoporous domains. The latter are characterized by a narrow pore size distribution peaking at around 9–11 nm. They seem to exist in all studied materials. The results of the BJH calculations, however, also indicate the presence of larger pores. This is in agreemen<sup>t</sup> with the transition of the two branches of the adsorption isotherms at p/p<sup>0</sup> = 0.1 MPa and the SEM micrographs. Thermoporosimetry studies confirmed both the presence of mesopores and coexistence of significantly larger pores (Figure 4). All pore-size distributions have their maximum at about 50 nm diameter. The results of the nitrogen sorption experiments, thermoporosimetry data, and SEM micrographs sugges<sup>t</sup> that phosphorylation at the studied low DSP values has only a marginal impact on morphological features. This includes specific surface and pore-size distribution.

**Figure 4.** Normalized pore-size distributions as obtained by thermoporosimetry.

### *2.6. Aerogel Stability and Moisture Sorption during Storage*

The abundance of hydroxyl groups renders cellulosic surfaces very sensitive towards moisture sorption. While this feature is highly desired in some applications like wound dressings [15], it is the opposite for lightweight cellulose aerogels. Their high volume fractions of interconnected nanopores exceeding sometimes 99 vol.% render them very fragile, in particular in moist environments. Capillary condensation is a phenomenon that occurs in hydrophobic materials as well, but is amplified by huge

internal surfaces of high hydrophilicity. Water condensation in capillaries give rise to the formation of strong inward forces alongside a capillary gradient adjacent to the water menisci formed. These forces are caused by differences in specific energies of the involved phases. Their strength is reversely correlated with pore radii, which renders open-nanoporous, hydrophilic and soft materials particularly sensitive towards shrinkage. Considering both the anhydrous conditions in the last stages of solvent exchange and scCO2 drying, as well as the hydrophilicity of the studied cellulosic materials, it was assumed that the aerogels would gain weight immediately once exposed to air. However, this was not the case. It turned out that the aerogels prepared from the non-derivatized source materials lost about 5–7 wt.% instead within 60 min after scCO2 drying (Figure 5). This is surprising at a first glance and needed a more thorough investigation since this weight loss could be explained solely by degassing of CO2 and replacement by air. Considering a bulk density of 71.2 mg cm<sup>−</sup><sup>3</sup> for the CL aerogels for example and a cellulose skeletal density of 1.56 g cm<sup>−</sup>3, the corresponding pore volume would be about 0.95 cm<sup>−</sup>3. Assuming ideal gas behavior, standard conditions and occupation of the entire pore volume by CO2, its quantitative replacement by dry air (80% N2 and 20% O2) would cause a weight loss of ca. 1.22 mg. This would be equivalent to 1.7% loss related to the bulk density calculated from weight and spatial dimensions shortly after scCO2 drying. Even if water sorption would have caused the difference between measured bulk density (71.2 mg cm<sup>−</sup>3) and theoretical density (55.25 mg cm<sup>−</sup>3)—calculated from the original cellulose content of the CL gel (3 wt.%) and the volume loss throughout aerogel preparation (45.7 vol.%)—the maximum possible weight loss would be 2.21 wt.% only. Therefore, it is reasonable to conclude that small quantities of ethanol were still present in the scCO2 dried aerogels, which is in agreemen<sup>t</sup> with olfactory impressions. Headspace-GC/MS investigations confirmed this assumption. They revealed that—depending on the cellulosic material processed into aerogels—considerable amounts of ethanol can remain after scCO2 drying. While scCO2 extracted virtually all ethanol from the pores of bacterial cellulose aerogels of similar sample size and shape within one hour drying time, more than the ten-fold amount of ethanol remained for the significantly denser CL aerogels. Phosphorylation, even at low degree of substitution, seems to disfavor ethanol sorption as evident from the slight weight losses after scCO2 drying.

**Figure 5.** Weight [%] of the materials recorded over 60 min after scCO2 drying (left). Release of ethanol from freshly scCO2 dried cotton linters aerogels (CL) and bacterial cellulose aerogels (BC) under headspace (80 ◦C) GC-MS conditions (right).

Long-term dimensional stability and moisture sorption of all aerogels were studied for a time period of 84 days and controlled levels of relative humidity. Weight gain and shrinkage values revealed that moisture uptake occurs for all studied aerogels mainly during the first 24 h (data not shown) and is largely completed after 48 h. This was independent of cellulose type, degree of phosphorylation, and relative humidity (Table 5) except for the highest value of 98%RH. Here, water sorption increased by 2.2 wt.% for the CL sample from 23.6% after 2 days to 25.8% after 84 days. The amount of water adsorbed was confirmed to depend on the air moisture content. While at 30% RH, water uptake was

negligible for all samples, it was moderate at 65% RH (5.0–6.6%) but strong at 98% RH (23.6–24.2%). Corresponding to the amounts of adsorbed water, all samples suffered from significant shrinkage, specifically within the first two days of storage. Here, the extent of shrinkage increased strongly in the order 30%RH (24–28%) < 65%RH (64–73%) < 98%RH (90–93%). Storage in anhydrous environment as accomplished by placing the samples in a closed compartment over P4O10 is not recommended. Removal of structural water and hornification cause the aerogels to shrink at an extent comparable to that observed for 30% RH.

**Table 5.** Volume and weight changes of studied aerogels in dependence on relative humidity and storage time (% of initial mass and volume right after scCO2 drying). Numbers in brackets indicate the 95% confidence interval (*n* = 4).


As a result of water uptake and shrinkage, the bulk densities of the prepared aerogels increased as well. While it was to a minor extent only when storing them at 0% and 30% RH, three- (65% RH) to 17-fold (98% RH) higher values were obtained for the more humid environments. Phosphorylation rendered the aerogels somewhat more sensitive towards moisture sorption and shrinkage.
