Accepted Manuscript

Title: Corrosion rate estimations of microscale zerovalent ironparticles via direct hydrogen production measurements

Author: Milica Velimirovic Luca Carniato Queenie SimonsGerrit Schoups Piet Seuntjens Leen Bastiaens

PII: S0304-3894(14)00057-0DOI: http://dx.doi.org/doi:10.1016/j.jhazmat.2014.01.034Reference: HAZMAT 15697

To appear in: Journal of Hazardous Materials

Received date: 18-7-2013Revised date: 19-12-2013Accepted date: 20-1-2014

Please cite this article as: M. Velimirovic, L. Carniato, Q. Simons, G. Schoups, P.Seuntjens, L. Bastiaens, Corrosion rate estimations of microscale zerovalent ironparticles via direct hydrogen production measurements, Journal of Hazardous Materials(2014), http://dx.doi.org/10.1016/j.jhazmat.2014.01.034

This is a PDF file of an unedited manuscript that has been accepted for publication.As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain.

Page 1 of 30

Accep

ted

Man

uscr

ipt

1

Highlights

• The hydrogen gas generated by anaerobic corrosion of different ZVIs was measured. • In aquifer-free conditions mZVIs have ~10-30 times lower corrosion rate than nZVIs. • Under natural environment higher corrosion rate of ZVIs was observed. • Estimated corrosion rates can serve as an indication of ZVIs life-time. • Inhibitory effects of the ZVI corrosion reaction products were interpreted.

Page 2 of 30

Accep

ted

Man

uscr

ipt

2

Corrosion rate estimations of microscale zerovalent iron

particles via direct hydrogen production measurements

Milica Velimirovica,b, ‡, Luca Carniatoc, ‡, Queenie Simonsa, Gerrit Schoupsc, Piet Seuntjensa,b,d,

Leen Bastiaensa,*

aFlemish Institute for Technological Research (VITO), Boeretang 200, 2400, Mol, Belgium;

bUniversity of Antwerp, Department of Bio-Engineering, Groenenborgerlaan 171, 2020 Antwerp,

Belgium;

cDepartment of Water Management, Delft University of Technology, PO Box 5048, 2600 GA Delft,

The Netherlands;

dGhent University, Department of Soil Management, Coupure links 653, 9000 Gent, Belgium.

*Corresponding author: Tel: +32 14 33 5634; fax: +32 14 583 0523.

‡ These authors contributed equally.

E-mail addresses: Milica Velimirovic: milica.velimirovic@vito.be; Luca Carniato: L.Carniato@tudelft.nl; Queenie Simons: queenie.simons@vito.be; Gerrit Schoups:

G.H.W.Schoups@tudelft.nl; Piet Seuntjens: piet.seuntjens@vito.be; Leen Bastiaens: leen.bastiaens@vito.be.

Abstract

In this study, the aging behavior of microscale zerovalent iron (mZVI) particles was investigated by

quantifying the hydrogen gas generated by anaerobic mZVI corrosion in batch degradation

experiments. Granular iron and nanoscale zerovalent iron (nZVI) particles were included in this

study as controls. Firstly, experiments in liquid medium (without aquifer material) were performed

Page 3 of 30

Accep

ted

Man

uscr

ipt

3

and revealed that mZVI particles have approximately a 10-30 times lower corrosion rate than nZVI

particles. A good correlation was found between surface area normalized corrosion rate (RSA) and

reaction rate constants (kSA) of PCE, TCE, cDCE and 1,1,1-TCA. Generally, particles with higher

degradation rates also have faster corrosion rates, but exceptions do exists. In a second phase, the

hydrogen evolution was also monitored during batch tests in the presence of aquifer material and

real groundwater. A 4-9 times higher corrosion rate of mZVI particles was observed under the

natural environment in comparison with the aquifer free artificial condition, which can be attributed

to the low pH of the aquifer and its buffer capacity. A corrosion model was calibrated on the batch

experiments to take into account the inhibitory effects of the corrosion products (dissolved iron,

hydrogen and OH-) on the iron corrosion rate.

Keywords: mZVI, H2 production, apparent corrosion rate, life-time

1. Introduction

Reductive dechlorination of chlorinated aliphatic hydrocarbons (CAHs) with zerovalent iron (ZVI)

particles has been shown as a promising technique for in-situ remediation of contaminated

groundwater [1-3]. The long-term performance of ZVI remedial systems and consequently site

cleanup time, depends on the continuous effectiveness of the iron surface to serve as an electron

donor in reducing CAHs [4]. Under anaerobic conditions, the reduction mechanism for CAHs by

ZVI is through direct oxidative corrosion of zerovalent iron by water to ferrous ion and hydrogen

gas, and reductive dechlorination of CAHs to their nonchlorinated analogues and chloride ions [3,5].

ZVI corrodes by the oxidative action of groundwater by overall reaction where one mole of

hydrogen gas is generated for every mol of iron corroded [6,7]:

Fe0 + 2H2O Fe2+ + 2OH- + H2 (1)

As a consequence of the possible conversion of Fe(OH)2 to Fe3O4, an additional 0.33 moles of

hydrogen might be produced for every mole of iron corroded under field conditions [7]. The actual

corrosion pathway is even more complex as the precipitates formed are strongly dependent on

geochemical conditions such as groundwater composition, pH, redox potential, alkalinity and

Page 4 of 30

Accep

ted

Man

uscr

ipt

4

dissolved oxygen. However, estimation of the corrosion rate is important as ZVI corrosion

negatively impacts the reactivity of ZVI particles (1) by forming a corrosion layer on the surface of

ZVI which reduces the exchange rate of electrons between ZVI and the CAHs, and (2) by depletion

of ZVI, and as such of electrons. To date, precise estimation of the longevity and long-term

performance of ZVIs is challenging.

The aging behavior due to corrosion of different ZVI particles was investigated previously [6-11].

According to previous studies, the estimated life-time of granular particles ranged from several years

to several decades [4,6,12]. On the other hand, nanoscale zerovalent iron (nZVI) particles had

relatively shorter life-time compared to granular ZVI [13,14]. Estimated life-times of nanoscale

zerovalent (nZVI) particles after in-situ application were just a few weeks [10,11,13,14] and up to

one year for Ni/Fe nanoparticles [15] as well as crystalline nZVI [16]. According to the limited data,

mZVI particles are expected to have an intermediate life-time [7,17]. However, the effective life-

time of ZVI particles after in-situ application is a point of discussion in many studies. The

correlation of long-term performance and longevity of granular and microscale particles was

investigated most of the time using column tests or via hydrogen pressure build-up in long-term

batch studies [7,18]. The aging rate of particularly nZVIs was estimated via direct hydrogen

evolution in batch tests [11,19].

Several models were proposed to estimate the longevity of permeable reactive barriers, where

granular iron was used. Kouznetsova et al. [20] proposed a phenomenological model where the

decline in iron reactivity is a function of space and time, without describing the iron corrosion

processes. More detailed geochemical modeling of iron deactivation was proposed by different

research groups [21-23]. For nZVI, empirical rate laws describing the iron consumption due to

corrosion and contaminant degradation, were developed by Liu and Lowry [11].

The aim of this study was to fill the knowledge gap and assess the life-time of mZVI particles via

hydrogen (H2) production measurements and calculation of corrosion rates. In a first step H2

production was quantified for 17 different types of ZVI in standardized batch reactors with synthetic

groundwater polluted by CAHs [24] and converted to ZVI corrosion rates. Correlations between the

ZVI corrosion rate and particle sizes, specific surface areas and CAH-degradation rates were

Page 5 of 30

Accep

ted

Man

uscr

ipt

5

determined. Next, the impact of aquifer material and real CAH-polluted groundwater on the apparent

H2 production rate and consequently on the corrosion of selected ZVI particles was studied. Iron

corrosion rates were determined via zero-order fitting but also via numerical modeling [25], where

inhibitory effects of the iron corrosion reaction products (OH-, dissolved iron and dissolved H2) on

the corrosion rates were taken into account. Finally, life-times of the different ZVIs were estimated

based on the obtained corrosion rates.

2. Materials and Methods

2.1. Zerovalent iron particles

mZVIs were supplied by Höganäs and BASF, while granular ZVI and nZVI particles by Gotthart

Maier and Nanoiron/TODA, respectively. More details are summarized in Table 1 and SI1.

2.2. Batch experiments with synthetic groundwater

Batch experiments were carried out as described earlier [24] when reporting on CAH-degradation

rates for ZVIs in synthetic groundwater (SI2). Life-times of ZVIs were determined via pressure and

hydrogen (H2) gas measurements in the same reactors. For nZVIs, samples were taken 0, 3, 6, 8 and

22 days after the start of the test, for other ZVI particles after 0, 14, 28, 49 and 105 days. Pressure

build-up was measured manually using a BLAUWE LIJN® S2401-10A pressure manometer

(EURO-INDEX, Netherlands). Before performing CAH-analysis, 2000 µL headspace-samples were

taken directly from the batch reactors with an air-tight syringe for H2 analyses using a gas Trace GC

MPT-10286 (Interscience) equipped with a HayeSepQ column (Alltech Associates, Inc.) and a

thermal conductivity detector. Calibration was based on standards ranging from 2 to 60 % hydrogen

spiked directly in a N2-headspace taking into account the low solubility of H2 in water.

2.3. Batch experiments with contaminated aquifer and groundwater

For three ZVI types an additional set of similar experiments was set-up using real groundwater

(taken 8 m below ground surface, SI2) and aquifer material originating from a CAHs contaminated

site in Belgium. To mimic in-situ conditions after iron injection, batch tests contained 45 g of aquifer

material and 25 g of real groundwater supplied with 50 g kg-1 of a selected granular and microscale

ZVI or 5 g kg-1 of nanoscale ZVI, leaving a 90 ml headspace. The experiments were set-up in an

Page 6 of 30

Accep

ted

Man

uscr

ipt

6

anaerobic glove box (nitrogen) and consisted of 3 replicates. The vials were incubated at

groundwater temperature (12 0C) on an Edmund Bühler SM 30-control shaker (125 rpm min-1). Each

series included a control set using the same conditions but without ZVI particles.

In time, the CAH-concentrations, pressures and hydrogen measurements were performed as

described for the tests with synthetic groundwater. ORP and pH were measured using a redox/pH

meter (Radiometer) at the first and last sampling point in the test conditions with mZVI. For nZVIs,

pH and ORP were measured 0, 6 and 22 days after the beginning of experiment.

2.4. Apparent corrosion rate and ZVI life-time calculations

Hydrogen production data were used for calculating corrosion rates. Accordingly to Reardon [7],

apparent corrosion rates are estimated accounting for the hydrogen entering in gas and liquid phase.

From the ideal gas law, the moles of hydrogen in the gas phase at each sampling event were

calculated as:

(2)

where VH2 is the corresponding volume of H2 in the gas phase (L), R is the ideal gas constant

(8.314 J K-1 mol-1) and T is temperature during the measurements (293.15 K). A pressure, P of 1 atm

(ideal gas pressure) was used to calculate the moles of hydrogen in the reactor. Pressure build-up in

the batch reactors was measured and ranged from 1 to 1.1 atm, while up to 1.3 atm was only

observed for batch reactors containing nZVI particles at the last time point. The VH2 was calculated

directly from the H2 standards prepared in vials containing the same volume of the synthetic

groundwater as batch reactors.

The rates of hydrogen production in the gas phase (Rgas) were calculated fitting a zero order model

to the estimated moles H2(gas) at each sampling event :

Rgas = - d[moles H2(gas)] / dt (3)

The moles of hydrogen in the aqueous phase at the end of the experiment were calculated using

Henry’s law:

moles H2(aq) = KHPH2MH2O (4)

where MH2O is the mass of water in kilograms. At 20 °C, the Henry’s law constant (KH), calculated

from the solubility of hydrogen in water [26], is equal to 10-5.098 for hydrogen pressures expressed in

moles H2(gas) = PVH2 / (RT)

Page 7 of 30

Accep

ted

Man

uscr

ipt

7

kPa. PH2 is the final overpressure (PH2=Pinitial-Pend) measured in the headspace. The rate of hydrogen

production in the aqueous phase (Raq) was calculated dividing moles H2(aq) by the duration of the

experiment tend:

Raq = moles H2(aq) / tend (5)

The apparent corrosion rate (Rapp, mmol of Fe kg-1 d-1) was calculated summing Rgas and Raq and

assuming that 1 mol of hydrogen in the reactor is produced for every mole of iron corroded with no

conversion of Fe(0H)2 to Fe3O4 [6,27,28]:

(6)

where MFe is the mass of iron in kilograms. Finally, the corrosion rate was normalized to the

specific surface area of the iron. Assuming that iron particles will be completely consumed in the

reaction with groundwater [29,30], calculated iron corrosion rates can be used to estimate life-time

of iron particles, namely by dividing the iron content by the calculated corrosion rate [28].

Considerable amount of hydrogen might also enter in the iron lattice, depending on the iron

characteristics (e.g. elemental composition) and on the hydrogen pressure in the headspace.

Corrected corrosion rates might be estimated by adding to Rapp the rate of hydrogen entering the iron

[7]:

2

0.5Hcorr app sR R k P= + (7)

where ks is the Sieverts rate constant [31] and 2HP is the average hydrogen overpressure. To

determine the value of ks the headspace pressure must be decreased at least once during the

experiment, producing a discontinuity in the apparent corrosion rate. The value of ks that removes

the discontinuity in the corrected corrosion rate corresponds to the estimated ks [7]. Since this test

was not performed, we report only the apparent corrosion rates, which might underestimate the

corrosion rates for some iron types.

2.5. Modeling approach

For three iron types, the H2 and the pH measurements were used to estimate the parameters of a

PHREEQC (version 2) batch model [25] that simulates the corrosion rate in the liquid phase and the

hydrogen volume in the headspace. The GAS_PHASE keyword in PHREEQC was used to simulate

the batch system, keeping the total gas volume equal to the headspace (60 ml in the synthetic

Rapp= (Rgas+ Raq)/ MFe

Page 8 of 30

Accep

ted

Man

uscr

ipt

8

groundwater experiments and 85 ml in the real groundwater experiments) and allowing hydrogen

and carbon dioxide to enter in the gas phase (at the beginning of the experiment no hydrogen is

present). Therefore, the pressure built up due to hydrogen production was taken into account in the

model. The information about the aquifer composition and iron content were used as initial

condition for the liquid phase (100 ml for the synthetic experiment and 30 ml for the real

groundwater experiments). The corrosion rate in the liquid phase was described using the following

formula [21]:

⎧ ⎫⎡ ⎤⎛ ⎞⎪ ⎪= −⎨ ⎬⎢ ⎥⎜ ⎟⎝ ⎠⎪ ⎪⎣ ⎦⎩ ⎭

0max 1 ,0IrCIrC app Fe

IrC

IAPR R M

K (8)

where RIrC is the actual corrosion rate (mmol L-1 d-1), Rapp is the apparent corrosion rate (mmol of Fe

kg-1 d-1), MFe0 is the amount of iron present in the bottle (kg L-1), KIrC is an equilibrium constant for

the iron corrosion reaction and IAP is the ion activity product of iron corrosion reaction (following

equation 1, IAP is equal to {Fe2+}{H2}{OH-}2, where braces stand for activity). Thus, the corrosion

rate slows down when the IAP increases and when the mass of iron present MFe0 in the liquid phase

is consumed. The rate expression in Eq. 8 accounts for the inhibitory effect of high pH, hydrogen

and iron concentrations on the corrosion rate, not the corrosion layer formed on the ZVI.

Subsequently, Eq. 8 rate model will be referred as inhibition model.

Inhibitory effect of high pH’s on batch tests has been reported for similar experiments [11]. Their

study concluded that iron corrosion rate has a first-order dependency on dissolved H+ at pH less than

8 and a zero-order dependency at higher pH. The inhibition model accounts for this observed shift.

Moreover, the influence of the aquifer composition on the IAP, and therefore the corrosion rate, is

also accounted for the model, since dissolved species (dissolved iron, dissolved hydrogen and pH)

are calculated from geochemical equilibrium reactions included in the PHREEQC database (SI3-

SI5). The precipitation of carbonate minerals (such as calcite and iron hydroxy carbonate) was not

modeled, even if their precipitation will likely occur and will lower the corrosion rate. A reasonable

approximation was to assume the liquid phase in equilibrium with amorphous iron hydroxide, a

mineral that certainly forms in iron/water systems [7]. Using this approximation the average pH

value was reasonably reproduced for 4 out of 6 experiments.

Page 9 of 30

Accep

ted

Man

uscr

ipt

9

The Rapp and the equilibrium constant KIrC were calibrated on the available hydrogen and pH

measurements for each batch experiment (both with synthetic and real groundwater). The calibration

was performed minimizing the following objective function:

φ=

= −∑ 2, ,

1

( )n

i meas i simi

C C

(9)

where Ci,meas is the measured concentration (hydrogen in the headspace or pH) , Ci,sim is the model

simulated counterpart and n is the total number of measurements (varying from a minimum of 5 to a

maximum of 8). Minimization was performed using the SCE-UA algorithm [32]. To be consistent

with the pH measurements, hydrogen concentrations were log-transformed in Eq. 9. The initial

aquifer compositions and physical conditions used in the model are reported in SI2.

3. Results and Discussion

3.1. H2 production rate and estimated apparent corrosion rates of ZVIs

Zero-order fits of H2 (ml) produced in the ZVI batch reactors over time are shown in SI6. Calculated

Rapp with corresponding estimated iron life-times are presented in Table 1. Most reactors exhibited

initially a linear increase in H2 content.

The Rapp of the examined mZVIs ranged between 0.20 ± 0.01 and 4.36 ± 0.04 mmol kg-1 d-1, with an

average of 1.74 mmol kg-1 d-1. The lowest Rapp corresponds to the FeH7, which is also among the

least reactive irons towards contaminants. Highly reactive FeQ2 iron has the highest corrosion rate

(4.36 ± 0.04 mmol kg-1 d-1), followed by FeH14 (3.14 ± 0.02 mmol kg-1 d-1), FeH8 (3.03 ± 0.04) and

HQ (2.88 ± 0.03 mmol kg-1 d-1) among other mZVI particles. The calculated corrosion rates are

lower than previously reported for electrolytic irons (d50 < 150 µm) according to the H2 pressure

build up measurements and accounted amount of H2 gas that directly enters the iron lattice [8].

For a number of ZVIs, a clearly decreased production was observed near the end of the 105 days

experimental time. This is especially the case for carbonyl irons MS200, MS200+ and HQ where the

corrosion rate slowed down and the H2 content reached a plateau after 49 days. This is consistent

with previously published data where a plateau in H2 production and consequently iron passivation

is explained by accumulation of carbonate-containing precipitates [33]. In this case, only the first 4

Page 10 of 30

Accep

ted

Man

uscr

ipt

10

data points were used for calculation of the Rapp, resulting in overestimated corrosion rates.

Interestingly, MS200+ is more reactive than MS200 [24], yet its initial corrosion rate is lower.

Deviations from a linear increase in H2 evolution are also evident for FeH11 (R2 = 0.703) and FeH12

(R2 = 0.714). The origin of the lag phase might be a consequence of the high oxygen content (SI1) of

these particles [34]. Odziemkowski et al. [35] suggested that the corrosion reaction occurs in four

stages and during the first three phases a variable H2 evolution rate might be observed before

stabilization observed after 10 days. Herein, most of the production rates did not show a substantial

variation after 14 days (SI6, with exception for FeH11 and FeH12), indicating that a steady state

condition was reached.

The Rapp of reference granular material FeA4 was 0.36 ± 0.06 mmol kg-1 d-1. This is slightly higher

than or equal to reported corrosion rates from 0.1 to 0.7 mmol kg-1 d-1 for granular iron supplied by

Master Builders [6] and 0.3 mmol kg-1 d-1 for Connelly iron [7], respectively.

For the Nanofer25s the highest Rapp (145 ± 4.33 mmol kg-1 d-1) was recorded. The Rapp of RNIP was

approximately 5 times lower than that of Nanofer25s. Differences in corrosion rates of Nanofer25s

and RNIP might be explained by different synthesis method [10,16]. The large surface area of nZVIs

particles may be another reasons for their fast consumption in batch reactors. Moreover, the Rapp of

RNIP corresponds to previously reported corrosion data (35.4 to 41.1 mmol kg-1 d-1) for nZVI [19].

Finally, the Rapp of mZVIs, compared to granular ZVI and nZVI, identify mZVI as prospective

particles for in-situ reactive zones treatments.

3.2. Effect of particle size and surface area on apparent corrosion rates of ZVIs

H2 production measurements in batch reactors revealed that generally particle size influences the

iron corrosion rate, which confirms findings reported by Noubactep et al. [14]. An inverse

correlation was found between Rapp and ZVI particle sizes (Fig. 1), although the correlation factor

was relatively low (R2=0.52).

Comparing to all examined ZVIs, mZVI particles supplied by Höganäs showed lower corrosion rate

as a result of their larger particle size. Particles with similar size, FeH3 (D50 = 84 µm), FeH6 (D50 =

98 µm) and FeH7 (D50 = 96 µm), showed similar corrosion rates. Corrosion rates of BASF irons and

Höganäs particles were generally within the same range.

Page 11 of 30

Accep

ted

Man

uscr

ipt

11

Finally, corrosion rates for nZVIs are one order of magnitude higher than for mZVIs with

intermediate corrosion rate (1 to 4 mmol kg-1 d-1) as a consequence of smaller particle size.

A better correlation was observed between the corrosion rate of ZVI particles and their particle

surface area (Fig. 2). As expected, particles with higher surface area displayed the highest corrosion

rate. Exceptions from the correlation analysis are particles (FeA4, SM and FeH11-FeH14) with high

carbon content (SI1), and as a consequence its high sorption properties [34] and/or limited or no

reactivity [34] due to the high oxygen content (SI1). Moreover, high carbon content can affect

surface area measurements [34]. As such, iron corrosion rate are a complex function of different

factors like particle size, specific surface area of the particles, as well as iron composition.

3.3. Correlation between apparent ZVI corrosion rates and CAH degradation rates

Surface area normalized iron corrosion rates (RSA, SI7) from this study were plotted against

previously presented surface area normalized degradation rate coefficients [24] (kSA, SI7) for PCE,

TCE, cDCE and 1,1,1-TCA (Fig. 3). A significant effect (p<0.05) of surface normalized iron

corrosion rates on surface area normalized degradation rates is observed, using a one-way analysis

of variance (ANOVA). However, satisfactory correlation coefficient (R2 = 0.68) was only observed

for cDCE.

FeH3 and FeH4 particles with the highest kSA coefficients are among the particles with the highest

RSA values, namely 3.05x10-7 and 8.39x10-7 mol m-2 h-1, respectively. This observation is consistent

with other studies where in some cases the degradation rate constants were found to be dependent on

the iron corrosion rate [11,36]. On the other hand, it is possible that H2 and examined CAHs do not

compete for the same reactive sites [22]. The normalized corrosion rates for nZVIs are in the same

order of magnitude as most mZVIs.

3.4. Impact of aquifer and real groundwater on H2 production rate

In the presence of aquifer material and real groundwater, H2 production based iron corrosion rates

were determined for three selected ZVIs. FeH4 and BASF HQ were selected as representative

reactive mZVIs, while Nanofer25s as nZVI with the highest reactivity and the fastest corrosion rate

in the aquifer free conditions. Fig. 4 shows the H2 evolution over time measured in aquifer batch

Page 12 of 30

Accep

ted

Man

uscr

ipt

12

reactors supplied with FeH4, BASF HQ and Nanofer25s in similar conditions where different stages

could be distinguished.

Taking into account the data collected the first 14 days, corrosion rates calculated using a zero-order

approximation for FeH4 and BASF HQ were 10.5 and 54.1 mmol kg-1 d-1, respectively. Afterwards,

the corrosion rate of these particles dropped approximately one order of magnitude. A three-step H2

production was observed for Nanofer25s with a corresponding initial corrosion rate of 3.44 x 103

mmol kg-1 d-1. The next six days, the calculated corrosion rate was 10 times lower (3.31 x 102 mmol

kg-1 d-1). Finally, a plateau was observed, suggesting that no further significant corrosion was

ongoing. The initial corrosion rates of particles in batch degradation tests with aquifer materials

were 6 to 24 times higher than the condition with artificially CAH-polluted groundwater as a

consequence of the initially low pH in the slurry (pH = 5.11). Faster iron passivation in the real

aquifer system might also be explained by dissolved carbonate species (e.g. HCO3-) responsible for

accelerated ZVI corrosion [22]. For TCE and cDCE, both present in the test with synthetic

groundwater and real aquifer system, faster degradation rates were observed in the natural

environment, explicable by the lower pH [11].

3.5. Iron corrosion rates determined via numerical modeling

A tool to predict corrosion rate of the iron particles under different conditions was developed in this

study. A limitation of the zero-order model is its inability to describe the observed plateau in the H2

content observed for some batch experiments (see Fig. 5). The model proposed with Eq. 8 is able to

adapt the iron corrosion rate as the liquid phase becomes saturated with hydrogen, dissolved iron

and OH- and aims reproducing the observed plateau in the H2 content.

The comparison between the measured and simulated pH and hydrogen productions with synthetic

and real groundwater is shown in Fig. 5.

From the H2 measurements it can be seen that the H2 production slowed down at later time, except

for the last measurements of the FeH4 batch experiment with synthetic groundwater (Fig. 5(a)). A

good fit to the hydrogen measurements was obtained for HQ and Nanofer25s tests (Fig. 5(c)-(f)). A

reasonable fit was obtained for FeH4 with real groundwater (Fig 5(b)) and for the first

measurements of the synthetic test performed with the same iron type (Fig 5(a)). Excluding this

Page 13 of 30

Accep

ted

Man

uscr

ipt

13

case, the value of the equilibrium constant reported in table 2 varies between 1.55 x 10-18 to 6.47 x

10-18. Low values of the equilibrium constant allow the simulation of the observed plateau in the H2

content at the end of most experiments. For FeH4 iron with synthetic groundwater a slowdown in

the hydrogen production was not observed and the equilibrium constant was higher, resulting in a

zero order fit to the hydrogen measurements. The value measured at the end of the FeH4 batch with

synthetic groundwater might represent an outlier, producing an unrealistic estimation of the

equilibrium constant. The observed higher hydrogen production in batch reactors with lower pH

(real groundwater) agrees with the formulation of the corrosion rate proposed in Eq. 8.

The simplified geochemistry used in the model, here simulated using the equilibrium reactions of the

PHREEQC database and assuming the liquid phase in equilibrium with amorphous iron hydroxide,

did not allow the reproduction of pH dynamics but only the prediction of a constant pH value. This

constant value agrees reasonably with the pH measurements of the FeH4 batch experiment with

synthetic groundwater (Fig. 5(a)), HQ with synthetic groundwater and real groundwater (Fig. 5(c)

and 5 (d)) and Nanofer25s with synthetic groundwater (Fig. 5(e)). However, the pH values measured

at the end of the tests 5(b) and 5(f) were largely overestimated by model simulations. Even

considering the liquid phase in equilibrium with magnetite (which releases acidity), the pH did not

decrease below 7.56 (result not shown). These low pH measurements can only be reproduced with a

zero corrosion rate, which is not plausible based on the hydrogen measurements. It is hypothesized

that in the experiments performed with real groundwater the soil matrix was able to buffer the pH

increase due to the cation exchange capacity. As can be seen from Fig. 5 the initial pH values in the

real groundwater batch reactors is much lower than those in the synthetic groundwater batch

reactors. For FeH4 the initial groundwater pH was 4.38, which indicates a strongly acid soil. While

corrosion proceeds part of the OH- released could be compensated by H+ releases previously sorbed

in the soil. Additional analyses are required to provide exact explanations of the low observed pH’s

for FeH4 and Nanofer25s, for example by buffering the batch reactions with sodium borate to

determine if the aquifer material is responsible for the low observed pH’s.

It is interesting to note that in Fig. 5(f) the hydrogen production reach a plateau despite the low pHs

during the experiment. This suggests again that Eq. 8 is appropriate to model corrosion reaction

Page 14 of 30

Accep

ted

Man

uscr

ipt

14

because can also account for the effect of hydrogen and dissolved iron on the iron corrosion rate.

Further improvements might include the addition of cation exchange processes for the experiment

performed with aquifer material and the description of the passivation layer, by using detailed

information on the mineral composition of the corrosion ring.

In Table 2, the calibrated corrosion rate parameters and the iron corrosion equilibrium constants are

reported. As expected, the higher hydrogen production observed for the batch experiments

performed with real groundwater gave an estimate of higher corrosion rates, which can be attributed

to the higher groundwater buffer capacity in the experiments performed with aquifer material.

3.6. Translation of apparent corrosion rates into life-times of ZVIs

The Rapp reported in Table 1 were extrapolated to life-times of the studied ZVI particles. The

estimation was obtained by dividing the iron mass present in the batch tests by the estimated initial

corrosion rate of the particles assuming no passivation occurs. This can be considered as a worst

case scenario for estimated life-time of ZVIs, as corrosion rates slow down over time due to the

formation of passivating precipitates.

The extrapolated life-times under aquifer free conditions according to the initial corrosion rate of

reference granular material FeA4 was 137.5 years. The shortest life-time within mZVIs was

estimated for highly reactive FeQ2 (11.24 years), followed by FeH14 (15.61 years) and FeH8

(16.22 years). Estimated life-time of FeH4 iron further studied in the presence of aquifer and real

groundwater was 25.93 years. The shortest life-time among all examined ZVIs was reported for

Nanofer25s (0.34 years) and RNIP (1.62 years). The estimated life-time of Nanofer25s corresponds

to previously reported life-time of 8 to 80 days [8,11].

As reported, in the presence of rather acid aquifer and real groundwater initial corrosion rate was

higher than for the tests with artificial groundwater polluted by CAHs. The same trend was observed

for the extrapolated life-times. Considering only the initial corrosion rate (Fig. 4), before passivation

occurs, estimated life-time of FeH4, HQ and Nanofer25s were 4.66, 0.91 and 0.01 years,

respectively. However, taking into account the corrosion rate of the second corrosion phase, after the

initial high corrosion phase, the estimated life-time of FeH4, HQ and Nanofer25s increased till

Page 15 of 30

Accep

ted

Man

uscr

ipt

15

33.13, 13.51 and 0.15 years, respectively. This is significantly higher than estimated for the initial

corrosion rate.

4. Conclusions

This study was performed (1) to determine the corrosion rates for mZVI particles in batch reactors

and (2) to extrapolate the results in time to estimate worst case life-times of mZVI under field

conditions.

The obtained results suggest that the initial corrosion rate of ZVIs depends on particle properties (the

particle size, specific surface area of the particle, particle composition), as well as on environmental

factors (pH, buffer capacity). Using a zero-order model, Rapp of ZVIs were calculated. This may be

underestimation of the real corrosion rate of ZVI particles as the hydrogen entering the lattice was

not accounted for. On the other hand, it may also be an overestimation of the real corrosion rates as

the initial apparent hydrogen production rates were used, which was shown to be reduced after a

while. Via numerical modeling inhibitory effects of the iron corrosion reaction products can be taken

into account. To allow more accurate prediction of the corrosion rate of ZVI in liquid medium, iron

speciation within the passivating corrosion layer on the ZVI particle should be investigated more in

detail.

In contrast to aquifer-free conditions, higher initial iron corrosion rates were estimated for the

batches performed with aquifer and real groundwater, where low pH levels were measured during

the experiments. Low pH levels might be maintained by the soil buffer capacity and can promote

iron corrosion. In contrary, limited buffer capacity will lead to the increase in pH and lead to the

precipitation of ferrous hydroxide. Summarizing this experiments as a worst case scenario, under

field conditions the estimated life-time of mZVIs using only initially high corrosion rates was 0.9-

4.7 years for mZVIs and only 0.01 years for nZVI. When using corrosion rates from a later

experimental phase, estimated life-time of mZVIs was 13.5-33.1 years, and 0.15 years for nZVI.

Reality is expected to be in between the two scenarios.

Page 16 of 30

Accep

ted

Man

uscr

ipt

16

The presented data on extrapolated life-times of ZVI particles can serve as a first worst case

indication of the life-times. To determine the actual life-time of particles under in-situ conditions,

further studies in continuous aquifer systems are needed.

Acknowledgments

The work was developed in the framework of the European Union project AQUAREHAB (FP7 -

Grant Agreement Nr. 226565). Special thanks to Höganäs, Sweden for providing mZVI particles.

Page 17 of 30

Accep

ted

Man

uscr

ipt

17

References

[1] R.W. Gillham, S.F. O'Hannesin, Metal-catalysed abiotic degradation of halogenated organic

compounds. In: Proceedings of IAH Conference Modern Trends in Hydrogeology (10-14 May 1992)

Hamilton, Ontario, Canada, 1992, pp.94-103.

[2] R.W. Gillham, S.F. O'Hannesin, Enhanced degradation of halogenated aliphatics by zerovalent

iron, Ground Water. 32 (1994) 958–971.

[3] I.J. Matheson, P.G. Tratnyek, Reductive dehalogenation of chlorinated methanes by iron metal,

Environ. Sci. Technol. 28 (1994) 2045-2053.

[4] Farrell, J. M. Kason, N. Melitas, T. Li, Investigation of the longterm performance of zero-valent

iron for reductive dechlorination of trichloroethylene, Environ. Sci. Technol. 34 (2000) 514-521.

[5] W.A. Arnold, A.L. Roberts, Pathways and kinetics of chlorinated ethylene and chlorinated

acetylene reaction with Fe(0) particles, Environ. Sci. Technol. 34 (2000) 1794-1805.

[6] E.J. Reardon, Anaerobic corrosion of granular iron: Measurement and interpretation of hydrogen

evolution rates, Environ. Sci. Technol. 29 (1995) 2936-2945.

[7] E.J. Reardon, Zerovalent Irons: Styles of Corrosion and Inorganic Control on Hydrogen Pressure

Buildup, Environ. Sci. Technol. 39 (2005) 7311-7317.

[8] E.J. Reardon, R. Fagan, J.L. Vogan, A. Przepiora, Anaerobic Corrosion Reaction Kinetics of

Nano-sized Iron, Environ. Sci. Technol. 42 (2008) 2420-2426.

[9] R.A. Gillham, Discussion of nano-scale iron for dehalogenation, Ground Water Monit. R. 23

(2003) 6-8.

[10] Y. Liu, S.A. Majetich, R.D. Tilton, D.S. Sholl, G.V. Lowry, TCE dechlorination rates,

pathways, and efficiency of nanoscale iron particles with different properties, Environ. Sci. Technol.

39 (2005) 1338-1345.

Page 18 of 30

Accep

ted

Man

uscr

ipt

18

[11] Y. Liu, G.V. Lowry, Effect of Particle Age (Fe0 Content) and Solution pH On NZVI

Reactivity:� H2 Evolution and TCE Dechlorination, Environ. Sci. Technol. 40 (2006) 6085-6090.

[12] J. Klausen, P.J. Vikesland, T. Kohn, D.R. Burris, W.P. Ball, A.L. Roberts, Longevity of

granular iron in groundwater treatment processes: solution composition effects on reduction of

organohalides and nitroaromatic compounds, Environ. Sci. Technol. 37 (2003) 1208-1218.

[13] F. He, D.Y. Zhao, Manipulating the size and dispersibility of zerovalent iron nanoparticles by

use of carboxymethyl cellulose stabilizers, Environ. Sci. Technol. 41 (2007) 6216–6221.

[14] C. Noubactep, S. Caré, R.A. Crane, Nanoscale metallic iron for environmental remediation:

prospects and limitations, Water Air Soil Poll. 223 (2012) 1363-1382.

[15] B. Schrick, J.L. Blough, A.D. Jones, T.E. Mallouk, Hydrodechlorination of trichloroethylene to

hydrocarbons using bimetallic nickle-iron nanoparticles, Chem. Mater. 14 (2002) 5140-5147.

[16] Y. Liu, H. Choi, D. Dionysiou, G.V. Lowry, Trichloroethene hydrodechlorination in water by

highly disordered monometallic nanoiron, Chem. Mater. 17 (2005) 5315-5322.

[17] K.-F. Chen, S. Li, W.-X. Zhang, Renewable hydrogen generation by bimetallic zerovalent iron

nanoparticles, Chem. Eng. J. 170 (2011) 562–567.

[18] O.J. Suk, S.-W. Jeen, R.W. Gillham, L. Gui, Effects of initial iron corrosion rate on long-term

performance of iron permeable reactive barriers: column experiments and numerical simulation, J.

Contam. Hydrol. 103 (2009) 145-156.

[19] Y. Zhuang, S. Ahn, A.L. Seyfferth, Y. Masue-Slowey, S. Fendorf, R.G. Luthy, Dehalogenation

of Polybrominated Diphenyl Ethers and Polychlorinated Biphenyl by Bimetallic, Impregnated, and

Nanoscale Zerovalent Iron, Environ. Sci. Technol. 45 (2011) 4896–4903.

[20] I. Kouznetsova, P. Bayer, M. Ebert, M. Finkel, Modelling the long-term performance of zero-

valent iron using a spatio-temporal approach for iron aging, J. Contam. Hydrol. 90 (2007) 58-80.

Page 19 of 30

Accep

ted

Man

uscr

ipt

19

[21] K.U. Mayer, D.W. Blowes, E.O. Frind, Reactive transport modeling of an in situ reactive

barrier for the treatment of hexavalent chromium and trichloroethylene in groundwater, Water

Resour. Res. 37 (2001) 3091–3103.

[22] S.-W. Jeen, K.U. Mayer, R.W. Gillham, D.W. Blowes, Reactive Transport Modeling of

Trichloroethene Treatment with Declining Reactivity of Iron, Environ. Sci. Technol. 41 (2007)

1432-1438.

[23] L. Carniato, G. Schoups, P. Seuntjens, T. Van Nooten, Q. Simons, L. Bastiaens, Predicting

longevity of iron permeable reactive barriers using multiple iron deactivation models, J. Contam.

Hydrol. 142–143 (2012) 93–108.

[24] M. Velimirovic, P.-O. Larsson, Q. Simons, L. Bastiaens, Reactivity screening of microscale

zerovalent irons and iron sulfides towards different CAHs under standardized experimental

conditions, J. Hazard. Mater. 252-253 (2013) 204-212.

[25] D.L. Parkhurst, C.A.J. Appelo, User's guide to PHREEQC (Version 2) - a computer program

for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. U.S.

Geological Survey, Water-Resources Investigations Report 99-4259, 1999.

[26] J.A. Dean (1992) Lange’s Handbook of Chemistry. 14th ed. McGraw- Hill Inc., New York.

[27] B.C. Reinsch, B. Forsberg, R.L. Penn, C.S. Kim, G.V. Lowry, Chemical transformations during

aging of zero-valent iron nanoparticles in the presence of common groundwater dissolved

constituents. Environ. Sci. Technol. 44 (2010) 3455-3461.

[28] Y. Liu, T. Phenrat, G.V. Lowry, Effect of TCE concentration and dissolved groundwater solutes

on NZVI-promoted TCE dechlorination and H2 evolution. Environ. Sci. Technol. 41 (2007) 7881-

7887.

[29] M.M. Scherer, S. Richter, R.L. Valentine, P.J.J. Alvarez, Chemistry and microbiology of

permeable reactive barriers for in situ groundwater clean up. Rev. Environ. Sci. Technol. 30 (2000)

363-411.

Page 20 of 30

Accep

ted

Man

uscr

ipt

20

[30] M.S. Odziemkowski, L. Gui, R.W. Gillham, D.E. Irish (2000) The role of oxide films in the

reduction of N-nitreosodimethylamine with reference to the iron groundwater remediation

technology. In: Hebert, K.R., Lillard, R.S., MacDougall, B.R. (Eds), Oxide Films, Proc. Internat.

Symp. The Electrochemical Society, Inc., Pennington, NJ. Vol. 2000-4, 357-368.

[31] A. Sieverts, W. Danz, Solubility of D2 and H2 in Palladium, Z. Physik. Chem. 34 (1936) 158.

[32] Q. Duan, S. Sorooshian, V. Gupta, Effective and efficient global optimization for conceptual

rainfall-runoff models, Water Resour. Res. 28 (1992) 1015-1031.

[33] J.T. Nurmi, P.G. Tratnyek, Electrochemical studies of packed iron powder electrodes: Effect of

common constituents of natural waters on corrosion potential, Corros. Sci. 50 (2008) 144-154.

[34] M. Velimirovic, P.-O. Larsson, Q. Simons, L. Bastiaens, Impact of carbon, oxygen and sulfur

content of microscale zerovalent iron particles on its reactivity towards CAHs, Chemosphere 93

(2013) 2040-2045.

[35] M.S. Odziemkowski, T.T. Schuhmacher, R.W. Gillham, E.J. Reardon, Mechanism of oxide film

formation on iron in simulating groundwater solutions. Corros. Sci. 40 (1998) 371-389.

[36] Y. Jiao, C. Qiu, L. Huang, K. Wu, H. Ma, S. Chen, L. Ma, D. Wu, Reductive dechlorination of

carbon tetrachloride by zero-valent iron and related iron corrosion, Appl. Catal. B-Environ. 91

(2009) 434–440.

Page 21 of 30

Accep

ted

Man

uscr

ipt

21

Figure Caption

Table 1. Information on iron samples used in the batch study with calculated corrosion rates and

iron particle life-times. All data on iron concentration and H2 production are averages of triplicates.

Fig. 1. Power trend line for the calculated apparent iron corrosion rate (Rapp) relative to particle size

(D50) of zerovalent iron.

Fig. 2. Correlation between BET surface area (m2 g-1) and calculated apparent iron corrosion rate

(Rapp). The linear log-log relationship (R2 = 0.999) was observed between 11 of 17 particles.

Particles excluded from the correlation analysis (black color ellipse).

Fig. 3. Surface normalized degradation rate coefficient (kSA) plotted as a function of surface area

normalized iron corrosion rate (RSA).

Fig. 4. H2 evolution rate measured in batch reactors containing aquifer material and microscale ZVIs

(FeH4 and BASF HQ, each loaded at a dose of 50 g kg-1) (A) and H2 evolution rate measured in

batch reactors containing aquifer material and 5 g kg-1 of Nanofer25s (B).

Fig. 5. Measured (cross marks) and simulated cumulative hydrogen volumes produced in the

headspace (solid line) and measured pH values measured in the liquid phase (asterisks) and

simulated (dashed line) for three types of iron (FeH4, BASF HQ and Nanofer25s).

Table 2. Estimated corrosion rates by fitting iron corrosion rates and hydrogen inhibitory terms (Eq.

8) to hydrogen and pH measurements.

Page 22 of 30

Accep

ted

Man

uscr

ipt

22

Iron manufacturer

Iron name PSD[D10, D50, D90]a BETb

Iron concentration

H2 production

R2 (n)c Rapp

d Estimated lifetime

[μm] [m2 g-1] [g L-1] [ml day-1]

[mmol kg-1 d-

1] [years]

Gotthart Maier (DE) FeA4 300-1300

mm

1.15 50.04 4.34 x 10-2 0.95 (5)

0.36 137.5

Höganäs (SE)

FeH3 FeH4 FeH6 FeH7 FeH8 FeH11 FeH12 FeH13 FeH14 FeQ2

36, 84, 168 22, 41, 62 41, 98, 162 44, 96, 158 34, 63, 97 6, 19, 38 6, 17, 32 7, 18, 34 21, 79, 162 8, 26, 44

0.06 0.09 0.09 0.12 0.35 3.98 3.62 1.50 11.40 0.40

50.00 49.98 49.97 49.99 49.97 49.64 50.01 49.99 49.99 50.02

5.21 x 10-2 2.24 x 10-1 7.73 x 10-2 2.38 x 10-2 3.62 x 10-1 3.18 x 10-1 1.38 x 10-1 6.12 x 10-2 3.72 x 10-1 5.17 x 10-1

0.98 (5) 0.92 (5) 0.97 (5) 0.99 (5) 0.82 (5) 0.70 (5) 0.71 (5) 0.99 (5) 0.97 (5) 0.89 (5)

0.46 1.89 0.65 0.20 3.03 2.68 1.15 0.49 3.14 4.36

106.9 25.93 75.57 246.9 16.22 18.33 42.54 101.1 15.61 11.24

BASF (DE)

BASF MS200 BASF MS200+ BASF SM BASF HQ

2.1, 4.2, 7.2 1.7, 3.7, 7.1 1.4, 2.5, 4.1 0.6, 1.2, 2.4

0.36 0.47 0.48 0.82

50.00 50.01 49.98 50.01

3.72 x 10-1 3.45 x 10-1 2.56 x 10-2 6.61 x 10-1

0.77 (4) 0.89 (4) 0.94 (5) 0.87 (4)

1.59 1.67 0.21 2.88

30.77* 29.39* 236.7 17.06*

NANOIRON (CZ)

Nanofer25s D50< 0.05e

25.0c 5.150 1.78 x 100 0.92 (5)

145 0.337

TODA (JP) RNIP D50< 0.07e 4.97 4.740 3.47 x 10-1 0.9

0 (5)

30.3 1.619

aParticle Size Distribution measured by laser diffraction with a Sympatec Helos/Rodos dry particle size analyzer. bBET : Specific Surface Area analyzed with a Micromeritics Flowsorb II according to the Brunauer-Emmett-Teller (Single point measurement). cCorrelation coefficient (number of data points). dApparent corrosion rate. eProducers data. *Iron passivation was observed within the time frame of the batch test.

Page 23 of 30

Accep

ted

Man

uscr

ipt

23

Synthetic groundwater Natural groundwater

Iron type Rapp [mmol kg-1 d-1] KIrC Rapp

[mmol kg-1 d-1] KIrC

FeH4 1.3 1.05 x 10-12 18.57 1.55 x 10-18 BASF HQ 13.2 2.17 x 10-18 137.2 6.47 x 10-18 Nanofer25S 64.04 3.80 x 10-18 1678.8 3.31x 10-18 Rapp - apparent corrosion rate KIrC - equilibrium constant

Page 24 of 30

Accep

ted

Man

uscr

ipt

24

y = 5.820x‐0.47

R² = 0.5160.1

1

10

100

1000

0.01 0.1 1 10 100

R app(m

mol kg‐1d‐

1 )

D50 (µm)

Nanoscale particles

BASF

Höganäs

Microscale particles Granularparticles

Page 25 of 30

Accep

ted

Man

uscr

ipt

25

R² = 0.999

0.1

1

10

100

1000

0.01 0.10 1.00 10.00

R app(m

mol kg‐1d‐

1 )

BET (m2 g‐1)

R² = 0.999

0.1

1

10

100

1000

0.01 0.10 1.00 10.00

R (m

mol kg‐1d‐

1 )

BET (m2 g‐1)

mZVI‐Höganäs

mZVI‐BASF

nZVI

Granular ZVI

Page 26 of 30

Accep

ted

Man

uscr

ipt

26

1.0E‐06

1.0E‐05

1.0E‐04

1.0E‐03

1.0E‐02

1.0E‐08 1.0E‐07 1.0E‐06

k SA(L m

‐2h‐

1 )

RSA (mol m‐2 h‐1)

PCE

1.0E‐06

1.0E‐05

1.0E‐04

1.0E‐03

1.0E‐02

1.0E‐08 1.0E‐07 1.0E‐06

k SA(L m

‐2h‐

1 )

RSA (mol m‐2 h‐1)

TCE

1.0E‐06

1.0E‐05

1.0E‐04

1.0E‐03

1.0E‐02

1.0E‐08 1.0E‐07 1.0E‐06

k SA(L m

‐2h‐

1 )

RSA (mol m‐2 h‐1)

cDCE

Page 27 of 30

Accep

ted

Man

uscr

ipt

27

1.0E‐05

1.0E‐04

1.0E‐03

1.0E‐02

1.0E‐08 1.0E‐07 1.0E‐06

k SA(L m

‐2h‐

1 )

RSA (mol m‐2 h‐1)

1,1,1‐TCA

mZVI‐Höganäs mZVI‐BASF nZVI Granular ZVI

Page 28 of 30

Accep

ted

Man

uscr

ipt

28

y = 1.2722xR² = 1

y = 6.4972xR² = 1

y = 0.1811x + 14.908R² = 0.9708

y = 0.443x + 88.928R² = 0.7931

0

20

40

60

80

100

120

140

160

0 20 40 60 80 100 120

H2(m

L)

days

(A)50 g kg‐1 FeH4

50 g kg‐1 BASF HQ

y = 40.418xR² = 1

y = 3.8945x + 36.523R² = 1

y = 0.1413x + 59.278R² = 0.9428

0

10

20

30

40

50

60

70

80

0 5 10 15 20 25

H2(m

L)

days

(B)5 g kg‐1 Nanofer25s

Page 29 of 30

Accep

ted

Man

uscr

ipt

29

0 20 40 60 80 1000

5

10

15

20

25

30

35

mL H

2 headspace

days

(a) FeH4 synthetic groundwater

0 20 40 60 80 1000 20 40 60 80 1003

4

5

6

7

8

9

10

0 20 40 60 80 1000

5

10

15

20

25

30

35

days

(b) FeH4 real groundwater

0 20 40 60 80 1000 20 40 60 80 1003

4

5

6

7

8

9

10

pH

0 20 40 60 80 1000

5

10

15

20

25

30

mL H

2 headspace

days

(c) BASF HQ synthetic groundwater

0 20 40 60 80 1000 20 40 60 80 1003

4

5

6

7

8

9

10

0 20 40 60 80 1000

30

60

90

120

150

days

(d) BASF HQ real groundwater

0 20 40 60 80 1000 20 40 60 80 1003

4

5

6

7

8

9

10

pH

0 5 10 15 200

10

20

30

40

50

60

mL H

2 headspace

days

(e) Nanofer25s synthetic groundwater

0 5 10 15 200 5 10 15 203

4

5

6

7

8

9

10

0 5 10 15 200

20

40

60

80

days

(f) Nanofer25s real groundwater

0 5 10 15 200 5 10 15 203

4

5

6

7

8

9

10

pH

Page 30 of 30

Accep

ted

Man

uscr

ipt