From the journal Environmental Science: Atmospheres Peer review history

20-Oct-2023

Dear Dr lbadaoui-darvas:

Manuscript ID: EA-ART-09-2023-000140
TITLE: Molecular scale mechanism of deposition ice nucleation on Silver Iodide

Thank you for your submission to Environmental Science: Atmospheres, published by the Royal Society of Chemistry. I sent your manuscript to reviewers and I have now received their reports which are copied below.

I have carefully evaluated your manuscript and the reviewers’ reports, and the reports indicate that major revisions are necessary.

Please submit a revised manuscript which addresses all of the reviewers’ comments. Further peer review of your revised manuscript may be needed. When you submit your revised manuscript please include a point by point response to the reviewers’ comments and highlight the changes you have made. Full details of the files you need to submit are listed at the end of this email.

Please submit your revised manuscript as soon as possible using this link:

*** PLEASE NOTE: This is a two-step process. After clicking on the link, you will be directed to a webpage to confirm. ***

https://mc.manuscriptcentral.com/esatmos?link_removed

(This link goes straight to your account, without the need to log on to the system. For your account security you should not share this link with others.)

Alternatively, you can login to your account (https://mc.manuscriptcentral.com/esatmos) where you will need your case-sensitive USER ID and password.

You should submit your revised manuscript as soon as possible; please note you will receive a series of automatic reminders. If your revisions will take a significant length of time, please contact me. If I do not hear from you, I may withdraw your manuscript from consideration and you will have to resubmit. Any resubmission will receive a new submission date.

The Royal Society of Chemistry requires all submitting authors to provide their ORCID iD when they submit a revised manuscript. This is quick and easy to do as part of the revised manuscript submission process. We will publish this information with the article, and you may choose to have your ORCID record updated automatically with details of the publication.

Please also encourage your co-authors to sign up for their own ORCID account and associate it with their account on our manuscript submission system. For further information see: https://www.rsc.org/journals-books-databases/journal-authors-reviewers/processes-policies/#attribution-id

Environmental Science: Atmospheres strongly encourages authors of research articles to include an ‘Author contributions’ section in their manuscript, for publication in the final article. This should appear immediately above the ‘Conflict of interest’ and ‘Acknowledgement’ sections. I strongly recommend you use CRediT (the Contributor Roles Taxonomy, https://credit.niso.org/) for standardised contribution descriptions. All authors should have agreed to their individual contributions ahead of submission and these should accurately reflect contributions to the work. Please refer to our general author guidelines https://www.rsc.org/journals-books-databases/author-and-reviewer-hub/authors-information/responsibilities/ for more information.

I look forward to receiving your revised manuscript.

Yours sincerely,
Prof. Nønne Prisle
Associate Editor, Environmental Sciences: Atmospheres

************

Reviewer 1

The authors investigated ice formation on the AgI surface in vapor using the combination of molecular dynamics and grand canonical Monte Carlo simulations and proposed a mechanism of deposition ice nucleation. Heterogeneous ice nucleation on solid surfaces is an important process in the field of environmental science, however, I don’t recommend publication of this manuscript.

1) The authors proposed a four-step mechanism for deposition ice nucleation on AgI: “a) Water molecules adsorb on the surface, forming nanodroplets. b) The supercooled water nanodroplets merge into a continuous multilayer when they grow to about molecular layers thick. c) The layer continues to grow until the critical thickness for freezing is reached. d) The critical ice cluster continues to grow.” It is likely that a) and b) are artifacts arising from simulation setting. Figure 3 shows that thousands of water molecules adsorb on the AgI surface only within 1 ns. This rate of adsorption seems too high. If the adsorption rate is low and longer MD simulations are performed, a single layer of water would be formed instead of nanodroplets. The rate of adsorption in the simulation is determined not only by the pressure but also by the frequency of MC trial for water insertion. Did the authors tune the frequency of MC trial so that it reproduces the experimental rate of adsorption?

2) The deposition nucleation mode is not the only pathway for ice formation in the gas phase. In most cases, in the atmosphere below the freezing point, snow is produced by a multi-step process in which water droplets are first formed around dust particles, and then ice nucleation occurs. This is a typical example of Ostwald's step rule. Whether the crystallization process is classified to adsorption nucleation or multi-step nucleation due to Ostwald's rule depends on whether the liquid phase is observed or not. Since a liquid water layer forms in the proposed mechanism, it is no longer deposition nucleation. Multi-step nucleation is observed at low supercooling and deposition nucleation is observed at high supercooling. The temperature examined in this study, 253 K, is too high for deposition nucleation. I think that the authors merely observe the well-known multi-step nucleation of ice in their simulations.

3) Figure 2b shows that crystalline structures form in the system with small amount of adsorbed water when the temperature is low (T = 233, 223, and 213 K). It is needed to perform long simulations in this temperature range to investigate deposition ice nucleation. A very different behavior from that found in this study may be observed.

4) I thought that the row of N(L1) in Table 1 shows the number of water molecules in layer 1 at the time shown in the second row of the table. The N(L1) value is 247 at 755 ps for run 4. There is no panel corresponding this result in Figure 3.

5) Page 5: “Figure 5a-c illustrates the build up of hexagonal patterns over time”. It is not clearly seen in Figure 5. Please show molecules or hydrogen bonds forming hexagonal patterns with a different color.

6) The critical radius depends not only on the ice/water surface free energy but also on the ice/AgI surface free energy, but the latter is neglected in Eq. 1.

Reviewer 2

In this article, authors performed MD simulations of deposition ice nucleation using AgI substrate. Their findings are unique such that water molecules adsorb in patches or clusters and then these clusters merge to become a continuous water layer are not well known before. These results provide a new understanding of how deposition ice nucleation would occur at molecular resolution. The paper is well written, and I recommend publication after addressing the following minor comments.

It is not clear the selection of temperature range selected for the simulation (section 3.1). The range used is 213 to 253K. This range is used because there were no experimental data (Fig 7a) outside this temperature range? In reality, the AgI particles will not be smooth spheres but will have irregular morphology or surface. Are simulations involving such heterogeneous structures considered? How results would be affected if these structures were used in the simulations? Recently David et al published a paper (also cited here #25 in the reference section) and performed MD simulations. This previous paper highlights the importance of the pore condensation and freezing mechanism of ice formation, and the traditional view of the deposition ice nucleation mechanism is questionable. I suggest authors discuss this David et al paper in the context of the results presented in this article (end of section 3). Please elaborate more on the atmospheric implications “water adsorption” note (end of section 4). How existing ice nucleation theories can be improved further?

Reviewer 3

Please see the attached PDF file

[This text has been copied from the PDF response to reviewers and does not include any figures, images or special characters.]

Response to referee reports
for manuscript ID EA-ART-09-2023-000140
Molecular scale mechanism of deposition ice nucleation on
Silver Iodide
Golnaz Roudsari, M´aria Lbadaoui-Darvas, Andr´e Welti, Athanasios Nenes, Ari Laaksonen,
Novemver 2023
Dear Editor,
We would like to thank all the referees for their reviews of our manuscript and have prepared a
revised version of the manuscript. The changes in the revised manuscript are highlighted in magenta.
1 Response To Referee 1
The authors investigated ice formation on the AgI surface in vapor using the combination of molecu-
lar dynamics and grand canonical Monte Carlo simulations and proposed a mechanism of deposition
ice nucleation. Heterogeneous ice nucleation on solid surfaces is an important process in the field of
environmental science, however, I don’t recommend publication of this manuscript.
1) The authors proposed a four-step mechanism for deposition ice nucleation on AgI: “a) Water
molecules adsorb on the surface, forming nanodroplets. b) The supercooled water nanodroplets
merge into a continuous multilayer when they grow to about molecular layers thick. c) The layer
continues to grow until the critical thickness for freezing is reached. d) The critical ice cluster
continues to grow.” It is likely that a) and b) are artifacts arising from simulation setting. Figure 3
shows that thousands of water molecules adsorb on the AgI surface only within 1 ns. This rate of
adsorption seems too high. If the adsorption rate is low and longer MD simulations are performed,
a single layer of water would be formed instead of nanodroplets.
Response: Similarly to [1] and [6], we use Si = 3 to speed up the simulations, otherwise they
would simply take too long to observe the adsorption. Whether or not the nanodroplets would form
at lower humidities is a matter of wettability of the surface. Silver iodide is rather hydrophobic [8],
which would suggest droplet-wise rather than film-wise adsorption, especially if the surface contains
isolated hydrophilic patches as is the case in reality [8]. In our simulations we did not introduce such
adsorption sites that would preferentially collect water molecules. In any case, whether the initial
adsorption is droplet-wise or film-wise is a secondary point as the film is eventually formed in any
case, and ice nucleation takes place within the film and not within the nanodroplets. We have now
clarified this in the revised manuscript in Sec.3.5, Page 7:
The formation of nanodroplets at lower humidities depends on the wettability of the surface.
Given the hydrophobic nature of silver iodide [8], it suggests a tendency for droplet-wise rather than
film-wise adsorption. It should be noted that whether the initial adsorption is droplet-wise or film-
wise is a secondary point as the film is eventually formed on the surface. Importantly, the critical
1
event of ice nucleation occurs within the formed film and not within the nanodroplets as we observed
in our MD simulations. This emphasizes the ultimate formation of the film as the crucial factor in
ice nucleation.
The rate of adsorption in the simulation is determined not only by the pressure but also by the
frequency of MC trial for water insertion. Did the authors tune the frequency of MC trial so that it
reproduces the experimental rate of adsorption?
Response: We begin our answer to this comment by stressing that the supersaturation used
in the GCMC/MD simulations in this work is not to be confused with the physical vapor pressure
and the resulting rate of adsorption is never used in a quantitative manner nor even mentioned
throughout the manuscript. The target supersaturation as well as the MC/MD ratio are simple
tuning parameters that are chosen for the following reasons:
• The supersaturation is chosen to set the computational efficiency, keeping in mind the poor
parallelizability of the GCMC/MD simulations with atomistic models. Note that even well
parallelizable coarse grained GCMC/MD simulations [1] use supersaturations of 2.5 and 3.0.
• In a GCMC/MD simulation the Monte Carlo steps are used to add/delete water molecules while
the molecular dynamics steps drive molecular motion in space based on realistic Newtonian
dynamics. In simple terms, the higher the MC/MD ratio the more realistic is the resulting
adsorbed layer structure. The positions of freshly added water molecules are more efficiently
equilibrated if the ratio is high, suggesting that our adsorbed layer structure is closer to the
equilibrium structure of the real adsorbed layer at high MC/MD ratios.
We agree with the reviewer that the adsorption rate is influenced by the frequency of MC tri-
als. Our experimentation involved testing various MC/MD ratios, namely 500/100, 500/200, and
500/225, and the outcomes are detailed in both the Supporting Information and the manuscript.
Figures S2 and S3 depict the results for MC/MD ratios of 5 and 2.5, respectively, at a vapor pressure
of 60 kPa. Furthermore, simulations were carried out at a MC/MD ratio of 5 for a vapor pressure
of 6 kPa, which is more pertinent to atmospheric conditions. It is noteworthy that, for the MC/MC
ratio of 5, we observed the adsorption and desorption of water molecules. However, a complete
adsorption of a nanodroplet or layer was not observed. Furthermore, In the latest study conducted
by Lbadaoui-Darvas et al. [6], a comparison was made between a ratio of 20 and 5. The results did
not show a significant difference.
2) The deposition nucleation mode is not the only pathway for ice formation in the gas phase.
In most cases, in the atmosphere below the freezing point, snow is produced by a multi-step process
in which water droplets are first formed around dust particles, and then ice nucleation occurs. This
is a typical example of Ostwald’s step rule. Whether the crystallization process is classified to
adsorption nucleation or multi-step nucleation due to Ostwald’s rule depends on whether the liquid
phase is observed or not. Since a liquid water layer forms in the proposed mechanism, it is no
longer deposition nucleation. Multi-step nucleation is observed at low supercooling and deposition
nucleation is observed at high supercooling. The temperature examined in this study, 253 K, is too
high for deposition nucleation. I think that the authors merely observe the well-known multi-step
nucleation of ice in their simulations.
Response: With silver iodide as ice nucleus, 253K is not too high for deposition ice nucleation
as has been demonstrated in the laboratory measurements by [15] and as shown in our Fig. 7:
ice nucleation occurs well below water saturation. Regarding snow production, it is initiated by
heterogeneous freezing of micron-sized supercooled cloud droplets. This is quite different to ice
nucleation starting from a four-molecule thick adsorption layer that is initially more liquid- than
ice-like. Not only is the amount of water completely different in these two cases, also the timescales
are different.
2
We have attached a movie of the trajectory of one of our adsorption simulations to the revised
manuscript to present our observation of the multi-step mechanism. We hope this will be satisfactory
for the referee.
3) Figure 2b shows that crystalline structures form in the system with small amount of adsorbed
water when the temperature is low (T = 233, 223, and 213 K). It is needed to perform long simu-
lations in this temperature range to investigate deposition ice nucleation. A very different behavior
from that found in this study may be observed.
Response: As we mentioned in the manuscript, the GCMC/MD simulations are poorly par-
allelizable, therefore performing long simulations at lower temperatures such as 233 K and 213 K
suggested by the referee is not feasible as they would require even more time due to the slowed
kinetics, However, we conducted the GCMC/MD simulations at the mentioned temperatures for
540 ps time, showing that as the temperature decreased, the rate of adsorption decreased (Please
see Fig. 1).
In addition, as previously mentioned in the earlier response, laboratory measurements indicate
that ice nucleation on AgI clearly occurs below water saturation at 253 K, and there doesn’t appear
to be a qualitative difference as the temperature decreases, at least not one that can be conclusively
identified from the experimental data.
Figure 1: Temporal evolution of water adsorption on AgI (0001). The adsorbate is segmented into
layers, each having a thickness of 5 ˚A. The vapor uptake is simulated at a) 213 and 233 K with P=
60 kPa.
4) I thought that the row of N(L1) in Table 1 shows the number of water molecules in layer 1 at
the time shown in the second row of the table. The N(L1) value is 247 at 755 ps for run 4. There
is no panel corresponding this result in Figure 3.
Response: We agree with the referee that the value in N(L1) for the number of ice in the first
layer and Fig.3 did not match. We corrected the mistake and updated Fig.3 to the new timesteps
(Please see Fig. 2). In addition, we corrected the mistake pointed out by the referee in Table 1 in
the revised manuscript. We attached here the updated Fig.3 and Table 1.
5) Page 5: “Figure 5a-c illustrates the build up of hexagonal patterns over time”. It is not
clearly seen in Figure 5. Please show molecules or hydrogen bonds forming hexagonal patterns with
a different color.
Response: Thank you for pointing this out. We changed the color to represent the figure better
in the revised manuscript. The edited figure can be seen in Fig. 3:
3
Figure 2: Temporal evolution of water adsorption on AgI (0001) in 4 parellel simulations. The
adsorbate is segmented into layers, each having a thickness of 5 ˚A. The vapor uptake is simulated
at 253 K with P= 60 kPa.
Table 1: Time until first layer adsorption, number of adsorbed water molecules and their surface
coverage percentage is reported for the first (L1), second (L2), third (L3) and fourth (L4) layer at
T = 253K
Run Time (ps) N (L1) N (L2) N (L3) N (L4)
a 246 137 48 20 8
11.80% 4.10% 1.70% 0.70%
b 377 303 77 43 14
26.48% 6.73% 3.75% 1.22%
c 743 312 111 54 25
27.5% 9.72% 4.72% 2.20%
d 567 296 71 48 8
25.87% 6.20% 4.10% 0.70%
6) The critical radius depends not only on the ice/water surface free energy but also on the
ice/AgI surface free energy, but the latter is neglected in Eq. 1.
Response: The critical radius is the same for homogeneous freezing and heterogeneous freezing
at a given temperature. The volume of the critical cluster becomes smaller with lower contact
angles, but the radius remains the same. We do not know the actual value of the contact angle
(which contains information on the ice/AgI surface free energy via the Young equation), but if it is
90 degrees or below, then the critical cluster fits into the four-molecule adsorption layer. We have
now clarified this in the manuscript in Sec.3.4, Page 6.
It should be noted that the precise value of the contact angle, which provides information about the
ice/AgI surface free energy through the Young equation, is unknown to us. However, if the contact
angle measures 90 degrees or less, it indicates that the critical cluster aligns with the four-molecule
adsorption layer.
4
Figure 3: Atomistic details of the first adsorbed layer evolution on the AgI surface during the
GCMC/MD simulation at a) 250, b) 1000 and c) 2000 ps. AgI is colored in silver, oxygen and
hydrogen atoms of the water molecules are colored in red
5
2 Response To Referee 2
Comments to the Author In this article, authors performed MD simulations of deposition ice nucle-
ation using AgI substrate. Their findings are unique such that water molecules adsorb in patches
or clusters and then these clusters merge to become a continuous water layer are not well known
before. These results provide a new understanding of how deposition ice nucleation would occur at
molecular resolution. The paper is well written, and I recommend publication after addressing the
following minor comments.
Response: We thank referee 2 for the favorable review of our manuscript. In the revised
manuscript we tried to address all of the issues in the comments. Also, we tried to clarify the
misconceptions by adding clearer explanations where they were required.
1) It is not clear the selection of temperature range selected for the simulation (section 3.1). The
range used is 213 to 253K. This range is used because there were no experimental data (Fig 7a)
outside this temperature range?
Response: We consider this temperature range because it aligns with the experimental con-
ditions used in our research. In addition, the study of ice nucleation on the AgI (0001) has been
conducted at T= 253 K and T = 242 K in the work of [16] and [2]. With present-day computing
power, going below 213 is a technical challenge as the simulations slow down considerably when
temperature is lowered.
2) In reality, the AgI particles will not be smooth spheres but will have irregular morphology or
surface. Are simulations involving such heterogeneous structures considered? How results would be
affected if these structures were used in the simulations?
Response: This is of course an interesting question. The study of ice nucleation (immersion
freezing) using classical molecular dymaincs on defective and wedge shape AgI has been conducted
in the work of [12] and [11]. The results showed the ice nucleation hindered or slowed down on
the defective and wedge-shaped structures of silver iodide compared to the perfect AgI (0001),
however, the ice nucleation occurred in the simulation systems. Conducting GCMC/MD simulations
of defected AgI requires a rather time consuming set of completely new simulations, so we prefer to
leave it out from this manuscript and return to it in a future paper.
3) Recently David et al published a paper (also cited here 25 in the reference section) and
performed MD simulations. This previous paper highlights the importance of the pore condensation
and freezing mechanism of ice formation, and the traditional view of the deposition ice nucleation
mechanism is questionable. I suggest authors discuss this David et al paper in the context of the
results presented in this article (end of section 3).
Response: We thank the referee for this comment. We added an explanation about the men-
tioned paper at the end of Sec.3.2 of the revised manuscript (page 4).
The formation of liquid water as an intermediate pathway to ice nucleation has also been observed
on a nonporous slab, a model of a porous slab of silica with a pore in the work of David et al. [1].
However, while they observed ice nucleation on the surfaces with pores, the nano-porous silica surface
was unable to promote ice nucleation.
4) Please elaborate more on the atmospheric implications “water adsorption” note (end of section
4)
Response: We have added the atmospheric implications “water adsorption” note at the end of
Sec.4 of the revised manuscript, Page 8.
These results indicate an alternative route to deposition ice nucleation involving an intermediary
liquid phase. Given the significant radiative influence of cirrus clouds on climate [9], traditional
ice formation parameterizations based on deposition ice nucleation should be substituted with models
that integrate an adsorption mechanism in cirrus cloud models [7].
6
How existing ice nucleation theories can be improved further?
We have added the following discussion to the revised manuscript, Sec.2.3, Page 4.
The classical nucleation theory (CNT) stands as the most widely adopted explanation for hetero-
geneous nucleation. CNT outlines the conditions necessary for the formation of a critical droplet or
ice particle on a surface from a metastable phase, such as supersaturated vapor or supercooled liquid.
It assumes a single-step process without accounting for pre-critical interactions between the surface
and the vapor/liquid phase. An encouraging theoretical framework for addressing the aforementioned
challenges in CNT is adsorption nucleation theory (ANT) [3, 4]. By incorporating multilayer ad-
sorption into the model, ANT is able to resolve the issue of neglecting pre-critical surface-water
interactions. Our findings support ANT as we identified intermediate adsorption of liquid water
and interactions between water molecules and the AgI surface, suggesting a pathway to deposition
freezing.
We hope this will be satisfactory.
7
3 Response To Referee 3
The authors study the “deposition ice nucleation” from water vapor on AgI using combined GCMC/MD
simulation and the classical atomistic TIP4P/2005 water model. They identify four steps of this
process. Although this is not the first study of ice on AgI, the work provides several interesting
observations. However, there are issues which should be addressed before the paper is ready for
publication.
Response: We thank referee 3 for the favorable review of our manuscript. In the revised
manuscript we tried to address all of the issues in the comments. Also, we tried to clarify the
misconceptions by adding clearer explanations where they were required.
1. Details about GCMC are not given. What was the acceptance ratio? Were molecules in-
serted/deleted piece-wise? If not (which seems to me probable because of huge supersaturation
needed), the simulation is expected to be very inefficient, with a small acceptance ratio for both
insertion and deletion. The common trick is to consider “fractional molecules” as intermediate steps;
see, e.g., DOI = 10.1021/ct7000039,10.1063/1.4914461. Also, how the velocities were set after a MC
insertion?
Response: We generated Fig. 4 to illustrate the acceptance ratio of insertion and deletion of the
Monte Carlo (MC) simulations over the course of the simulation time. As depicted in Fig. 4, during
the initial phase of the simulation, both the acceptance ratio of insertion and deletion were relatively
high, approximately around 45% and 34% for pressure 60 kPa and 6kPa, respectively. However, as
the simulation progressed towards its end, the acceptance ratio decreased, reaching approximately
10% for the insertion ratio and around 3% for the deletion ratio for the P=60 kPa. This decline can
be attributed to the condensation of water molecules on the AgI surface in the later stages of the
simulation. Consequently, the deletion ratio decreased due to the condensation of water molecules on
the AgI surface. The insertion/deletion ratio fluctuates for the P= 6 kPa as the simulation time for
this system was short. During the insertion of an atom or molecule in a simulation, its coordinates
are randomly selected within the existing simulation cell or region. Subsequently, new velocities
for the atom are chosen from a temperature distribution specified by T. Optionally, the effective
temperature for these new velocities can be adjusted using the tfac insert keyword. The tfac insert
value chosen in our simulation is 1.66 which gives a good results in terms of equilibrating the system
with the target temperature see 7. The insertion point is determined by placing the center of mass
of the molecule at the specified location, and the orientation of the molecule is randomly determined
by rotation around this central point. We are aware of the issues of computational efficiency and we
will optimize our method based on the suggestions of reviewer 3. Note that this work is only the
3rd attempt to simulate deposition freezing and not simple adsorption using molecular dynamics
or related techniques, therefore the methodology is far from being perfect, nevertheless, qualitative
conclusions that are reached from this work are valid.
We have added the following discussion about in the revised manuscript Sec.2.2, page 2:
The insertion/deletion step can be performed more efficiently by Continuous Fraction Compo-
nent (CFC) Monte Carlo method. Within moves of the CFC reaction, the insertion or deletion of
molecules is discretized. This is achieved by gradually activating or deactivating their interactions
with the rest of the system over multiple Monte Carlo (MC) moves [13, 10]. However, in this work,
we follow the same insertion/deletion method of David et al. and Lbadaoui-Darvas et al. [1, 6]
The authors say (about the GCMC algorithm): “Water molecules are added and deleted accord-
ing to the Metropolis algorithm until a stable concentration is reached.” Since the supersaturation
is huge (60 kPa is the water vapor pressure at 86 ◦C), the attached water (ice) just keeps growing
(Fig. 3). Which “stable concentration” do you mean?
Response: We agree with referee 3 that the supersaturation is exaggerated, but as in earlier
8
Figure 4: Acceptance ratio of insertion and deletion in GCMC/MD simulations at P=60 kPa.
works simulating deposition nucleation [1], [6] this is done to speed up the simulations. In particular,
the GCMC/MD method used here is not parallelizable and therefore simulations at atmospherically
relevant supersaturations are not feasible. However, the adsorption of the first 4 layers was stable
in all of our simulations according to the Fig. 3 of submitted manuscript. In addition, Fig. 6 shows
the number of water molecules as a function of time. The number of water molecules increases to
reach a plateau indicating that saturation has been reached.
A GCMC insertion of a full (not fractional) molecule with a high chemical potential (supersat-
uration) deposits a molecule on a surface with enhanced probability and also with a high potential
energy. The potential energy is relaxed in the following MC or MD steps, but removing a molecule
is not probable. Please provide the ratio of insert to removed molecules in the course of your GCMC
simulation at both pressures considered. I expect that this ratio will be high; if this is the case, one
can replace GCMC essentially by any method attaching molecules to the growing surface. In other
words, I do not see any advantage of the GCMC method here. The Metropolis GCMC algorithm
has been derived for systems in thermal and chemical potential equilibrium. Thus, the GCMC at
slightly subsaturated conditions should be able to predict the number of (above mentioned) ice-like
layers in equilibrium. However, an efficient insertion/deletion algorithm would be needed. It could
also have some sense to simulate slightly supersaturated systems so that the layers have enough time
to relax and rearrange before more water gets adsorbed.
Response: We thank the reviewer for pointing out these technical aspects and possibilities to
improve the methodology of studying deposition freezing in situ using molecular simulations. Please
note that this paper is only the third one using molecular simulations for deposition freezing after
[1] and [6]. That means the methodology is far from being perfect. We attach the required graphs
and will consider in the future to optimize our methods along the lines pointed out in this comment.
However, we would like to point out that GCMC/MD is useful for studying deposition freezing,
which is a combination of two processes: i) the adsorption of water vapor on ice nucleating surfaces
and ii) the subsequent freezing of the adsorbed layer. GCMC is well-known technique to study
adsorption with a plethora of examples relevant to atmospheric systems [5]. MD on the other hand
has been widely used to study crystallization [14]. The combination of the two allows for studying
the adsorption upon the addition of molecules to the vapor phase.
9
Figure 5: Acceptance ratio of insertion and deletion in GCMC/MD simulations at P=6 kPa.
2. I do not understand (the interpretation of) the Kelvin equation. This equation (aka Ost-
wald–Freundlich) as well as the classical nucleation theory give the following expression for the
critical nucleus:
r∗ =
2·vσn
s
μs − μn
, (1)
where n = nucleusphase, s = supersaturatedphase, μ = chemical potential. For s = gas, one may
write μs − μn = kTlnS, where S = psat (ideal gas approximation); here psat denotes the saturated
vapor pressure above a flat surface. This looks similar to Eq. (1). However, the authors say that
σi/w ”the surface tension between ice and water”. The Kelvin equation is still a reasonably good
approximation for nucleation of ice from supercooled water, but why quantity Si = exp[ (μw −μi)
kT ] is
called ”saturation”?
Response: It can be shown that the chemical potential difference in freezing of ice in water equals
kT times the logarithm of Pw/Pi where the P’s denote saturation vapor pressures of supercooled
water and ice, respectively. We now clarify this in the revised manuscript in Sec.3.4, Page, 5 as
follows:
Si = Pw/Pi, where the P ’s denote saturation vapor pressures of supercooled water and ice at the
temperature T , respectively. The calculations show that the dependence of Sw on T for a fixed radius
of the critical embryo is similar as the experimentally observed temperature dependence of the onset
Sw for deposition ice nucleation on AgI particles.
3. (Sec. 2.2, 1st par) Timestep 5 fs is too long. E.g., the rotational and translational temperatures
differ by 8 K (with leap-frog kinetic energy). Is this a typo?
Response: We have corrected this typo in the revised manuscript. The timestep was 0.5 fs.
Thanks!
10
Figure 6: The changes in the number of water molecules throughout the GCMC/MD simulation.
Figure 7: Temperature fluctuation throughout GCMC/MD simulation
4. The ice-like (at AgI) and water-like (close to the surface) layers depicted on Fig. 2 reflect
premelting; see, e.g., DOI = 10.1063/1.4895399.
Response: The ice-like structures or interfacial ice in the hydration layer on the surface of
silver iodide is recognized by the LICH-TEST algorithm, and the percentage of ice is reported in
Fig. 2. The premelting layer denotes a thermodynamically stable disordered structure present at
the interface of an otherwise ordered crystalline solid. It occurs when the temperature is close to,
but below, the bulk melting temperature. However, in our MD simulations, the temperatures were
between 213-253 K.
11
5. At a formal level: I think that multilayer adsorption at slightly subsaturated conditions (the
authors mention 4 layers for S = 0.95, reflecting a small water/AgI contact angle) should not be
called “heterogeneous ice nucleation”. Similarly, in pores and subsaturated environments, one may
observe capillary condensation with r∗ < 0 (concave menisci) according to the Kelvin equation.
Response: We do not equate multilayer adsorption and heterogeneous nucleation, our message
is that multilayer adsorption with at least four layers precedes ice nucleation. Please note that at
temperatures below 268 K, water vapor is supersaturated with respect to ice when the saturation
ratio with respect to supercooled water is 0.95 (see blue dashed line in Fig. 7a).
References
[1] R. O. David, C. Marcolli, J. Fahrni, Y. Qiu, Y. A. P. Sirkin, V. Molinero, F. Mahrt,
D. Br¨uhwiler, U. Lohmann, and Z. A. Kanji. Pore condensation and freezing is responsible for
ice formation below water saturation for porous particles. Proc. Natl. Acad. Sci., 116(17):8184–
8189, 2019.
[2] G. Fraux and J. P. K. Doye. Note: Heterogeneous ice nucleation on silver-iodide-like surfaces.
J. Chem. Phys., 141(21):216101, 12 2014.
[3] A. Laaksonen. A unifying model for adsorption and nucleation of vapors on solid surfaces.
J. Phys. Chemis. A, 119(16):3736–3745, 2015. PMID: 25831213.
[4] A. Laaksonen and J. Malila. An adsorption theory of heterogeneous nucleation of water vapour
on nanoparticles. Atmos. Chem. Phys., 16(1):135–143, 2016.
[5] M. Lbadaoui-Darvas, G. Garberoglio, K. S. Karadima, M. N. D. S. Cordeiro, A. Nenes, and
S. Takahama. Molecular simulations of interfacial systems: challenges, applications and future
perspectives. Mol. Simul., 49(12):1229–1266, 2023.
[6] M. Lbadaoui-Darvas, A. Laaksonen, and A. Nenes. Deposition freezing, pore condensation
freezing and adsorption: three processes, one description? Atmospheric Chemistry and Physics,
23(17):10057–10074, 2023.
[7] X. Liu and J. E. Penner. Ice nucleation parameterization for global models. Meteorol Z (Berl),
14(4):499–514, 09 2005.
[8] C. Marcolli, B. Nagare, A. Welti, and U. Lohmann. Ice Nucleation Efficiency of AgI: Review
and New Insights. Atmos. Chem. Phys., 16(14):8915–8937, 2016.
[9] A. V. Matus and T. S. L’Ecuyer. The role of cloud phase in earth’s radiation budget. Journal
of Geophysical Research: Atmospheres, 122(5):2559–2578, 2017.
[10] F. Moucka, D. Bratko, and A. Luzar. Electrolyte pore/solution partitioning by expanded grand
canonical ensemble Monte Carlo simulation. J. Chem. Phys., 142(12):124705, 03 2015.
[11] G. Roudsari, O. H. Pakarinen, B. Reischl, and H. Vehkam¨aki. Atomistic and coarse-grained
simulations reveal increased ice nucleation activity on silver iodide surfaces in slit and wedge
geometries. Atmospheric Chemistry and Physics, 22(15):10099–10114, 2022.
[12] G. Roudsari, B. Reischl, O. H. Pakarinen, and H. Vehkam¨aki. Atomistic simulation of ice
nucleation on silver iodide (0001) surfaces with defects. The Journal of Physical Chemistry C,
124(1):436–445, 2020.
12
[13] W. Shi and E. J. Maginn. Continuous fractional component monte carlo: An adaptive biasing
method for open system atomistic simulations. J. Chem. Theory and Comput., 3(4):1451–1463,
2007. PMID: 26633216.
[14] G. C. Sosso, J. Chen, S. J. Cox, M. Fitzner, P. Pedevilla, A. Zen, and A. Michaelides. Crystal
nucleation in liquids: Open questions and future challenges in molecular dynamics simulations.
Chem. Rev., 116(12):7078–7116, 2016.
[15] A. Welti, K. Korhonen, P. Miettinen, A. A. Piedehierro, Y. Viisanen, A. Virtanen, and A. Laak-
sonen. Spin modification for low-temperature experiments. Atmos. Meas. Tech., 13(12):7059–
7067, 2020.
[16] S. A. Zielke, A. K. Bertram, and G. N. Patey. A molecular mechanism of ice nucleation on
model agi surfaces. J. Phys. Chem. B, 119(29):9049–9055, 2015. PMID: 25255062.
13

02-Dec-2023

Dear Dr lbadaoui-darvas:

Manuscript ID: EA-ART-09-2023-000140.R1
TITLE: Molecular scale mechanism of deposition ice nucleation on Silver Iodide

Thank you for submitting your revised manuscript to Environmental Science: Atmospheres. I am pleased to accept your manuscript for publication in its current form. I have copied any final comments from the reviewer(s) below.

You will shortly receive a separate email from us requesting you to submit a licence to publish for your article, so that we can proceed with the preparation and publication of your manuscript.

You can highlight your article and the work of your group on the back cover of Environmental Science: Atmospheres. If you are interested in this opportunity please contact the editorial office for more information.

Promote your research, accelerate its impact – find out more about our article promotion services here: https://rsc.li/promoteyourresearch.

We will publicise your paper on our Twitter account @EnvSciRSC – to aid our publicity of your work please fill out this form: https://form.jotform.com/211263048265047

How was your experience with us? Let us know your feedback by completing our short 5 minute survey: https://www.smartsurvey.co.uk/s/RSC-author-satisfaction-energyenvironment/

By publishing your article in Environmental Science: Atmospheres, you are supporting the Royal Society of Chemistry to help the chemical science community make the world a better place.

With best wishes,

Prof. Nønne Prisle
Associate Editor, Environmental Sciences: Atmospheres

Reviewer 3

The authors have clarified the issues. I recommend the revised paper for publication.