Diffusion of water in clinopyroxene
Elizabeth Ferriss , Terry Plank, and David Walker
*
Lamont-Doherty Earth Observatory, Columbia University; [email protected]; poster ID: V51A-2751
*
Introduction
Modeling diffusivity
Results
Motivating question: Why are some volcanoes more explosive than others?
Ultimate goal: Develop a clock to measure magma ascent rates.
Project goal: Measure diffusivity of water in clinopyroxenes.
Diffusion profiles were obtained through the sides of an uncut block using FTIR. Each point
represents the average water content though the entire thickness of the sample.
Water diffusion in Kunlun diopside is roughly isotropic (slightly faster // [100]*) and slower
than in previous studies of natural diopside.
Temperature (°C)
700
log(diffusion coefficient in m2 /s)
Measured water diffusion profiles along three directions for sample heated for 3 days at 1000 °C.
E represents the direction of the infrared beam. Dotted line represents initial water.
600
1 hour
−11
1 day
−12
1 week
−13
1 month
−14
1 year
−15
10 years
−16
7
8
9
4
10
11
10 / Temperature in K
0.6
NEW
no sample
cutting
required
0.5
0.4
water
(O-H stretch)
0.2
0.1
4000
3800
Kunlun Mts., China
1.5-4 mm-side blocks
3600
3400
3200
wavenumber (cm −1 )
0.4
0.2
y ( µm)
4000
3500
3000
−1
Waven. (cm )
initia
200
100
30
0.8
20
0.6
0.4
10
0.2
0
−500
0
y ( µm)
500
0
−500
0
y ( µm)
500
Water (ppm)
0.6
1
Area / Initial
ing
t
a
e
h
e
r
o
a bef
e
r
a
k
a
e
p
l
300
Peak area (cm 2)
Absorbance (cm −1)
Water diffusion profiles are produced from Fourier transform infrared spectra (FTIR)
whole-block
with changing
diffusivity
D*=D*0(C/C0)
−10
900
40
−1
0
x (mm)
1
// [100]*, E // [010]
−0.5
0
y (mm)
0.5
// [010], E // [001]
−0.5
0
z (mm)
0.5
−11
−12
−13
−14
−15
−16
8
9
10
10 4 /Temp. (K)
11
7
8
9
10
10 4 /Temp. (K)
11
7
8
9
10
10 4 /Temp. (K)
11
7
10
40
−1
0
x (mm)
1
// [100]*, E // [010]
−0.5
0
y (mm)
0.5
// [010], E // [001]
−0.5
0
z (mm)
0.5
// [001], E // [010]
1 hour
The diffusivity was modeled with 1D, 3D, and whole-block
models using both a constant diffusivity D and a
concentration-depended apparent diffusivity D*, where
D* is proportional to water concentration (D*=D*0(C/C0).
The whole-block model with concentration-dependent D*
fits the data best in all cases.
1 year
synth. Fe−free diopside [5]
synth. Fe−poor diopside [6]
basanite diopside [7]
10 years
8
904 °C
3.5 days
D*: concentration-dependent
D
D
0.28 0.51
0.36 0.56
0.60 0.70
9
10
10 4 /Temp. (K)
100 years
11
1 2 3 4
1.5 Sample center
before heating
1
0.5 After 3 days
at 1000 °C
Wavenumber (cm −1 )
812 °C
6 days
1-dimensional models
3-dimensional model
whole-block model
1 week
1 month
0
4200 4000 3800 3600 3400 3200
r2 values of model fits
D*
1 day
Cr−diopside H2−D 2 [4]
Peak #
The individual O-H peak shifted after heat treatment above
900 °C, suggesting changes in speciation that likely
contribute some error.
20
600
without comp. dependence
cpx field evidence [1]
Jaipur diopside [2]
San Carlos olivine [3]
Cr−diopside in H2 [4]
// [001], E // [010]
30
750
Kunlun diopside (D*0)
7
D*
0.50 0.58
0.55 0.61
0.62 0.62
904 °C
6 days
D
D*
0.56 0.67
0.57 0.71
0.68 0.71
1000 °C
3 days
D
D*
0.20 0.52
0.26 0.65
0.67 0.73
30
References: [1] Wade et al. 2008; [2] Woods et al., 2000; [3] Demouchy & Mackwell,
2000; [4] Hercule & Ingrin, 1999; [5,6] Sundvall et al., 2009; [7] Xia et al., 2000
20
10
0
−1
3000
Best fit
// [001], E // [010]
10
75 hours at 1000 °C
0.3
water (ppm), whole−block
75 hours at 1000 °C
0.7
water (ppm), D*=D*0 (C/C 0 )
absorbance normalized to 1 cm
diopside
Better fit
whole-block
// [010], E // [001]
1100
20
FTIR measures absorbance
through entire sample thickness
Furnace
800 to 1000 °C
fO2 ~QFM
500
Mediocre fit
3-D model
// [100]*, E // [010]
30
0
Required if
water is lost
from center
0.8
0
Useful if
sample sliced
after heating
40
0
Methods
0
−500
Poor fit
1-D models
water (ppm), 1D model
75 hours at 1000 °C
800
−10
As the magma rises
and the pressure
drops, water
diffuses out
of the phenocryst
and may provide
a clock for
the ascent rate.
Clinopyroxene crystals
form containing uniform
water in equilibrium
with
magma
at high
H2O
pressure.
900
water (ppm), 3D model
75 hours at 1000 °C
H2O
1000
timescale for diffusion (X=500 µm=(Dt)1/2 )
1200
// [001]*
40
0
x (mm)
1
// [100]*, E // [010]
−0.5
0
y (mm)
0.5
// [010], E // [001]
−0.5
0
z (mm)
0.5
// [001], E // [010]
30
20
10
0
−1
0
x (mm)
1
−0.5
0
y (mm)
0.5
−0.5
0
z (mm)
0.5
timescale for diffusion
(X=500 µm=(Dt)1/2 )
Diffusion clock
// [010]
Absorbance (cm−1 )
(Legend, references shown in Results)
2
Huge range in diffusivities from previous work
log10 (diffusivity in m /s)
We developed a new approach, the whole-block model, that takes this averaging effect into
account and allows the determination of diffusivities in three dimensions without cutting the
sample after heat treatment.
// [100]*
Temp. (°C)
Future work
1. Perform similar experiments on augite megacrysts, high-Ti augite, phenocrysts, and
synthetic material with systematically varying chemistry
2. Re-hydrate Kunlun blocks and measure diffusivities
3. Perform complete analysis of FTIR peaks
4. Apply results to determining magma ascent rates