The manuscript entitled “Rapid hydration and weakening of anhydrite under stress: Implications for natural hydration in the Earth’s crust and mantle” by Heeb and coauthors addresses the role of the hydration reaction in anhydrite to gypsum under effective pressure and differential stress to constrain the role of stress in the progress of the reaction and its rheological evolution. The experimental data set and particularly the microstructural characterization of starting and reactant products is outstanding.

I fully agree with the authors in the importance of hydration reactions in many several geodynamic scenarios, particularly in the context of active settings.

The main characteristic of hydration reactions is the increase in the volume of the solids. Although it is mentioned few times in the text, it is not given the relevance it has, as it probably explains the behavior of the reaction, particularly in the case of deforming reactant rock. The access of water to hydrate the rock requires the existence of a network of fluid pathways, which become sealed as the reaction progresses, given the large increase in volume of the solid during the reaction. In an static environment, this may be a contributing factor to halt the progression of the reaction. In the case of a dynamic environment, the continuous introduction of microfracturing, damage and pore space keeps the reaction going as it provides pathways for fluid to access unreacted material. The experiments done by Heeb et al. gain more relevance when addressed from this point of view. Perhaps it is no surprise that the reaction progresses faster or earlier in the case of a dynamic or stressed environment (thus strained) in comparison with a static set up. The experiments presented illustrate this very well.

But aside from considering this in the introduction and presentation of the tests, I also find several things in how the experiments are presented that need clarifying. In some cases the description of experimental procedure is confusing, in other perhaps more information is needed. The main issue is regarding the description of the tests, particularly the differences between the different sets of experiments, in some cases because there are several things mixed or because of the terminology used. For instance, the difference between “wet” and SSDC tests (steady-state differential compaction). Both require pore fluid pressure, the main difference resides in the magnitude of the effective pressure. But I’m not sure that this difference justifies their classification as different set, as there are other factors, mainly time. These also undergo large differences in time at pressure, the SSDC having almost a magnitude longer at stress that the “wet” ones. The graph in fig 3 shows the mechanical data in all tests, but it does not show the differences in confining pressure, and also in effective pressure, where there is a large difference between Ò1 and Ò7 or Ò8. It would be better for the reader to show separately the curves at different effective pressure, for instance, or color coding tests under similar conditions.

The usage of “initiating the strain rate” is somehow confusing. I take it to mean initiating the loading, which is what it is normally used. But one needs to bear in mind that one thing is the moving rate of the piston, which is easily kept constant, another is keeping the strain rate constant, which requires to recalculate the speed at which the piston moves to keep strain rate constant. I’m aware the differences may be minimal at low strain but they will increase as the shortening of the samples builds up. And by the look of some of the samples, we are probably closer to the latter than the former.

With regards to the procedure of the testing, in some of the experiments, there are instances where the piston is stopped, but the differential stress is kept constant, which is odd, because as soon as you stop the loading, the sample will start to relax reducing the differential stress. In relaxation tests, the strain rate will vary with time, in that case orders of magnitude depending on the times at which the sample is left to relax. How exactly is conducted this part of the experiments is not clear in the description, but in Fig 7 there is a segment in the loading curve where effective stress is kept constant while strain keeps accumulating, for several hours. The time is perhaps not too relevant for the amount of strain, but it certainly is for the progress of the reaction, since at this stage the sample remains stressed, thus, in conditions potentially favouring the progress of the reaction.

The name given to the tests Ò1 and Ò2 is confusing:  “steady-state differential compaction under fluid pressure” (line 219). As written it can be understood as the fluid pressure what produces the compaction, when in fact it is the effective pressure which does. If the “strain rate” is put on hold, probably meaning that the piston is stopped, then the reaction is progressing during relaxation at high effective stress, in this case provided by the initial 100+ MPa differential stress, which will reduce very likely over time. The time at which the samples are left loaded is important (I assumed is tssdc in table 2) as despite some relaxation of stresses, they presumably will still be high enough to favour compaction.

However in line 260 it is said that the piston is stopped before fluid pressure was applied, is that correct?? If that is the case, then it is possible that a lot of fracturing is induced by reducing effective pressure. This part requires more detail. If it is how it is said, then certainly “steady-state” is not the right word to describe this type of experiment.

The microstructural work is outstanding and complements very well the lab work, once the procedures are presented more clearly. I would also suggest to rebrand the names of the tests to reflect more objectively the type of test.

Once the flow of events in the running of the tests is clear, I would probably put more emphasis on the fact that the time at high stress is the key factor to enhance further hydration. 

I have some other minor comments that I make in order of appearance.

In the graphical abstract I wonder whether the first and last sketches are oriented similarly to the three middle sketches or that the shear sense in the fault in the sketch to the right is wrong, as it does not agree with the major faults in the middle three drawings. As far as one can tell, there are no reverse fault movements in the sample cylinder under vertical shortening.

Line 88, I would rewrite “… influencing the reaction activity and kinetics of hydration to anhydrite…”

Line 90, I would replace “material-specific characteristics (petrography)” by “microstructure”

Section 1.2 Mechanisms of anhydrite hydration

Somewhere in the text, the volume increase related to hydration needs to be given, perhaps this is the section. This is a key parameter and may potentially control the progress of the reaction.

3.1.2 Mechanical data

I would not consider in the stress-strain curves anything that happens before the linear elastic behavior as they are probably artifacts, more to do with the assembly of the sample and the adjustment of the loading column than with the internal deformation of the sample. And any of this can be seen in the graph in Fig. 3 anyway.

Line 280, That “gypsum infill implies an extensional component to the kinematics of these structures” very much depends on whether the net solid volume increase exceeds the porosity generated during fracturing. If not, then it is not strictly necessary. That is one reason why the volume increase during gypsification needs to be given. And consequently, an estimate of the porosity generated during fracturing (which perhaps will be easily estimated using the SEM micrographs or even EBSD images).

Fig. 5

The size of some of the gypsum grains is very large and will certainly biased the CPO when using the complete data set. Is it the same with 1 point per grain as used for anhydrite?

Lines 309-321

CPO data in anhydrite is shown as point per grain, while gypsum is the complete data set. I wonder how much that plays a role in the interpretation in this paragraph.

4.2.1. Rapid hydration of anhydrite under stress

Other than stress is the fact that there might be some microfractures produced due to high effective pressure, which is what ultimately allows the fluid to permeate the samples and have access to the inside of the specimens. This is not considered here and it is key, given that as it is said, there are fractures and grain size reduction by comminution.

In this regard, it is important to clarify whether the fluid pressure to SSDC tests is applied after loading to +100 MPa or before.

Line 385, I wouldn’t use “quasi-elastic stress-strain behavior” to refer to the inelastic part of the curves.

Line 390, “re-application of strain rate”: reloading of the sample from x differential stress to maximum differential stress.

Fig. 7

I would indicate in the graph the amount of time at SSDC.

Fig. 8.

If sigma 1 is vertical, then the shear sense in the sketch to the right (reverse fault) is wrong.

Line 468-470. That the weakening of sample Ò2 comes from the appearance of gypsum instead of the fracturing is difficult to ascertain from outside, really.

Line 474, A consequence of hydration under stress is the weakening of the sample during deformation.

Line 476, tests Ò4 and Ò7 are not at slow strain rates, but the highest according to table 2.

Lines 482-486, “A stronger connected shear fracture network developed until the onset of isotropic principal stress conditions…” If these experiments were initiated at high differential stress, I can’t see how they are evolving at isotropic principal stress conditions.

Line 498, “Chemical potential depends on a ‘weighted’ mean stress, which means that the magnitude and orientation of stress have a relatively minor impact”. Surely the magnitude is important, since influences the mean stress. Is this quote correct?

Line 546-549,  ...” fluid migration through shear zones facilitated highly localized eclogitization of anhydrous (granulite) crust along these zones and can result in transient mechanical weakening, brittle deformation and earthquakes”.

I’m not sure this is an adequate analogue system for the gypsification. Gypsum is much weaker that anhydrite, I’m not so sure eclogite is that much weaker than granulite. Also, the weakening in the granulite case is the consequence of the fracturing, not necessarily of the hydration itself, unless the eclogites were deformed. If the latter is the case, it needs to be mentioned. 

Please find my comments in the attached PDF file. 

Kind regards,

James Gilgannon