Clay Minerals (1998) 33, 87-101
Reduction of nitrate to ammonium by
sulphate green rust" activation energy and
reaction mechanism
H. C. B. H A N S E N
AND C. B E N D E R
KOCH
Chemistry Department, Royal Veterinary and Agricultural University, Thorvaldsensvej 40,
DK- 1871 Frederiksberg C., Denmark
(Received 29 August 1996; revised 9 February 1997)
A B S TRACT: Iron(II)-containing minerals are potential inorganic nitrate reductants in soils and
sediments. Specifically, synthetic green rust (Fe~IFeIII(OH)12SO4.yH20,GR) reduces nitrate to
ammonium. The reaction of GR with two different nitrate salts, NaNO3 and Ba(NO3)2, has been
compared. The reaction stoichiometry and the reaction order with respect to Fe(II) in GR does not
change in the examined temperature range (15-50~ irrespective of the nitrate salt used. Activation
energies of 83.9 + 7.6 kJ mo1-1 and 90.5 ___6.9 kJ mo1-1 have been determined for the reaction of
GR with NaNO3 and Ba(NO3)2, respectively. However, for the latter reaction the rate of reaction is
increased 40 times. Based on X-ray and M6ssbauer investigations, this acceleration of the reaction
rates is attributed to the forced exchange of sulphate with nitrate in GR interlayers caused by
precipitation of BaSO4 outside the GR particles, a reaction which does not occur in the presence of
NaNO3. This difference in anion exchange behaviour is confirmed by anion exchange experiments
with the redox-inactive GR-analogue, pyroaurite. A reaction model initiated by nitrate
electrostatically bound at positively charged sites of GR is proposed.
Detailed knowledge of the redox chemistry of
nitrogen is imperative for quantification of the
global N cycle and thus for assessment of terrestrial
nutrient balances and atmospheric chemistry.
Nitrate is the dominating inorganic N compound
in aerobic terrestrial environments. This anion is
only weakly adsorbed to soil minerals, and thus has
a high mobility in soils and sediments. Therefore, if
not assimilated into the biomass, nitrate may leach
to groundwater, streams or lakes, causing groundwater pollution and eutrophication of aquatic
environments. These effects are non-existent if
nitrate is denitrified to volatile nitrogen forms
(dinitrogen, nitrous oxide) or reduced to the less
mobile ammonium ion under anoxic conditions. If
denitrification were not active, large amounts of
nitrogen would accumulate on land. Unfortunately,
the different nitrate reduction pathways are not
known in great detail, and the amount of nitrate
being reduced and the distribution of the products
among dinitrogen, nitrous oxide and ammonium is
poorly quantified causing rather imprecise estimates
on this part of the nitrogen cycle.
Organic matter is the common reducing agent in
microbial catalysed nitrate reduction, although
Fe(II) and reduced sulphur compounds may serve
as reductants in autotrophic nitrate reduction
(Postma et al., 1991; Rrdelsperger, 1989; Krlle et
aL, 1983). These compounds also may be active in
abiotic, non-enzymatic nitrate reduction, e.g. Fe(II)
bound in minerals such as pyrite, siderite,
hydroxides and silicates have been suggested to
reduce nitrate (Lind & Pedersen, 1976; Ernstsen &
Morup, 1992; Verdegem & Baert, 1985; Postma,
1990; Buresh & Moraghan, 1976). An inverse
relationship between the content of Fe(II) and
nitrate is often observed in subsoils and sediments
where microbial activity is expected to be low
9 1998 The Mineralogical Society
88
H. C B. Hansen and C Bender Koch
(Lind & Pedersen, 1976; Emstsen, 1990). Thus
abiotic nitrate reduction cannot be excluded a
priori. In an attempt to identify geochemically
important mineral reductants, we have found that
Fe(II)Fe(III) hydroxides (green rusts, GRs) are
capable of nitrate reduction in laboratory systems
(Hansen et al., 1996; Koch & Hansen, 1997).
Green rusts belong to the pyroaurite-group of
layered hydroxides (Brindley & Bish, 1976). They
are layered hydroxides composed of positively
charged trioctahedral Fe(II)-Fe(III) hydroxide
sheets, sandwiching layers of charge compensating
anions
and
water
molecules
II
In
[Fe6_xFe;/
(OH)12]x+.[,4z,,yH20]. In this formula A
is an n-valent interlayer anion. In general, carbonate
or sulphate are the two most common interlayer
anions in the pyroaurite group (Drits et al., 1987).
The x coefficient in the formula of GRs is variable,
but in the presence of excess Fe(II) in the synthesis
solution, it takes values close to two (Hansen et al.,
1994; Koch & Hansen, 1997). The number of water
molecules in the intedayers is not well confined. In
general it varies with the layer charge, the relative
humidity and the type of interlayer anion (Cavani et
aL, 1991). In pyroaurite, Mg substitutes for Fe(II)
in the GR formula. Pyroaurite compounds
commonly consist of platy crystals having more
or less perfectly hexagonal outlines. The GR is also
expected to exhibit this morphology (Bigham &
Tuovinen, 1985; McGill et al., 1976). The platiness
is indirectly confirmed by the strong preferred
orientation observed in the X-ray diffractogram of
GR smears. There is a lack of combined
compositional and thermodynamic data on GRs
and the E-pH stability field is thus poorly defined.
Estimates of the free energy of formation for
sulphate-GRs indicate that at equilibrium conditions
GRso, can be expected to be stable in strongly
reducing, non-acid conditions (Hansen et al., 1994).
Green rusts have been reported in a number of
natural environments (Koch & Hansen, 1997;
Trolard et al., 1996).
In previous papers it has been shown that nitrate
is reduced to ammonium by GRso, at pH 7-8.5
and magnetite is formed (Hansen et al., 1996; Koch
& Hansen, 1997):
n nI
Fe4Fe2 (on)12(SO4)(s) + 1NO3- + ~OH-
SO2- + 88
+ 2Fe3On(s) + 688
(1)
For the synthetic GRso, material used in these
studies the rate law was first order with respect to
both nitrate and Fe(II) in GR (Fe(II)6R):
d[NH +]
= k. [Fe(II)]cR. [NO;I
dt
(2)
where k = 4.93 x 10 -5 _ 0.39 x 10 -5
1 mo1-1 s -1. Using measured values for concentrations in natural pore-waters, we estimate that in
some natural environments this abiotic reaction
could cause nitrate reduction rates of similar
magnitude as microbially catalysed reduction
rates. Many details of the reaction remain to be
investigated, in particular the molecular reaction
mechanism, the effect of the Fe(II):Fe(III) ratio in
GR on the reaction rate and mechanism, and the
specific role of the GR interlayer environment in
promoting nitrate reduction. Another important
aspect concerns the influence of common pore
water solutes (e.g. silicic acid, organic substances,
metal cations) on the reduction rate and the
speciation of the iron oxide and nitrogen products.
Provided that ammonium is always the main
product, the reaction would generally be nitrogen
conserving.
In this paper the temperature dependence of the
rate of the title reaction is analysed, and the
activation energy of the reaction is determined.
The role of the GR metal hydroxide sheets in the
nitrate reduction is discussed, based on a comparison of the kinetics of reaction of GRso, with
sodium nitrate and barium nitrate, and comparison
of the sulphate-nitrate exchange patterns observed
for GRso, and a Mg-substituted GR analogue
(pyroaurite).
EXPERIMENTAL
Synthesis o f G R s o ` and reaction with nitrate
GRso, was synthesized, isolated and subsequently
reacted with nitrate according to procedures given
by Koch & Hansen (1997). These initial steps
comprise precipitation of 1 mmole of GRso, by
aerial oxidation of FeSO4 solutions at pH 7,
isolation and washing of the precipitate in a glove
box and redispersion of the GRso, in 200 ml of Arflushed water in the reaction vessel. The reaction
with nitrate was carried out at constant pH,
controlled by a constant-pH titrator (pH between 7
and 8.5) and using 0.1 u NaOH as the titrant. The
consumption of base generally followed the
stoichiometric requirement (Koch & Hansen,
1997). The reaction vessel was thermostatically
Nitrate reduction by green rust
controlled between 15 and 50~ _ 0.1~ In all
experiments the initial concentration of nitrate was
200 mg 1- t of NO~--N (14.28 raM) added either as
NaNOs or Ba(NO3)2. At fixed time intervals the
reaction suspensions were sampled and analysed for
NH4, Fe(II) in solution (Fe(II)soO, and Fe(II) in
solution + GRso, (Fe(II)cR+~ol) (Koch & Hansen,
1997). The solid phase(s) were examined by X-ray
diffraction (XRD), scanning electron microscopy
(SEM) and Mi~ssbauer spectroscopy on samples
withdrawn from the reaction suspension using Arflushed syringes.
89
& Hansen, 1997). These comprise determinations of
Fe(II) by the 1,10-phenanthroline method (Fadrus &
Maly, 1975) and NH] by gas diffusion into
indicator solutions (Ramsing et aI., 1980). For
determination of Mg and Fe in pyroaurite, 25 mg of
the solid was dissolved in 1 M HC1 followed by
atomic absorption spectroscopic analysis.
A Mrssbauer spectrum of the product of the
reaction between GRso, and Ba(NO3)2 was
obtained at room temperature using a conventional
constant acceleration spectrometer. The isomer
shifts of the components are given relative to the
centroid of the spectrum of natural Fe foil at room
temperature. The sample was stored in a deep
Synthesis o f pyroaurite and reaction with nitrate
freeze prior to measurement. Samples for XRD
Sulphate-intedayered pyroaurite (PYso,) can be were prepared by retaining the solid on a 0.22 gm
used as a non-reactive model compound for GRso,. Millipore filter followed by percolation with a few
As it does not contain Fe(II), it is not active ml of a 1:1 (v/v) glycerol:water solution to prevent
towards nitrate reduction, but it is expected to fast oxidation of GR (Hansen, 1989). The glycerol
exhibit similar anion exchange properties. paste was smeared onto glass plates and scanned at
Therefore, by studying similar reactions with a rate of 1~ 20.min-1 using Fe filtered Co-Ks
PYso, as with GRso,, it may be possible to separate radiation and a Philips PWI710 goniometer.
the effect of ion-exchange reactions from those of Crystallite sizes were estimated using the Scherrer
redox reactions, and thus aid in the interpretation of formula with k = 0.9 (Cullity, 1978) and peak
widths corrected for instrumental broadening.
the overall reaction mechanism.
Scanning electron micrographs were obtained
Sulphate-interlayered pyroaurite was synthesized
by air oxidation of Fe(II) in a solution of MgSO4 at using a Philips XL20 microscope and Pt-sputtered
constant pH 9 and 35~ The synthesis followed the samples. Elemental analysis was obtained using an
procedure given by Hansen & Koch (1995) except energy-dispersive X-ray (EDX) spectrometer.
Diffuse reflectance FTIR spectra were obtained
that Fe(II) was added as (NH4)zFe(SO4)z-6HaO, Mg
as MgSO4.7H20, and 1 M NaOH was used as using a Perkin Elmer 2000 spectrometer and a
titrant. All handling was carried out under Ar and Harris 'praying mantis' accessory. Sample mounts
all solutions were flushed with Ar to prevent CO2 consisted of 1 mg of air-dry sample mixed with
from entering the synthesis setup causing inclusion 50 mg of KBr. Spectra were averaged over 25
of carbonate in the pyroaurite interlayers. Exchange single scans recorded from 4000-400 cm -1 with a
of sulphate with nitrate in PYso, was attempted mirror velocity of 0.20 cm s -1 and strong Nortonusing NaNOs or Ba(NOs)z and following the same Beer apodization.
All chemicals were pro analysis quality or better
procedures as described for reaction with GRso,.
However, the pH of the exchange solutions was and double deionized water was used throughout.
kept around 10 to prevent dissolution of the
pyroaurite. The solid reaction products were
RESULTS
examined by XRD and Fourier transform infrared
(FTIR) spectroscopy. Suspension samples were
Reaction between GRso ~ and sodium nitrate
withdrawn using Ar flushed syringes, the solids
isolated on 0.22 gm Millipore filters and dried in a
The change in concentrations of Fe(II)GR and
dessicator over silica gel and in the presence of NI-I~ with time can be fitted to expressions for firstorder reactions (Fig. 1A). Stoichiometric data and
soda time.
rate of reaction for reduction of nitrate to
ammonium by GRso, in the presence of NaNO3
Analytical procedures
at temperatures from 15 to 50~ is given in
The determinations of Fe(II) and NI-I~ followed Table 1. The XRD of the final reaction product
the same procedures as previously described (Koch shows that GRso, (Fe(II):Fe(III) = 2:1) is oxidized
H. C. B. Hansen and C. Bender Koch
90
A
12.0 -
-0.8
-0.6
8.0.
"5"
IE
E
-0.4 ~
z
4,0 84
-0.2
'3.0
I
J
1000
B
I
i
0.0
j
2000
3000
Time (rain)
4000
5000
-0.8
12,0
Reaction between GRso, and barium nitrate
0.6
8.0
g
g
0.4
-l'-
z
--
4.0
.0,2
0.0
I
40
I
I
80
120
Time (rain)
'
0.0
1
160
is calculated.
From X-ray diffractograms of samples withdrawn
from the reaction mixture, no significant changes in
the layer structure of the unreacted GR can be
observed (Fig. 2A). For the initial 630 rain of
reaction a constant 003 spacing of 1.09 nm is
found. However, an increase in the width of the
basal spacings indicates a decrease in the thickness
along the c-axis from 52 to 33 nm during a period
of 630 min. The absence of distinct non-basal
diffraction peaks excludes determination of changes
in sizes along a and b axes. Previously we have
shown by M6ssbauer spectroscopy that the local
coordination of Fe in the unreacted GR phase
remains unchanged (Koch & Hansen, 1997).
200
FZG. 1. Examples of the kinetics for reaction between
GRso, and NaNO3 (A) and GRso, and Ba(NO3)2 (B) at
pH 8.00 and 25~ First-order fits of rate data for
consumption of Fe(II)oR (o) and production of
ammonium (41') are shown in full line.
to magnetite (Fe(II):Fe(IlI) = 1:2) (Fig. 2A), and
thus half of the Fe(II) in GR passes on to Fe304.
The average ratio between the consumption of
Fe(II)oR for nitrate reduction ([Fe(II)]oR.r~act =
89
and the production of NH4 is equal to
8.17 and is thus close to the ideal ratio of eight.
None of the ratios deviate significantly from eight,
indicating that no changes in reaction (eqn. 1) are
taking place within the temperature range studied.
There is a significant increase in rate of reaction
with increase in temperature. The temperature
dependence of the first-order rate constant (kobs)
is fitted to the Arrhenius equation (Fig. 3) from
which an activation energy of 83.9 • 7.6 kJ mo1-1
The reduction of nitrate to NtG by GRso, is ~40
times faster when nitrate is added as the barium salt
than when it is added as the sodium salt (cf.
Tables l and 2). The reaction between GRso" and
Ba(NO3)2 can also be described as a first-order
reaction, at least with respect to Fe(II)oR (Fig. IB).
The average ratio between consumption of
Fe(II)oR ..... t and formation of NI-I~4 is 8.17 _
0.51, which is close to the ideal ratio of eight. From
the rate constants determined at the three temperatures studied, an activation energy of 90.5 ___ 6.9
kJ mol -~ is estimated, very similar to the value
found for the GRso -NaNO3 reaction (Fig. 3). Thus,
both the stoichiometry of the reaction and the
reaction pathway appear not to depend on which
salt of nitrate is used.
The M6ssbauer spectrum of the reaction product
(Fig. 4) can be fitted with two sextets (A and B
components) having hyperfine parameters of 48.4
T, 0.27 n u n s i and 45.2 T, 0 . 6 5 m m s 1 for the
magnetic hyperfine field and isomer shift, respectively. The quadrupole shift of both sextets is
negligible. These parameters are comparable to
those determined for the products of the GRso,NaNO3 reaction at 250 K (Koch & Hansen, 1997)
and close to those of pure magnetite (Murad &
Johnston, 1984). The spectra of the magnetites have
identical B:A-area ratios (approximately 1.7) and
exhibit broadened lines of the B-component (widths
approx. 0.35 and 0.7 mm s-1 of the A- and B
components, respectively). This broadening could
be due to defects and substitutions.
The morphology of the magnetite crystals
produced in the GRso,-Ba(NO3)2 reaction (Fig. 5)
91
Nitrate reduction by green rust
TABLE 1. Stoichiometry and rate constants for the reaction between GR and N a N O 3 ([NO3]in~t = 14.28 rnM).
Experiment
Temp.
pH
[Fe(II)]6a, starta A[Fe(II)]aR, obsb A[NI~obs c
(~
FEN51
GRLJ2
Average
GRLJ3
FEN47
GRLJ4
FEN48
15
20
25 g
30
35
40
50
Ratio d
(n~)
8.39
8.50
8.30
8.06
7.90
8.00
8.00
10.87
7.53
6.63
10.72
10.42
10.86
12.60
kobsc
r2 ~
(s - l ) x t0 -5
3.97
7.27
6.30
9.96
10.04
10.28
12.60
0.281
0.400
0.387
0.650
0.700
0.575
0.692
7.06
9.09
8.14
7.66
7.17
8.94
9.10
0.488
0.610 f
1.161
2.39
1.84
8.05
19.3
0.9957
0.9906
0.9914
0.9511
0.9877
a The concentration of Fe(II) in GR at t = 0.
b The change in concentration of Fe(II)oR during reaction.
c The total production of NH4 + during reaction.
d The amount of Fe(II)~R which has been oxidized (=Fe(II)6R. . . . t = 0.5A[Fe(II)]6~ obs) ratioed to the amount of
NH~ produced.
Pseudo first-order rate constant determined by non-linear regression of [Nt~4] vs. time data, and the
corresponding r 2 (coefficient of regression).
f Determined from the initial rate of reaction.
g Average of five experiments (Hansen et al., 1996) (pH 8.15-8.39). Individual values are shown in Fig. 3.
deviates from those observed in the GRso,-NaNO3
reaction (Fig. 11 in Koch & Hansen, 1997). The
BaSO4 is found as relatively large p.m-sized
particles embedded in chain-forming magnetite
(Fig. 5A). Although Ba could be detected in all
analyses o f magnetite crystals by means of EDX, it
was in very low and variable amounts. Based on
XRD and M6ssbauer data, we do not think that
substitution o f Ba in magnetite is very extensive.
The magnetite crystals form chains of particles
with a rather broad size distribution and quite
c o m m o n l y n u c l e a t i n g and growing from the
surface of other particles (Fig. 5B). This change
in morphology indicates that at the higher reaction
rate, surface nucleation becomes more important
for the growth of magnetite.
The G R s o , u n d e r g o e s significant structural
changes during reaction with Ba(NO3)2, quite
different from the reaction with NaNO3. Most o f
the 1.09 nm layer spacing has transformed into
0.77 nm when only - 2 0 % o f the GRso, has reacted
(cf. Fig. 1B and 2B). Similar changes are observed
for the higher-order basal reflections. This behaviour indicates a shrinking of the anion- and watercontaining interlayer associated with ion exchange
of sulphate for nitrate. Notice also, that peaks due
to baryte (BaSO4) are found after a small extent o f
reaction (20 min, Fig. 2B).
TABLE 2. Stoichiometry and rate constants for the reaction between GR and Ba(NO3)2
Experiment
Temp.
(~
pH
GR16
GR18
GR23
GRLJ5
GRSO4BN
GRLJ7
25
25
25
25
15
10
7.80
7.12
8.00
8.28
9.29
8.15
a - e See legend to Table 1.
[Fe(II)]oR, ~taaa A[Fe(II)]GR. obsb A[NH~]obse
(nlM)
10.77
6.23
11.54
4.45
9.08
10.58
10.63
6.0l
11.34
4.45
8.66
8.88
0.710
0.34l
0.675
0.269
0.520
0.564
([NO3]init
=
14.28 mM).
Ratio d
kobse
(S- l ) • 10 4
7.49
8.81
8.40
8.27
8.33
7.88
2.75
4.36
3.70
2.32
1.08
0.490
r2 c
0.9159
0.9798
0.9805
0.9834
0.9567
0.9901
92
H. C. B. Hansen and C. Bender Koch
A
(006)
0.546~M
(003)
1.09
I
(009)
M
0.365
~!,~
M~
~M
I
I
I
4300 min
I
I
I
I
if)
I
e"
I
_J
630 rain
I
I
|
I
I
'
10
I
i~-"?~'
i
0 min
30
40
50
'
20
I
Degrees 20
B
(oo3)'
(oo6)'
(oo9)'
1.08
0.544
0.363
J
'
'
~B
M
' u RI, - , t ~ ._ ; .,i t ~. B B
T~AD,~
-
I
30O min
....'
~-
'
I
,
' "' '
I
I
-.~
~ ~
.-,,
L_ . _j~_ . . . . .
I ' i ' I
~"
I
120 min
I
....
~i
'
I
20 min
r-.
10
20
30
Degrees 20
40
'
0min
i
50
FIG. 2. Examples of XRD patterns of solids after different times of reaction between GRso, and (A) NaNO3
(experiment FEN45) or (B) Ba(NO3)2 (experiment GRLJ8), both at initial nitrate concentrations of 14.28 mM,
25~ and pH around 8. Indices for GR reflections refer to the hexagonal system; spacings are given in nm. Peaks
marked with 'B' are due to baryte, and those marked with 'M' are due to magnetite. The two sets of GR basal
spacings denoted ' and " in Fig. 2B refer to two different interlayer-ion forms of GR. (Intensities may vary due to
variable amounts of solid in the glycerol smears).
Nitrate reduction by green rust
-8-
93
R e a c t i o n o f pyroaurite with nitrate
The content of Mg and Fe in the synthetic
sulphate pyroaurite (PYso,) was 18.6 wt% and
16.6 wt%, respectively. Assuming an ideal stoichiom e t r y o f the p y r o a u r i t e , the f o r m u l a is
Mg4.32Fe~I.~s(OH)12(SO4)0.sa-yH20, g i v i n g a
Mg:Fe(III) ratio of 2.57. The 003 distance of
PYso, is 1.10 nm, i.e. the same as observed for
GRso,. On reaction with NaNO3 no significant
change in basal reflections is observed during the
l l 0 0 m i n , period for which the exchange was
investigated (Fig. 6A). The width of the basal
spacings increases slightly during this period, e.g.
the width (corrected for instrumental broadening) of
the 003 peak changes from 0.93 to 1.09 ~ 20. The
FTIR spectrum of the PYso, shows prominent
vibrations due to interlayer sulphate (HernandezMoreno et al., 1985) (Fig. 7A). The v3 vibration is
split, with a peak maximum at 1118 cm -1 and two
shoulders at 1155 cm - t and 1164 cm l, respectively; a weak v4 vibration is located at 621 cm -1.
The absorption at 983 cm -I is probably due to a
forbidden Vl vibration. An absorption at 1362 cm -1
can be assigned to an impurity of interlayer
-9-
-10-
-11
t,
-12 -
-13
'
3.0
I
'
I
'
I
3.2
3.4
T-1 x10-3(K"~)
3.6
FIG. 3. Change in In kobs (in s -1) with reciprocal
temperature for reactions between GRso, and NaNO3
(e) and Ba(NO3)2 (O), both at initial nitrate concentrations of 14.28 mM. Data from Tables 1, 2 and
Hansen et al. (1996). Straight lines are calculated by
linear regression.
!
!
9
!
!
9
I
1.000
"~
E=
0.995
t'-
I-
O.990
0.985
,
-12.0
,
-8.0
,
I
-4.0
,
I
0.0
VELOCITY
,
,
4.0
,
I
8.0
,
12.0
(MM/S)
FIG. 4. MSssbauer spectrum of magnetite formed by reaction of GRso, with Ba(NO3)2 (experiment GRLJ5). A
and B subspectra and sum of spectra are shown.
94
H. C. B. Hansen and C. Bender Koch
Nitrate reduction by green rust
carbonate (Hernandez-Moreno et al., 1985). No
changes are seen with respect to location and
intensity of the vibrations of the sulphate group
with time of ion exchange. Nitrate is clearly present
in the samples after treatment with NaNO3 as seen
from the v3 vibration at 1385 cm - l (HernandezMoreno et al., 1985). The intensity of the
absorption does not change with time (Fig. 7A),
indicating that adsorption of nitrate into the
interlayer is not extensive, in accordance with the
observations from XRD. Thus the absorption
probably arises from nitrate adsorbed at outer
particle surfaces and trapped in the interparticle
porosity.
A completely different reaction pattern is
observed when PYso, is treated with Ba(NO3)2.
The X-ray trace of the sample withdrawn just 2.5
min. after the addition of Ba(NO3)2 shows that the
003 basal spacing has decreased from 1.10 nm to
0.99 nm, but also that peak widths have increased
considerably (Fig. 6B). The sharp reflections are
due to baryte which shows that interlayer sulphate
from PYso, has combined with Ba2§ to form a wellcrystallized BaSO4 precipitate. With prolonged time
of reaction the 003 basal spacing is further
decreased to 0.81 nm after 68 h, and a decrease in
peak widths is also observed. The FTIR spectra
demonstrate that the coordination of sulphate
changes during reaction; the v3 peak in PYso, is
split into three distinct absorptions (1209, 1112,
1081 cm -Z) characteristic of baryte (Nyquist &
Kagel, 1971). Furthermore, a strong v3 absorption
at 1385 cm -1 indicates that nitrate is exchanged
into pyroaurite in amounts much higher than when
NaNO3 is used as the source of nitrate (Fig. 7B).
The single Vl + v4 nitrate combination band at
1767 cm i indicates that nitrate occurs as a free ion
in the pyroaurite interlayer (Lever et al., 1971).
DISCUSSION
Activation energy f o r the reduction o f nitrate
by GRso,
The nitrate reduction has a relatively high
activation energy of 85 kJ mo1-1. It is not possible
95
to relate the activation energy to a specific reaction
step as the detailed molecular reaction mechanism
is unknown at present. However, diffusional
processes, such as migration of anions in the GR
interlayer, appear not to be rate controlling, as they
commonly exhibit significantly smaller activation
energies in the order of 10-20 kJ mo1-1 (Laidler,
1987). This is also corroborated by the experimental finding that increasing of the stirring rate
has no influence on the observed reaction rate.
Nitrite reacts much faster with GRso, than does
nitrate (Hansen et al., 1994). Hence, if nitrate is
reduced through a nitrite intermediate, the reduction of nitrate to nitrite could be the ratedetermining step associated with the high energy
of activation.
The activation energy does not depend on
whether NaNO3 or Ba(NO3)2 was used. Thus the
structural changes induced by the use of Ba(NO3)2
probably has no influence on the reaction pathways.
The activation energy of microbial nitrate reduction
to ammonia is not known. For microbial denitriflcation, a decrease in temperature from 20 to 10~
lowers the reaction rate by a factor of 1.5-6
(USEPA, 1994; Lewandowski, 1982). For a similar
temperature decrease, the rate of the GR-facilitated
nitrate reduction is reduced 3.4 times. Thus,
microbial denitrification and abiotic nitrate reduction to ammonium cannot be distinguished from the
temperature dependence of the reaction rate. This
result also implies that when microbial nitrate
reduction is low due to low microbial respiration
rates (<5~
nitrate reduction by GRso, will also
be low.
The structural chemistry o f nitrate reduction
Green rust is the active substance facilitating the
reduction of nitrate. Thus nitrate has to come in
contact with GR for reduction to take place (Hansen
et al., 1996; Koch & Hansen, 1997). The rate of
nitrate reduction increases almost 40 times when
Ba(NO3)2 is used instead of NaNO3. We will
attempt to attribute this difference to an increase in
the number of reactive associations of the Fecontaining octahedral sheet and nitrate.
FTG. 5. Scanning electron micrographs of reaction products from the reaction between GRso, and Ba(NO3)2
(experiment GRLJ5). (A) Large baryte crystal in matrix of smaller magnetite crystals. (B) Details of magnetite
chains showing extensive surface nucleation.
1t. C. B. Hansen and C. Bender Koch
96
A
(003)
1.10
t-
e-
(006)
0.56
J
J
(009)
0.372
1100 min
I
I
I
I
I
I
63 min
J
0 min
I
'
10
I
20
'
I
'
30
Degrees 20
I
I
40
50
B
B
0.81
B
B
BB
4100 min
II
I
I
,
t~
I
I
"
0.87
,
60 min
09
C
C
,
I
i
II
II
0.99
/'
|
I
[
~,
I
I
,
I
I
I
,
,
I
2.5 min
(003)
(006)
I. i o
(009)
0 min
I
10
'
1
20
'
I
30
Degrees 20
'
I
I
40
50
FIG. 6. XRD patterns of PYso, after different times of reaction with (A) NaNO3 or (B) Ba('NO3)2, both at an
initial nitrate concentration of 14.28 naM. Indices for GR reflections refer to the hexagonal system; spacings are
given in nm. Peaks marked with B are due to baryte.
Nitrate reduction by green rust
97
A
t'N
'
t~O
II I
I I
3
t-,.
t--
0 min
63 min
1100 min
'
2000
I
'
I
'
I
'
800
1600
1200
Frequency (cm-~)
I
400
B
r
I I
I I
I I
~
o
~\
C/)
~
~
I,,I
E
\
I I
I I
I--
0 min
4100 min
'
2000
I
'
I
'
I
1600
1200
800
Frequency (cm-1)
'
I
4OO
FIG. 7. F T I R spectra o f P Y s o , after reaction with (A) NaNO3 and (B) Ba(NO3)2 for different times.
98
H. C. B. Hansen and C. Bender Koch
The rapid reduction of the doo3 distance and the
precipitation of baryte observed when GRso, reacts
with Ba(NO3)2 indicates that nitrate exchanges for
sulphate in the interlayer. This ready anion
exchange is confirmed by the experiments with
the pyroaurite analogue. Treatment of PYso, with
Ba(NO3)2 also causes layer shrinkage, and the FTIR
spectrum of the product shows that sulphate in the
interlayer has been replaced by nitrate (Figs. 6B,
7B). However, when NaNO3 is used, no interlayer
shrinkage is seen in either the GRso, nor PYso,,
and only small amounts of nitrate are absorbed into
or on the latter as evidenced by the IR spectra
(Fig. 7A). The faster shrinkage of the interlayer in
GRso, than in PYso, on treatment with Ba(NO3)2
may be due to the higher layer charge in the former.
There is strong evidence that baryte crystallizes
outside the GR crystals, as the baryte crystals are
big (Figs. 2B, 6B and 5B), and because a decrease
in interlayer spacing is observed on treatment with
Ba(NO3) 2. Baryte precipitation in bulk solution is
expected to extract interlayer sulphate which is
replaced by nitrate (forced anion exchange). The
initial unreacted suspension of GRso, contains
approximately 0.5 mM Fe(II) due to dissolution of
GRso, (Koch & Hansen, 1997). During the
dissolution, some sulphate is also liberated into
solution:
FenFenI(OH)12S04 ~ Fe304 + 3Fen + 4OH+ SO2- + 4H20
(3)
Therefore an initial sulphate concentration of
0.25 mM is expected. The solubility product of
baryte is 10 -9.97 (Wagman et al., 1982). With an
initial Ba2+ concentration of 7.1 mM the equilibrium
sulphate concentration is around 10-5 mM, and a
much stronger diffusion gradient exists for sulphate
from the interlayer to the outer solution in presence
of Ba2§ than without.
The observed anion exchange behaviour of
nitrate and sulphate in absence of Ba2+ is in
agreement with reported anion affinity sequences
for pyroaurite-type compounds (Miyata, 1983).
Reported 003 spacings for nitrate forms of
pyroaurite-type compounds vary between 0.73 and
0.91 nm, but values in the range 0.85-0.90 nm are
by far the most frequent (e.g. Meyn et al., 1990;
Lal & Howe, 1981; Miyata, 1983), i.e. the 003
spacings are always smaller than for the corresponding sulphate forms, in agreement with our
observations. The differences observed in layer
thickness probably correspond to different orienta-
fions of nitrate in the interlayer (Kruissink et al.,
1981). Thus the small 003 spacing of the nitrateinterlayered GR (Fig. 2B) indicates that nitrate is
oriented similar to carbonate in the interlayer, i.e.
with the C3 axis perpendicular to the hydroxide
sheet. In the pyroaurite sample the plane of the
nitrate ion is probably inclined with respect to the
hydroxide sheets causing an increased 003 spacing.
Based on these findings, we suggest a model for
the reaction in which the reactive complex is a
nitrate ion electrostatically attracted to the surface
of the metal hydroxide sheet close to an Fe(III) ion
(the site of the positive layer charge). In the
reaction with NaNO3, we anticipate that the initial
GR only can exchange nitrate for sulphate on the
exterior basal and edge sites (Fig. 8). During
reaction the GR plates become thinner, but
nothing is changed in the interior of the crystals.
However, when an efficient sink for sulphate is
introduced, as in the Ba(NO3)2 system, nitrate is
forced into the interlayer space which facilitates
formation of the reactive complex throughout the
crystal. During anion exchange, the thickness of the
crystals decreases from 35 to 21 nm as determined
from the widths of the 003 reflections in X-ray
traces at time zero and after 20 min (Fig. 2B) by
use of the Scherrer formula. This is close to the
expected shrinkage caused by a shift in 003 spacing
from 1.09 nm to 0.77 nm when sulphate is
exchanged with nitrate at constant number of
hydroxide sheets per crystallite. Thus the 40 times
higher rate in the Ba(NO3)2-GRso, reaction cannot
be ascribed to a significant increase in the external
surface area possibly caused by a much smaller
particle size. Reaction (oxidation and dehydroxylation) will locally decrease the layer charge and
cause swelling which exposes the layers near the
edges to the solution. This is depicted as a frayed
edge in Fig. 8.
Nitrate in the interlayers of the Ba(NO3)2exchanged form do not react instantaneously as it
was possible to detect the basal spacing from the
nitrate form of GR (Fig. 2B). This is supported by
the fact that nitrate forms of GR can be synthesized
and are sufficiently stable to be characterized by
XRD (Lewis, 1997; Gancedo et al., 1983). Hence,
the rate of reaction observed in Ba(NO3)2 experiments probably expresses a value close to the
maximum reduction rate of nitrate in GR. It may be
anticipated that the rate of reaction per unit area is
constant. Assuming that the GR crystals have platy
morphology (Bigham & Tuovinen, 1985; McGill et
Nitrate reduction by green rust
AA
i
zx
SO,~-
9
NO~
k
A
A
A
A
A
~AAAAAAAAAAAAAAA
'~ A A A A A A A A A A A A A A
I
AAAAA~AAAAA~AAA/k
.............................~ v v v
'
I
A
k
k
A
k
k
k
k.~
^ ^ ^ ^ ^ ~ ^ ^ ^ ^ ~ ^ ^ ^ *
I
~AAAAAAAAAAAAAAAA
. . . . . . . . . . ~
. .v. v. v
. . . . . . . . .v. v. . .
vvv
I
GR S O 4
vvvv4
;z/
-,..........................,A A A G ~ b A A A A A A A A A A A A ,
/
f
~,
". .
AA
99
NO~
0~),
,~............................
.
~
~
=
NO;
+Ba2+
~ NO~
4" NO~
FIG. 8. A model to explain the difference in reduction rates between NaNO3-GRso, and Ba(NO3)2 -GRso,
reactions based on nitrate interlayer-accessibility from crystal edges (see text). (Interlayer water has been
omitted).
al., 1976), and that the edges of the GRso, particles
constitute an insignificant portion of the total
surface area, the outer and total surface areas,
respectively, can be calculated by:
Outer surface area: 2 x A
(crystal thickness
Total surface area: L
d003
)
x A x 2
where A is the area of a plate surface, and the
factor 2 signifies that the reaction is assumed to
occur at both surfaces of the octahedral sheet. For
the GRso, in Fig. 2B the ratio between total and
outer surface areas is 35/1.09 - 32. Thus, if the
available surface area for nitrate reduction is
restricted to the outer particle surfaces when
NaNO3 is used, but is increased by the internal
surface area in the reaction with Ba(NO3)2, the
reaction rate is expected to be about 30 times
higher in the latter reaction. This rough estimate is
in reasonable agreement with the experimentally
determined increase in reaction rate of 40 times.
One important determinant of GR-nitrate reaction
rates thus appears to be related to the exchangeability of the GR interlayer ion by nitrate. Divalent
interlayer anions are in general not readily
exchanged for nitrate, and therefore the carbonate
form of GR, which may be a dominant form of GR
in anoxic, calcite-containing soils and sediments, is
expected to react in a similar way to the sulphate
form, or slower, as carbonate is held more tightly in
the interlayer (Miyata, 1983). However, GRs
containing monovalent anions (e.g. chloride,
organic anions) or pillars (e.g. silica polymers),
and/or having relatively high outer total surface
areas should be more accessible to nitrate and thus
lead to higher nitrate reduction rates.
CONCLUSIONS
The overall stoichiometry of the reaction between
nitrate and GRso, (eqn. 1) is independent of
temperature in the range 15-50~ and does not
depend on the nitrate salt (NaNO3 or Ba(NO3)2).
The energy of activation for the reaction between
NaNO3 and GRso, is 83.9 ___ 7.6 kJ mo1-1. The
active nitrate reducing sites are located at the GR
surfaces and the accessibility of nitrate to these sites
controls the rate of reaction when sulphate for
nitrate exchange is slower than the nitrate
reduction, ceteris paribus. A model is proposed
where GRso, is oxidized by nitrate from the particle
surfaces.
ACKNOWLEDGMENTS
Thanks to H. Nancke-Krogh and L.K. Jensen for
carefully carrying out most of the chemical analyses,
and to F. Kragh for help in obtaining electron
micrographs. The project was funded partly by Grant
13-4852 from the Danish Research Council.
H. C. B. Hansen and C. Bender Koch
100
REFERENCES
Bigham J.M. & Tuovinen O.H. (1985) Mineralogical,
morphological, and microbiological characteristics
of tubercles in cast iron water mains as related to
their chemical activity. Pp. 239-250 in: Planetary
Ecology (D.E. Caldwelll, J.A. Brierly & C.L.
Brierly, editors), Van Nostrand Reinhold Comp.,
New York.
Brindley G.W. & Bish D.L. (1976) GR: a pyroaurite
type structure. Nature 263, 353.
Buresh R.J. & Moraghan J.T. (1976) Chemical reduction
of nitrate by ferrous iron. J. Environ. Qual. 5,
320-325.
Cavani F., Trifiro F. & Vaccari A. (1991) Hydrotalcitetype anionic clays: Preparation, properties and
applications. Catal. Today, 11, 173-301.
Cullity B.D. (1978) Elements of X-ray diffraction. 2nd
ed. Addison-Wesley Publ. Comp.
Drits V.A., Sokolova T.N., Sokolova G.V. & Cherkashin
V.I. (1987) New members of the hydrotalcitemanasseite group. Clay Clay Miner. 35, 401-417.
Ernstsen V. (1990) Nitrate reduction in morainic till.
NPo-research Rep. B2, Ministry of the Environment,
Copenhagen, 52 pp. (in Danish).
Ernstsen V. & Morup S. (1992) Nitrate reduction in
clayey till by Fe(II) in clay minerals. Hyperf
Interact. 70, 1001-1004.
Fadrus H. & Maly J. (1975) Suppression of iron(III)
interference in the determination of iron(II) in water
by the 1,10-Phenanthroline method. Analyst, 100,
549-554.
Gancedo J.R., Martinez L. & Oton J.M. (1983)
Formation of green rust in NHaNO3 solutions.
Anal. Quim. 79, 470-472.
Hansen H.C.B. (1989) Composition, stabilization, and
light absorption of Fe(II)Fe(III) hydroxy carbonate
(Green Rust). Clay Miner 24, 663-669.
Hansen H.C.B., Borggaard O.K. & Sorensen J. (1994)
Evaluation of the free energy of formation of
Fe(II)-Fe(III) hydroxide-sulphate (green rust) and
its reduction of nitrite. Geochim. Cosmochim. Acta,
58, 2599-2608.
Hansen H.C.B. & Koch C.B. (1995) Synthesis and
characterization of pyroaurite. Appl. Clay Sci. 10,
5-19.
Hansen H.C.B., Koch C.B., Nancke-Krogh H.,
Borggaard O.K. & Sorensen J. (1996) Abiotic nitrate
reduction to ammonium: Key role of green rust.
Environ. Sci. Technol. 30, 2053-2056.
Hernandez-Moreno M.J., Ulibarri M.A., Rendon J.L. &
Serna C.J. (1985) IR characteristics of hydrotalcitelike compounds. Phys. Chem. Mineral. 12, 34-38.
Koch C.B. & Hansen H.C.B. (1997) Reduction of nitrate
to ammonium by sulfate green rust. Adv.
GeoEcology, 30, 373-393.
K611e W., Werner P., Strebel O. & B6ttcher J. (1983)
D e n i t r i f i k a t i o n in e i n e m r e d u z i e r e n d e n
Grundwasserleiter. Vom Wasser 61, 125-147.
Kruissink E.C., van Reijen L.L. & Ross J.R.H. (1981)
Coprecipitated nickel-alumina catalysts for methanation at high temperature. J. Chem. Soc. Faraday
Trans. L 77, 649-663.
Laidler K.J. (1987) Chemical Kinetics. 3rd ed. Harper &
Row, New York.
Lal M. & Howe A.T. (1981) Studies of zinc-chromium
hydroxy salts. I. Thermal decomposition of
[ZnzCr(OH)6]X.nH20, where X - = F-, CI-, Br-,
1
2-I-, 2~-CO3
, and NO3-. J. Solid State Chem. 39,
368-376.
Lever A.B.P., Mantovani E. & Ramaswamy B.S. (1971)
Infrared combination frequencies in coordination
complexes containing nitrate groups in various
coordination environments. A probe for the metalnitrate interaction. Can. J. Chem. 49, 1957-1964.
Lewandowski Z. (1982) Temperature dependency of
biological denitrification with organic materials
addition. Water Res. 16, 19-22.
Lewis D.G. (1997) Factors influencing the stability and
properties of green rests. Adv. GeoEeology, 30,
345-372.
Lind A. & Pedersen M.B. (1976) Nitrate reduction in the
subsoil. Danish J. Plant Soil Sci. 80, 82-99.
McGill I.R., McEnaney B. & Smith D.C. (1976) Crystal
structure of green rust formed by corrosion of cast
iron. Nature, 259, 200-201.
Meyn M., Beneke K. & Lagaly G. (1990) Anionexchange reactions of layered double hydroxides.
lnorg. Chem. 29, 5201-5207.
Miyata S. (1983) Anion-exchange properties of hydrotalcite-like compounds. Clay Clay Miner. 31,
305-311.
Murad E. & Johnston J.H. (1984) Pp. 507-582 in:
Mfssbauer Spectroscopy Applied to Inorganic
Chemistry, Vol. 2 (G.J. Long, editor), Plenum
Press, New York.
Nyquist R.A. & Kagel R.O. (1971) Infrared Spectra of
Inorganic Compounds. Acad. Press, New York.
Postma D. (1990) Kinetics of nitrate reduction by
detrital Fe(II)-silicates. Geochim. Cosmochim. Acta,
54, 903-908.
Postma D., Boesen C., Kristiansen H. & Larsen F.
(1991) Nitrate reduction in an unconfined sandy
aquifer: Water chemistry, reduction processes, and
geochemical modeling. Water Res. Res. 27,
2027-2045.
Ramsing A., Ruzicka J. & Hansen E.H. (1980) A new
approach to enzymatic assay based on flow injection
spetrophotometry with acid-base indicators. Anal.
Chim. Acta, 114, 165-181.
R6delsperger M. (1989) Natural denitrification processes in the aquifer. Pp. 159-161 in: Contaminant
Transport in Groundwater (H.E. Kobus & W.
Kinzelbach, editors), Balkema, Rotterdam.
Nitrate reduction by green rust
Trolard F., Abdelmoula M., Bourri~ G., Humbert B. &
G~nin, J.-M.R. (1996) Mise en 6vidence d'un
constituant de type 'rouilles vertes' dans les sols
hydromorphes. Proposition de l'existence d'un
nouveau min6ral: la 'foug~rite'. CR. Acad. Sci.
Paris Ser. lla, 323, 1015-1022.
USEPA (1994) Process chemistry and kinetics of
biological denitrification. Pp. 101-110 in: Nitrogen
Control, Technomic Publ. Co., Lancaster.
101
Verdegem L. & Baert L. (1985) Losses of nitrate
nitrogen in sandy and clayey soils. 2. A qualitative
and quantitative approach to the chemical NO~-N
reduction in reduced subsoils. Pedologie, 25, 39-54.
Wagman D.D., Evans W.H., Parker V.B., Schurnm H.,
Halow I., Bailey S.M., Chumey K.L. & Nuttall R.L.
(1982) The NBS tables of chemical thermodynamic
properties. Phys. Chem. Ref. Data VoL 11, Suppl.
No. 2, Nat. Bur. Standards, Washington, USA.