pnas.2408248121.sapp.pdf

Supp

orting

Information for

Glycine synthesis from nitrate and glyoxylate mediated by ferroan brucite: a

new integrated pathway for

abiotic amine synthesis.

Chimiak, L.

; Hara, E.

; Sessions, A.

; Templeton, A.S

Department of Geological Sciences, University of Colorado, Boulder, CO

Department of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA

orresponding author

: Laura Chimiak

Email:

laura.chimiak@colorado.edu

This PDF file includes:

Supporting

text

Table S1

Figures S1 to S

SI References

Supporting

Information Text

Materials and methods

Synthesis

Syntheses were carried out in 30

mL serum vials, in an

anaerobic chamber (‘Coy chamber’). All

glassware was combusted at 450



C for eight hours prior to use. Brucite was synthesized from iron(II)

chloride tetrahydrate (Sigma Aldrich, Part#220299, 98% purity), magnesium hexahydrate (Sigma Aldrich,

Part#M9272, 99.0

102.0% purity), Iron (0) (VWR, Part#

AA00170

14, Lot#

G19X048), 12N hydrochloric

acid (Sigma Aldrich, Part#258148, 37%), potassium hydroxide (Sigma Aldrich), and water from a MilliQ

filtration system (18.3



hereafter, ‘water’). Water was vigorously bu

bbled for 30 minutes with pure N

(Airgas), put in the Coy chamber with a loose cap, and allowed to equilibrate for at least 24 hours prior to

use. For ferroan brucite synthesis, a 1.0 M magnesium hexahydrate solution and a 0.5 M iron chloride

tetrahydrate solution with 0.02 M Fe(0) and 0

.0008% HCl were made and equilibrated in a Coy chamber

for 24 hours prior to use. These were combined in a 1:1 volume to create a metal solution and 5.0 mL

were added to each serum vial. Brucite was precipitated by adding 6

mL of 2N KOH to each vial and

gently swirling the vials. The pH adjusted to 10 with 2N KOH. Minerals were left for one hour to settle after

which the aqueous phase was decanted and replaced with water that had been adjusted to pH 10 with 2N

KOH. This proc

ess was repeated 2 times, after which the pH was measured on an electronic pH meter

(Horiba LAQUAtwin pH

22 digital pH meter) and adjusted to pH 9.5 if needed. The magnesium hydroxide

brucite control followed the same protocol but used a 0.75M magnesium he

xahydrate solution instead of

the metal solution.

Directly after brucite synthesis, a 0.5 mL aqueous aliquot was drawn using a 25

gauge needle (BD, part

#305125) and 1.0 mL syringe (BD, Part #309628) for an initial pH measurement on a Horiba LAQUAtwin

22 digital pH meter.

Following 24 hours, a second 0.5 mL aqueous aliquot was drawn, and a pH measurement was taken. The

pH was adjusted to 9.0 when samples were below this value

Reactant stock solutions were first constituted as separate solutions from water with glyoxylate

monohydrate (Sigma Aldrich, Part G10601, Lot WXBD7215V), potassium nitrate (Sigma Aldrich, Part#

221295), and ammonium chloride (VWR). To create reactant solutions,

5 mL of a 1.6 M glyoxylate

monohydrate stock solution and 5 mL of a

1.6 M nitrogen solution (

e.g.,

ammonium chloride or potassium

nitrate) were combined into a combusted glass vial. 2.50 mL of the reactant solution was added to the

brucite

containing vials via borosilicate glass pipette. Vials were capped with 20 mm blue butyl stoppers

(Fischer Science,

Item#

194

5347, multiple lots) and crimped with aluminum crimps (Sigma Aldrich,

multiple lots). Prior to use, butyl stoppers were washed three times with a 2N NaOH solution and

autoclaved.

Sampling

After two weeks, a 2.0 mL aliquot was taken from each sample using a using a 25

gauge needle and 3.0

mL syringe (BD, 309657). Prior to each extraction, the syringe was flushed with Argon (Airgas). The

aliquot was subsampled into three 2.0 mL microfuge tube

s (Eppendorf, EP022363344): 0.5 mL for a pH

measurement, 0.5 mL for NMR analysis, and 1.0 mL to be desalted and analyzed via LC

and GC

MS.

After six weeks, sample vials were taken into a Coy chamber, shaken, and decanted into a 50 mL Falcon

tube (Corning, CLS352070). Falcon tubes were subsequently capped, removed from the Coy chamber,

and put on a centrifuge at 1800 rpm for 3 minutes. Samples

were then transported back into the

anaerobic chamber, uncapped, and the aqueous phase was decanted into a new 50 mL Falcon Tube.

Aqueous fractions were subsampled into 2.0 mL microfuge tubes as follows: pH samples (0.5 mL), NMR

and IC samples (1.0 mL), a

nd two sets of archived samples (2.0 mL). The remaining aqueous fraction

was saved in the original Falcon tubes for cation columns. Mineral fractions had 1.0 mL subsampled into

microfuge tubes for column chemistry. The rest of the mineral fractions were sa

ved in the initial tubes for

XRD. All samples were capped inside of the anaerobic chamber, transported outside of the chamber, and

immediately placed in a



C freezer.

Amine Purification

Amine samples were purified and desalted for GC

MS and LC

MS analysis.

Prior to use, 5 mL Dowex X8

50W 50

100 mesh columns (Part: 44509, Sigma Aldrich, hereafter ‘Dowex’) were prepared by washing

with at least the following: 25 mL water, 25 mL 2N NH

OH (diluted from 37% NH

OH, Sigma Aldrich,

multiple lots), 25 mL water, 25 mL 4N HCl (diluted from 28

30% HCl, Sigma Aldrich, Part# 258148,

Multiple Lots), and finally, water until pH reached 5.

Aliquots of sample set aside for LC

and GC

analyses were thawed and immediately partitioned into 1.0 mL samples for concentration analysis and

the rest for speciation and isotope analysis. The 1.0 mL samples were frozen until needed. Thawed

samples wer

e acidified to pH 2 or lower and loaded dropwise on a Dowex column. Columns were rinsed

with 25 mL of water, which was collected as a load fraction in a 50 mL Falcon tube. Then, a 40 mL glass

vial was placed under the column and amines were eluted with 25

mL of 2N NH

OH, which was collected

and dried to completion under N

. Following its dry down, samples were reconstituted in 0.5 to1.0 mL

water and partitioned between LC

MS and GC

MS analyses.

Analyses

NMR

NMR was conducted on samples in 10% D

(Sigma Aldrich)/90% H

O on a Bruker 400 MHz

spectrometer using 5 mm Bruker SampleJet NMR tubes (Wilmad, Part WG

1000

SJ, Multiple Lots). A

subset of samples was analyzed using a 10% D

O, 0.1% DSS solution in water (Sigma Aldrich, Part

373773, diluted with D

O above). Glycine, glycolic acid (Sigma Aldrich), glyoxylate, diglycine, triglycine,

tartarate, and DL

threo

hydroxyaspartic acid peaks were identified NMR spectra through standards

using a 10% D

O, 0.1% DSS solution in water (Figures S3

9).

These standards were added to tubes of

the sample solutions that had already been analyzed via NMR such that the matrix would be identical to

samples measured. Peaks were identified via subtraction. To constrain peaks from imine species,

glyoxylate was equ

ilibrated with NH

Cl for 24 hours in the matrix from a brucite sample that had had no

organics added (Sample 24) and had previously been processed for NMR. This imine sample was run in

10% D

O and peaks were compared to the initial brucite sample’s NMR and

the NMR from the glyoxylate

that had polymerized.

Samples were prepared for NMR using methods from Barge

et al

(201

). In short, 0.050 mL of 2N NaOH

was added to a 2.0 mL a microfuge tube containing 0.5 mL of each 2

week sample or 1.0 mL of each 6

week sample. Tubes were centrifuged at 1800 rpm for 3 minutes. When a mineral precipitated, the

sample’s supernatant was dec

anted into another microfuge tube containing 0.050 mL of 2N NaOH. This

process was repeated until no mineral precipitated from the sample. At this point, samples were frozen

until they neede

d to be analyzed via NMR. For NMR analysis, samples were thawed, and 0.400 mL was

pipetted into an NMR tube to which 0.040 mL of D

O or D

O with a DSS spike was added.

An aliquot of samples prepared for NMR were analyzed on a Thermo Scientific Dionex ICS

5000 HPIC

system in anion mode for NO

and glyoxylate and in cation mode for NH

. For anion mode, samples were

first diluted to 1 part in 100 in water and analyzed against a suite of standards including NO

and

glyoxylate at concentrations ranging from 10

M to 1000

M. Standards were analyzed both in water at

1000

M and in the same matrix as the samples. Samples and standards were measured on the IC using

an isocratic

gradient of 36.00 mM KOH

pumping at 1 mL/min, a Dionex IonPac AS25 column, and UV

Vis

detection at 214 cm

. For cation mode, samples were diluted to between 3 and 40 parts in 1000 in a 3

mM methylsulfonic acid solution. Standard solutions consisted of NH

ranging from 2

M to 2000

M in a

3 mM methylsulfonic acid solution. Samples and standards were analyzed on the same data on the IC

using an isocratic gradient of 3 mM methylsulfonic acid pumping at 1 mL/min, a Dionex Ionpac CS12A

column, and conductivity detection.

MS analysis was performed on a Thermo Fisher Dionex UltiMate 3000 UHPLC system with an ISQ

EM single quadrupole mass spectrometer and Newchrom AH column (Newchrom, Part#, NAH

46.250.0510). Eluents were acetonitrile (Supelco OmniSolv, Part# 62258, Multip

le Lots), water, and formic

acid (Sigma Aldrich, Part# 33015, Lot# STBH6997). For preparation, all lines leading to the column were

washed with methanol for 10 minutes at a flow rate of 1.0 mL/min and then with the eluent for another 10

minutes at a flow r

ate of 1.0 mL/min. At the start of each measurement day, a 45

minute run of 95%

acetonitrile and 10% water was performed to remove any bubbles in the line. Measurements were made

with an isocratic pump method with 10.0% acetonitrile, 89.9% water, and 0.1%

formic acid. Standards

used to confirm retention times and key masses for glycine, diglycine (Sigma Aldrich, Item: 50199),

triglycine (Sigma Aldrich, Item: 50239), iminodiacetate, and DL

threo

hydroxyaspartic acid (Sigma

Aldrich, Item: h2775). These comp

ounds were prepared in water at concentrations of 6.6 mM to 6.6 x 10

mM. Samples were run in duplicate for 20 minutes each. Analyses were performed using Chromeleon

Software (Thermo Fischer).

For GC

MS analysis, samples were derivatized as N

trifluoroacetyl

methyl esters. To do so, they were

dried under N

in GC vials. A procedural blank with water and a glycine standard were concurrently

derivatized. Vials were placed in an ice bath, 400 uL of methanol (Sigma

Aldrich, Part# 322415) were

added to them, and then 100 uL of acetyl chloride (Sigma

Aldrich, Part

# 00990, Lot#BCCG5173) was

added dropwise while the vial was continuously swirled. Vials were then capped and heated to 70



C for 1

hour. Vi

als were allowed to cool, uncapped, and samples were dried to completion under N

. To these,

400 uL of hexane (Sigma Aldrich) and 200 uL of trifluoroacetic anhydride (Part# 106232, Lot# BCCH1476)

were added. Vials were capped and heated to 80



C for 30 minutes. Samples were dried to 25% of their

volume, 1.0 mL of hexane was added, and then samples were dried until ~50 uL remained. These

samples were transferred into 200 uL inserts (Thermo Scientific, Part # C4012

529) to which 100 uL of

hexane w

as added. The

glycine standard was not transferred into an insert but instead diluted 4 mM and

serially diluted by factors of 10 down to 4 x 10

mM.

MS analyses of sample’s amine fractions N

trifluoroacetyl

methyl esters for 2

week samples were

performed on a Q

Exactive Orbitrap Mass Spectrometer with a 30 m TG

5SILMS column (Thermo

Scientific, Part# 26096

1420). Six

week samples were measured on

a Thermo Trace 1310 connected to a

TSQ 9000 with a 30 m Restek Column (Rxi

XLB, Part# 13708). All measurements were run with a

temperature ramp that started at 50



C and increased to 120



C. For standards, this final temperature was

held for 5 minutes. For s

amples, it was held for 25 minutes. Standards were run at the beginning and end

of measurement sessions and analytical blanks of hexane were run throughout. To decrease changes of

contamination, two hexane blanks were run after the initial standard measurements.

Analyses of GCMS data was done using XCalibur software (Thermo Fisher) and the NIST webbook for

peak identification. Excel was used to analyze

concentration data.

XRD

To prepare samples for analysis, minerals were thawed in a Coy chamber, pulled into a 5 mL syringe, and

filtered using a polypropylene syringe filter holder with a 0.20

m hydrophilic filter (Millipore, Part#

GTTP02500, Lot# RODB51339) backed by a glass fiber filter (Whatman, Part# 1825

024, Lot# 9752460).

Filter paper with the mineral on it was transferred into a weigh boat, lightly capped with aluminum foil, and

left to

dry overnight in a desiccator in a Coy chamber. Once dried, samples were crushed using

an agate

mortar and pestle and transferred to a zero blank XRD plate (MTI Corporation, Item#ZeroSi24D10C1US)

in an anaerobic holder (Bruker, Item#

A100B33). To prevent the minerals from orienting in one direction,

samples were spread on the plates by using the narrow side of a spatula to lightly cut into the powder and

gently push it in a direction. This motion was done at multiple orientations until

the sample was spread out

over the plate. At this point, the holder was sealed and removed from the Coy cha

mber for immediate

XRD analysis on a Bruker X

Ray Diffractometer with a 0.1 mm divergence slit and the air scatter

suppressor seated on top of the sample holder. Samples were analyzed from a 2



of 5



to 65



or 75



with

a step increment of 0.0206 and a 2.0 second dwell time. An analytical blank of the sample holder with no

mineral was run to collect a background spectrum. Data were analyzed on Diffrac.Suite software (Bruker)

and on Excel by comparing output spectra

to mineral spectra on RRUFF

(

)

Raman Spectroscopy

After being analyzed

on the XRD,

Raman spectroscopy was conducted

on a subsample of the dried and

filtered mineral

at the Raman Microspectroscopy Laboratory, Department of Geological Sciences,

University of Colorado

Boulder (RRID:SCR_019305) on a Horiba LabRAM HR Evolution Raman

spectrometer equipped with a 100mW 532 nm excitation laser

with both a 50x and 100x objective, a

spectral range from 10

3000 cm

and a laser power of 25%. Specific points for spectral analysis were

selected in a manner to sample a

ll visually distinct mineral phases at least once. Minerals were identified

by comparison to known standards on the RRUFF database

(2)

Fourier

Transform Infrared Spectroscopy

A subsample of each dried and filtered mineral analyzed on XRD was mixed

with KBr powder and

pressed into pellets. These pellets were measured on a Thermo

Nicolet Nexus 670 FTIR spectrometer

with 128 spectral measurements per sample. Peak positions were compared to various literature sources

including the NIST database

(3)

Dilution and concentration corrections

Amine concentrations were corrected to be presented as millimolar values. To this end, glycine

concentrations were calculated according to the calibration curve on the GC

MS. Then, background

concentrations were subtracted. For aqueous samples, all samples

were taken from an initial 1 mL, so no

further corrections were required. Most mineral samples had overlying aqueous solution, amounts of

minerals hydrolyzed varied, and only some of the amine fraction was partitioned for GC analysis. To this

end, concent

rations for mineral samples were further corrected as follows. Firstly, the glycine

concentration in the sample was multiplied by the inverse of the fraction partitioned for GC analysis (

e.g.

if ½ was set aside for analyses, the concentration was multiplied by 2). Then, the contribution from the

aqueous portion was subtracted. This was found by taking the initial volume of solution in the mineral

sample (between 0 and 1 mL) and multiplying it

by the concentration of glycine per mL in the aqueous

samples. F

inally, mineral samples were normalized to 1 mL, which was accomplished by multiplying by

the inverse the of total volume of minerals in milliliters. These steps are summarized in Equation S1:

퐶

푓푖푛푎푙

(

퐶

푢푛푐표푟푟

−

퐶

푏푘푔푟푑

푓

푎푚푖푛푒

−

푉

푎푞

퐶

푎푞

)

푉

푚푖푛

(Eqn. S1)

where C

final

is the final glycine concentration in mol/mL, C

uncorr

is the uncorrected measured concentration

of glycine in mol/mL, C

bkgrd

is the background concentration of glycine in the GC

MS, f

amine

is the fraction

of the amine sample used in the measurement,

is the volume of aqueous solution purified with the

mineral fraction in mL, C

is the background corrected concentration of glycine in the aqueous sample in

mol/mL, and V

min

is the volume of minerals that were hydrolyzed and purified.

A secondary check on hydrolyzed mineral samples was performed by using a portion of the samples for

XRD that had been put in an anaerobic chamber where they were filtered, rinsed with degassed MilliQ

water, and dried. These samples were purified in the sam

e manner as the previous aqueous and mineral

samples and measured on a GC

MS. These samples, shown in Figure S4d, had slightly higher glycine

concentrations than the samples used in the manuscript thereby confirming that the glycine was sorbed

on the miner

al and not created by an aerobic reaction between constituents in the aqueous and mineral

phases during sample processing.

All samples were assigned standard deviation by adding in quadrature the standard deviation from two to

four replicate measurements of the area under the glycine peak in a given sample with that of the

procedural blank. As mineral fractions only had one me

asurement, the standard deviation of glycine

concentration in these were assumed equal their aqueous counterparts. For total glycine concentration

error, therefore, the standard deviation of the aqueous measurement was added in quadrature with itself.

Iron Concentration Calculation

In the work here, the limiting

reagent is always the concentration of Fe(OH)

, which at most has a 125

mM concentration. This concentration is found by taking the 5 mL of 0.75 M Fe

0.33

,Mg

0.66

that is initially

added to vials. We assume that the Fe

will be in brucite before the overlying water is washed. If all 1.25

mmol of Fe

remains, this in 20 mL of solution gives the 62.5 mM concentration used in the calculations

below when determining the maximum possible yield of glycine.

In all samples, the imine must be reduced to an amino acid via a 2

electron transfer, so for each glycine

made, at least 2 equivalents of Fe

must be oxidized, resulting in a maximum of 31.25 mM glycine

production possible. However, samples that use nitrate also require an 8

electron transfer to reduce

nitrate into ammonia. Due to all samples having the same amount of brucite, the nitrate syste

m samples

are 4

fold more Fe

limited, so at most could produce a 6.25 mM glycine concentration. That one sample

achieves this, rather than forming high abundances of glycolate, leads to the possibility that the high local

negative change from the nitrate while bound to iron sites on the brucite might have led to less glyoxylate

reducing to form glycolate and therefo

re being available for reductive amination.

Mineralogical analyses

Ferroan brucite is more stable above pH 8.5

(

), so samples from the ammonia system, whose final pH

were circumneutral (Table 1), had a noisier XRD pattern due to partial dissolution of the mineral during

the experiments. These

experiments’ XRD spectra have only one potential additional feature at 2

of 35°,

which could indicate a minor magnetite component (Figure S12). Samples from the nitrate system have

multiple new peaks including those at 2

of 12°, 23°, 34°, 46°, 60°, and 61°, which suggest the presence

of pyroaurite or iowaite (

(OH)

[CO

]·4H

(OH)

·4H

) and those at 30°, 36°, and

43° which suggest the presence of magnetite (

)

(Figure S4). Peaks at 2

of 18° and 38° could

indicate magnetite and pyroaurite or iowaite respectively; however, they are also indicative of ferroan

brucite so are not unique identifiers.

FTIR analysis also detects the minor new mineral components the nitrate systems and not the ammonia

systems (Figure S13). FTIR spectra of ammonia system minerals matched those of the magnesium

brucite and the unreacted ferroan brucite controls. Conversely,

spectra of nitrate system minerals had

peaks at 837, 1320, and 1630 cm

which correspond to Fe(NO

)

(

)

and at 1107 and 1373 cm

which

corresponds to the Fe

and NO

components as determined by comparison to spectra of Fe(NO

)

complexed to NiO

(

)

(Figure S12). The broad FTIR peak near 568 cm

also support the presence of

magnetite

(

)

(Figure S13). While the bulk minerals in the ammonia system predominantly had brucite,

those in the nitrate system sequester nitrate ions as a nitrate

substituted pyroaurite, which demonstrates

a clear interaction with the mineral and the reactant ions.

Raman spectra were measured on the oxidized mineral samples to perform an investigation of minor

mineral components that might be overshadowed during the above analyses. Raman spectra of ammonia

system mineral samples broadly matched those of the ferroan b

rucite control and showed no strong

indication of organics present on the mineral surfaces studied (Figure S14a). The oxidized samples had

peaks at 214, 280, 384, 415 and a shoulder at 600 cm

, which matched spectra of hematite

(

7,8

)

(Figure

S13a). An additional peak at 684 cm

could indicate the presence of magnetite

(

7,8

)

(Figure S14a).

On the other hand, nitrate samples had variable spectra depending on the grain type. Most grains in the

oxidized nitrate samples matched those of the ammonia samples with 214, 280, 391, 598 cm

peaks

matching spectra of hematite and a 655 cm

peak indicative of magnetite

(

7,8

)

. However, the dark grains

showed different spectra with peaks at 224 and 526 cm

, which match spectra of lepidocrocite

(

)

(Figure

S14b). This mineral is associated with organics as indicated the peaks from a C

C stretching mode at 918

and 951 cm

, a CH

bending mode at 1443 cm

, and C

O stretch at 1637 and 1662 cm

, and a CH

stretching mode at 2905 cm

. Furthermore, the spectrum has minor peaks at

1057 1092 (NO

stretching

mode) and 1443 (NO

stretching mode), which indicate the presence of nitrate (Figure S14b).

Mechanism

We posit that in ammonia systems, the canonical reductive scheme amination occurs (Figure 4, Main

text). Ammonia attacks the carbonyl carbon is the

C on glyoxylate to form 2

amino

hydroxyacetic

acid, which is in equilibrium with both glyoxylate and 2

minoacetic acid. Either 2

amino

hydroxyacetic

acid or 2

iminoacetic acid can be reduced to glycine (Gomez

et al

, 2002). In nitrate systems, nitrate is first

reduced to nitrite from which it can either become nitrous oxide or hydroxylamine and then ammoni

(Wong

et al

, 2023). The ammonia can then follow canonical reductive amination and results in glycine

that is associated with the mineral such that only acid hydrolysis fully liberates it. Due to the association

between organics, nitrogen species, and FeO(OH), we posi

t that this binding occurs at Fe

sites (Figure

4b main text, Figure S14).

As noted in the text the ammonia systems produce more glycine than should be feasible given the

electrons available in ferroan brucite for reductive amination. Given that a small amount of glycine is also

detected in both the aqueous phase of the magnesium

brucite control and also in the non

mineral

controls in Barge

et al

(2020), we assume that this result points to a real side reaction. The NMR and IC

results in this study demonstrate that formate is present. Barge

et al

(2020) also find this in their wor

The Leuckart reaction uses ammonium formate or formamide to perform reductive amination as depicted

in Schemes S1 and S2 (11). Here, formate acts as the reductant. While this synthesis typically requires

temperatures higher than those used in this experiment (120°C and above), the presence of certain metal

complexes can reduce these temperatures to roughly 50°C (11,12).

Here, at 6 weeks, the ammonia system experiments have a pH below 9.2

—

the pK

of ammonium

—

and

have excess ammonium, glyoxylate, and formate. Consequently, if the minerals present can in the

ammonium system can allow Leuckart reaction to proceed at roughly 20°C even at a slow rate, it could

explain the additional 38 ± 15 % and 21 ±

4 % yields relative to the 100 mM glyoxylate and 100 mM

ammonia here. This proposed mechanism is speculative, and further work needs to occur in order

ascertain the mechanism by wh

ich glycine is formed when iron cannot provide adequate electrons and

formate is present.

Table

Concentrations of glycine, aqueous phase reactants, and pH of samples at 6 weeks. Glyox refers to glyoxylate. Yield versus Fe

total glycine that

could be produced from the 62.5 mM Fe(OH)

, which is 31.25 mM glycine in ammonia systems and 6.25 mM glycine in nitrate systems.

Source

mineral

wks

[NO

]

0 wks

(mM)

[NO

]

6 wks

(mM)

[NH

]

0 wks

(mM)

[NH

]

2 wks

(mM)

[NH

]

wks

(mM)

[Glyox]

0 wks

(mM)

[Glycine]

aqueous

(mM)

Dev

[Glycine]

mineral

(mM)

[Glycine]

total

(mM)

yield vs

glyox

and N

(%)

yield vs

(%)

Fe,Mg(OH)

7.81

1E+02

2E+01

3E+01

1E+02

4.3E+01

2.0E+00

8.8E+00

5.2E+01

52 ± 4

167 ± 13

Fe,Mg(OH)

7.89

1E+02

2E+01

3E+01

1E+02

5.1E+01

7.7E+00

1.8E+01

6.9E+01

69 ± 15

222 ± 50

Fe,Mg(OH)

9.07

1E+02

5E+01

6E+01

1E+02

2.9E

4.7E

7.3E

1.0E

0.1 ± 0.1

1.6 ± 0.3

Fe,Mg(OH)

8.28

1E+02

5E+01

6E+01

1E+02

1.1E

4.5E

3.0E+01

3.1E+01

3.1 ± 0.2

50 ± 1

Mg(OH)

9.43

1E+02

4E+01

1E+02

2.3E

3.5E

0.0

2.7E

0.4 ± 0.1

0.4 ± 0.2

None

Fe,Mg(OH)

9.41

0.0

3.7E

N/A

0.0

Fig. S1.

Depiction of methods used in this study. Processes on the right

hand side of the dotted line are

performed in anaerobic conditions under a 95% N2/5% H2

atmosphere. Raman and FTIR are exceptions

to this and are performed in aerobic conditions. All processes on the lefthand side of dotted line are

performed in aerobic conditions.

Figure c

reated with BioRender.com

Fig. S2.

MS of

trifluoroacetic

methyl ester of glycine at (a) 2 weeks on the Orbitrap GC

IRMS

where glycine elutes at 4.60 minutes and at (b) 6 weeks on a GC

MS where glycine elutes at 3.65

minutes. Panel (c) is the same data as (b) but zoomed in to see the nitrate s

ystem and control peaks.

Panel (d) is from mineral samples with no overlying aqueous phase. Aq samples are from the aqueous

phase and mins samples are from hydrolyzed minerals. In panel (b), 45 and 57% of the glycine in the

mineral samples for the NH

syst

ems were from overlying aqueous phase. This contribution has not been

removed from the figure as the additional glycine explains the peak shape (i.e., the column overloading).

All NH3 system samples are in green,

system samples are in blue,

magnesium brucite controls are

in plum, and ferroan brucite controls are in orange. Mineral samples are lighter in color than their aqueous

counterparts.