Aqueous synthesis of highly functional, hydrophobic, and chemically recyclable cellulose nanomaterials through oxime ligation

Cellulose nanofibril (CNF) materials are candidates for the sustainable development of high mechanical performance nanomaterials. Due to inherent hydrophilicity and limited functionality range, most applications require chemical modification of CNF. However, targeted transformations directly on CNF are cumbersome due to the propensity of CNF to aggregate in non-aqueous solvents at high concentrations, complicating the choice of suitable reagents and requiring tedious separations of the final product. This work addresses this challenge by developing a general, entirely water-based, and experimentally simple methodology for functionalizing CNF, providing aliphatic, allylic, propargylic, azobenzylic, and substituted benzylic functional groups. The first step is NaIO4 oxidation to dialdehyde-CNF in the wet cake state, followed by oxime ligation with O-substituted hydroxylamines. The increased hydrolytic stability of oximes removes the need for reductive stabilization as often required for the analogous imines where aldehyde groups react with amines in water. Overall, the process provides a tailored degree of nanofibril functionalization (2–4.5 mmol/g) with the possible reversible detachment of the functionality under mildly acidic conditions, resulting in the reformation of dialdehyde CNF. The modified CNF materials were assessed for potential applications in green electronics and triboelectric nanogenerators.

In this work, the reversible formation of oximes is proposed as a new functionalization of cellulose nanofibers. Though end group functionalization of cellulose and nanocellulose is a pretty wellknown topic, the authors investigate several surface functionalizations by using different commercial or noncommercial hydroxylamines and demonstrate the generality of the proposed reaction approach, its viability and effectiveness. I think that from the "chemical" point of view the work is sound and the results are clear. In my opinion, the work is interesting but needs revision to meet the requirements of novelty and urgency to be published in Nature Communications.
-First of all, hydroxylamine binding is a method for titrating the available aldehyde groups. This somehow decreases the novelty of the work. Then, the introduction of dialdehyde groups onto surface anhydro glucose units in cellulose nanofibers is not new too. This work makes an excellent presentation of several surface functionalization of CNFs by the proposed reaction method. Further discussion is needed to support the true advantage of the reaction with hydroxylamines: are they effective where other methods failed?
-Relying on NaIO4 oxidation, the authors introduce a high amount of aldehyde groups on the CNFs. ~4.5 mmol per g of CNF means 4.5 mmol aldehyde groups/6.2 mmol of cellulose are there, so 36% of AGU units have been oxidized. It is quite a high degree of oxidation. CNFs original structure is not preserved. I do not understand the usefulness of such an extensive skeleton functionalization. In the introduction, the authors declare that the challenges identified in this work include: "preserving the nanostructure of CNF, enabling high accessibility of the functional groups, and overcoming moisture sensitivity of the transformation". Indeed, it is accepted that a low degree of substitution/functionalization (DS<0.37) will preserve the nanomorphology of nanocellulose (D.-Y. Kim, Y. Nishiyama, and S. Kuga, "Surface acetylation of bacterial cellulose," Cellulose, vol. 9, no. 3-4, pp. 361-367, 2002). I think the authors should discuss more and perform the necessary characterizations also on samples with lower oxidation extent, but more preserved cristallinity.
-Microscopy investigation of the functionalized samples needs to be performed, either by AFM or scanning electron microscopy.
-Reversibility: how is the appearance of fibers after the removal of oximes? -How do the authors know that CNFs are composed of 18 cellulose chains? Are there experimental data to support this statement? -The first scheme of supporting information is wrong: the correct product is hydroxylamine hydrochloride, not amine hydrochloride.
-Please explain the following sentence: "In the case of more rigid hydrophobic substituents (e.g., R1-R5, R10), the R-ON-CNF wet cake stiffens during the progression of the reaction. In contrast, Me-and pentyl-functionalized DA-CNF wet cakes stayed soft and flexible." -"Whether the difunctional hydroxylamine reacted with one or two aldehydes was assessed by using triflates as a counterion instead of halides." It would be much more reliable to look at the bifunctional reagents by XPS technique.
-The authors decided to study kinetics of oxime removal following CF3-phenyl group dissolution by NMR spectroscopy. They also report an example with a pentafluorobenzene appendage on oximes. In my experience these rings are susceptible to nucleophile attack, did the authors check their water stability? Reviewer #2 (Remarks to the Author): This is an interesting paper, showing the preparation and characterization of DA-CNF and various R-ON-CNFs, and application of R-ON-CNFs to triboelectric generators, from some new aspects. Therefore, this manuscript can be publication in Nature Communications after revisions. 1) Since the R-ON-CNFs still contained some aldehyde groups, not only DA-CNF films but also R-ON-CNF films may be discolored after drying at 93C. This discoloration of aldehyde-containing films should be addressed in terms of aldehyde content.
2) It may be difficult in this study, but the yields (based on the starting dry CNF weight) and molecular weights of DA-CNFs prepared under various conditions should be measured or discussed. DA-CNFs may partly decrease in the yield by oxidation and washing processes.
3) Not only the conversion ratio from aldehyde groups to oximes (i.e., 73% at 1.5 eq. of Bn-ONH3+Cl-in page 5) but also the efficiencies of oxime formation (i.e., oximed amine/added amine by mol) should be added in Table 1).
4) It is better also to describe the reaction efficiency of NaIO4 in the DA-CNF preparation. Namely, the molar ratio of NaIO4 consumed for aldehyde formation/that added in the oxidation. 5) Page 7: The reaction conditions in acidic acetone/water at 55 degree C for 24 h may be harsh for R-ON-CNF. The low yield losses (<1%) were described in page 7, but some depolymerization of oxidized cellulose molecules may be unavoidable during de-oxime reactions, which should be addressed. 6) Figure 1: Since C.I decreased to almost zero after oxidation for 7.5 h, did this oxidation occur also in the crystalline inside CNFs without reaction selectively on the CNF surfaces? 7) Figure 2, green spectrum: Is it true that the C-H peak is located at ~3250 cm-1? 8) Figure 3b: The changes in water uptakes were consistent with water contact angles on the films? 9) Figure 3a, S4, S5, and S6: It is better to include photographs of the films in this manuscript to show the colors or transparencies. 10) Figure S6: Why the films have such high haze values? Many pores with air/material interfaces are present in the films? What are the factors influencing the haze values of the films? 11) Figure S1: What is the signal at ~50 ppm owing to? 12) Figure S1 and Figure 2c: Since the signal pattern at 60-110 ppm in Figure S1 is similar to that in Figure 2C, this sample may have significant amounts of aldehyde groups (or the oximation ratio may be too low). Please explain this matter. Or it is better to show another C13-NMR spectrum of R-ON-CNF with a much higher degree of oximation. 13) Signal assignments should be added to Figure S3 and all 1H-and 13C-NMR spectra in Appendix. The solvents used in the 1H-and 13C-NMR measurement are also better to be described in the figures.
Reviewer #3 (Remarks to the Author): The present manuscript with the title "In water synthesis of highly functional, hydrophobic, and chemically recyclable cellulose nanomaterials through oxime ligation" describes an interesting study on the chemical modification of nanostructured cellulosic materials. The manuscript is well structured, the work concisely described, the experimental procedures can be repeated based on the protocols and the illustrations are of satisfying quality.
Nevertheless, I believe that the work is not suited for publication in Nature Comm. There are two main reasons. First, the element of novelty, a prerequisite to publication, is largely missing. As the authors state themselves, periodate oxidation of carbohydrate (the very old Malaprade oxidation) is very well-known and has been extensively applied also to (nano)cellulosic materials. In most cases, those studies did not target the "dialdehyde cellulose" directly, but used it as an intermediate for further functionalization. Schiff bases as well as non-substituted and substituted hydrazones and oximes have been synthesized from dialdehyde cellulose and other dialdehyde polysaccharides. Despite placing the emphasis more on the properties of the products, the manuscript does not offer any novel aspects here. Second, there are several aspects of the work which seem to indicated the studies have not yet fully matured (see below), and there are several claims which seem a bit exaggerated.
-line 3 (title): the products are certainly less hydrophilic than cellulose, but to call them hydrophobic would be quite an exaggeration (cf. for instance to hydrocarbons). The product still possesses a wealth of hydrophilic functions , and in case of the hydrochlorides or other salts are even ionic. -line 89: the statement regarding atom economy would require a solid support by numbers to demonstrate that the proposed periodate oxidation / oxime approach is superior to the alternatives -line 91: the statement of "harsh conditions" needed for functional group / substituent removal is not fully conclusive: a 0.1 M NaOH used for removal of acetyl/acyl groups is not harsher than the 1 M HCl used in the present work to cleave the oximes. -line 105: this statement is not supported by a literature survey: the "dialdehyde" structure obtained by periodate oxidation is in very many cases used to have a reactive moiety for further modification (mostly by reacting with N-or C-nucleophiles). This further modification of the aldehydes is certainly not "less common". -line 112: current reduction methods rather use aminoboranes which are selective towards imines and leave the aldehydes/hemiacetals unaffected. -line 114: the evolution of hydrogen gas is strongly depending on the conditions and can be largely avoided by deeper temperatures, higher pH, alcoholic co-solvents, etc. -line 122: it is correct that for oxime cleavage often ketone or aldehyde traps are used to scavenge the release hydroxylamine. However, these traps are bound to an appreciable extent by the dialdehyde polysaccharides. If this occurs by hemiacetal bonds, the reaction is reversible and the binding temporary, if covalent bonds are formed (acid-catalyzed aldol), the scavenger is consumed and the (surface) structure and chemistry of the polysaccharide irreversibly altered. This chemistry thus needs much further work to become fully reliable. -line 143: the amount of acetal linkages in rather low. Moreover, if they exist they would not react in the proposed way with hydroxylamines. -line 157: the use and fate of the "CNF wet-cake" is not fully clear. Why would it not disintegrate in the aqueous surroundings? Is the contact time with the periodate reagent (and the resulting extent of oxidation) everywhere in the cake the same? -line 183: the problem with increasing iodate concentrations is also the increasing instability of periodate (of which the mechanism is unknown) -lines 209-212: the reason for the selection of these substituents remains somewhat unclear. It appears as if just the commercially available O-substituted hydroxylamines were tested. This demonstrates the general applicability of the procedure, but would not support claims that there was any "targeted" selection.
-line 220/387 (e.g.): the reagents should be called O-substituted hydroxylamines, but not "ether of hydroxyl amine" -lines 270/276/281-282: the "recycling of CNF" is quite an exaggeration. First, after cleavage, the product is not cellulose anymore, but periodate oxidized cellulose -the second modification step, the oxime formation, might be reversible, the first one, periodate oxidation, is not. Second, the conditions required for oxime cleavage (1M HCl, hydroxylamine scavengers), are rather detrimental to cellulose integrity. The acidity causes hydrolytic cleavage, which is already well noticeable at that pH and reaction times, and the scavenger might react with the dialdehyde cellulose and modify it (see above). -line 396: is such a high degree of oxidation / modification really compatible with the claim that only the surface of the CNF is modified? -line 410: the experiments used holocellulose, which contains certain amounts of hemicelluloses which are more reactive than cellulose in periodate oxidations. This aspect is not considered, but it might imply that a certain amount of modified CNF is actually modified hemicellulose.

Reviewer #1 In this work, the reversible formation of oximes is proposed as a new functionalization of cellulose nanofibers. Though end group functionalization of cellulose and nanocellulose is a pretty wellknown topic, the authors investigate several surface functionalizations by using different commercial or noncommercial hydroxylamines and demonstrate the generality of the proposed reaction approach, its viability and effectiveness. I think that from the "chemical" point of view the work is sound and the results are clear. In my opinion, the work is interesting but needs revision to meet the requirements of novelty and urgency to be published in Nature Communications.
Comments:

First of all, hydroxylamine binding is a method for titrating the available aldehyde groups. This somehow decreases the novelty of the work. Then, the introduction of dialdehyde groups onto surface anhydro glucose units in cellulose nanofibers is not new too. This work makes an excellent presentation of several surface functionalization of CNFs by the proposed reaction method. Further discussion is needed to support the true advantage of the reaction with hydroxylamines: are they effective where other methods failed?
Response: We agree that the novelty and advantages of the developed method were not adequately highlighted in the old version of the manuscript. Indeed, oxidation of cellulose with NaIO4 is a known and powerful method to perform the functionalization of cellulose fibers with aldehyde groups, opening vast opportunities for further modifications. However, it is hard to adopt the developed protocols directly translated to nanocellulosic substrates, especially on colloidally stable and highly charged cellulose nanofibers, such as TEMPO-oxidized and carboxymethylated CNF. In fact, we found only two works on NaIO4 oxidation of charged CNF. In one of the works, the authors performed oxidation of carboxymethylated CNF in a dilute suspension and purified the product via dialysis lasting for 3 days (Cellulose (2017) 24:3883-3899). In another work, oxidation was performed on TEMPO-CNF hydrogel, isolated from the oxidation via centrifugation. It was used as a paper additive without characterizing the structural properties of oxidized nanofibers (Carbohydrate Polymers 250 (2020) 116941). To the best of our knowledge, the reported method is the first example of a simple, fast, and tunable oxidation of colloidal stable CNF. We managed to avoid cellulose aggregation and tedious separation by 1. using cellulose nanofibers containing hemicellulose (holo-CNF) 2. conducting the reaction on preformed holo-CNF wet cakes. In holo-CNF, cellulose nanofibers are covered by hemicellulose chains, which allows for efficient defibrillation with a small number of negative charges on the nanofibers' surface (e.g., TEMPO-CNF). The high content of negative charges during periodate oxidation slows down the oxidation kinetics, presumably via electrostatic repulsion of IO4anions (which we refer to in the manuscript: "Using TEMPO-CNF in this transformation is impractical due to low reaction rate. The reason is repulsion of the IO4-anion by the negatively charged carboxyls at TEMPO-CNF surfaces."observed when trying to perform the reaction on TEMPO-CNF wet cake.
While hydroxylamine is used as a titrating agent to determine the number of aldehyde groups in dialdehyde celluloses, O-substituted hydroxylamines have not been implemented as functionalization reagents for cellulose materials. Generally utilized methods rely on the functionalization of oxidized cellulose with amines coupled with in situ reductions of the formed imines to secondary amines for hydrolytic stability. This work uses the increased hydrolytic stability of oximes vs. imines to perform direct functionalization without a reduction step while keeping the possibility for post-life defunctionalization. Moreover, using salts of O-substituted hydroxylamines allowed us to install even highly hydrophobic functionalities on oxidized CNF in water, which would be difficult to achieve otherwise. Other water-based systems for functionalization of CNF that are relevant to include silylation and amide coupling. A main disadvantage of amide coupling is the use of stoichiometric amounts of coupling agents; and the formation of a stable amide bond, making the transformation irreversible. There are also several concerns regarding silylation, which can be performed in water. Trialkyl silanes are prone to hydrolysis and oligomerization, which result in less defined functionalization, where oligomeric siloxanes can be either covalently attached or absorbed on the surface of cellulosic material. In addition, due to several reactive sites on trialkylsilanes, it is challenging to control cross-linking (ChemSusChem,2015,8,2681-2690, Journal of Applied polymer science, 2012. Moreover, the defunctionalization of the materials obtained by the above methods was not studied.
The presented method is extremely experimentally simple, where CNF wet-cake is directly placed in a water solution of a corresponding reagent, allowing for easy isolation of the modified material and direct reuse of the solution containing unreacted reagents.
We introduced substantial changes in the abstract and in the introduction of the manuscript to stress the novelty and advantageousness of the developed process.

Relying on NaIO4 oxidation, the authors introduce a high amount of aldehyde groups on the CNFs ~4.5 mmol per g of CNF means 4.5 mmol aldehyde groups/6.2 mmol of cellulose are there, so 36% of AGU units have been oxidized. It is quite a high degree of oxidation. CNFs original structure is not preserved. I do not understand the usefulness of such an extensive skeleton functionalization. In the introduction, the authors declare that the challenges identified in this work include: "preserving the nanostructure of CNF, enabling high accessibility of the functional groups, and overcoming moisture sensitivity of the transformation". Indeed, it is accepted that a low degree of substitution/functionalization (DS<0.37) will preserve the nanomorphology of nanocellulose (D.-Y. Kim, Y. Nishiyama, and S. Kuga, "Surface acetylation of bacterial cellulose," Cellulose, vol. 9, no. 3-4, pp. 361-367, 2002). I think the authors should discuss more and perform the necessary characterizations also on samples with lower oxidation extent, but more preserved cristallinity.
Response: We thank the reviewer for raising this crucial point. Indeed, 4.5 mmol/g is a very high degree of oxidation, where the oxidation proceeds not only on the surface of CNF. We performed the reaction on a highly oxidized substrate for our substrate scope (salts of O-substituted hydroxylamines) mainly to make the quantification of the transformation more accurate. However, as shown in Figure  1 and Table S1, the extent of the oxidation is easily tuned by changing the oxidant concentration or the reaction time. We examined the properties of the film with a lower degree of oxidation (in the manuscript, we refer to it as DA-CNF-2h). In addition, we performed the functionalization of this film (Bn-ON-CNF-2h) and defunctionalization via hydrolysis. We present the obtained results in Figure 3c.
To clarify, we mention the reason and choice of such a high degree of substitution in the main text of the manuscript (highlighted in green): "In addition, for substrate scope studies, the reaction was driven to high DF to facilitate characterization of the final products. However, the designed methodology allows for simple modulation of DF, where lower DF (mainly surface modification of nanofibrils) is desirable when mechanical properties of the material are of the greatest importance.

2) It may be difficult in this study, but the yields (based on the starting dry CNF weight) and molecular weights of DA-CNFs prepared under various conditions should be measured or discussed. DA-CNFs may partly decrease in the yield by oxidation and washing processes.
Response: We agree with the reviewer that it would be exciting to see the changes in molecular weight as this would give us intel into the mechanistic features of the oxidation, possibly answer the question on is the hemicellulose oxidized before cellulose, etc. We tried to evaluate this with SEC, however, we had real issues obtaining reliable data due to issues dissolving the films according to standard protocols. The only data point we could get was after 4.5 h of oxidation, see figure below. This data suggests that during oxidation, we have degradation of the cellulose substrate. However, this is still too uncertain to draw this conclusion. For one, oxidation leads to the hemiacetal formation, which undoubtedly will lead to intra-molecular acetal-formation that changes the hydrodynamic volume. In addition, we had problems with solubilizing the sample. As such, we are unsure which fraction we analyze, i.e., is the high molecular fraction not dissolved? To answer these questions, we need to develop a new solubilizing system and combine that with both NMR and viscosity measurements, which we believe falls outside of this study's scope.
Regarding to the yield of the oxidation step, we refrain of such statements in the manuscript, because of necessity to perform the reaction at higher scales to provide reliable and reproducible data. Our preliminary results revealed that 85-95 wt% of the material can be recovered. Table 1).

Response:
We have now added the requested information in Table 1 (highlighted in green in the manuscript).

4) It is better also to describe the reaction efficiency of NaIO4 in the DA-CNF preparation. Namely, the molar ratio of NaIO4 consumed for aldehyde formation/that added in the oxidation.
Response: We agree with the reviewer that consumption of NaIO4 can also be used to determine the degree of oxidation of dialdehyde cellulose. However, it has been shown that hydroxylamine titration and NaIO4 consumption methods show a relatively good correlation. In light of this, we limited ourselves to only the hydroxylamine titration method (Thermochim. Acta 2001, 369 (1-2), 79-85).

5) Page 7: The reaction conditions in acidic acetone/water at 55 degree C for 24 h may be harsh for R-ON-CNF. The low yield losses (<1%) were described in page 7, but some depolymerization of oxidized cellulose molecules may be unavoidable during de-oxime reactions, which should be addressed.
Response: Indeed, partial depolymerization of cellulose may occur under acidic conditions. However, this factor is hard to assess due to problems with determining the changes in molecular weight by SEC. However, we did not observe any film coloration after oxime hydrolysis reactions, which is an indirect evidence of a low degree of cellulose degradation. We have added the following statement to the manuscript (highlighted in green): "Even though partial depolymerization of cellulose may occur under aqueous acidic conditions, we did not observe any coloration of the film after oxime detachment, indicative of limited cellulose degradation."

6) Figure 1: Since C.I decreased to almost zero after oxidation for 7.5 h, did this oxidation occur also in the crystalline inside CNFs without reaction selectively on the CNF surfaces?
Response: Yes, we interpret this data as oxidation occurs inside CNF and not only on the surface at prolonged reaction times. We have added the following statement in the manuscript (highlighted in green): "We found that after 7.5 hours, crystallinity of the samples was completely lost, suggesting that the oxidation occurred not only on the surface of CNF, but within inner regions as well."

7) Figure 2, green spectrum: Is it true that the C-H peak is located at ~3250 cm-1?
Response: Yes, according to the literature (Socrates G. Infrared and Raman characteristic group frequencies: tables and charts, 3rd ed. edn. J. Wiley (2001)) C-H stretching in alkynes occurs in the region of 3200-3400 cm -1 .

8) Figure 3b: The changes in water uptakes were consistent with water contact angles on the films?
Response: We agree with the reviewer that it is a crucial point to mention. We chose to assess water uptake rather than contact angles because hydrophobicity is a bulk property and is better reflected by water uptake measurements than contact angle measurements. This relates to the fact that the surface roughness of the samples has a significant effect on contact angles, making conclusions regarding the samples' hydrophobicity difficult to draw. We add the following statement to the manuscript (highlighted in green): "Water sorption of cellulosic materials is a bulk property; therefore, we studied the water uptake for CNF, DA-CNF, and some R-ON-CNF films rather than water contact angles (which can also be influenced by the surface roughness), presented in the table on Figure 3b".

9) Figure 3a, S4, S5, and S6: It is better to include photographs of the films in this manuscript to show the colors or transparencies.
a. b.
c. Figure S7. Pristine and modified CNF films on the surface (left) and lifted a few cm from the surface (right) for a. Holo-CNF, b. DA-CNF, c. Bn-ON-CNF.

10) Figure S6: Why the films have such high haze values? Many pores with air/material interfaces are present in the films? What are the factors influencing the haze values of the films?
Response: The haze can be caused by surface roughness and by the presence of pores in the material. As can be seen from Figure S6, by preparing samples with smoother surfaces, haze can be significantly decreased (e.g., Me-ON-CNF haze reduces from 80 to 11% for the samples with rough and smooth surfaces, respectively, Table S3). Figure S1: What is the signal at ~50 ppm owing to?

11)
Response: Signal around 50 ppm is a spinning side band originating from the signal of carbon atoms in the aromatic ring. The main band of these protons appears in the region 125-130 ppm. Figure S1 and Figure 2c: Since the signal pattern at 60-110 ppm in Figure S1 is similar to that in Figure 2C,

this sample may have significant amounts of aldehyde groups (or the oximation ratio may be too low). Please explain this matter. Or it is better to show another C13-NMR spectrum of R-ON-CNF with a much higher degree of oximation.
Response: We apologize for the confusion. Figure S1 and the right-hand side of Figure 2c are identical spectra, but Figure S1 also shows signal assignment. Figure 2c shows two NMR spectra. The left spectrum corresponds to DA-CNF, and the spectrum on the right side of the figure corresponds to Bn-ON-CNF.There is a striking difference between the two spectra (appearing of new signals corresponding to the aromatic ring and oxime linkage and a decrease of intensity of the region corresponding to oxidized cellulose 60-110 ppm). Figure S3 and all 1H-and 13C-NMR spectra in Appendix. The solvents used in the 1H-and 13C-NMR measurement are also better to be described in the figures.

Response:
We thank the reviewer for this comment. We agree that signal assignment would be very beneficial, however, in this case, a scaling up of the process would be necessary to enable the isolation of all the products. Since only a crude mixture was assessed for the reaction presented in Figure S3, we would like to refrain from the full assignment and will highlight the presence of the product resulting from the trapping of the released O-substituted hydroxylamine with acetone (signals around 1.9 ppm). We also include the information regarding the solvent used.

Reviewer #3
The present manuscript with the title "In water synthesis of highly functional, hydrophobic, and chemically recyclable cellulose nanomaterials through oxime ligation" describes an interesting study on the chemical modification of nanostructured cellulosic materials. The manuscript is well structured, the work concisely described, the experimental procedures can be repeated based on the protocols and the illustrations are of satisfying quality.

Nevertheless, I believe that the work is not suited for publication in Nature Comm. There are two main reasons. First, the element of novelty, a prerequisite to publication, is largely missing. As the authors state themselves, periodate oxidation of carbohydrate (the very old Malaprade oxidation) is very well-known and has been extensively applied also to (nano)cellulosic materials. In most cases, those studies did not target the "dialdehyde cellulose" directly, but used it as an intermediate for further functionalization. Schiff bases as well as non-substituted and substituted hydrazones and oximes have been synthesized from dialdehyde cellulose and other dialdehyde polysaccharides. Despite placing the emphasis more on the properties of the products, the manuscript does not offer any novel aspects here.
Response: We appreciate the possibility of clarifying some of the concerns expressed by the reviewer regarding the novelty and advantages of the developed method. Although NaIO4 oxidation of cellulose fibers is a known process, NaIO4 oxidation of nanocelluloses is far less studied. Oxidation directly on nanocellulose fibers increases the degree of oxidation (due to a large amount of OH-group exposed) and the homogeneity of aldehyde groups within the substrate. Unlike cellulose fibers, where inner parts of fibers are less available for oxidation, cellulose nanofibrils are individualized, allowing for better chemical accessibility. While there are reports regarding the oxidation of cellulose nanocrystals (CNC), the oxidation of cellulose nanofibers (CNF) is very rare. We found only two papers on NaIO4 oxidation of CNF; in both cases, CNF was pre-modified either via carboxymethylation or TEMPO oxidation. The oxidized materials were used as additives to improve the properties of papers rather than studied as individual materials. In addition, separation of the product from the solution of an oxidant is difficult (dialysis for several days was implemented in one of the works). In our work, we developed a novel and direct method to perform fast, efficient, and experimentally simple oxidation of CNF. The key parameters for the successful oxidation are: 1) use of CNF with low charge (holo-CNF vs. TEMPO CNF) to avoid anionic repulsion between IO4and COOgroups on cellulose, which we observed when trying to perform the oxidation on TEMPO-CNF wet cake; 2) perform the reaction on the preformed wet cake to avoid separation problems and aggregation of CNF. To our knowledge, Osubstituted hydroxylamines have not been implemented as functionalization reagents for cellulose modification. Most reported methods used amines to produce imines followed by (or performed in situ) reduction. In this work, we raised a scientific question of whether improved hydrolytic stability of oximes will remove the stabilization step via reduction while still providing stable covalent attachment. Our results show successful covalent attachment of various functional groups to CNF. Moreover, a complete detachment of the installed functional groups was performed under mild reaction conditions.
It is essential to mention that both steps are very experimentally simple. The CNF wet cake is placed into a beaker containing water solution of either oxidant (first step) or a corresponding salt of Osubstituted hydroxyl amine (second step), allowing for simple product isolation and removal of the unreacted derivatizing agent.

Second, there are several aspects of the work which seem to indicated the studies have not yet fully matured (see below), and there are several claims which seem a bit exaggerated. -line 3 (title): the products are certainly less hydrophilic than cellulose, but to call them hydrophobic would be quite an exaggeration (cf. for instance to hydrocarbons). The product still possesses a wealth of hydrophilic functions, and in case of the hydrochlorides or other salts are even ionic.
Response: Yes, in the final material, there are still hydrophilic functional groups, which come from the unreacted cellulose fragments. However, our studies on water uptake demonstrated that some substrates were significantly hydrophobized (water uptake decreased from 231% for CNF to 6% for pentafluoro functionalized CNF, Figure 3b). With regards to hydrochlorides, these are not present in the final materials since salts of O-substituted hydroxylamine hydrochlorides form uncharged oximes once reacted with aldehyde groups of dialdehyde CNF.

-line 76: not all modifications of CNF suffer from moisture sensitivity. There have been general approaches to reactions of CNF proceeding in the surface-confined water layer (Beaumont et al. Nature Comm. 2021) or at least being fully water-tolerant (Hettegger et al. Chem.Sus.Chem. 2015 and 2016).
Response: We appreciate that the reviewer brings up this point. True, some methodologies suggest it occurs in a water medium. For example, in the presented paper by Beaumont et al. Nature Comm. 2021 et al. on acetylation in the surface-confined water layer. However, this method cannot be applicable to the modification of CNF wet cakes (ca. 85 wt%). In addition, stating that silane chemistry is fully water-tolerant is inaccurate. Silanes are known to react in the presence of water, forming Si-OH substrates that undergo further condensation, resulting in the formation of nano-clusters.
(Colloids and Surfaces A: Physicochem. Eng. Aspects 312 (2008) 83-91) Comparably, oxime ligation is high yielding regardless of water content. The chemistry is orthogonal, meaning it is the intended reaction that occurs, and no other side reaction that might be misinterpreted as a successful reaction outcome.
The following alterations to the manuscript have been made to clarify this and adhere to the reviewer's comments.
"…For these transformations solvent exchange is generally required, even though there are acetylation methods which tolerate residual water." "Silylation, while being promising example of water-based modification pathway, can be complicated due to the propensity of trialkyl silanes to hydrolysis and oligomerization which can result in less defined functionalization, where oligomeric siloxanes can be either covalently attached or absorbed on the surface of cellulosic material."

-line 89: it is not fully clear what "prefunctionalization" means. TEMPO oxidation, periodate oxidation or acyl transfer catalyzed by imidazole would not require any preceding steps.
Response: We apologize for the confusion caused by this statement. We have now removed it from the manuscript.

-line 89: the statement regarding atom economy would require a solid support by numbers to demonstrate that the proposed periodate oxidation / oxime approach is superior to the alternatives.
Response: We refer to atom economy regarding amide coupling, which is another method to perform modifications of oxidized CNF (TEMPO-CNF) in water. These protocols require stoichiometric amounts of coupling agents (e.g., 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide, EDC), which produces a stoichiometric amount of waste. While in the case of oxime formation, the only by-product is HCl. We have introduced the following clarification in the manuscript (highlighted in green):