Indirect determination of partial depolymerization reactions in dialdehyde celluloses (DAC) by gel permeation chromatography of their oxime derivatives

Owing to a supposed quantitative transformation, oximation of dialdehyde cellulose (DAC) with hydroxylamine hydrochloride is commonly employed in chemical DAC analysis, e.g., for the determination of the degree of oxidation (DO) by titration or elemental analysis. In this study, this modification was utilized for the indirect determination of molecular weight distributions (MWD) by gel permeation chromatography (GPC). The presumably quantitative conversion of aldehyde groups in DAC to the corresponding oxime also breaks up the intermolecular and intramolecular hemiacetal crosslinks, which were associated with solubility issues in the DMAc/LiCl solvent system in previous studies. The limits of the procedure and the material's stability during oximation were investigated. For samples with a DO up to approximately 9% a good applicability was observed, before at higher DO values residual crosslinks led to solubility problems. The oximation/GPC protocol was used to examine the development of the MWD in the early stages of DAC formation under different reaction conditions. The time-dependent partial depolymerization of the polymer backbone was observed. Furthermore, the stability of DAC towards different pH conditions ranging from strongly acidic to strongly alkaline was tested. The depolymerization of DAC in alkaline media occurred with concomitant degradation of aldehyde moieties. In turn, DAC proved to be remarkably stable in acidic and neutral solutions up to a pH of 7.


Introduction
The current worldwide ambitions to limit the use of fossil-fuel-based materials have increased the research interest in more sustainable, bio-based alternatives (Heidbreder et al. 2019;MacLeod et al. 2021;Song et al. 2009).Due to its abundance and peculiar properties, cellulose was identified as a promising feedstock for this endeavour (Li et al. 2021).However, the range of potential applications of cellulose is restricted to the predetermined properties set by the repeating glucopyranose units' molecular Vol:.( 1234567890) structure (Wang et al. 2016).These can be overcome by mechanical treatments or reorganization of the native cellulose via dissolution/regeneration procedures to some extent.To significantly alter the materials' properties, covalent chemical modifications of the cellulose backbone are needed (Jedvert and Heinze 2017).Historically, the preferred reactions for that purpose were etherifications and esterifications of the alcohol functionalities, resulting in a variety of bio-based polymers with proven industrial relevance (Edgar et al. 2001).
Another long-known modification strategy, with a clear recent uptick in research interest, is the selective glycol cleavage of the glucopyranoses´ vicinal hydroxyl C2/C3 moiety by periodate oxidation (Bobbitt 1956).This reaction can be performed in water and at low temperatures, thus in principle allowing for a mild and comparably sustainable derivatization approach.Notably, there are known issues with the price and (eco)toxicity of the employed periodate oxidant (Liimatainen et al. 2013) which can, however, be overcome by the recycling protocols reported (Arndt et al. 2020;Janssen and Blijlevens 2003;Koprivica et al. 2016).The obtained material is commonly known as "dialdehyde cellulose" (DAC).It offers a rich follow-up chemistry which builds on the introduced aldehyde functionalities.Consequently, this modification strategy was exploited to produce various cellulose-based materials with potential applications in a wide area of material sciences, which were recently summarized in a thorough comprehensive review (Dalei et al. 2022).However, handling and further modification of DAC are not straightforward.For example, its aging behaviour (Munster et al. 2017) and instability towards alkaline conditions cause degradation due to beta-elimination reactions (Potthast et. al, 2009;Hosoya et al. 2018;Liu et al. 2020).Furthermore, the interconversion between the different types of masked aldehydes and the interchain and intra-chain crosslinking via hemiacetal and hemialdal linkages are well known (Simon et al. 2022a;Spedding 1960).
These peculiar properties of DAC also introduce difficulties in the analytical characterization of the obtained materials.In many cases, DACs are characterized only by the degree of oxidation (DO)-the percentage of the modified glucopyranose units in the cellulose backbone.The DO was determined by UV-Vis spectroscopy (Maekawa and Koshijima 1984), solid state CP MAS 13 C NMR (Leguy et al. 2019) and IR spectroscopy (Simon et al. 2022a) or indirectly after oximation of DAC by titration or elemental analysis (Zhao and Heindel 1991;Larsson et al. 2008).Nonetheless, the results were shown to be comparable to some extent (Simon et al. 2022a;Leguy et al. 2019).Changes in the materials' properties upon oxidation to increasing DOs have been systematically investigated.Thereby the focus was on crystallinity, thermal properties (Siller et al. 2015) or the changes in morphology (Lindh et al. 2014).However, there are still several blind spots in DAC analytics that could not be properly examined yet.One of these is the change of the degree of polymerization (DP) during DAC preparation, which represents an important marker in polymer research in general and sets the scope for many applications of cellulosederived materials.Owing to the known instability of the newly introduced moieties and the occurrence of both alkali-induced and radical side reactions, degradation during preparation was expected in the older literature (Painter 1988).However, further insights into the severity of the depolymerization or the influence of different reaction conditions could so far not be obtained.As the DP significantly influences the properties of cellulosic materials and of polymers in general, this represents a considerable analytical knowledge gap.
Notably, attempts have been reported to measure the DP of different DAC samples by adapting techniques from classic cellulose analytics.Their applicability is limited, however.Calculations of the DP from the intrinsic viscosity of DAC dissolved in cupriethylenediamine solutions were hampered by the complete destruction of the newly introduced moieties and subsequently the polymer backbone owing to instability of oxidized glucopyranose units in the highly alkaline medium (Zaccaron et al. 2020).The obtained values can instead only be used to roughly estimate the DO (Calvini et al. 2004(Calvini et al. , 2006)).Also, the commonly used DMAc/LiCl solvent system for cellulose gel permeation chromatography (GPC) is unsuitable for DACs with a higher DO since the in situ-formed hemiacetal crosslinks lead to solubility problems (Potthast et al. 2015).So far, it was only possible to measure periodate-oxidised celluloses up to a relatively low DO of approximately 3-5% before insolubility was observed due to excessive crosslinking (Potthast et al. 2009;Siller et al. 2014).Furthermore, for the soluble samples, the molecular weight distribution (MWD) was shifted to significantly higher values owing to inter-chain crosslinking (Siller et al. 2014) thereby further complicating the evaluation of depolymerization.For DAC fractions with a very high DO, a GPC and asymmetrical flow field-flow fractionation (AF4) protocol was reported utilizing their water solubility (Sulaeva et al. 2015).These isolated water-soluble fractions consist of highly disperse, compact polysaccharides which are partially cross-linked to large aggregates, differing significantly from the cellulosic starting material (Munster et al. 2017).Thus, they do not represent a proper model to follow the partial depolymerization reactions as observed in low-and medium-DO DACs.
We tried to circumvent these analytical problems by transforming the aldehyde groups masked by crosslinks to open-chain derivatives (Fig. 1).The aim was to render DAC accessible to conventional cellulose GPC analysis with the DMAc/LiCl solvent system up to higher DO values.To reliably assess the partial depolymerization reactions during DAC formation using this indirect approach, several issues must be considered: Foremost, the applied derivatization protocol must result in a quantitative transformation to assure measurement of single chains rather than crosslinked aggregates.Degradation of DAC during the follow-up reaction must be excluded.Additionally, the derivatized DAC should have solution properties similar to cellulose to avoid problems during dissolution, separation, and in the GPC detector unit.The molecular weight of the modified glucopyranose units should also not differ significantly from the cellulose repeating unit, to exclude systematic errors in the comparison of the MWDs.While all these requirements are commonly assumed to be met The measurable MWDs are shown in red and the hypothetical, non-measurable chromatogram of the unmasked dialdehyde cellulose is shown with dashed lines (the Gaussian distributions obviously do not represent the actual molecular weight distributions and are used for the sake of simplicity).Starting from a cellulose model substrate (1A) with its molecular weight distribution (2A), periodate oxidation leads to a selective cleavage of the C2/C3 vicinal diols.Besides a formal introduction of two aldehyde functionalities (1B) also partial depolymerization due to degradative side reactions can occur (2B).However, there is an immediate masking of the aldehyde moieties through hydrates, hemialdals and hemiacetals (1C), note: out of many possible structures (Munster et al. 2017) only a C2-hydrate and an intermolecular C6-C3 hemiacetal are shown.The resulting crosslinking shifts the MWD observed by GPC to higher values (2C) or even results in insolubility in the eluant.By a subsequent quantitative transformation to DAC oximes (1D), these crosslinks can be removed, and the observed chromatograms (2D) provide a reasonable approximation of the initial-although not directly accessible-openchain DAC structure and molecular weight distribution (2B) for a wide variety of DAC modifications, some previous literature (Maekawa and Koshijima 1991) and one of our recent publications focusing on reductive aminations raised some doubts in this regard (Simon et al. 2023a).The three most intensively studied derivatization strategies for DAC involve the reaction with N-nucleophiles (Maekawa and Koshijima 1991), the reduction to dialcohol cellulose (Kasai et al. 2014;Duran et al. 2016) and the further oxidation to dicarboxycellulose (Maekawa and Koshijima 1984).Among them, the modifications with N-nucleophiles allow for the supposedly mildest reaction conditions, limiting possible degradation during the derivatization.For our study, we used the DAC-oximes prepared from hydroxylamine (Fig. 1D), which are also commonly applied in determining DO values using titration techniques or elemental analysis.The reaction proceeds under slightly acidic conditions, where the risk for beta-elimination reactions is manageable (Zhao and Heindel 1991;Larsson, et al. 2008).
In this work, we showcase the indirect oximation/ GPC approach to determine the occurring partial depolymerization during the early stages of DAC formation and discuss its scope and limits.Cotton linters was used as a model cellulose substrate and the influence of reaction time, DO and periodate concentration on the MWD was analyzed.Furthermore, the degradation of both DAC and DAC-oximes at different pH values was investigated.The oximation process was monitored for up to one week to ensure quantitative transformation of the aldehyde moieties and check the stability of the polymer backbone under various reaction conditions.The analysis approach is schematically represented in Fig. 1.

Chemicals and reagents
Cotton linters was disintegrated before use in deionized water using a commercial kitchen blender.The key chromatographic parameters of the starting material, i.e., number-average molecular weight (M n ), weight-average molecular weight (M w ), Z-average molecular weight (M z ) and dispersity (Ɖ, M w /M n ), were independently determined several times and are listed in the supplementary information.All reagents and chemicals were purchased from commercial sources in the highest available quality and were used without further purification unless stated otherwise.The prepared dialdehyde celluloses were purified by thorough washing with water and did not show any signs of residual oxidizing iodine compounds in occasionally conducted colorimetric tests (Simon et al. 2023b).The buffer solutions used in the experiments (HOAc/NaOAc; citrate buffer; NH 3 /NH 4 OAc; NaH 2 PO 4 /NaOH) were prepared in house according to common protocols.The exact buffer composition is listed in the supplementary information.The buffers exhibited pH values ± 0.1 from the stated values according to control by a pH meter (SevenEasy pH, Mettler-Toledo AG).

Methods
Gel permeation chromatography (GPC) was performed according to the reported procedures (Potthast et al. 2015;Siller et al. 2014), using a size exclusion/ multi-detector (GPC-MALLS-RI) system including a MALLS detector (Wyatt Dawn DSP, Wyatt Inc., 488 nm laser) coupled with a refractive index detector (Shodex RI-71, Showa Denko K.K.), four Waters GPC columns (Styrage HMW 6E, 7.8 mm i.d., 300 mm length, 15-20 µm), one Agilent GPC guard column (PL gel, 7.8 mm i.d., 50 mm length, 20 µm) and a Bio-Inert 1260 Infinity II pump (Agilent) with automatic injection (HP Series 1100 autosampler, Agilent).N,N-Dimethylacetamide/lithium chloride (0.9%, w/v; filtered through a 0.02 µm filter) was used as the mobile phase.100 µL sample volumes were injected for each measurement with a 45 min run time at a flow rate of 1 mL/min.The MWD and the statistical moments were calculated based on a refractive index increment of 0.140 mL/g for cellulose in N,Ndimethylacetamide/lithium chloride (0.9% w/v) at 488 nm.The raw data was processed with Astra 4.73 (Wyatt Technologies) and GRAMS/AI 7.0 software (Thermo Fisher Scientific).
For sample preparation an aliquot containing approx.30 mg of the frozen, wet (H 2 O) cellulosic materials was transferred to a 20 mL screw cap bottle.After addition of DMAc (4 mL) the bottle was placed on a shaker for solvent exchange and sample activation for at least 24 h.Thereafter, DMAc was removed by filtration, before 2 mL of DMAc/LiCl solution (9% w/v) was added.The bottle was vortexed for 20 s and placed on a shaker until the solid had visually Vol.: ( 0123456789) dissolved (1 day to 1 week).For series 1b, where the samples remained partly insoluble in DMAc/LiCl (9% w/v) after longer periods, ethyl isocyanate (EIC, 0.15 mL) was added to facilitate solubilization based on a procedure by Berthold et al. (2004).The same procedure did not enhance the solubility of stronger oxidized samples, e.g., in samples 1c.The solutions were diluted 1:3 (v/v) with pure DMAc, filtered through 0.45 µm polytetrafluoroethylene (PTFE) syringe filters and stored at 4 °C until GPC analysis.The cotton linters starting material was independently measured for reference with every series of experiments conducted.
Fourier-transformation infrared (FTIR) spectra were obtained on a Frontier FTIR spectrophotometer from PerkinElmer in conjunction with the attenuated total reflection (ATR) technique.The DAC samples were dried in the open at ambient conditions for at least 48 h before measurement.The parameters for all measurements included a 4 cm −1 resolution, the 4000-650 cm −1 spectral range and 32 scans per sample.
The degree of oxidation in the DAC samples was determined according to our previous work, from the FTIR spectrum combined with partial least squares regression (Simon et al. 2022a).Our publicly provided model (Simon et al. 2022b) was used to predict the degree of oxidation from the recorded FTIR spectrum.

Experiment 1: Stability of the DAC polymer during oxime formation
Cotton linters (1 g, 6.2 mmol of anhydroglucose units (AGUs)) was disintegrated in deionized H 2 O (1 L) using a kitchen blender and filtered off.The wet cellulose was transferred to a 100 mL Schott bottle covered with aluminium foil, and a solution of NaIO 4 (1a = 0.66 g, 3.1 mmol, 0.5 eq per AGU; 1b = 1.3 g, 6.2 mmol, 1 eq per AGU; 1c = 2.6 g, 12.4 mmol, 2 eq per AGU) in H 2 O (100 mL) was added.The suspension was vortexed and placed on a laboratory shaker for 20 h at ambient temperature (20 to 25 °C).Thereafter, the solid was filtered off under vacuum, using a Buchner funnel equipped with a textile filter, and thoroughly washed with H 2 O (500 mL).
The obtained, still wet DAC was split into eight aliquots (approx.0.125 g of dry matter each) and transferred into 20 mL screw cap glass bottles.One aliquot was air-dried for DO determination using IR spectroscopy.To the seven other aliquots, 20 mL of a stock solution of NH 2 OH HCl (0.86 g, 12.4 mmol, 2.3 eq per AGU) dissolved in an acetate buffer (140 mL, pH 4, 0.5 M) was added.The suspensions were vortexed and placed on a laboratory shaking plate at ambient temperature for different times (19 to 170 h).The solids were filtered off under vacuum, using a Buchner funnel equipped with a textile filter, and thoroughly washed with H 2 O (200 mL).The collected, still slightly wet DAC-oxime samples were stored at -24 °C until GPC analysis.

Experiment 2: Influence of reaction time and NaIO 4 concentration on DAC depolymerization
Cotton linters (2.5 g, 15.4 mmol of AGU) was disintegrated in deionized H 2 O (1 L) using a kitchen blender and filtered off.The wet cellulose was split into 10 aliquots (approx.0.25 g dry matter, 1.5 mmol of AGU) and transferred into 20 mL screw cap glass bottles wrapped with aluminium foil.A NaIO 4 stock solution (2a = 1.65 g, 7.7 mmol, 0.5 eq per AGU; 2b = 3.3 g, 15.4 mmol, 1 eq per AGU; 2c = 6.6 g, 30.8 mmol, 2 eq per AGU) in H 2 O (200 mL) was prepared and 20 mL were added to each glass bottle.The suspensions were vortexed and put on a laboratory shaking plate at ambient temperature (20 to 25 °C) for different reaction times (20 min to 24 h).The remaining solids were filtered off under vacuum, using a Buchner funnel equipped with a textile filter, and thoroughly washed with H 2 O (200 mL).The collected, still wet DAC samples were split into two aliquots (approx.0.125 g dry matter) and each transferred to a 20 mL screw cap glass bottle.One part was air-dried for DO determination using IR spectroscopy.A stock solution of NH 2 OH HCl (20 mL, 2.15 g, 30.8 mmol, 4 eq per AGU) dissolved in acetate buffer (200 mL, pH 4, 0.5 M) was added to the other aliquot.The suspensions were homogenized using a vortex apparatus and placed on a laboratory shaking plate at ambient temperature (20 to 25 °C) for 24 to 36 h.The solid was filtered off under vacuum, using a Buchner funnel equipped with a textile filter, and thoroughly washed with H 2 O (200 mL).
The collected, still slightly wet DAC-oxime samples were stored at -24 °C until GPC measurement.

Experiment 3: Stability of the DAC and its oximes at different pH values (buffer solutions).
Cotton linters (20 g, 123 mmol of AGU) were disintegrated in deionized H 2 O (5 g per 1 L, 4 runs) using a kitchen blender and filtered off.In a 1 L Schott bottle wrapped with aluminium foil, NaIO 4 (6.6 g, 30.8 mmol, 0.25 eq per AGU) was dissolved in H 2 O (500 mL) before the wet cellulose was added and the bottle was filled with H 2 O to the 1 L mark.The suspension was homogenized by thorough shaking by hand and thereafter put on a laboratory shaking plate at ambient temperature (20 to 25 °C) for 23 h.Subsequently, the remaining solids were filtered off under vacuum, using a Buchner funnel equipped with a textile filter, and thoroughly washed with H 2 O (3 L).After washing, vacuum was applied to the filter unit until visually no more H 2 O was removed from the solid matter.The wet DAC was split into two aliquots (approx.10 g dry matter each).
For Experiment 3a, the prepared DAC was separated into 20 aliquots (approx.0.5 g dry matter each) and transferred into 20 mL screw cap bottles.To ten of these samples, 20 mL of different buffer solutions were added (see ESI).The suspensions were vortexed and placed on a laboratory shaking plate at ambient temperature for 22-23 h.Thereafter, the solid was filtered off under vacuum, using a Buchner funnel equipped with a textile filter, and thoroughly washed with H 2 O (250 mL).The treated DAC samples were split into two parts (approx.0.25 g dry matter each).One part was air-dried for DO determination using IR spectroscopy.The other aliquot was transferred to a 20 mL screw cap bottle and a stock solution of NH 2 OH HCl (20 mL, 2.15 g, 30.9 mmol, 1 eq per AGU) dissolved in acetate buffer (200 mL, pH 4, 0.5 M) were added.The suspensions were vortexed and placed on a laboratory shaking plate at ambient temperature (20 to 25 °C) for 24 h.The solids were filtered off under vacuum, using a Buchner funnel equipped with a textile filter, and thoroughly washed with H 2 O (200 mL).The collected, still slightly wet DAC-oxime samples were stored at -24 °C until subsequent dissolution for GPC analysis.
For Experiment 3b a solution of NH 2 OH HCl (4.3 g, 61.7 mmol, 1 eq per AGU) dissolved in a NaOAc/ HOAc buffer (500 mL, pH 4, 0.5 M) was prepared in a 1 L Schott bottle before.The second aliquot of never-dried DAC (approx.10 g dry matter) was added.Subsequently, the bottle was filled with H 2 O to the 1 L mark, thoroughly shaken by hand to homogenize the suspension and put on a laboratory shaking plate at ambient temperature (20 to 25 °C) for 22 h.Thereafter the solid content was filtered off under vacuum, using a Buchner funnel equipped with a textile filter and thoroughly washed with H 2 O (2 L).After washing, vacuum was applied until visually no more H 2 O was removed from the solid matter.The wet DAC-oxime was split into 20 aliquots (approx.0.5 g dry matter), transferred to 20 mL screw cap bottles and stored at 4 °C before further treatment.To ten of the samples, 20 mL of different buffer solutions were added (see ESI).The suspensions were vortexed and put on a laboratory shaking plate at ambient temperature (20 to 25 °C) for 22 -23 h.Thereafter, the solid content was filtered off under vacuum, using a Buchner funnel equipped with a textile filter and thoroughly washed with H 2 O (250 mL).The collected, still slightly wet DAC-oxime samples were stored at -24 °C until GPC analysis.

Oximation of dialdehyde cellulose: limits and stability for GPC applications
The applicability of the oxime derivatization for the indirect determination of MWDs of DAC samples was examined.Three DAC samples with different degrees of oxidation were prepared (1a-c, Fig. 2 and Table S1-2).The oximation of DACs was performed under the commonly applied, mild reaction conditions (pH 4, acetate buffer, room temperature, excess NH 2 OH HCl) (Zhao and Heindel 1991).The reaction times for the oximation were varied to investigate the stability and quantitative conversion of the aldehyde groups by screening for possible changes in the GPC chromatograms over time.
With determined DO values of approximately 7.1% (1a), 9.2% (1b) and 14.1% (1c), all three prepared DAC samples can be characterized as being low-to medium-oxidized specimens.Despite the relatively small differences in the DO in the derived DAC-oximes, significant differences in the dissolution behavior in the DMAc/LiCl solvent system were already observed.While all samples from group 1a gave low-viscosity solutions, 1b exhibited somewhat reluctant dissolution behavior and could only be solubilized after addition of ethyl isocyanate (EIC; Berthold 2004).Series 1c could not be dissolved at all, even after addition of EIC, thus preventing GPC analysis.Furthermore, there were differences in the chromatograms of the soluble samples 1a and 1b.The peak shape of the lower-DO derivatives (1a) resembled that of the cotton linters starting material, only shifted to lower-MW values (Fig. 2A).The 1b chromatograms showed a significant peak broadening, reflecting higher dispersity (Đ), with fractions in the high molecular weight area even exceeding the highest DP fractions of the cotton linters starting material (Fig. 2B).The differences between the two series of experiments were also discernible in the respective conformation plots, i.e., the double-logarithmic plot of the radius of gyration versus molecular weight.This correlation between hydrodynamic radius (retention time) and molar mass data from light scattering data (calibration line) of series 1a did not change significantly compared to unmodified cotton In group 1a (DO = 7.1%), a quantitative transformation was already observed after 19 h with MWDs resembling the cotton linter starting material (dashed line), although shifted to lower values.In group 1b (DO = 9.2%), even after 170 h of oximation time no quantitative transformation was observed and the presence of residual crosslinks was evident from high molecular weight fractions, exceeding the MWD of the cotton linter starting material (dashed line).Furthermore, the chromatograms showed a higher dispersity compared to 1a or the starting material.The calculated molecular mass averages (M n , M w , M z ) for 1a remained stable over the whole investigated timeframe (170 h), thus disfavoring degradation reactions of the DAC oximes in the reaction solution (C).In contrast, a slight decrease of all calculated molecular mass averages over time was observed for experimental group 1b (D) Vol:. ( 1234567890) linters (see ESI, Table S1 and Figure S1).However, the scaling parameter of the conformation plot of the oximes 1b decreased compared to the unmodified cotton linters (see ESI, Table S2 and Figure S2), indicating a more compact polymer structure in solution.A more compact structure points towards increased crosslinking between the cellulose chains, most likely stemming from hemiacetal functionalities.
Contrary to our expectations which were based on literature reports (Zhao and Heindel 1991), we observed problems in the quantitative transformation of all DAC moieties to their oxime derivatives in samples 1b and 1c.The oximation protocol had initially been developed for fully water-soluble periodateoxidized dextran.(Zhao and Heindel 1991) The heterogenous reaction conditions needed for DAC with a lower DO presumably had a negative influence on the aldehydes´ reactivity.Furthermore, there was only a slight influence of the reaction time on the MWDs.For 1a a complete conversion was observed already after overnight treatment (1a-1, t = 19 h) and no significant degradation was visible.This was evident from the almost identical MWDs and the stability of the molecular mass averages over time (Fig. 2C).In group 1b, two phases in the course of the oximation were observed.After the initial attack on the more reactive or more accessible aldehyde moieties, which enabled dissolution in DMAc/LiCl (1b-1, t = 19 h), some crosslinks prevailed.These crosslinks are evident from the broadening of the MWD and the presence of high-molecular weight fractions (Fig. 2B).Although the molecular weight of the generated oximes slightly decreased with increasing reaction duration (Fig. 2D), the dispersity remained constant.Even when treating the DAC for one week with hydroxylamine hydrochloride (1b-7, t = 170 h), the MWD indicated remaining high-molecular weight polymers in solution.Thus, the oximation was not quantitative and residual intermolecular crosslinks remained.Assuming that in all three conducted experiments the same aldehyde moieties, including their masked forms (hydrates, hemiacetals and hemialdals), were present, the different reactivity must be attributed to supramolecular phenomena.A plausible-though not experimentally validated-explanation for these observations could be the formation of strongly crosslinked sub-areas in the higher oxidized DAC samples, which are sterically inaccessible for the subsequent oximation.

Development of the degree of polymerization of cellulose during periodate oxidation
After evaluating the applicability of the oximation with hydroxylamine to indirectly analyze the molecular weight of dialdehyde celluloses, the GPC protocol was applied to investigate the development of the MWD during periodate oxidation over time at room temperature using different concentrations of NaIO 4 (2a = 0.5 eq; 2b = 1 eq; 2c = 2 eq).Thereby, the focus was on the early stages of the reaction and the correlation between reaction time, DO and DP (Figs. 3  and 4).Again, solubility issues were observed for samples with a DO higher than 8 to 9%.Samples exceeding this threshold were obtained in the longer overnight oxidations for the experiments involving 1 eq and 2 eq of NaIO 4 (2b-8 to 2b-10 and 2c-8 to 2c-10).Nonetheless, in the early stages, up to a monitored reaction time of 3 h, the DAC-oxime samples in all three experiments proved to be soluble.Thus, the GPC results well allowed for a comparison of the reaction conditions.The respective results are summarized in the ESI in Figures S3-5 and Tables S3-5.
As expected from the literature (Painter 1988), the instability of the introduced oxidized moieties led to measurable depolymerization over time in all trials.The decrease showed an exponential behavior and affected all calculated mass averages equally (M w , M n and M z ; Fig. 3A).Thus, the degradation does not seem to have a preference for high-or low-molecular weight fractions in the MWD.Especially the relatively strong effect on the DP in the early stages of the reaction was remarkable in all trials.For example, the decrease in M w in experiment 2c in the first 20 min of the reaction was 47.8 kDa.In turn, continuing the reaction for additional 160 min led to a decrease in M w of just 25.2 kDa (Fig. 3B).In the only sample set from overnight reactions, which was soluble in DMAc/LiCl (2a-8 to 2a-10), the molecular weight of the analysed oximes remained constant over time, indicating a stabilization of the degree of polymerisation.Noteworthy, there was still a measurable increase in the determined DO values, suggesting that the stop of the depolymerization was not connected to a halt of the oxidation.The measured DO values showed an increase proportional to the reaction time (Fig. 4).While this trend was evident, it should be noted that the FTIR method used for the DO determination suffers from a high standard deviation in the lower DO areas which are relevant in this study (Simon et al. 2022a).Both DP decrease and DO increase were proportional to time, which makes an interdependency of these values likely.However, since accurate quantification of very mildly oxidized DAC with the commonly applied analytical techniques is a known challenge, a quantitative evaluation of this relation was not further attempted.

Stability and stabilization of DAC under different pH conditions
In the last series of experiments, the DAC oximation/GPC protocol was employed to investigate the stability of a mildly oxidized DAC sample (DO = 5.9 ± 0.1%) towards different pH conditions.So far degradation has been reported especially for alkaline solutions following a beta-alkoxy-elimination mechanism (Potthast et al. 2009;Hosoya et al. 2018).However, for cellulose and other oxidized celluloses there are also known instabilities towards hydrolysis under acidic conditions.The acid-sensitivity and hydrolyzability of the glycosidic bond is influenced by electronic effects from substituents and oxidized groups at the adjacent glucopyranoside units.Also, for highly oxidized, water-soluble DAC samples, changes of the DP in acidic solutions were observed (Munster et al. 2017;Sulaeva et al. 2015).Acetals, hemiacetals, hemialdals-and alike masked aldehyde functions in general-get less stable at more acidic media and the interconversion between those species in dynamic equilibria becomes faster.Thus, we were interested in benchmarking the stability of DAC with regard to the pH conditions.Aliquots of one sample were treated in 10 different pH solutions from strongly acidic to strongly alkaline, set and maintained by buffer solutions (see ESI for details) for 24 h, before transformation to the DAC-oxime and GPC analysis.
In DAC chemistry, information on the degradation behavior is essential to choose proper reaction conditions to avoid-or at least limit-depolymerization.Certain chemicals in frequently applied reactions may require harsher reaction conditions the effects of which on the DAC may not be clear from the outset.For example, NaBH 4 used in the reduction to dialcohol cellulose (Kasai et al. 2014;Duran et al 2016) or for reductive aminations (Simon et al. 2023a) requires a pH of approx.10 to avoid a rapid side reaction of the reductant with the solvent under H 2 formation.However, for the same pH ranges also the degradation of DAC was reported, employed for instance for the isolation of cellulose nanocrystals from DAC (Liu et al. 2020).Thus, depolymerization is likely to occur also during DAC reductions-although seldom considered in the literature.
In our experiments with DAC (see ESI Figure S6 and Table S6), no significant depolymerization was observed up to a pH of 7 (Fig. 5).Only for the sample treated with 0.5 M HCl (pH < 0) a slight broadening of the MWD was observed.Similarly, the determined DO values remained stable in the acidic and neutral pH regions.By contrast, the samples treated with alkaline solutions exhibited significant depolymerization, which concurred with a significant decline in the determined DO values (Table S6).Concomitantly, an evident change in the morphology of the samples from cotton-like to an appearance reminding of cellulose nanocrystals occurred (Liu et al. 2020).There was a difference in the results of the samples treated with ammonia-containing solutions and the samples treated with a 0.5 M Na 2 HPO 4 / NaOH buffer or 0.5 M NaOH.While the dialdehyde moieties were quantitatively destroyed in the latter according to FTIR (Table S6), a small amount remained in the ammonia-treated DAC samples.As the pH 11 buffer and the 0.5 M ammonia solution had similar pH values, apparently the degradation was also affected by the employed base.Although we did not investigate this difference any further, a stabilization of the DAC moieties by intermediately formed imine structures seems plausible, because beta-elimination from aldehyde (carbonyl) structures is much faster than from their N-heteroanalogous (imine, oxime, hydrazone) derivatives.
Additionally, the DAC-oximes 3b (prepared from the DAC sample used in experiments 3a) were examined regarding their stability against different pH solutions.Besides altering the properties of DAC, derivatives thereof are also known to stabilize the material towards degradation reactions.The Fig. 5 Comparison of the molecular weight average (M w ) of DAC (3a; DO = 5.9%) and the derived DAC-oxime (3b) after treatment in different solutions.Both materials proved to be stable in acidic to neutral conditions up to a pH of 7 and showed strong degradation in strongly alkaline solutions.
Owing to the known exchange reaction of oximes with water leading to the reformation of crosslinks, M w values for series 3b were shifted to slightly higher values.For the same reason no significant stabilization of the DAC-oximes in highly alkaline solutions was observed Vol.: (0123456789) modification reduces the acidity of the hydrogen in beta-position to the carbonyl moiety and largely prevents the associated beta-elimination.
In the conducted series of experiments (see ESI Figure S7 and Table S7) the expected stabilization was not evident from the MWDs.In essence, the same behavior as for the untreated DAC was observed.There were no significant changes in the acidic and neutral pH media and strong depolymerization in the alkaline solutions occurred.Compared to experiments 3a (DAC), only a slight stabilization could be achieved through the oxime derivatization, indicated by the statistical moments.In contrast to the previous series, the morphology did not change visually for the samples treated with bases (3b-7-3b-10).Instead, yellowing was observed.Obviously, the investigated oxime modification proved to be insufficiently permanent for the stabilization study, owing to the occurrence of oxime hydrolysis / exchange reactions (Subbotina et al 2022).While oximes are generally relatively stable DAC-derivatives, in the presence of water they are in equilibrium with the free aldehydes.Thus, in the presence of strong bases, the alkali-induced degradation of the released aldehyde moieties continuously withdrew them from the equilibrium, eventually resulting in a depolymerization similar to the non-derivatized DAC.Owing to the high sensitivity of the GPC protocol the exchange became already apparent after storage of the DACoximes in the wet state at 4 °C and in the samples treated in acidic or neutral solutions.Compared to the untreated DAC-oxime the treated samples showed GPC statistical moments shifted to higher values and molecular weight fractions higher than in the starting material (Fig. 5).Noteworthy, the samples treated with ammonia-containing solutions again exhibited an unusual behavior.The MWDs narrowed remarkably (Đ = 1.2) due to still unknown exchange and stabilization phenomena caused by the ammonia or formed imines.For comparison, the cotton linter starting material and the untreated DAC-oximes exhibited dispersities ranging from approximately 1.5 to 1.7 (see ESI Figure S7 and Table S7).

Conclusions and outlook
By performing the well-studied conversion of DAC to its oxime derivatives, insolubility issues owing to intermolecular crosslinks-preventing GPC analysis of DAC-were overcome.Consequently, we obtained deeper insights into depolymerization reactions during periodate oxidation of cotton linters as a cellulose model compound.As expected according to older DAC literature, cellulose degradation upon oxidation turned out to be quite severe, with a reduction of the molecular weight averages by approximately 50% already at rather low degrees of oxidations of 6 to 7%.This DO range is widely considered to be obtainable under mild conditions and to be restricted to the surface.Furthermore, the depolymerization was found to be proportional to time and periodate concentration.Although not further investigated in this study, the depolymerization also influences the properties of DAC-derived materials.This might be especially a problem when applying periodate oxidation to nanocellulosic substrates, since their inherent strong mechanical properties are sought to be preserved throughout chemical modifications (Heise et al. 2021, Heise et al. 2022).
While the oxime modification did not lead to any considerable degradation even at prolonged reaction times, unexpected problems with the quantitative conversion of the aldehyde groups to the oximes were encountered, even for DAC samples with relatively low degrees of oxidation (starting at DO values around 9%). Future investigations will resume this issue and focus on an optimization of the oximation conditions to allow for an extension of the protocol to higher DO values.Notably, this study was conducted on a GPC-system optimized for analysis of neat celluloses.Also, adaption and optimization of the GPC setup might render higher-DO specimens accessible to analysis.
However, when the experiments are systematically conducted, the protocol provides new insights into the development of the MWD of DAC.In this contribution, we highlighted the depolymerization during DAC formation and the (in)stability in different pH solutions.Nonetheless, there are several aspects and variables influencing the degree of polymerization of DAC that still need to be addressed.We especially hope that an improved and more general protocol for DP analysis in DAC research will lead to the development of strategies limiting or even preventing its unwanted depolymerization reactions.

Fig. 1
Fig. 1 Schematic of the conducted chemical modifications (1A-D) and the associated development of the MWD (2A-D).The measurable MWDs are shown in red and the hypothetical, non-measurable chromatogram of the unmasked dialdehyde cellulose is shown with dashed lines (the Gaussian distributions obviously do not represent the actual molecular weight distributions and are used for the sake of simplicity).Starting from a cellulose model substrate (1A) with its molecular weight distribution (2A), periodate oxidation leads to a selective cleavage of the C2/C3 vicinal diols.Besides a formal introduction of two aldehyde functionalities (1B) also partial depolymerization due to degradative side reactions can occur

Fig. 2
Fig. 2 Molecular weight distributions (A, B) according to GPC for the long-time DAC oximation experiments 1a and 1b.In group 1a (DO = 7.1%), a quantitative transformation was already observed after 19 h with MWDs resembling the cotton linter starting material (dashed line), although shifted to lower values.In group 1b (DO = 9.2%), even after 170 h of oximation time no quantitative transformation was observed and the presence of residual crosslinks was evident from high molecular weight fractions, exceeding the MWD of the cotton linter

Fig. 3 A
Fig. 3 A Course of the GPC statistical moments for the timedependent periodate oxidation of cellulose with 0.5 equivalents of NaIO 4 .All three calculated averages (M n , M w , M z ) show an exponential decrease before reaching a plateau after longer reaction times.B Comparison between the molecular weight