&EPA

U nited States
Environmental Protection
Agency

Technical Review Workgroup for Metals and Asbestos:
Bioavailability Committee. Mineralogical Report.
XAS Data and Linear Combination Fitting Results

Bradley W. Miller and Kirk G. Scheckel

13 August 2012


-------
Introduction

To enhance our understanding and capabilities to protect human health and safeguard the natural
environment, the application of molecular-level spectroscopic techniques that are highly sensitive and
non-destructive to sample integrity would provide definitive answers to complex environmental
questions. One such atomic-level technique, X-ray absorption spectroscopy (XAS), works by
bombarding an element of interest with a beam of high-energy particles from a synchrotron radiation
source to excite and expel outer-shell electrons of the particular element of interest. When the outer-
shell electrons are expelled, they emit an energy called fluorescence that can be measured by computer-
controlled detectors. The data collected by the detector yield characteristic spectra that provide
information such as oxidation state, number and type of nearest neighboring atoms, coordination
environment, and interatomic bond distances. XAS can be used to probe most phases of matter
including crystalline or amorphous solids, liquids, and gases thus making XAS one of the most versatile
research tools to fully investigate the molecular nature of a wide variety of substances. XAS is an in-
situ technique meaning one can analyze samples taken directly from the field without any alterations
that may skew true results. This type of research enhances our understanding of the fate and transport of
toxic elements in the environment.

X-ray absorption spectroscopy (XAS) has been used in many different studies to examine
contaminates such as Pb in soils (Cotter-Howells et al., 1994, 1999; Ryan et al., 2001; Scheckel and
Ryan 2004). The use of XAS can determine the speciation of element and quantify via comparison to
reference spectra with statistical analyses such as linear combination fitting (LCF) or principle
component analysis (PCA) to predict the mineralogical identification of the element (Beauchemin et al.,
2002; Scheinost et al., 2002; Scheckel and Ryan, 2004). Speciation refers to its chemical form or
species. This includes its redox state and physicochemical characteristics that are relevant to
bioavailability. This information can be used in conjunction with additional experiments to predict the
reaction of an element of interest in the environment or human body. The speciation and bioavailability
of a metal play a significant role in the risk assessment of contaminated media.

This mineral ogical report contains the result of XAS analyses with LCF predictions of the As
minerals present from nine samples including residential soil, orchard soil, an agricultural soil and
mining wastes. XAS analyses have been performed on more than 11 reference arsenic minerals and have
been included in this study. The minerals used for the LCF predictions include the As minerals most
commonly found under oxidizing and reducing conditions in soil environments and at the sites where
the materials were collected.


-------
Materials/ Methods

X-Ray Absorption Spectroscopy

X-ray absorption spectroscopy data were collected on samples from nine sites at the Materials
Research Collaborative Access Team 10-BM beamline, Advanced Photon Source (Argonne National
Laboratory). All samples were fractured with a mortar and pestle, passed through a 250 |im sieve,
pressed into a 1 cm pellet, and mounted on Kapton tape. Data was collected using a 4-element Vortex
florescence detector with several layers of aluminum foil shield to suppress florescence from other
elements such as iron in the samples. Arsenic concentrations < 20 mg As kg"1 were determined to be
below the detection limit of the Vortex detector in our experiments. Three As Ka (11874 eV) spectra
were collected in fluorescence mode at room temperature for every soil with a NaAs(V) reference
sample for calibration.

Data analysis was conducted using Athena software (Ravel and Newville 2005). Each replicate
scan was calibrated against the NaAs(V) reference (11874 eV), merged, normalized, and converted to k
space. Linear combination fitting (LCF) was used to identify the As species in each soil samples. The
LCF models were performed using the normalized, derivative, and chi(k) spectra of the soil samples and
reference standards. There were 14 reference minerals included in the LCF models (Table 1). The
reference minerals include a mix of synthetic and natural minerals received from the Smithsonian
National Museum of Natural History. The XAS spectra of the 14 reference minerals are shown in Figure
1. The LCF models predicts the As speciation in each soil as percentages of the reference minerals.

The results of the LCF analyses generate a model with the best fit (indicated by the lowest R-
factor and reduced chi square values). The pH and elemental concentration of Method 3051a extractable
elements in each sample was consulted when assessing the LCF predictions of As minerals present. In
some cases, the LCF model predicted mineral phases unlikely to be the present. If the LCF model
predicted As minerals that were not appropriate (e.g. Yukonite was predicted but 0 mg Ca kg"1 soils was
reported from 3051a extractions and sample pH was very acidic) then the mineral phase is very unlikely
to be present. Therefore, LCF models were perform again without the predicted mineral (Yukonite in
this example) and the LCF model was repeated.


-------
Results and Discussion

XAS Analyses

The As XAS spectra, both normalized and derivative data, are found in Figure 2. Analyses of
the samples collected from surface soil horizons or from mining activities had strong peaks at binding
energies around 11875 eV. This demonstrates that As(V) was the dominant As oxidation state in the
samples. The best results of the LCF model predictions (indicated by the lowest R- factor and reduced
chi square values) that most samples are dominated by arsenate sorbed to ferrihydrite or other iron
minerals (Table 2a). The LCF models also predicted the concentrations of As minerals in each sample
(Table 2b).

The LCF predicted that most samples contaminated with pesticides were dominated by As(V)
sorbed to ferrihydrite (Table 1). Our synthetic As(V) sorbed to ferrihydrite has a strong peak around
11874.5 eV (Figure 1) which corresponds to the peaks in the samples (Figure 2). Many soils in the US
are moderately to highly weathered. Therefore, these soils have higher concentrations of secondary
minerals like kaolinite (alumina silicate minerals) and Fe-oxi(hydr)oxide precipitates like ferrihydrite.
Most of the finely sieved reddish brown or yellowish samples appear to be dominated with Fe-minerals
or Al-minerals respectively. Arsenic has a high affinity for Fe minerals. Thus, the LCF prediction that
most of the samples that were contaminated with arsenical pesticides are bound to ferrihydrite was
expected and is supported by previous research.

The samples collected from or affected by mining sites have more than one As species present
and from less common As minerals. Generally, As mineral with arsenite have strong peak at 11871 eV,
and As(III)-S bonds are formed around 11867 eV (Figure 2). Sample IKJ 583 with significant
concentrations of Pb and S were predicted to have minerals with these elements, Beudantite (PbFe3+3
(As04)(S04)(0H)6). Scorodite and orpiment were among the most abundant phases predicted in soils after
As(V) sorbed to ferrihydrite (Table 2). The LCF model predicted that the sample Asarco-Ruston
contained the mineral Lollingite (FeAs2). This mineral is typically found in highly reducing
environments or as an ore component.

Brief Conclusions

All of the samples were collected from oxidized environments, are dominated by the more stable
As(V) phases and stable iron minerals. Only the samples from mine operations had reduced As minerals
present and at concentrations less than 400 mg kg"1 soil. The few samples that had high concentrations


-------
of reduced As minerals, thus have the potential to be oxidized and leach As, were taken directly from or
affected by mining activities.

Planned work

New As minerals, are being added to the pool of reference standards used during the LCF modeling
(Table 1). These include synthetic yukonite, mimetite (PbsfAsO^Cl, an analogue of the pesticide once
widely used), As(V) and As(III) adsorbed to synthetic A1 minerals will be analyzed in the summer of
2012. If the new binding energies (E0) of the synthetic minerals falls within 3 eV of the E0 of the new
reference materials, we will repeat the LCF model with the new reference minerals.

Works Cited

Beauchemin S, Hesterberg D, and Beauchemin M. 2002. Principal component analysis approach for
modeling sulfur K-XANES spectra of humic acids. Soil Sci Soc Am J 66:83-91.

Cotter-Howells JD, Champness PE, Charnock JM, and Pattrick RAD. 1994. Identification of
pyromorphite in mine-waste contaminated soils by ATEM and EXAFS. Eur J Soil Sci 45:393-402.

Cotter-Howells JD, Champness PE, and Charnock JM. 1999. Mineralogy of Pb-P grains in the roots of
Agrostis capillaris L-by ATEM and EXAFS. Mineral Mag 63:777-89.

Prietzel, J, Botzaki, A, Tyufekchieva, N, Brettholle, M, Thieme, J, and Klysubun, W. 2011. Sulfur
Speciation in Soil by S K-Edge XANES Spectroscopy: Comparison of Spectral Deconvolution and
Linear Combination Fitting. Environ Sci Technol 45:2878-2886.

Ryan, JA, Zhang, PC, Hesterberg, D, Chou, J, and Sayers, DE. 2001. Formation of chloropyromorphite
in a lead-contaminated soil amended with hydroxyapatite. Environ Sci Technol 35:3798-3803.

Scheckel, KG, and Ryan, JA. 2004. Spectroscopic speciation and quantification of lead in phosphate-
amended soils. J Environ Qual 33:1288-1295.

Scheinost AC, Kretzschmar R, and Pfister S. 2002. Combining selective sequential extractions, X-ray
absorption spectroscopy, and principal component analysis for quantitative zinc speciation in soil.
Environ Sci Technol 36:5021-8.


-------
Figures and Tables

Table 1. List of natural and synthetic As bearing minerals used for linear combination fits (LCF) using
XAS normalized and derivative [j,(E) spectra as well as chi(k) function to predict As phases in the soil
samples. Syn = Synthetic. TBD = To Be Determined at future experiments at Advance Photon Source.

Mineral

Chemical Elements

As Species

Edge (E0)

Arsenopyrite

FeAsS

As (III)

11865.84

Orpiment Cryst

As2S3

As(III)S

11866.67

Realgar

As4S4

As(III)S

11866.89

Lollingite

FeAs2

As(III)

11867.49

Mackinawite

Fe(Ni)S0.9

As(III)

11867.59

Fougerite

(Fe2+,Mg)6Fe3+2(OH) 18-4H2OAs3

As(III)0

11868.42

Arsenolite NMNH
94146

AS2O3

As(III)

11868.48

As(III) Ferrihydrite

FeOOH'0.4(H2O) As(3)

As(III)0

11868.68

Beudantite NMNH
B13898

PbFe3+3 (As04)(S04)(0H)6

As(V)0

11872.66

Scorodite

Fe3+As04 •2H20

As(V)0

11873.11

Sodium Arsenate

NaAs

As(V)0

11874.00

As(V) Ferrihydrite

FeOOH'0.4(H2O) As(5)

As(V)0

11874.61

Yukonite NMNH
6481

Ca7Fe3+12 (As04)io(OH)20 -15H20

As(V)0

11875.69

Yukonite (syn)

Ca7Fe3+12 (As04)io(OH)2(I -15H20

As(V)0

TBD

As(V) AlOH (syn)

(As04)-A10H

As(V)0

TBD

As(V) Kaolinite (syn)

(As04)A12(Si205)(0H)4

As(V)0

TBD

As(V)
Montmorillonite (syn)

(AsO3)(Na,Ca)0 33(Al,Mg)2
(Si4O10)(OH)2 nH20

As(V)0

TBD

As(III) AlOH (syn)

(As03)-A10H

As(III)0

TBD

As(III) Kaolinite

(syn)

(As03)A12(Si205)(0H)4

As(III)0

TBD

As(III)
Montmorillonite (syn)

(AsO3)(Na,Ca)0 33(Al,Mg)2
(Si4O10)(OH)2 nH20

As(III)0

TBD

Mimetite (syn)

Pb5(As04)3Cl

As(V)0

TBD

Hydroxlmimetite
(syn)

Pb5(As04)30H

As(V)0

TBD


-------
Table 2. Results of linear combination fitting (LCF) models with arsenic source, concentration (3051a extractable), and linear combination
fitting (LCF) models. A) Predictions of mineral present (%); B) concentrations of mineral present (mg kg"1).



LCF Analyses %

Soil Name

As Source

RBA As

As mg/kg*

R-factor

Reduced chi

Beudantite

As(V) Ferrihydrite

Scorodite

Lollingite

Asarco-Ruston

Smelter

Mouse

162

0.0170

0.0220

-

76%

-

24%

Barber Orchard MSI

Pesticide

Mouse

283

0.0355

0.0442

100%

Barber Orchard MS4

Pesticide

Mouse

353

0.0253

0.0302

100%

Barber Orchard MS5

Pesticide

Mouse

391

0.0414

0.0526

100%

Barber Orchard MS8

Pesticide

Mouse

375

0.0335

0.0398

100%

Hl-Hilo

Pesticide

Mouse

641

0.0119

0.0172

-

64%

36%

-

HSJ 583

Mining

Swine

249

0.0119

0.0176

-

61%

39%

-

IKJ 583

Mining

Swine

3913

0.0095

0.0145

8%

67%

25%

-

Mohr Orchard

Pesticide

Mouse

332

0.0048

0.0071

100%



LCF Analyses mg/kg

Soil Name

As Source

RBA As

As mg/kg*

R-factor

Reduced chi

Beudantite

As(V) Ferrihydrite

Scorodite

Lollingite

Asarco-Ruston

Smelter

Mouse

162

0.0170

0.0220

-

123.63

-

38.76

Barber Orchard MSI

Pesticide

Mouse

283

0.0355

0.0442

282.81

Barber Orchard MS4

Pesticide

Mouse

353

0.0253

0.0302

352.65

Barber Orchard MS5

Pesticide

Mouse

391

0.0414

0.0526

390.85

Barber Orchard MS8

Pesticide

Mouse

375

0.0335

0.0398

375.27

Hl-Hilo

Pesticide

Mouse

641

0.0119

0.0172

-

409.89

231.13

-

HSJ 583

Mining

Swine

249

0.0119

0.0176

-

152.71

96.58

-

IKJ 583

Mining

Swine

3913

0.0095

0.0145

331.43

2610.05

971.71

-

Mohr Orchard

Pesticide

Mouse

332

0.0048

0.0071

331.64


-------
Figure 1. XAS scans of standards used for linear combination fit models. A) Normalized data and B)
smoothed derivative of normalized XAS data used for linear combination fits models. Three vertical
lines are at 11867, 11871 and 11875 eV.

A

O
Csl

Ld

X

TD



"O

V_l-errihydnle.r
Vukonite.merge.Nk/

MNH B

lerge
NH.6481

11850 11860 11870 11880 11890

E (eV)

11900

11910

1 1920


-------
Figure 2. A) Normalized XAS data of soils used for linear combination fit models. B) Smoothed
derivative of normalized XAS data of standards used for linear combination fits models. Three
vertical lines are at 11867, 11871 and 11875 eV.

A

all marked groups



Ill







1 1

1 1 1 1

As

|'||
Asarco— Ru

, , ,

iion.merge

i

-







"Jf

x-—.











As

Barber Ore

lard MS1 merge

-



As

Barber Ore

hard MS4 m

erge















V

As

Barber Ore

nard MS5 m

erge

-



As

Barber Ore

hard MS8 rr

erge

-











-









¦X









. V

~ As

HSJ.583.merge

-

-



V

As

IKJ583. merge

-













I



As

Mohr Orchard,merge

-

-



. .

, , , ,

, , ,

, , ,

, , ^



, , ,

LIT

LO

o

11860 11870 1 1880 1 1890

E (eV)

1900

1 1 91 0

all marked groups

T3

in

LO

I

o
I

1860

i i I i i i i I i i i i I i i i i

iii

ii

i i

¦





As

Asarco— Ru

ston .merge

-

-





\ As

Barber Orchard MS1 rr

erge







Vxj











/ *



Barber Ore

aard MS4 rr

erge

-





^ As

Barber Orchard MS5 rr

erge

-





y As

Barber Ore

nard MS8 rr

erge

-





As

HI — H i I o. m e

rge

-

-





As

HSJ. 583. merge

-





















As

IKJbbi.merge



-





x As

Mohr Orchard.merge

-

¦











-



, ,

, ,

,

I 870

1 1900

11910

E (eV)


-------