Sediment-Peridotite Reaction Controls Fore-Arc Metasomatism and Arc Magma Geochemical Signatures

Round 1

Reviewer 1 Report

Review

Geoscience Manuscript 1341149

 

The manuscript “Sediment-peridotite reaction controls fore-arc metasomatism and arc magma geochemical signatures” by Forster and colleagues is a very interesting and well-written manuscript that aims to further our understanding of how sediments subducted with the oceanic lithosphere melt and metasomatize infertile peridotite in the fore-arc region. To explore this topic, the authors perform reaction experiments between sediments and peridotite over a pressure and temperature interval that corresponds to the subducting slab surface. LA-ICP-TOFMS is used to study the transfer and retention of trace elements between dunite and the metasomatic layers. The authors show that the sediments react with the depleted peridotite to form a layered reaction front and that the chemical characteristics of the metasomatic layers closely resemble the trace element characteristics of high-K2O magmas. Using the experimental products conducted over a range of time intervals, the growth rates of the metasomatic front are estimated.

 

This manuscript builds upon previous studies published by Forster and colleagues and is an important contribution to our understanding of processes leading to the transfer of material from the subducting slab into the overlying mantle wedge and how these influence the geochemical characteristics of arc volcanoes and post-collision High-K magmas.

My recommendation is that the manuscript is published with only minor revisions.

 

 

General comments

This manuscript is in great form and requires very little editing prior to publication. My comments are mainly related to terminology and a discussion on the broader feasibility of this mechanism to influence arc geochemistry at a global scale considering the wide range of subduction parameters (slab dip angle, slab age, subduction zone thermal structure, and amount of sediments) that vary between different arcs.

 

The only section that seems confusing to me was the use of the term “partition coefficient” to show the trace element enrichment between the metasome and the unreacted dunite. Apart from a few elements (Ni and Fe as mentioned on L202) the trace element enrichment in the metasome is being derived from the sediment. Thus, it would make more sense to call this a “partition coefficient” if the reaction zone was normalized to count rates in the sediment residue. It seems more appropriate to call this value an “enrichment factor” or some other term that doesn’t imply partitioning between two phases (or many phases if it is a bulk D) since the original mineralogy is no longer present. If the D ^{reactionzone/dunite} resemble bulk partition coefficients, it would strengthen your argument that these are reasonable values if you calculated a range of bulk D’s from previously published partition coefficients for olivine, spinel, and clinopyroxene and a high-K melt. Including these in figure 4a would allow the reader to assess the reliability of the technique and the subsequent instantaneous fractional melting calculations.

On this same topic, I found it confusing when the authors were discussing the retention of HREE within the sediment residue due to garnet and this was reflected in the low “partition coefficient” for HREE (L297). Two points worth mentioning: 1) The “partition coefficients” for the HREE are all greater than 1 which is inconsistent with this statement; 2) It seems inappropriate to use your calculated “partition coefficients” to describe this process when the “partition coefficients” really describe the relative enrichment between the metasome and the dunite.

           

My other comment relates to the feasibility of this model to control volcanic arc trace element geochemistry or if it is a part of a larger system including the mechanism listed on L38-41. I completely agree with the interpretations presented in this study and I think they are significant results. However, I question whether subducted sediments melt with 100% efficiency in all arc settings. For this mechanism to retain the majority of the HFSE and impart a HFSE depleted arc signature for all arc-front volcanoes, the entire sedimentary cover must completely melt within a relatively restricted window. Is this a feasible assumption considering that sediments can be 300 to 500 m thick? What are the temperature gradients within the sediments as the slab descends into the mantle? Would the entire sedimentary package reach ~675°C before the subducting slab reaches the peridotite wet solidus considering the wide range of slab velocities? Would the melting front propagate downward into the sedimentary package? I don’t expect the authors to answer these questions, but they are ideas that I pondered while reading this manuscript and I would bet that future readers would have similar questions.

 

Were the authors concerned with the abrupt shift in the presence of glasses between the two experimental methods? It seemed a little unusual to have glasses present in nearly all the experiments performed using the piston cylinder-apparatus (1-3 GPa) while glasses were only present in one of the experiments performed using a belt apparatus (4-6 GPa)?

 

Specific comments

L109 The dunite has >95% olivine here but on L100 is indicates >97%.

L 130   There are two references included at the end of this paragraph

L131    How were the glass beads prepared?

L164    How can you know that the sediments partially melted at 1 GPa if none of the experiments were conducted at that pressure?

L176    Should there be a period at the end of this sentence?

L214    The phases listed in the caption are not labeled on the images. However, the labels are included in Figure 1 but they are not included in the caption.

L282    What is the significance of Cl- and CO2-bearing hot springs in the front of the volcanic arc? Does it support this model? It would be good to elaborate on this point.

L459    What does the M represent in this equation?

L472 What else could have been lost in the fluid phase? Would the loss of a significant portion of FeO impact the fO₂ of the system?

 

 

Author Response

 Review 1

The manuscript “Sediment-peridotite reaction controls fore-arc metasomatism and arc magma geochemical signatures” by Forster and colleagues is a very interesting and well-written manuscript that aims to further our understanding of how sediments subducted with the oceanic lithosphere melt and metasomatize infertile peridotite in the fore-arc region. To explore this topic, the authors perform reaction experiments between sediments and peridotite over a pressure and temperature interval that corresponds to the subducting slab surface. LA-ICP-TOFMS is used to study the transfer and retention of trace elements between dunite and the metasomatic layers. The authors show that the sediments react with the depleted peridotite to form a layered reaction front and that the chemical characteristics of the metasomatic layers closely resemble the trace element characteristics of high-K2O magmas. Using the experimental products conducted over a range of time intervals, the growth rates of the metasomatic front are estimated.

This manuscript builds upon previous studies published by Forster and colleagues and is an important contribution to our understanding of processes leading to the transfer of material from the subducting slab into the overlying mantle wedge and how these influence the geochemical characteristics of arc volcanoes and post-collision High-K magmas.

My recommendation is that the manuscript is published with only minor revisions.

 General comments

This manuscript is in great form and requires very little editing prior to publication. My comments are mainly related to terminology and a discussion on the broader feasibility of this mechanism to influence arc geochemistry at a global scale considering the wide range of subduction parameters (slab dip angle, slab age, subduction zone thermal structure, and amount of sediments) that vary between different arcs.

The only section that seems confusing to me was the use of the term “partition coefficient” to show the trace element enrichment between the metasome and the unreacted dunite. Apart from a few elements (Ni and Fe as mentioned on L202) the trace element enrichment in the metasome is being derived from the sediment. Thus, it would make more sense to call this a “partition coefficient” if the reaction zone was normalized to count rates in the sediment residue. It seems more appropriate to call this value an “enrichment factor” or some other term that doesn’t imply partitioning between two phases (or many phases if it is a bulk D) since the original mineralogy is no longer present. If the D ^{reactionzone/dunite} resemble bulk partition coefficients, it would strengthen your argument that these are reasonable values if you calculated a range of bulk D’s from previously published partition coefficients for olivine, spinel, and clinopyroxene and a high-K melt. Including these in figure 4a would allow the reader to assess the reliability of the technique and the subsequent instantaneous fractional melting calculations.

On this same topic, I found it confusing when the authors were discussing the retention of HREE within the sediment residue due to garnet and this was reflected in the low “partition coefficient” for HREE (L297). Two points worth mentioning: 1) The “partition coefficients” for the HREE are all greater than 1 which is inconsistent with this statement; 2) It seems inappropriate to use your calculated “partition coefficients” to describe this process when the “partition coefficients” really describe the relative enrichment between the metasome and the dunite.

We changed “partition coefficients” to “enrichment factors” as this describes the process more correctly. Indeed, we are looking at the metasomatic enrichment of the dunite infiltrated by a siliceous ultrapotassic melt that is not in equilibrium with the dunite.

My other comment relates to the feasibility of this model to control volcanic arc trace element geochemistry or if it is a part of a larger system including the mechanism listed on L38-41. I completely agree with the interpretations presented in this study and I think they are significant results. However, I question whether subducted sediments melt with 100% efficiency in all arc settings. For this mechanism to retain the majority of the HFSE and impart a HFSE depleted arc signature for all arc-front volcanoes, the entire sedimentary cover must completely melt within a relatively restricted window. Is this a feasible assumption considering that sediments can be 300 to 500 m thick? What are the temperature gradients within the sediments as the slab descends into the mantle? Would the entire sedimentary package reach ~675°C before the subducting slab reaches the peridotite wet solidus considering the wide range of slab velocities? Would the melting front propagate downward into the sedimentary package? I don’t expect the authors to answer these questions, but they are ideas that I pondered while reading this manuscript and I would bet that future readers would have similar questions.

Yes, I guess depending on slab velocity, angle, and sediment thickness there would be everything from completely melting to partially or nearly no melting.

Were the authors concerned with the abrupt shift in the presence of glasses between the two experimental methods? It seemed a little unusual to have glasses present in nearly all the experiments performed using the piston cylinder-apparatus (1-3 GPa) while glasses were only present in one of the experiments performed using a belt apparatus (4-6 GPa)?

It is a result of the higher pressure at similar temperature. We discussed this in our publication from 2019: Förster, M.W., Foley, S.F., Marschall, H.R., Alard, O., Buhre, S. 2019. Melting of sediments in the deep mantle produce saline fluid inclusions in diamonds. Science Advances, 5(5), 1-6.

Specific comments

L109 The dunite has >95% olivine here but on L100 is indicates >97%.

Corrected L109 to >97%.

L 130   There are two references included at the end of this paragraph

Corrected.

L131    How were the glass beads prepared?

Added this information.

L164    How can you know that the sediments partially melted at 1 GPa if none of the experiments were conducted at that pressure?

This should read 2-3 GPa. Changed accordingly.

L176    Should there be a period at the end of this sentence?

Added a semicolon.

L214    The phases listed in the caption are not labeled on the images. However, the labels are included in Figure 1 but they are not included in the caption.

Moved to Figure 1.

L282    What is the significance of Cl- and CO2-bearing hot springs in the front of the volcanic arc? Does it support this model? It would be good to elaborate on this point.

The fluids are themselves indicative of reaction processes in the mantle, however, it is not much known about their origin.

L459    What does the M represent in this equation?

M is the modal proportion of each phase, added into text.

L472 What else could have been lost in the fluid phase? Would the loss of a significant portion of FeO impact the fO₂ of the system?

We do not expect the significant loss of other elements to a fluid phase as the mass balance shows that those elements are close to the expected concentrations. From trace element mapping we also see that trace elements lost from the sediment are detected within the reaction zone. A significant loss of FeO could potentially impact the fO₂ in an unbuffered experiment, however, our fO₂ is dictated by the inner graphite capsule and we see the same reaction zone across multiple experiments with various FeO losses, so impact of FeO loss is probably very low.

Reviewer 2 Report

The pape entitled "Sediment-peridotite reaction controls fore-arc metasomatism and arc magma geochemical signatures" submitted by Förster and co-workers presents the results of high pressure experiments to constrain the composition of metasomstic layers produced by the infiltration of sediment-derived reactive melts into depleted peridotites in convergent margins and discuss the implications for the generation of potassic magmas in this petrogenetic environment.

English style is clear although some sentences need rephrasing or clarification. The topic is very interesting and the experimental approach is correct. However, I have found significant deficiencies in assumptions and methods for the geochemical modeling that must be corrected before acceptance of the work. 

Below I comment my major concerns (see annotated pdf for additional comments)

Major comments 

Lines 231-233: In these lines it is indicated that trace element count rates from LA-ICP-TOFMS of the reaction zone in the 3 GPa/850 ºC experiment were normalized to count rates in the dunite to give a qualitative estimate of element partition coefficients. I don't agree with this procedure for estimating partition coefficients as it implies to consider the reactive metasomatic bands as the liquid in equilibrium with olivine. The correct approach would be to explain these ratios (for each single band)  in terms of the Gresens-Grant theory and to plot them into isocon diagrams to evaluate losses and gains and to determine which parameters were conserved  (volume, total mass, abundance of compatible elements as Ni (?)). 

Could you extend this analysis to major and minor elements?

Lines 233-235: These findings imply to assume conservation of the total mass. This must be justified with a Gresens-Grant analysis.

Lines 240-244 and Figure 4: There is a confusion. In the text it is indicated that trace element composition of melt which infiltrated the dunite to form the metasome was calculated using the instantaneous fractional melting equation from [43] for 1-50 % melt in the 3 GPa/850 ºC experiment; however in Figure 4 it is plotted the calculated metasome composition. They are different things, one would be the initial infiltrating melt, the other the final composition that is a mix of a reactive liquid component and a residual olivine component. To calculate this latter it is necessary to apply reactive fluid flow models; see examples of numerical approaches in Keller and Katz (2016; Journal of Petrology, 57: 1073–1108), Zhao, et al. (2020, Miner Petrol 114: 141–159), Borghini et al (2020; Chemical Geology, 532, 119252), Vernieres et al. (1997; J. Geophys. Res., 102 , pp. 24771-24784).

Lines 298-301: I agree with the ratios representing a metasomatic enrichment, but they can not be interpreted as partition coefficients.

Minor comments 

Figure 5: This Figure needs clarification. The unit in y-axis is distance/time, but distance and thickness are also plotted in the diagram.

Comments for author File: Comments.pdf

Author Response

Review 2

The pape entitled "Sediment-peridotite reaction controls fore-arc metasomatism and arc magma geochemical signatures" submitted by Förster and co-workers presents the results of high pressure experiments to constrain the composition of metasomstic layers produced by the infiltration of sediment-derived reactive melts into depleted peridotites in convergent margins and discuss the implications for the generation of potassic magmas in this petrogenetic environment.

English style is clear although some sentences need rephrasing or clarification. The topic is very interesting and the experimental approach is correct. However, I have found significant deficiencies in assumptions and methods for the geochemical modeling that must be corrected before acceptance of the work. 

Below I comment my major concerns (see annotated pdf for additional comments)

Major comments 

Lines 231-233: In these lines it is indicated that trace element count rates from LA-ICP-TOFMS of the reaction zone in the 3 GPa/850 ºC experiment were normalized to count rates in the dunite to give a qualitative estimate of element partition coefficients. I don't agree with this procedure for estimating partition coefficients as it implies to consider the reactive metasomatic bands as the liquid in equilibrium with olivine. The correct approach would be to explain these ratios (for each single band)  in terms of the Gresens-Grant theory and to plot them into isocon diagrams to evaluate losses and gains and to determine which parameters were conserved  (volume, total mass, abundance of compatible elements as Ni (?)). 

Could you extend this analysis to major and minor elements?

We added an isocon plot and Gresens-Grant analysis for each metasomatic layer.

Lines 233-235: These findings imply to assume conservation of the total mass. This must be justified with a Gresens-Grant analysis.

The total mass is conserved, all gains are balanced by a decrease of FeO and MgO in the reaction zone. We show this in the Gresens-Grant analysis.

Lines 240-244 and Figure 4: There is a confusion. In the text it is indicated that trace element composition of melt which infiltrated the dunite to form the metasome was calculated using the instantaneous fractional melting equation from [43] for 1-50 % melt in the 3 GPa/850 ºC experiment; however in Figure 4 it is plotted the calculated metasome composition. They are different things, one would be the initial infiltrating melt, the other the final composition that is a mix of a reactive liquid component and a residual olivine component. To calculate this latter it is necessary to apply reactive fluid flow models; see examples of numerical approaches in Keller and Katz (2016; Journal of Petrology, 57: 1073–1108), Zhao, et al. (2020, Miner Petrol 114: 141–159), Borghini et al (2020; Chemical Geology, 532, 119252), Vernieres et al. (1997; J. Geophys. Res., 102 , pp. 24771-24784).

We have re-phrased this part and don’t present the composition of the metasome. We now present the composition of the infiltrating melt. Comparison with the measured melt composition of a ~30 % melt of the same sediment from a different experiment shows that it plots within the expected 1-50 % range for sediment melts that metasomatise the mantle. Intriguingly it is also the same compositional range observed in ultrapotassic rocks.

Lines 298-301: I agree with the ratios representing a metasomatic enrichment, but they can not be interpreted as partition coefficients.

Agree, changed accordingly.

Minor comments 

Figure 5: This Figure needs clarification. The unit in y-axis is distance/time, but distance and thickness are also plotted in the diagram.

We checked the figure again and re-phrased parts of the caption. The metasome thickness and distance of infiltrated melt are treated as the same because the metasome formed by a (high-Si) melt that crystallized and reacted out while at higher T, the melt stays liquid and does not react out.

Comments from pdf file

Please, explain better or rephrase; this quantity is meaningless if a time interval is not indicated; units are usually volume/time  or mass/time

Rephrased.

Plese specify  brevelly the three parts after the two points. Otherwise, pleace, replace the two points by a point mark and use the following structure: "First...  . Second... .Third (or Finally)... .

Done.

Please, correct as follows: [9, 10, 11, 12]

Please, correct as indicated. Correct it in other parts of the text

Changed accordingly.

Which elements?

Added.

Please correct as [13, 23]

Changed accordingly.

Please, indicate the elements

Added.

Probably the point mark after  Gl should be deleted

Deleted abbreviation, “Gl” now as “glass” in table.

sediment solidus (white arrow in Fig. 1)

Changed.

which  elements?

Added.

I agree with the ratios representing a metasomatic enrichment, but they can not be interpreted as partition coefficients.

Agree, changed accordingly.

Plese check the width of the Table for avoiding to use 2 rows for sediment composition

Changed to save space.

 

Round 2

Reviewer 2 Report

Dear authors, 
I consider that my comments have been properly answered and the paper can be accepted. However, I still have a pair of questions:

1) If Ni and S are immobile, the latter cannot be sourced from sediment (lines 496-498).
2) The axes  of diagram from Figure A4b need units.

Best regards 

Author Response

1) If Ni and S are immobile, the latter cannot be sourced from sediment (lines 496-498).

This is correct, this was wrong to put into one sentence to begin with. Ni is only sourced from the dunite as the dunite is S free, so only Ni can behave immobile and it combined with S that comes from the sediment.

2) The axes  of diagram from Figure A4b need units.

It is now included.