Hydrolysis of Al3+ in Aqueous Solutions: Experiments and Ab Initio Simulations
, Viviana Mollica-Nardo 2, Claudia Foti 1, Rosina Celeste Ponterio 2, Franz Saija 2, Sebastiano Trusso 2, Jiri Sponer 3, Giuseppe Cassone 2,* and Ottavia Giuffrè 1,*Round 1
Reviewer 1 Report
The manuscript entitled “Hydrolysis of Al3+ in aqueous solution: experiments and ab initio simulations” by Fausta Giacobello, Viviana Mollica-Nardo, Claudia Foti, Rosina Celeste Ponterio, Franz Saija, Sebastiano Trusso, Jiri Sponer, Giuseppe Cassone, and Ottavia Giuffrè represents a combined experimental-computational study of aluminum +III cation speciation in water. The study seems to be well performed and brings some new insight into the topic. However, I am a bit puzzled by the short discussion of the (apparently) demanding molecular dynamics calculations and some lacking computational details, please see below. When these issues are solved, I will be happy to recommend this paper for publication in the Liquids journal.
1) The Authors mention on Page 4 that the system was hydrated by 128 water molecules. How was this performed? Did some pre-optimization take place?
2) For better comparison with DFT MD calculations, the Authors should calculate the structures treated with the polarizable continuum model also using BLYP-D3. Although I do not expect the results to change considerably, it is not clear to me why the Authors switch the method (MP2 vs. DFT) and the solvation approach (implicit vs. explicit) at once. This would be particularly important for comparing the bond lengths (see the next point).
3) The Authors performed 375 ps of DFT MD and discussed the obtained results within two paragraphs, which does not seem appropriate to me. The reader does not obtain any information on the dynamics of the structure: What is its structure (e.g., average bond lengths)? How readily is it formed? Is there some water exchange observed? Did the OH groups stay in one position or did proton transfer take place? Was the coordination number of the structure in Figure 2b/3b always five/six? Was there any difference observed for different starting points? Was the structure shown in Figure 4b always formed? It would be also interesting to compare the important bond lengths in the gas phase, polarizable continuum model and solution.
4) The Supporting Information document does not contain the optimized structures, it would be nice to have at least the structure in Figure 4a so that the calculations are reproducible.
5) Equation (2) is illegible.
6) The legend in Figure 1 seems to be wrong (e.g., Al3H-4).
Author Response
Query: “The study seems to be well performed and brings some new insight into the topic. However, I am a bit puzzled by the short discussion of the (apparently) demanding molecular dynamics calculations and some lacking computational details, please see below. When these issues are solved, I will be happy to recommend this paper for publication in the Liquids journal.”
Authors’ reply: We thank the Reviewer for her/his positive general comments on our work and for suggesting some improvements that we have fully taken into account during the revision process of our manuscript.
Query: “The Authors mention on Page 4 that the system was hydrated by 128 water molecules. How was this performed? Did some pre-optimization take place?”
Authors’ reply: We thank the Reviewer for giving us the opportunity to make this aspect clearer. All molecular configurations in liquid phase were set up and pre-equilibrated by means of extensive classical molecular dynamics simulations. Each simulation box was firstly equilibrated via a standard force-field (AMBER) for 1 ns by using periodic boundary conditions. This important information has been added in the revised version of the manuscript.
Query: “For better comparison with DFT MD calculations, the Authors should calculate the structures treated with the polarizable continuum model also using BLYP-D3. Although I do not expect the results to change considerably, it is not clear to me why the Authors switch the method (MP2 vs. DFT) and the solvation approach (implicit vs. explicit) at once. This would be particularly important for comparing the bond lengths (see the next point).”
Authors’ reply: Optimization of the investigated molecular structures was performed at the second order of the Møller-Plesset perturbation theory (MP2) because of the well-known reliability and accuracy provided by this theoretical framework. In fact, our main goal was not that of performing a direct comparison between the static quantum-mechanical calculations and the ab initio molecular dynamics (AIMD) ones but, rather, that of employing the state-of-the-art theoretical tools available and which were capable of reliably simulating the behaviour of the systems under investigation. As a matter of fact, no quantitative comparisons between, e.g., the bond lengths obtained with these different methods were performed in the original version of the paper. On the other hand, in order to add useful molecular information and have a one-to-one quantitative correspondence with the data stemming from AIMD simulations, we have now re-executed the static quantum-mechanical calculations at the lower level of theory suggested by the Reviewer (BLYP-D3) and implemented the respective results in the revised version of the manuscript, as shown in Figs. 2a, 3a, and 4a and throughout the text.
Query: “The Authors performed 375 ps of DFT MD and discussed the obtained results within two paragraphs, which does not seem appropriate to me. The reader does not obtain any information on the dynamics of the structure: What is its structure (e.g., average bond lengths)?”
Authors’ reply: The global time-length of the ab initio molecular dynamics (AIMD) simulations we performed is certainly large due to the necessity of having an adequate statistical treatment of the natural bias introduced when starting from specific initial molecular configurations and atomic velocities. To enrich the overall information stemming from those analyses and with the aim of comparing the AIMD findings with those emerging from quantum-mechanical geometric optimizations under implicit solvation, the relevant average bond lengths have now been determined and reported in the revised form of the paper, as shown in Figs. 2b, 3b, and 4b and throughout the text.
Query: “How readily is it formed? Is there some water exchange observed? Did the OH groups stay in one position or did proton transfer take place?”
Authors’ reply: The average time necessary for forming stable AlCl2+ complexes surrounded by four water molecules is lower than 5 ps. An even shorter value is recorded for the solvation process occurring in presence of AlClOH+. Those data are now reported in the main text of the revised manuscript whilst the curves showing the relevant instantaneous distances of the oxygen atoms of the involved water molecules with respect to the Al atomic positions as a function of time are displayed in newly prepared Figures, which are reported in the revised Supplementary Materials file (see Figures S1 and S2).
By contrast, due to the higher molecular cooperation required in rearranging the respective hydration shells, characteristic temporal windows necessary for the whole complexation of the Al3(OH)45+ are longer (i.e., larger than 15 ps), as now displayed in a newly prepared Figure reported in the Supplementary Materials file (Figure S3). Finally, as far as the the OH groups of the AlClOH+ and the Al3(OH)45+ species are concerned, they remain stable in their initial position, relative to the Al one, over the whole trajectories here explored. All these important information have now been included in the revised version of the paper.
Query: “Was the coordination number of the structure in Figure 2b/3b always five/six?”
Authors’ reply: Yes.
Query: “Was there any difference observed for different starting points?”
Authors’ reply: No significant differences have been observed.
Query: “Was the structure shown in Figure 4b always formed?”
Authors’ reply: Yes.
Query: “It would be also interesting to compare the important bond lengths in the gas phase, polarizable continuum model and solution.”
Authors’ reply: As explained in the previous points, we have prepared and added new Figures reporting relevant comparisons between the results emerging from the polarizable continuum model quantum-mechanical calculations and ab initio molecular dynamics (AIMD) simulations performed in solution (see Figures 2, 3, and 4 of the revised manuscript). As for the gas-phase bond lengths, we regret to say that they do not seem very relevant to the discussion faced in the current paper, which refers to condensed-phase systems and processes.
Query: “The Supporting Information document does not contain the optimized structures; it would be nice to have at least the structure in Figure 4a so that the calculations are reproducible.”
Authors’ reply: We thank the Reviewer for this suggestion. Coordinates in the xyz format of the structure shown in Fig. 4a of the main text are now reported in the Supplementary Material file (see Table S2) for the structure optimized at the MP2/6-311++G(2d,2p) level.
Query: “Equation (2) is illegible.”
Authors’ reply: Equation (2) has now been rewritten.
Query: “The legend in Figure 1 seems to be wrong (e.g., Al3H-4).”
Authors’ reply: For consistency with the manuscript text, as correctly suggested by the Reviewer, the speciation diagrams now report the same way of indicating the hydrolytic species, i.e., Al3(OH)4 and Al16(OH)32.
Reviewer 2 Report
The submitted manuscript is very well written and I find it suitable for publication in Liquids after revision. Although a lot has been already explained about the hydrolysis of Al(III), in this manuscript another piece to the whole picture has been added. I appreciate the use of MP2 instead of DFT as well as the ab initio molecular dynamics simulations. My comments can be found below.
In the introduction the Authors can mention which methods were used for speciation of Al(III) complexes. Was it NMR or other spectroscopic techniques?
Line 20, rather not „microscopic” but „molecular”
Lines 46-47, those are not neutral molecules but ions, please add charges
Table 1, instead of “I” it should be “CNa”
Lines 126-128, please add references to the software
Lines 146 and 148, I suppose the number of water molecules results from the desired molar concentration. Could you please provide it?
Line 174, “multiple”-exactly how many?
What was the software used for MD simulations?
Were the MD simulations CPMD or BOMD?
Line 224, editorial, this should be an equation written in Word, not an image.
Author Response
Query: “The submitted manuscript is very well written and I find it suitable for publication in Liquids after revision. Although a lot has been already explained about the hydrolysis of Al(III), in this manuscript another piece to the whole picture has been added. I appreciate the use of MP2 instead of DFT as well as the ab initio molecular dynamics simulations.”
Authors’ reply: We are grateful to the Reviewer for the very positive evaluation and comments reported on our manuscript.
Query: “In the introduction the Authors can mention which methods were used for speciation of Al(III) complexes. Was it NMR or other spectroscopic techniques?”
Authors’ reply: We thank to the Reviewer for this important suggestion. The integration of the methods to the Introduction section provides a more general picture of the previous studies. In the Introduction, other methods such as 27Al NMR spectroscopy, mass spectrometry, Raman spectroscopy, IR spectroscopy, UV spectrophotometry, atomic absorption spectrometry, and coagulation processes for the study of Al hydrolysis have now been added in the revised version of the manuscript, along with the respective bibliography.
Query: “Line 20, rather not „microscopic” but „molecular””
Authors’ reply: We thank the Reviewer for this hint. We have replaced the term “microscopic” with the more correct term “molecular” in the abstract of the revised paper.
Query: “Lines 46-47, those are not neutral molecules but ions, please add charges”
Authors’ reply: Charges have been added to the mentioned ions.
Query: “Table 1, instead of “I” it should be “CNa””
Authors’ reply: In Table 1, “I” has been replaced with “CNa”.
Query: “Lines 126-128, please add references to the software”
Authors’ reply: The bibliographic note related to the used programs has been added.
Query: “Lines 146 and 148, I suppose the number of water molecules results from the desired molar concentration. Could you please provide it?”
Authors’ reply: Because of the high computational demand typical of ab initio molecular dynamics (AIMD) simulations, the exact reproduction of the experimental concentrations represents a task impossible to be achieved even by the most powerful and efficient supercomputers in the world. Thus, the number of water molecules employed in the current AIMD computations (i.e., 128) has been chosen to maximize the size of the simulation box and avoid spurious limited-size effects. It is worth mentioning that performing AIMD simulations of numerical samples composed by ≈ 400 atoms for 400-ps-long trajectories represents a sort of computational upper-bound for most of the high-performance computing (HPC) apparatus, as explicitly stated in the manuscript.
Query: “Line 174, “multiple”-exactly how many?”
Authors’ reply: For “multiple” we mean 5, as explained in the subsequent sentence. To avoid any misunderstanding, the exact number has explicitly been stated in the revised form of the manuscript.
Query: “What was the software used for MD simulations?”
Authors’ reply: We thank the Reviewer for giving us the opportunity to clearly state the used software for the AIMD simulations. In particular, we employed the CP2K software suite, as now explicitly mentioned in the Methods section of the revised paper along with the respective bibliographic citation.
Query: “Were the MD simulations CPMD or BOMD?”
Authors’ reply: We thank the Reviewer for giving us the opportunity of making this aspect clearer. We performed Born-Oppenheimer molecular dynamics (BOMD) simulations, as now explicitly reported in the revised manuscript.
Query: “Line 224, editorial, this should be an equation written in Word, not an image.”
Authors’ reply: Equation (2) has been rewritten.
Round 2
Reviewer 2 Report
The Authors have carefully revised their manuscript. This version can be accepted for publication.
