Biotite supports long-range diffusive transport in dissolution-precipitation creep in halite

. Phyllosilicates are generally regarded to have a reinforcing effect on chemical compaction by dissolution-precipitation creep (DPC) and thereby inﬂuence the evolution of hydraulic rock properties relevant to groundwater resources, geological repositories as well as fossil fuel reservoirs. We conducted oedometric compaction experiments on layered NaCl-biotite samples to test this assumption. In particular, we aim to analyse slow chemical compaction processes in the presence of biotite on the grain scale and determine the effects of chemical and mechanical feedbacks. We used time-resolved (4D) microtomo-5 graphic data to capture the dynamic evolution of the transport properties (cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58) porosity in layered NaCl-NaCl/biotite samples over 1619 and 1932 hours of compaction. Percolation analysis in combination with advanced digital volume correlation techniques showed that biotite grains inﬂuence the dynamic evolution of porosity in the sample by promoting a reduction of porosity in their vicinity. However, the lack of preferential strain localisation around phyllosilicates and a homogeneous distribution of axial shortening across the sample suggests that the porosity reduction is not achieved by pore collapse but by the precipitation of 10 NaCl sourced from outside the NaCl/biotite layer. Our observations invite a renewed discussion of the effect of phyllosilicates on DPC, with a particular emphasis on the length scales of the processes involved. We propose that, in our experiments, the diffusive transport processes invoked in classical theoretical models of DPC are superseded (cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58) complemented by chemo-mechanical feedbacks that arise on longer length scales. These feedbacks drive NaCl diffusion from the marginal pure NaCl layers into the central NaCl-biotite mixture over distances of several hundred µ m and several grain diameters. Such a mechanism was ﬁrst 15 postulated by Merino et al. (1983).


Introduction
Chemically controlled compaction influences groundwater resources, geological waste repositories and CO 2 sequestration as well as fossil fuel reservoirs.One of the major processes involved in chemical compaction in the Earth's upper crust is dissolution-precipitation creep (DPC), a diagenetic and low-grade metamorphic deformation process that significantly contributes to cementation and reduction of porosity in sedimentary rocks (Rutter, 1983;Green, 1984;Tada and Siever, 1989;Gratier et al., 2013).Due to its impact on the diagenetic evolution of sediments it is crucial to study how DPC contributes to the dynamic change of petrophysical rock properties during compaction.
DPC describes a sequential chemo-mechanical process in a non-equilibrium system consisting of a solid phase and its associated fluid under non-hydrostatic pressure conditions (Rutter, 1983;Gratier et al., 2013).The three successive steps in the sequence are i) dissolution of material at stressed grain contacts, ii) diffusive mass transport through an intergranular fluid and iii) local reprecipitation of dissolved material at low-stress sites (e.g.open pores, veins) (Rutter, 1983;Tada and Siever, 1989;Gratier et al., 2013).
Phyllosilicates have been recognised to have a reinforcing effect on the dissolution process (e.g.Heald, 1956;Weyl, 1959;Gratier, 1987) and act as loci for enhanced DPC.Whether this is due to enhanced reaction kinetics or effective transport pathways, is still under debate (Gratier, 1987) ::: and :::: may :::::: depend ::: on ::: the ::::::::::::: rate-controlling :::::: process.Macente et al. (2018) explored this effect using time-resolved x-ray microtomography to document chemical compaction in NaCl-biotite mixtures.They found that the increased porosity loss in the biotite-bearing layer did not lead to an increased strain localisation.These observations pointed towards infilling of porosity with material sourced outside the biotite-bearing layer rather than pore collapse, suggesting long-scale diffusive transport of dissolved NaCl.
Reviewing diffusive transport during DPC shows that in theory four pathways for material transport need to be considered.
The first sample (SBS) contained a central NaCl-biotite layer and two adjoining layers of pure NaCl as well as two layers of glass beads.The latter maintain permeability at the sample ends.For the second sample (SB) we increased the ratio of the NaCl-biotite layer and removed the top NaCl layer.A third pure NaCl sample (S1 : S) served as a reference.
Masses for all components were calculated to meet the dimensions of cylindrical samples with 5 mm diameter and a desired starting height of 8 mm.For the unconsolidated samples an initial porosity of ∼ 40% was assumed which was also included into the calculation.The absolute heights of the samples were determined after loading the oedometers with the granular aggregate.Analytical grade NaCl and natural biotite were chosen as main components for the samples, as the difference in the x-ray attenuation results in a sufficient contrast in the reconstructed µCT data.The granular NaCl was sieved to a grain size of 250-300 µm.Biotite of granodioritic origin (Lone Grove pluton, Texas, e.g.Zartman, 1964) was pre-processed by mineral separation techniques (conducted at GFZ Potsdam, who supplied the biotite mineral separate) and sieved to a grain size of 200-500 µm ::::::::: (maximum ::::: grain :::::::: diameter).Acid washed glass beads with a diameter of 212-300 µm were added as chemically inert and permeable top and base layers.
The individual sample components were introduced into the sample cell by wet loading.Saturated NaCl-brine was injected into the bottom of the sample cell.For the two biotite-bearing samples (SBS & SB) glass beads and NaCl were sequentially poured into the brine followed by a homogeneous slurry of NaCl and 20 wt% biotite.For the pure NaCl sample dry NaCl ::: (not :: a ::::: slurry) : was poured into the injected brine.All samples were covered at the top and bottom with a disc of filter paper to prevent blockage of the fluid inlet and outlet.

Experimental setup
The experimental setup was designed to run three oedometric compaction experiments simultaneously (Fig. 2).We adopted the basic setup from Macente et al. (2018) and added a thermally insulated environment, vibration damping, pressurised fluid 90 pumps and chemically inert glass beads layers to gain better control of parameters and establish a better characterised system.
Oedometer ::: The ::::::::: oedometer cells as described in Macente et al. (2018) were modified by sealing the sample cell with an oring around the piston in order to withstand fluid pressure of 0.5 MPa.The latter was applied in two different ways.For the SBS sample we used Cetoni neMESYS syringe pumps feeding a fluid transfer vessel that isolated the metal-corrosive brine.
The transfer vessel was composed of a silicone tube filled with saturated NaCl brine inside a pressure resistant glass column.

95
Saturation of the brine was guaranteed by the presence of solid NaCl in the reservoir.For the pure NaCl and the SB sample we used a plastic syringe that contained the corrosive brine itself and was driven by a pneumatic actuator.The experimental setup allowed maintaining a moderate pore fluid pressure to suppress gas bubbles while suppressing fluid advection.
The axial load was also applied by gas pressure driven pneumatic actuators when the oedometer cells were placed in loading frames following the construction of Macente (2017).
The experiments were run inside a thermally insulated box where the temperature was logged and found to be stable within ± 1.7 °C over the course of the experiments.For the acquisition of microtomographic data, the odoemeters were disconnected from the fluid and pneumatic capillaries and mounted into the microtomography scanner (see next section).
Upon initial loading, the unconsolidated granular samples were held at a constant effective load of 6.64 MPa : , ::::::::: 6.77 MPa and 10.5 MPa for samples : S : , SBS and SB+ S1 : , respectively for 60 minutes.After an initial compaction of 9 -18% a reference scan was acquired of the starting aggregate.

Data acquisition
During the experiment, the samples were scanned ex-situ on the x-ray microtomography instrument at the School of Geosciences, University of Edinburgh in regular intervals (for acquisition parameters see Table A2, appendix).To enable this, the oedometers were unloaded and temporarily removed from the loading frames.Nineteen scans (SBS sample)and : , ten scans (SB sample) ::: and ::: five ::::: scans : ( : S :::::: sample) : were acquired over a total duration of 1619 hoursand , : 1932 hours ::: and ::::: 1089 ::::: hours, respectively.At the beginning scans were collected in shorter time intervals to image the rapidly progressing deformation within the first 200 hours of the experiment.As the compaction slowed down, the intervals between each scan were gradually extended to monitor chemical compaction processes.The time-resolved 3D data series obtained in this way were then combined into three 4D data sets capturing the dynamic evolution of the porosity in the different samples.

Data processing
After each scan, Octopus ® software v. 8.9 (Dierick et al., 2004) was used to reconstruct the µCT data from the radiographic projections.The resulting stack of 2D images, which contains a virtual representation of half of the sample, covers a volume of 5 × 10 3 µm 3 .Each time step comprises two scans, vertically translated to cover the entire sample including the sample base and the top piston.Reconstructed image stacks were merged using the image processing software AVIZO ® v. 9.2.The relative shortening of the sample was calculated using

Gas supply
Oedometer cell
The height of each sample was determined from the merged microCT scans.The µCT data were prepared for analysis by pre-processing in Fiji-ImageJ (Schindelin et al., 2012).An uneven background of the reconstructed images was adjusted using the background correction plugin BaSiC (Peng et al., 2017) with the regularisation parameter λ D = 0.69 and λ F = 1.71 for the estimated dark-field and flat-field images, respectively.Denoising of the data was conducted by removing bright outliers (threshold of 50) of 1.5 px and using the Non-Local Means filter (Buades et al., 2011) with a standard deviation of σ = 3 and a smoothing factor of 1. Biotite grains and porosity were segmented as single classes from the images, applying the machine learning tool Trainable Weka Segmentation (Arganda-Carreras et al., 2017, classifier in supplementary material).The latter was also used to segment NaCl grains in the pure NaCl sample but not in the two biotite-bearing samples.Due to a lack of contrast between the grey scales of NaCl, glass beads and the :::: outer : rims of biotite grains it was not possible to use simple segmentation techniques.Instead we applied the Deep Learning Segmentation of Dragonfly software, version 2020.2 (Object Research Systems , ORS) :: to :::::::: discretely ::::::: segment ::: the ::::: NaCl.The resulting binarized image stacks were used for image analysis to quantify the porosity evolution and NaCl migration.

Porosity measurements
The evolving porosity was determined during compaction as 2-dimensional porosity which was measured along the direction of the loading axis as the relative area of the 2-dimensional binarized images.The results were plotted as the porosity on the abscissa and the number of slices along the loading axis on the ordinate (Fig. 13).Errors for this analysis were determined using the probability measured in Trainable Weka Segmentation.The absolute error of the porosity measurement was defined as the difference between the number of pixels belonging to the porosity with a certainty of ≥ 90% to the total number of porosity pixels measured from the segmented data.The absolute error is given in percent as it refers to the porosity which itself is a relative number.

NaCl redistribution measurements
In order to quantify the amount of NaCl migrating within the sample we used two different methods to isolate a pure NaCl signal from the rest of the sample.The first method is similar to the measurement of the 2D porosity.For each slice along the loading axis of a microtomography scan we ::::::: discretely : segmented the NaCl in the grey scale image using Trainable Weka Segmentation for the pure NaCl sample and the Deep Learning Segmentation of Dragonfly software for both biotite-bearing samples.From the segmented data we measured the relative area of NaCl in the 2-dimensional image and calculated its proportion relative to the cross section of the sample.The results were plotted as relative NaCl content on the abscissa and the number of slices along the loading axis on the ordinate (Fig. 14).
Our second approach towards quantifying :: the : NaCl redistribution in the sample was based on 3D volumetric measurements of segmented NaCl.The volumes of NaCl, biotite and porosity are measured relative to a subvolume in the compacting biotitebearing layer .::: (for ::::::: location :: of ::: the :::::::::: subvolume ::: see :::: Fig. ::: A1 :: in ::: the ::::::::: appendix).: We selected the subvolumes at fixed locations within the biotite-bearing domains of both samples that mimic the compaction of the respective layer.The upper and lower limits ::::: extent : of the subvolume were :::::: parallel :: to ::: the ::::::: loading :::: axis :::: was defined by two prominent biotite grains at the top and bottom of each biotite-bearing layer which were easy to identify with progressing compaction of the sample.The ::::::::::: Perpendicular :: to ::: that :::::::::: dimension, ::: the base of the subvolume was chosen as a 500x500 px square :: in ::: the ::::: centre, which is representative of the sample but excludes the contact area of the sample to the cell.This approach allowed us to measure the evolution of the NaCl volume within the biotite-bearing layer with progressing deformation, with biotite acting as an internal standard. .The results were plotted as NaCl volumes relative to a decreasing subvolume in the compacting biotite-bearing layer (Fig. 15).
With label analysis, also implemented in SPAM, we determined the rotation and rearrangement of single biotite grains with progressing deformation based on labelled and binarized data sets.For each microtomography scan we segmented the biotite grains as described in Sect.2.5 and applied a watershed algorithm to separate the grains into individual particles with an allocated label.Further, we calculated the eigenvalues and eigenvectors from each particle's moment of inertia, which is in the case of biotite directly related to the shape and orientation of the particle.The orientations are represented as maximum eigenvectors perpendicular to the basal planes of the grains and plotted as densities in Lambert projections with the vertical loading axis in the centre of the plot (Fig. 7).

Vertical shortening
The bulk compaction of the samples was monitored as vertical shortening and compaction rate over a total duration of 1089 hrs, 1619 hrs and 1932 hrs, for the reference sample (S1 : S), the SBS and the SB samples, respectively.As the compaction of the reference sample was stopped after 1089 hrs, the total shortening of this sample was with ∼ 17%, lower than the total shortening of the biotite-bearing samples.At comparable compaction stages of 1060 hrs and 1020 :::: 1020 :: hrs :::: and :::: 1060 : hrs, for the SB and SBS ::: and ::: SB sample respectively, the difference between the reference and the biotite-bearing samples was approximately ∼ 10% .:::: (Fig. :::: 4a).It is notable that the final scans in all three experiments were acquired before the compaction ceased.
Analysis of the bulk strain rates ( ˙ ) with progressing compaction support these findings.On a double logarithmic scale (Fig. 4b), strain rates for all three samples decreased approximately linearly and dropped in total by two orders of magnitude from ∼ 10 -6 s -1 to ∼ 10 -8 s -1 .Within the first 200 hours of compaction the strain rates decreased by one order of magnitude.
for the SB, SBS : , :: SB and pure NaCl sample, respectively.

Microstructures and compaction accommodation
Vertical slices through the geometrical centre of µCT data illustrate the evolution of the microstructure (Fig. 5).In the biotitebearing samples porosity reduction, a change of the cubic habit of the NaCl grains and the establishment of flat interphase boundaries between NaCl and biotite grains can be observed over 100 and 60 hrs for SBS and the SB sample, respectively.

Strain analysis
Digital Volume Correlation (DVC) was used to analyse a locally resolved strain field and associated strain rates.11. :::::: Overall, ::: the : dominating character of the volumetric strain was negative in all three samples, indicating compaction.
Furthermore, we calculated average strain rates from the locally resolved strains by dividing each calculated strain value by the time interval length given in seconds.The results are displayed in Table 2 and show an overall decreasing trend in all three samples.In the two biotite-bearing samples, deviatoric strain rate maxima were homogeneously distributed, independent of the layer composition.Comparison of these results to the pure NaCl sample showed no major difference.The observed rates are comparable to the bulk strain rates (see Sect. 3.1.1).
In order to locate the strain maxima in 3D and compare them to the position of the biotite grains, we plotted the deviatoric and volumetric strain data of the biotite-bearing sample SBS on top of the segmented biotite data (Fig. 12).This showed that in the early stages of our experiment, deviatoric strain maxima corresponded to the location of biotite grains as well as open pore space and pure NaCl clusters.Later on this correlation still existed, but was less distinctive.However, local minima and maxima of the volumetric strain did not correlate with the position of biotite grains.
NaCl reference sample.Therefore, we conclude that deformation was not localised in the biotite-bearing layer but distributed and accommodated across all layers.Macente et al. (2018) came to a similar conclusion, explaining their results with a stress bearing network of dynamic force chains in the pure NaCl domains which evolve in the granular material as response to a feedback of applied vertical loading and the increase in load-bearing cross sectional area due to local variations in the dissolution rate (Bruthans et al., 2014).Furthermore, they concluded that deformation was promoted by phase boundaries 340 in the biotite-bearing layer.Similar to Macente et al. (2018), we observed the highest porosity reduction in layers containing phyllosillicates.Compared to the pure NaCl layers, these layers lost ∼ 24% more in the SBS, and ∼ 41% more in the SB sample relative to the initial porosities :::::: porosity : in each layer.Combining the observations of higher porosity loss in the biotite-bearing layers with the evidence that compaction is not concentrated in those layers leads to a paradox that cannot be explained by the classical theory of DPC, which would suggest enhanced DPC to lead to strain localisation and thus a porosity loss.A possible solution would be diffusive material transport from a source outside the biotite-bearing layer into the pore space of that layer.
We were able to show that the NaCl content in the biotite-bearing layers of the SB and SBS sample increased with progressing deformation (cf.Figs. 14 & 15) and we interpret this salt to have been sourced from the pure NaCl layers.In the SB sample the largest effect can be observed at the layer interface from which a negative gradient emerged towards the top of the biotitebearing layer (Fig. 14b).In this case the effect of the NaCl migration is restrained by the breakdown of porosity after 170 hours which limits further migration of dissolved NaCl into the biotite-bearing layer from the suspected source in the pure NaCl layer at the bottom of the sample (Fig. 14b).In the SBS sample, we see no porosity breakdown and no gradient in the NaCl distribution emerge within the biotite-bearing layer.The porosity remains interconnected throughout the entire experiment, maintaining access to both NaCl layers as potential sources for salt.The biotite-bearing layer showed a consistently higher increase in NaCl than the marginal pure NaCl layers (Fig. 14c).Especially the upper NaCl layer did develop a pronounced gradient towards the interface with the biotite-bearing layer though, which could be evidence for a diffusive salt redistribution.
This is in accordance with field observations by Heald (1956), Mimran (1977) and Buxton and Sibley (1981), who report observations that challenge the classical theory of precipitation in the vicinity of dissolution sites and invoke larger transport distances in Sandstones, Chalk and Limestones, respectively.
4.2 What is the role of the biotite?
Our DVC analysis revealed that a proportion of the maxima in the grain-scale shear strains (cf.Figs. 9 & 10) corresponded to biotite-NaCl phase boundaries.Such phase boundaries are characterised by significant electrochemical effects (Walderhaug et al., 2006;Greene et al., 2009;Kristiansen et al., 2011), which likely ::: may : accelerate dissolution of NaCl.Visualising such interfaces from our data showed the efficiency of this process (cf.Fig. 6).At the same time, our label analysis showed that biotite grains did not rotate significantly (Fig. 7).Especially during the early stages of compaction, where the sample still had a high porosity, we would expect point loading to force biotite grains to realign.We interpret the fact that this did not happen as corroborating evidence for the efficiency of dissolution at biotite-NaCl phase boundaries.
An increased efficiency of dissolution along phase boundaries would imply that the biotite-bearing layers should compact preferentially.This is an effect that we clearly did not observe in our data (see previous subsection and Figs. 8, 9 & 10), which raises the question as to how the preferred dissolution is being balanced.We interpret this to happen in the following way: NaCl that is being dissolved at a biotite-NaCl phase boundary, or also along a NaCl grain boundary in the biotite-bearing layer, is only redistributed locally, within that layer, so that the net volume of that layer is preserved.In the absence of advective transport in our experiment, this is in line with classical DPC theory (e.g.Paterson, 1973;Raj, 1982;Rutter, 1983;Gratier, 1987;Groshong Jr, 1988).This effect is supported by the diffusive potential described above and the additional NaCl that migrates into the layer (see previous subsection), whereby the additional NaCl contributes to a load-bearing framework whose compaction rate is in sync with ::::::::::: approximates : the bulk sample's.
We do note that the biotite composition does not seem to have a first order effect on pressure solution at its interfaces: while the biotite that Macente et al. (2017;2018) B2 in the appendix for full compositions).Both cation sites are enclosed within the crystal structure of the biotite and are not exposed at the surface of the basal plane.As the latter is likely to be the reactive surface in our ::: the dissolution-precipitation creep experiments ::::::::: mechanism, the observed differences in composition do not :: the :::::::: chemical :::::::::: composition ::: are ::: not :::::::: expected :: to : affect the dissolution of matter :::::: process.
In summary, while biotite grains locally are effective facilitators for DPC irrespective of their composition, it also appears that the chemo-mechanical effect on the entire system is limited and probably outperformed by the trans-domain diffusion outlined above.
4.3 A detailed discussion of our DVC analyses

Magnitudes of local strains
Our DVC analyses resolve deviatoric and volumetric strains on the grain scale (Figs.9-12), and provide insights into the micromechanics of compaction in the various samples.Comparing the strains in the SBS with the SB sample showed the effect of the larger load that was used in the later compaction experiment; the maximum volumetric and deviatoric strains reached in the SB samples are about twice as high.

Character of local strains & their location relative to phyllosilicates
In all three samples the dominating local volumetric strain was negative.This trend is persistent throughout the experiments and in line with the bulk deformation and vertical shortening of the samples.Deviations from this trend occured in the glass bead layers (cf.paragraph below) and ::::::: occurred : at sites where porosity was reduced by precipitation of dissolved material.

Figure 1 .
Figure 1.Schematic sketches of sample configurations of a) the NaCl-biotite-NaCl sample (SBS), b) the NaCl-biotite sample (SB) and c) the pure NaCl reference sample (S1 : S) before deformation.Different shades of grey depict the single components but are not related to their appearance in the tomography scans.White angular patches represent the brine saturated pore space with arbitrary distribution of shape and size, light grey angular to square objects are cubic grains of analytical grade NaCl with a sieved grain size of 250-300 µm and black elongated angular shapes are biotite grains of 200-500 µm size.At the top and bottom of sample a) and b) dark grey circles describe acid washed glass beads of 212-300 µm diameter, which were inserted as chemically inert layer.The original samples have a diameter of 5 mm, note that the sketches are not to scale.

Figure 2 .Figure 3 .
Figure 2. Experimental setup for oedometric compaction experiments ::::::: conducted :: at ::: the :::::::: University :: of :::::::: Edinburgh.Three samples could be loaded simultaneously.Oedometer cells contained the cylindrical samples and were confined in straining frames during deformation.The load was applied by gas driven pneumatic actuators installed at the top of each frame.Cetoni neMESYS high-pressure syringe pumps were used to supply saturated NaCl brine via a fluid reservoir in order to maintain a pore fluid pressure sufficient to suppress gas bubbles in the samples.For the second experimental suit we replaced the fluid reservoir and high pressure pump with a brine filled syringe that was driven by another pneumatic actuator (not shown here).The oedometer cells were placed on acoustically dampened gabbro blocks to avoid external vibrations to reach the cells.

Figure 5 .Figure 6 .Figure 7 .Figure 8 .
Figure 5. Vertical slices through absorption µCT scans at different stages of compaction.::: The :: 3D ::::: model :: in ::: the :::: lower :::: right ::::: corner ::::: shows ::: the :::::: location :: of :: the :::::: section ::::: within ::: the ::: bulk ::::::: samples.Different shades of grey refer to different phases present in the samples (in black: brine filled pore space, dark grey: glass beads, grey: NaCl grains and light grey: biotite) The top and middle row show the NaCl-biotite samples SBS (a-d) and SB (e-h), respectively, the bottom row displays the pure NaCl sample (i-k).b), f) and j) show first signs of porosity reduction and indentation of NaCl grains which we interpret as indicators for active dissolution precipitation creep and which is continuing throughout the experiment.Note that the final scan of the SB sample (h) shows no remaining porosity in the biotite-bearing layers, whereas it is still clearly visible in the SBS (d) and the pure NaCl sample (k).

Figure 13 .
Figure 13.Porosity evolution of a) the pure NaCl, ::: SBS b) the SB ::::: sample : and c) the SBS ::: pure :::: NaCl sample with progressing deformation.The porosity was measured as 2-dimensional porosity for each slice along the loading axis.Different colours indicate different time steps, ranging from yellow to blue with progressing compaction.While a) shows a homogeneous decrease of the porosity within the sample, heterogeneities arise in b) and c) which show the compositionally layered samples : , :: a) ::::: shows : a ::::::::::: homogeneous :::::: decrease :: of ::: the ::::::: porosity ::::: within ::: the ::::: sample.The highest porosity loss occurs in biotite-bearing layers, resulting in compartmentalisation of the SB sample (cf.Sect.2.5).Note that the arrows on the right hand side mark the transition from the pure NaCl domain to the NaCl-biotite domain and colours are corresponding to the deformation stage as denoted in the key.

Figure 14 .
Figure 14.Evolving NaCl distribution in a) the pure NaCl ::: SBS, b) the SB and c) the SBS ::: pure ::::: NaCl sample with progressing deformation.The NaCl content was measured in 2D for each slice along the loading axis.While a) shows a homogeneous increase of the NaCl content within the sample, :: and : b) and c) show that the NaCl-content in the biotite-bearing layer increases more than in the rest of the sample, ::::: while : c) ::::: shows :: a :::::::::: homogeneous :::::: increase :: of ::: the :::: NaCl :::::: content ::::: within ::: the :::::: sample.Note that the arrows at the side mark the position of the interface between the biotite-bearing layer and the pure NaCl-layer, with their colours corresponding to the time steps as indicated in the key below.

Figure 15 .
Figure 15.Evolution of the relative volumes of NaCl, biotite and porosity in the biotite-bearing layers of a) the SBS and b) the SB sample, with progressing deformation.In both samples the biotite-content remains constant within the segmentation error ::::: (plotted :: as ::::: shade :: in ::: the ::::::: respective :::::: colour) while the NaCl content increases and the porosity decreases.This is persistent throughout the experiment.After 324 hours of compaction the increase of the NaCl content in the SB sample (b) stagnates, which corresponds to the breakdown of porosity in the biotite-bearing layer.

Figure 17 .
Figure 17.Possible transport length scales during dissolution-precipitation creep as proposed with in the scope of this work.In a) Diffusion

Table 2 .
Mean strain rates derived from image correlation.

Table B2 .
XRF analysis of mica used in our experiments (Bt-BS) and in Macente (2017) (Mc-AM).Values are given in weight%