Disproportionation of Rosin Driven by 4,4′-Thio-bis(3-Methyl-6-Tert-Butylphenol): Kinetic Model Discrimination

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper by Souto et al. reports a kinetic study of the catalyzed disproportionation of rosin. The experimental work appear well conducted and the report reasonably well written. Few typos (e.g. consider-> considered, depended->dependent) can be corrected on typesetting. 

We find it suitable to publication, once the following queries, in no particular order, have been solved:

1)        The position “dehydroabietic acid [is] the most stable compound due to its aromatic cycle in the terpene structure” is inconsistent with values of enthalpy of formation (which should be better reported) and does appear only superficially congruent with the standard discussion of aromaticity

2)        Eq. 1 and those in Table 3 appear poorly readable because of the occurrence of a Chinese-like character in place of concentrations. 

3)        Experimental data should be given in Tables in the supporting Information

4)        The substantial discussion of chemical aspects should be accompanied with schemes using chemical structures, not just their names.

5)        The paper lacks a Conclusion Section. Though not mandatory, it would add readability and interest in the paper.

6)        Models 2-5 seems very similar at the statistical level. No F-test is proposed to differentiate among them.

7)        Non-standard definition of Eq. (4) should be commented.

8)        Models 2-4 consider the transformation of palustric and neoabietic acids in abietic acid. This possibility would at most be deduced from the bottom-left inset of figure 2. In that case a discussion on the error od concentrations would be needed. Model 5 is unable to describe accumulation or slower consumprion of abietic acid. 

Comments on the Quality of English Language

ok

Author Response

Thanks for your constructive comments and valuable corrections and recommendations (in black italic font). We have carefully revised the manuscript according to your suggestions. The itemized responses (in blue) and the added text/references in the main manuscript (in red font) are attached as follows:

Comment 1

The paper by Souto et al. reports a kinetic study of the catalyzed disproportionation of rosin. The experimental work appears well conducted and the report reasonably well written. Few typos (e.g. consider-> considered, depended->dependent) can be corrected on typesetting. 

We find it suitable to publication, once the following queries, in no particular order, have been solved:

The position “dehydroabietic acid [is] the most stable compound due to its aromatic cycle in the terpene structure” is inconsistent with values of enthalpy of formation (which should be better reported) and does appear only superficially congruent with the standard discussion of aromaticity.

The reviewer is absolutely right. This sentence confuses thermodynamic stability with chemical stability, particularly against oxidation. Dehydroabietic acid is well known for its stability against oxidation and this is one of the main reasons for its industrial production. We have changed the explanatory phrase mentioned by the reviewer:

“…dehydroabietic acid, the most stable compound due to its aromatic cycle in the terpene structure.”

with the following one, citing an appropriate reference:

“..dehydroabietic acid, one of the most stable abietates under oxidative conditions [4]”.

Comment 2

Eq. 1 and those in Table 3 appear poorly readable because of the occurrence of a Chinese-like character in place of concentrations.

Thank you for your observation. We do not see any Chinese-like characters in our manuscript but usual characters for kinetic constants, concentrations, etc.; in any case, we have provided explanations for the symbols in equations and tables. Just before equation (1) we have written:

…Thus, all reactions rates are expressed following equation (1) with the catalyst concentration (C_cat) and the relevant acid concentration (C_acid) as main independent variables and k_it and k_ic as the thermal and catalytic terms kinetic constants:

And we have added, after Table 3, the following note:

Note: Where k_it and k_ic refer to kinetic constants for the thermal (t) component and the catalytic term (c) relevant to the reaction rate i (r_i), C_cat refers to the catalyst concentration (in mass percentage), C_ab is the abietic acid concentration, C_Dh is the dehydroabietic acid concentration, C_P is the palustric acid concentration, C_N is the neoabietic acid concentration, and C_Di is the dihydroabietic acid concentration, C_Pi is the pimaric acid concentration. R refers to the observed temporal evolution rate of any of the acids present in this kinetic model.

Comment 3

Experimental data should be given in Tables in the supporting Information.

Thank you for the query. Though all relevant experimental data is present in Figure 2, we have provided the evolution of the acid concentrations as a supporting information in a separated file, as requested by the reviewer, as well as added information regarding the fitting results.

Comment 4

The substantial discussion of chemical aspects should be accompanied with schemes using chemical structures, not just their names.

We are most grateful to the reviewer for this question. Though we have added a graphical abstract during the revision process, we have also modified figure 3 to include the chemical structures of the relevant acids, together with the reaction scheme and the kinetic equations.

Comment 5

The paper lacks a Conclusion Section. Though not mandatory, it would add readability and interest in the paper.

We agree with the reviewer, so we have added a short conclusion section at the end of the manuscript. It reads as:

4. Conclusions

              In this work, several phenomenological kinetic models have been tested to fit diverse runs of disproportionation of rosin with 4, 4’–thio-bis (3-methyl-6 tert-butylphenol) in the 260-280 °C interval and with diverse catalyst concentrations (from 0 till 2% w/w rosin). These kinetic models included disproportionation, isomerization, and dehydrogenation reactions. Diverse physicochemical and statistical criteria led to the selection of model 5, where abietic, neoabietic and palustric acids tranform into dehydroabietic acid by dehydrogenation, while abietic acid also disproportionates to di- and dehydroabietic acid and pimaric acid isomerizes to isopimaric acid. Thermodynamical reasons advocate for the isomerization on the grounds of enthalpy of formation values for pimaric acid, while abietic-type acids seem to transform towards dehydroabietic acid under kinetic control favored by the segregation of the created hydrogen out of the reaction liquid, thus shifting the dehydrogenation equilibria towards the products. Finally, we can conclude that, even if 4, 4’–thio-bis (3-methyl-6 tert-butylphenol) is less active than sulfur, it has an acceptable catalytic activity to be added to its antioxidant and bleaching action in rosin products.

 

 

Comment 6

Models 2-5 seems very similar at the statistical level. No F-test is proposed to differentiate among them.

Thank you for the question. In this case, the F-test is provided for all kinetic models under consideration as F values in Table 2, being this F estimated as explained in equation (4) for the 95% confidence. Models 2 to 5 are statistically very similar, with models 2 and 5 as the most statistically relevant, so only a minor difference in goodness-of-fit leads us to select model 5. This is a usual situation in several catalytic processes, where statistical model discrimination is not enough to determine the reaction network scheme. Furthermore, no evidence in the literature is found as kinetic model discrimination is not usually performed, being a model similar to model 2 used in most cases for hydrogenation over Ni catalysts and dehydrogenation over Pd-C catalysts. However, even if the mechanism determination was not the main target of this manuscript, but the selection of an adequate kinetic model, these phenomenological models open the door to further research into the scheme of chemical reactions taking place.

In any case, we have tried to add more information and comments in the Results and in the Discussion section. This latter has a new paragraph at the beginning:

3. Discussion

From a statistical perspective, models 2 and 5 are the most adequate both in terms of goodness-of-fit and considering the narrow intervals of error at 95% confidence for all kinetic parameters in both models. Therefore, it is complicated to select with accuracy the most adequate model on the grounds of statistics, even if model 5 is selected due to a slightly best fit. From a thermodynamic perspective, and considering thermodynamical control of the global transformation, both neoabietic acid and palustric acid trend to isomerize at a fast rate at temperatures higher than 200 °C to the more stable abietic acid (based on the respective standard formation enthalpies –DH_f⁰-: -615.5, -635.0 and -686.88 kJ/mol). Afterwards, the abietic acid is transformed into dehydro- and dihydroabietic acid by dehydration and disproportionation reactions. This is the network of reactions proposed in model 2 and by several authors for disproportionation and hydrogenation of rosin [18-20]. However, this perspective is in contrast with the fact that –DH_f⁰ for the dehydroabietic acid is -631.1 kJ/mol [22], higher than for abietic acid, which suggest that dehydroabietic acid is less stable from a thermodynamic perspective than abietic acid, but more stable than both palustric and neoabietic acids. Therefore, in a situation where isomerization, dehydrogenation and disproportionation coexist at high temperature and relatively low pressure (with formed hydrogen probably coming out from the reacting liquid), is not impossible to hypothesize that all abietic acids react to dehydroabietic acid while only abietic acid is affected by disproportionation, as stated by model 5, suggesting a kinetic control of the global reaction network. In both cases, pimaric acid (DH_f⁰ = –547.1 kJ/mol) turns into isopimaric acid (DH_f⁰ = –590.1 kJ/mol).

 

New literature:

 

21. Meyerriecks, W. Organic fuels: Composition and formation enthalpy, part I, wood derivatives, related carbohydrates, exudates and rosin. Pyrotech. 1998, 8.

 

Comment 7

Non-standard definition of Eq. (4) should be commented.

Thank you for the query. We have added the following comment after equation (4):

In particular, equation (4) refers to the F-value at a certain level of confidence as implemented in the non-linear regression algorithms in the Aspen Custom Modeler software.

To deepen in the discussion of goodness-of-fit parameters to avoid overparameterization, we have added calculated values of Akaike’s and Schwarz information criteria in Table 2 and discuss them in the Results section. These parameters are now defined in the Experimental Section (new equations 5 and 6). New references:

30. Cavanaugh, J. E. and Neath, A. A. The Akaike information criterion: Background, derivation, properties, application, interpretation, and refinements. Wiley Interdisc. Rev.: Stat. 2019 11(3), e1460.
31. Neath, A. A. and Cavanaugh, J. E. (2012). The Bayesian information criterion: background, derivation, and applications. Wiley Interdisc. Rev.: Stat. 2012 4(2), 199-203.

Comment 8

Models 2-4 consider the transformation of palustric and neoabietic acids in abietic acid. This possibility would at most be deduced from the bottom-left inset of figure 2. In that case a discussion on the error of concentrations would be needed. Model 5 is unable to describe accumulation or slower consumption of abietic acid.

Thank you for the observation. In fact, experimental errors in the determination of acid concentrations by GC-FID were in excess of 10%. No accumulation of abietic acid is observed in most cases in data displayed in Figure 2, except in subfigure 2e, so it is difficult to accept that abietic acid is too slowly disappearing at 270 °C, and not at 260 °C or 280 °C with 1% catalyst. Furthermore, no accumulation is observed in other cases, nor slow conversion of abietic acid. We must ascribe this strange trend, thus, to the random experimental errors, usually in the 10% range in GC-FIC when derivatization of compounds to render them volatile is employed, or to particular errors in the initial points of this experiment, that can be due to a slightly lower temperature at the beginning of the run or some problems issuing from the derivatization of the rosin acids with N,N-dimethylformamide dimethyl acetal in some samples in that particular run. In any case, we have added some comments in the paragraphs just before Table 2, trying to explain model selection based on statistical findings and random error possible in data.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Comments to the authors

This paper describes the study of a kinetic modelling related to the method of transformation of rosinic acid mixtures at high temperatures and in the presence of a catalyst that at the same time avoid oxidation or pirolisis at the conditions used. It is certainly a topic of interest, since the environmetal benefits in the best use of plant derivatives.

Because of the current topic, I think this paper is fine for publishing, but several modifications must be taken into account.

-       The structure of the document do not make easy the reading and understanding, until the final.

-        It seems that the experimental and methodology section have part of discussions.

-       Conclusions are not well enfizised, and do not correspond to the described in the abstract.

-       I would suggest to re-structure the sections and take care of form issues, like to define all acronyms since the first time mentioned (CG-FID, CG-MS, EV,COD, etc.).

-       There are several mistakes in the graphs, number of figures and tables that should be corrected, which are marked in the document.

-       As for the CG analyses, in some part should be mentioned that the analysed samples are all volatile, or what was the observed, to be sure that were correctly analyzed for obtaining the concentration of rosinic acids.

-       Also, could be more practical for the reading to consider the chemical structures of the dispropornation reactions of rosin acids

Comments for author File: Comments.pdf

Author Response

Thanks for your constructive comments and valuable corrections and recommendations (in black italic font). We have carefully revised the manuscript according to your suggestions. The itemized responses (in blue) and the added text/references in the main manuscript (in red font) are attached as follows:

Comment 1

This paper describes the study of a kinetic modelling related to the method of transformation of rosin acid mixtures at high temperatures and in the presence of a catalyst that at the same time avoid oxidation or pyrolysis at the conditions used. It is certainly a topic of interest, since the environmental benefits in the best use of plant derivatives.

Because of the current topic, I think this paper is fine for publishing, but several modifications must be taken into account.

      The structure of the document do not make easy the reading and understanding, until the final.

      Thank you for your insightful comments and your appreciation of our work. Considering this first query, we have added key information where needed to make our meaning clearer, adding explanations not only in the final experimental section but also before, in the results and in the discussion sections, if needed. Probably, to have the experimental section before results and discussion will be better in this case, but the template structure seems to be mandatory.

      Comment 2

It seems that the experimental and methodology section have part of discussions.

We have added some comments in those parts to facilitate the reading but, in any case, we have further commented the main facts in the discussion section as well, and rearrange the experimental and methodology sections to avoid discussion of results in that part of the manuscript. In any case, we prefer to explain the model discrimination methodology in the experimental section, as it belongs to this section in a natural way. In fact, we have added more statistical goodness-of-fit parameters widely accepted for model selection: the Akaike information criterion and the Bayesian or Schwarz information criterion.

Comment 3

Conclusions are not well emphasized, and do not correspond to the described in the abstract.

The reviewer is absolutely right. We have included a new, short conclusion section.

4. Conclusions

              In this work, several phenomenological kinetic models have been tested to fit diverse runs of disproportionation of rosin with 4, 4’–thio-bis (3-methyl-6 tert-butylphenol) in the 260-280 °C interval and with diverse catalyst concentrations (from 0 till 2% w/w rosin). These kinetic models included disproportionation, isomerization, and dehydrogenation reactions. Diverse physicochemical and statistical criteria led to the selection of model 5, where abietic, neoabietic and palustric acids tranform into dehydroabietic acid by dehydrogenation, while abietic acid also disproportionates to di- and dehydroabietic acid and pimaric acid isomerizes to isopimaric acid. Thermodynamical reasons advocate for the isomerization on the grounds of enthalpy of formation values for pimaric acid, while abietic-type acids seem to transform towards dehydroabietic acid under kinetic control favored by the segregation of the created hydrogen out of the reaction liquid, thus shifting the dehydrogenation equilibria towards the products. Finally, we can conclude that, even if 4, 4’–thio-bis (3-methyl-6 tert-butylphenol) is less active than sulfur, it has an acceptable catalytic activity to be added to its antioxidant and bleaching action in rosin products.

 

Comment 3

I would suggest to re-structure the sections and take care of form issues, like to define all acronyms since the first time mentioned (CG-FID, CG-MS, EV, COD, etc.).

Thank you for the observation. We have included the full names the first time the abbreviations appear for the benefit of the reader.

Comment 4

There are several mistakes in the graphs, number of figures and tables that should be corrected, which are marked in the document.

We have corrected these mistakes, thank you for the warning.

Comment 5

As for the CG analyses, in some part should be mentioned that the analyzed samples are all volatile, or what was the observed, to be sure that were correctly analyzed for obtaining the concentration of rosin acids.

Thank you for your query. The samples are quantitatively derivatized with N,N-dimethylformamide dimethyl acetal to render all rosin acids volatile by methylation so they can be analyzed with gas chromatography technique. Further information on the use of this reagent can be found in:

Stróżyńska, M., Schuhen, K. (2020). Derivatization of perfluorocarboxylic acids with N, N-dimethylformamide dimethylacetal prior to GC–MS analysis. Chromatographia, 83(3), 477-482.

Comment 6

Also, could be more practical for the reading to consider the chemical structures of the disproportionation reactions of rosin acids.

We totally agree with the reviewer. Therefore, figure 3 is newly made to enter all chemical structures, apart from the novel graphical abstract.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Several improvements have been operated on the manuscript, but I do not find them that convincing.

In particular, the statement that a model is preferred for a higher F, is only a minor support by the addition of two more indicators. The F test should have been applied to differentiate between the models. Likely one is selecting thinks that are equal at the experimental level. 

The F-test calculations is performed in a straightforward manner, so I do not find a good reason to avoid out, a part from recognising that model selection cannot be done. It is a negative,m but honest result, which I would think of greater scientific value than an unsupported selection.

The chemical scheme is wrong and must be corrected.

Comments on the Quality of English Language

ok

Author Response

We really appreciate the comments and suggestions of the reviewer and have modified the paper accordingly. The reviewer can find the changes in red font and highlighted in yellow to differentiate them from the previous modifications on the original version.

As he/she is really right, we have conducted an F-test between the better model in terms of lower sum of squares (SSR) and the second one -model 2-, finding no statistical difference. We have included comments in this sense in the results, discussion, conclusion and abstract sections, as both models are statistically indistinguishable.

We have also corrected Figure 3, depicting the reaction network of Model 5. Thank you very much for your efforts that have allowed us to improve our manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has been considerably improved, is more easy to understand, and conclusions correspond to the described in the document.

I do not have more observations and suggest its publication.

Author Response

Thank you very much for your kind appreciation of our work. It encourages us in our research.