Next-nearest neighbour contributions to the XPS binding energies and XANES absorption energies of P and As in transition-metal arsenide phosphides MAs 1-yPy having the MnP-type structure

Article abstract of DOI:10.1016/j.jssc.2008.06.034

X-ray photoeleetron spectroscopic (XPS) and X-ray absorption near-edge spectroscopic (XANES) measurements have been made for several series of metal arsenide phosphides MAs_1-yP_y, (M= Co, Fe, Cr) that adopt the MnP-type structure. The P and As XPS binding energies (BEs) and XANES absorption energies of the metal arsenide phosphides do not follow the trend observed for the simple binary phosphides or arsenides, a deviation that arises from the combination of nearest and nextnearest neighbour contributions acting on the P or As photoemission or absorption site. The P 2p_3/2 BEs and K-edge absorption energies are lower in MAs_1-yP_y than in MP because the P atoms are more negatively charged and because the P photoemission or absorption site is screened to a greater extent by less positively charged nearest-neighbour M atoms and more negatively charged next-nearest neighbour P atoms. The As L_3- and K-edge absorption energies are higher in MAs_1-yP_y than in MAs primarily because the As atoms are less negatively charged. The M charge has been evaluated from analysis of the M 2p XPS spectra and the M L- and K-edge XANES spectra.

Full text of DOI:10.1016/j.jssc.2008.06.034

ARTICLE IN PRESS
Journal of Solid State Chemistry 181 (2008) 2549– 2558
Contents lists available at ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Next-nearest neighbour contributions to the XPS binding energies and XANES
absorption energies of P and As in transition-metal arsenide phosphides
MAs1ꢀyPy having the MnP-type structure
Ã
Andrew P. Grosvenor, Ronald G. Cavell, Arthur Mar
Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
a r t i c l e i n f o
a b s t r a c t
Article history:
X-ray photoelectron spectroscopic (XPS) and X-ray absorption near-edge spectroscopic (XANES)
measurements have been made for several series of metal arsenide phosphides MAs1ꢀyPy (M ¼ Co,
Fe, Cr) that adopt the MnP-type structure. The P and As XPS binding energies (BEs) and XANES
absorption energies of the metal arsenide phosphides do not follow the trend observed for the simple
binary phosphides or arsenides, a deviation that arises from the combination of nearest and next-
nearest neighbour contributions acting on the P or As photoemission or absorption site. The P 2p3/2 BEs
and K-edge absorption energies are lower in MAs1ꢀyPy than in MP because the P atoms are more
negatively charged and because the P photoemission or absorption site is screened to a greater extent by
Received 20 December 2007
Received in revised form
11 June 2008
Accepted 12 June 2008
Available online 19 June 2008
Keywords:
XPS
XANES
less positively charged nearest-neighbour
M atoms and more negatively charged next-nearest
Next-nearest neighbour shift
Charge potential model
Phosphides
neighbour P atoms. The As L3- and K-edge absorption energies are higher in MAs1ꢀyPy than in MAs
primarily because the As atoms are less negatively charged. The M charge has been evaluated from
analysis of the M 2p XPS spectra and the M L- and K-edge XANES spectra.
& 2008 Elsevier Inc. All rights reserved.
Arsenides
1. Introduction
Effectively, how well the nuclear charge is screened within the
atom of interest depends on the magnitudes of these intraatomic
(kDqi) and interatomic (DSj¼i qj/rij) terms, the latter usually
Although X-ray photoelectron spectroscopy (XPS) is frequently
applied to examine surfaces, it is well established that XPS can
also provide information about the electronic structure of bulk
solids through analysis of line shapes and binding energies (BE).
A charge potential model has been developed to describe BE shifts
in terms of ground-state effects [1–3]:
playing a less significant role. Although the immediate coordina-
tion environment (the nearest neighbours around an atom)
constitutes the most important contribution to the interatomic
term, the more extended environment (the next-nearest neigh-
bours) may also affect the BEs, albeit to a lesser extent. For
example, when H is substituted by F in polymers, the a-C atom
n
X
(bound directly to F) experiences an increase in the C 1s BE of
D
E ¼ Ei ꢀ E0i ¼ k
Dqi þ
D
qj=rij
(1)
ꢁ3 eV because it becomes more positively charged and thus the
jai
magnitude of the intraatomic term is increased [5]. The b-C atom
Within different compounds, an atom i acquires a BE (Ei) that is
shifted relative to a reference energy Ei0 because of variations in
two possible factors: (i) the atomic charge (qi in kDqi, where k is a
constant representing interactions between valence and core
electrons), and (ii) the chemical environment (DSj¼i qj/rij), where
rij is the distance to neighbouring atoms j [1,2]. The second term,
often called the Madelung or crystal potential because of its
equivalence to the same concept in ionic solids, describes how the
charges of coordinating atoms, qj, surrounding the photoemission
site influence the BE of the resulting photoelectron [1,4].
(bound to the
BE of ꢁ1 eV because the Madelung potential it feels is modified by
the new charge on the -C [5]. Accordingly, this effect is called a
next-nearest neighbour ( - or secondary substituent) shift [6]. In
a-C) also experiences a positive, but smaller, shift in
a
b
addition to ground-state effects, after a photoelectron is ejected,
the production of a core hole causes electrons to relax towards the
atom, modifying the final-state energy and thereby the photo-
electron BE [6]. These final-state effects can be added to the
ground-state effects described by Eq. (1) through additional
terms, ꢀ(
D +D
EiIA1 EiIA2), that represent intraatomic (IA1) and inter-
atomic (IA2) relaxation, the former usually being negligible [6].
The electronic structure of solids can also be analysed by X-ray
absorption spectroscopy, which can be subdivided into X-ray
Ã
Corresponding author. Fax: +1780 492 8231.
E-mail address: arthur.mar@ualberta.ca (A. Mar).
0022-4596/$ - see front matter & 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2008.06.034

ARTICLE IN PRESS
2550
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
absorption near-edge spectroscopy (XANES) and extended X-ray
absorption fine structure (EXAFS) [7,8]. XANES probes the
absorption edge wherein core electrons are excited into bound
or continuum states, thereby giving information about atomic
charge and coordination environment [8]. EXAFS results from
constructive and destructive interference of photoelectrons
that are backscattered by atoms surrounding the absorption site
(in the case of single scattering); oscillations are observed in the
absorption spectrum, often at energies well above the absorption
edge. Analysis of these oscillations gives information about the
local structure surrounding the absorbing atom [7]. EXAFS
oscillations superimposed near the absorption edge often render
quantitative analysis of XANES spectra difficult. As is the case for
XPS BEs, X-ray absorption threshold energies depend on similar
factors to those described by Eq. (1) [9]. However, because
electrons are now promoted into bound or continuum states, they
will still provide partial screening to the excited atom. This
implies that absorption energies are less sensitive to relaxation
effects than are XPS BEs. For example, in the series of compounds
MnO, LaMnO3, and CaMnO3, the calculated Mn K-edge absorption
threshold energies increase as the occupancy of the Mn 3d band
and the Madelung potential change [9]. The shifts in these Mn
K-edge absorption energies are much larger than the shifts in the
Mn 1s XPS BEs because the absorption energies are not dampened
by relaxation effects [9]. Next-nearest neighbour effects have also
been identified in XANES [10].
BEs and absorption energies do not follow this linear trend and the
shifts are lower than expected. Here, the presence of two different
metals interacting through the metal–metal bonding network
induces a charge redistribution, the more electropositive metal M0
donating electron density not only to P but also to the more
electronegative metal M [12]. This next-nearest neighbour effect
reduces the Madelung potential operating on the P photoemission
or absorption site, lowering its energy, especially when the
concentration of the more electronegative metal M is high [12].
Because the next-nearest neighbour effect has been demon-
strated to be important in mixed-metal phosphides, it is of
interest to determine if it also applies to mixed-anion systems.
Several metal arsenide phosphide series MAs1ꢀyPy with the
MnP-type structure are known, wherein As and P atoms are
randomly distributed within the anion sites [16,17]. Here we
report XPS and XANES spectra of CrAs1ꢀyPy, FeAs1ꢀyPy, and the
new series CoAs1ꢀyPy and apply the charge potential model
(Eq. (1)) to explain shifts in their P 2p XPS BEs and P K-edge and As
L3- and K-edge absorption energies. The metal 2p XPS and L- and
K-edge absorption spectra are also presented.
2. Experimental
2.1. Synthesis
Recently we have used XPS and XANES to examine binary and
pseudobinary transition-metal phosphides MP and M1ꢀxM0xP, the
majority of which possess the MnP-type structure [11,12]. These
materials are of interest because of their magnetic properties and
catalytic activity [13,14]. The structure consists of an arrangement
of metal-centred distorted octahedra (Fig. 1), which form net-
works of strong metal–metal and weak P–P bonds [15], consistent
with the charge formulation M1+P1ꢀ [11]. Compounds of this
structure type are metallic [13]. Throughout the series of binary
phosphides MP, the P 2p3/2 XPS BEs and P K-edge absorption
energies decrease linearly as the difference in electronegativity
between M and P increases as we progress from CoP to CrP
(all MnP-type), to VP (NiAs-type), and finally to TiP (TiAs-type) [12].
As implied by Eq. (1), these shifts arise because as M becomes
more electropositive, P becomes more negatively charged [12].
In contrast, in the pseudobinary phosphides M1ꢀxM0xP, the P XPS
Starting materials were metal powders (Cr, 99.8%, Alfa-Aesar;
Fe, 99.9%, Cerac; Co, 99.8%, Cerac; V, 99.5%, Alfa-Aesar), red P
powder (99%, Alfa-Aesar), and ground As pieces (99.9998%, Alfa
Inorganics). Products were identified by their powder X-ray
diffraction (XRD) patterns collected on an Inel powder diffract-
ometer equipped with a CPS 120 detector. All samples were stored
in a glove box under Ar to limit exposure to air.
Binary monopnictides (CrP, FeP, CoP; CrAs, FeAs, CoAs, VAs)
were prepared through stoichiometric reaction of M with P or As
in sealed and evacuated fused-silica tubes. The tubes were heated
to 1323 K (for MP) or 1073 K (for MAs) over 2 days and held at this
temperature for at least 5 days before being quenched in water.
For MAs, addition of small amounts of I2 (Anachemia) improved
crystallinity of the products. Ternary pnictides MAs1ꢀyPy (M ¼ Cr,
Fe, Co; y ¼ 0.10, 0.25, 0.50, 0.75, 0.90) were prepared through
stoichiometric reaction of (1ꢀy) MAs and y MP in sealed and
evacuated fused-silica tubes. The tubes were heated to 1173 K over
24 h and held at this temperature for 5 days before being
quenched in water. To improve homogeneity, the samples were
reground under Ar, resealed in fused-silica tubes, and reheated
with the same heating profile; this procedure was repeated over
three (FeAs1ꢀyPy, CoAs1ꢀyPy) or four (CrAs1ꢀyPy) cycles. Powder
XRD confirmed that all samples were single-phase and had
adopted the MnP-type structure. Cell parameters for CrAs1–yPy
and FeAs1ꢀyPy agree with literature values [16,17], and those for
the new series CoAs1ꢀyPy interpolate smoothly from the end
members (Fig. 2), in accordance with Vegard’s law [18].
Pn
2.2. XPS analysis
Measurements were performed on a Kratos AXIS 165 spectro-
meter using monochromatic Al Ka X-rays. The resolution function
M
a
for this instrument with this X-ray source was determined from
analysis of the Fermi edge of cobalt metal (which is a component
in many of the samples studied here) to be 0.4 eV, and the BEs are
significant to two decimal places [11,12]. Samples were finely
ground under Ar, pressed firmly onto In foil (Alfa-Aesar), and
mounted on a Cu sample holder. (The grinding allows samples to
be more easily pressed onto the In foil with a flatter surface, which
c
b
Fig. 1. MnP-type crystal structure viewed down the b-axis. The small dark spheres
are M atoms and the large lightly shaded spheres are Pn atoms. The thin dashed
lines indicate the M–M bonding network whereas the thin solid lines indicate the
Pn–Pn bonding network.

ARTICLE IN PRESS
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
2551
As L-edge spectra were calibrated against a sample of Ge with the
first derivative of the L3 edge set to 1217 eV. The M L-edge spectra
were calibrated against the first derivative of the L3 edge of the
corresponding pure metal with energies of 574.1 eV (Cr), 706.8 eV
(Fe), and 778.1 eV (Co). The absorption edge values reported
herein represent the maximum of the first derivative peak, to
remove the influence of surface phosphates or arsenates formed. A
precision of 70.1 eV was estimated through comparison of
multiple analyses of the compounds studied.
112
110
108
106
104
102
100
98
5.9
5.8
5.7
5.6
5.5
5.4
5.3
5.2
5.1
c
a
5.0
V
3.6
2.4. As K- and M K-edge XANES analysis
3.5
3.4
3.3
3.2
3.1
3.0
2.9
2.8
b
XANES spectra of the As and M K-edges were collected at
Pacific Northwest Consortium/X-ray Operations and Research
Collaborative Access Team (PNC/XOR-CAT), Sector 20 at the
Advanced Photon Source (APS), Argonne National Laboratory with
the bending magnet (20BM) beamline. A silicon (111) double
crystal monochromator was used to provide a monochromatic
photon flux of ꢁ1011 photons/s, with a resolution of 1.4 eV at
10 keV and a beam size of approximately 1 ꢂ4.5 mm. Finely
ground samples were sandwiched between Kapton tape, and
positioned 451 to the X-ray beam. The fluorescence spectra were
measured with a Canberra 13-element fluorescence detector and
transmission spectra were measured with N2-filled ionization
chambers (the I0 ionization chamber contained a 50:50 mixture of
He and N2). Through the absorption edge, the X-ray energy was
increased by 0.15 eV per step. For the M K-edge spectra, a standard
of the elemental metal was positioned behind the sample and
analysed concurrently in transmission mode with N2-filled
ionization chambers, and the peak maximum of the first
derivative was calibrated to the accepted values of 5989 eV (Cr),
7112 eV (Fe), or 7709 eV (Co). The As K-edge spectra were
calibrated by collecting the Au L3 spectrum and setting the first
derivative to 11919 eV. The precision of the measurements was
estimated to be 70.1 eV based on multiple analyses of the
compounds and the absorption energies reported herein represent
the maximum of the first derivative peak. All XANES spectra were
analysed using the Athena software program [22].
96
94
92
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
CoAs CoP
y in CoAs1-yPy
Fig. 2. Plot of orthorhombic cell parameters vs y for the CoAs1ꢀyPy series, which
adopts the MnP-type structure.
is advantageous for the XPS analysis and for ensuring electrical
contact.) They were transported and inserted into the introduction
chamber of the spectrometer under Ar to reduce the possibility of
extensive surface oxidation. Samples were sputter-cleaned in situ
with an Ar+ ion beam (4 kV, 10 mA) to remove any surface oxides
or contaminants (e.g. C, I) formed. This sputtering process causes
slight reduction of As, manifested by a low-BE shoulder in the As
3d spectrum, but no effects were observed in the P or M spectra.
Annealing the samples in vacuum using temperatures as high as
700 K for ꢁ30 min or lowering the Ar+ ion sputtering current
(1 mA) did not eliminate this As reduction. Survey spectra
revealed that preferential sputtering of light elements did not
occur and that the surface and bulk compositions are similar. For
example, different batches of FeAs0.10P0.90 (nominal) had
a
composition of FeAs0.1(1)P1.0(1). High-resolution spectra of the M
2p, P 2p, As 3d, and C 1s core lines were collected with a pass
energy of 20 eV, a step size of 0.05 eV, a sweep time of 180 s, and
an energy envelope of 20–40 eV. Spectra were analysed with the
aid of the CasaXPS software program [19]. During data analysis,
the samples were calibrated using the C 1s line arising from
adventitious C with a fixed value of 284.80 eV, although this was
rarely required because the samples studied are metallic. To
remove the background arising from energy loss, a Shirley-type
function was applied.
2.5. Band structure calculations
To interpret the As and M K-edge XANES spectra, tight-binding
linear muffin-tin orbital band structure calculations were per-
formed on FeAs and FeAs0.50P0.50 within the atomic spheres
approximation (TB-LMTO ASA) [23]. Contributions from the Fe 4p,
Fe 3d, As 4p, and P 3p orbitals to the density of states were
extracted. Although FeAs and FeAs0.50P0.50 adopt the MnP-type
structure (space group Pnma (no. 62)), mixed occupancy of the As
and P atoms in the latter is difficult to treat. Instead, an ordered
model of FeAs0.50P0.50 was considered, with reduced symmetry in
space group Pcm21 (no. 26). Such a model emphasizes heteroatomic
P–As bonds which would be more prevalent in the heavily mixed
members of the FeAs1ꢀyPy series, in contrast to the predominantly
homoatomic As–As bonds which would be present in the As-richer
members, for which the FeAs band structure would be more
appropriate. The calculations were performed with 432k-points in
the irreducible portion of the Brillouin zone.
2.3. P K-edge, As L-edge, and M L-edge XANES analysis
P K-, As, L-, and M L-edge XANES spectra were collected at the
Canadian Light Source (CLS), Saskatoon, Saskatchewan, using the
spherical grating monochromator undulator beamline, 11-ID.1
[20]. The flux is ꢁ109 photons/s at energies near the P K-edge
(ꢁ2150 eV) and increases to ꢁ1011 photons/s below this energy
(1900 eV). The estimated resolution is ꢁ1 eV and a beam size of
approximately 50 ꢂ100
mm was used. Powdered samples were
3. Results
mounted on carbon tape and inserted in the vacuum chamber via
a load lock. Total electron yield and X-ray fluorescence yield
spectra were measured from 30 to 10 eV below the edge to ꢁ35 eV
(P), 50 eV (M), or 75 eV (As) above the edge at 0.1 eV per step. The
P K-edge spectra were calibrated against a sample of Na4P2O7,
with the peak maximum of the P K-edge set to 2152.4 eV [21]. The
3.1. Is As less electronegative than P?
In this report, electronegativity differences are identified as a
major factor influencing the shifts in energy and variations in

ARTICLE IN PRESS
2552
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
lineshape in the spectra of MP, MAs1ꢀyPy, and MAs. However, the
relative electronegativities of P and As on different electronega-
tivity scales are inconsistent (cf., Allred–Rochow, wP ¼ 2.06,
wAs ¼ 2.20 vs Pauling, wP ¼ 2.2, wAs ¼ 2.1) [24–26]. In previous
work [11,12], we have preferred to use Allred–Rochow electro-
negativities because they are derived from atomic properties and
thus more closely linked to binding and absorption energies,
instead of Pauling electronegativities, which are derived from
thermodynamic data [25,26]. Alternatively, we can examine the
experimental XPS BE values to determine the electronegativity of
As. For example, the skutterudites CoPn3 are known for Pn ¼ P, As,
Sb, and Co XPS spectra have been measured for CoP3 and CoSb3
[27]. The Co 2p3/2 BE decreases from 778.5 eV in CoP3 to 778.1 eV
in CoSb3 because the more covalent character in the Co–Sb bonds
leads to better screening of the Co nuclear charge [27]. We have
prepared CoAs3 (in a similar manner as CoP3 and CoSb3) and
found that the Co 2p3/2 BE is 778.3 eV, between those of CoP3 and
CoSb3. This implies that the electronegativity of As is intermediate
between P and Sb. If Allred–Rochow electronegativities are chosen
for Co (
are assumed to be linearly proportional to the difference in
electronegativity in the Co–Pn bond ( Co), then a
revised value of wAs ¼ 1.94 (vs 2.20 as above) is obtained. The
interpretation of spectra in the MAs1ꢀyPy compounds below will
thus be based on this revised value, which makes As less
electronegative than P.
w ¼ 1.70), P, and Sb (w ¼ 1.82) and if the Co 2p3/2 BE shifts
FeP
FeAs0.10P0.90
2p3/2
Dw ¼ wPnꢀw
3.2. P 2p XPS and P K-edge XANES spectra
2p1/2
The spectra of FeP and FeAs0.10P0.90 (Fig. 3) reveal a distinct
separation between the 2p3/2 and 2p1/2 spin–orbit coupled final
states, which can be fitted by component peaks having a full-
width at half-maximum of ꢁ0.7–0.8 eV with an intensity ratio of
2:1 (2p3/2:2p1/2), equal to the theoretical value [6]. In the mixed-
metal phosphides M1ꢀxM0xP examined previously, the component
peaks were much broader than those in the binary phosphides
because different local distributions of M and M0 atoms are
present in the first coordination shell around the P photoemission
site [12]. No such broadening is observed in FeAs0.10P0.90 relative
to FeP, probably because the different local distributions of
P and As atoms in the second coordination shell around the
P photoemission site are further away.
132
131
130
129
128
127
Binding Energy (eV)
Table
1 lists P 2p3/2 BEs for various MP and MAs1ꢀyPy
compounds. From multiple measurements on separately synthe-
sized samples or of the same sample examined at different times,
we deduce that the precision of these BEs is better than 70.10 eV
Fig. 3. High-resolution P 2p XPS spectra for FeP and FeAs0.10P0.90, with 2p3/2 and
2p1/2 component peaks labelled. The FeP spectrum, which was measured by the
authors of this work on a similar instrument, is taken from Ref. [11].
Table 1
P 2p3/2 and M 2p3/2 XPS binding energies (eV) and P K-, As L3-, and As K-edge absorption energies (eV) in MP, MAs1ꢀyPy, and MAs
Compound
P 2p3/2 BE
P K-edge
As L3-edge
As K-edgea
Co 2p3/2 BE
Fe 2p3/2 BE
Cr 2p3/2 BE
CoP
129.49b
129.34b
129.18b
129.11b
128.82c
128.51c
2142.9c
2142.8c
2142.6c
2142.5c
2142.1c
2141.7c
778.33b
FeP
706.91b
MnP
CrP
573.87b
VP
TiP
CoAs
1322.1
1322.0
1321.7
1321.5
1322.4
1322.3
1322.2
1322.1
1322.2
1322.2
1322.1
11866.7
11866.6
11866.4
777.91
FeAs
706.66
CrAs
573.80
VAs
CoAs0.10P0.90
CoAs0.25P0.75
CoAs0.50P0.50
CoAs0.75P0.25
FeAs0.10P0.90
FeAs0.25P0.75
FeAs0.50P0.50
FeAs0.75P0.25
FeAs0.90P0.10
CrAs0.10P0.90
CrAs0.75P0.25
129.34
129.35
129.39
129.40
129.20
129.24
129.24
129.28
129.33
128.99
129.06
2142.7
2142.7
2142.8
2142.9
2142.6
2142.6
2142.7
11866.9
11866.9
11866.9
11866.7
11866.9
11866.8
11866.6
11866.6
11866.6
11867.1
11867.0
778.22
778.20
778.16
778.11
706.81
706.85
706.77
706.76
706.76
2142.4
2142.4
1322.0
1321.8
573.81
573.90
a
Because the As K-edge absorption threshold energy is overlapped by the intense pre-edge peak and EXAFS peaks, as described in Section 3.3, caution must be applied
when interpreting these values.
b
Ref. [11].
c
Ref. [12].

ARTICLE IN PRESS
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
2553
(Table S1 in Supplementary Data). Fig. 4a shows a plot of BE vs
difference in electronegativity ( M). Although the P 2p3/2
129.5
129.3
129.1
128.9
128.7
128.5
CoP
CoAs
FeAs
P
y
Dw ¼ wPꢀw
1-y
1
2
3
4
P
P
BEs decrease linearly through the binary phosphide MP series
from CoP to TiP, as discussed previously [12], they do not
follow this trend in the mixed pnictides MAs1ꢀyPy, with the BEs
being lower than expected. Within a given MAs1ꢀyPy series, the
P 2p3/2 BEs actually deviate more with increasing P concentration.
Table 1 also lists P K-edge absorption energies from XANES, a
bulk-sensitive technique. A representative P K-edge spectrum for
CoAs0.10P0.90 is shown in Fig. 5a. A similar behaviour to that
observed by analysis of the P 2p BEs is revealed in a plot of
1-y
y
FeP
CrAs
1-y
y
5
6
7/8
9
CrP
MnP
10
11
VP
P K-edge absorption energies vs Dw (Fig. 4b), confirming that these
shifts arise from a bulk rather than a surface electronic effect.
3.3. As L- and K-edge XANES spectra
TiP
Although As 3d XPS spectra were also collected, their analysis
was complicated by the presence of a low-BE shoulder (arising
from slight reduction of As atoms during Ar+ ion sputtering, as
noted in Section 2.2) and by the partial overlap with Cr 3p signals
in CrAs and CrAs1ꢀyPy. To circumvent these difficulties, As L- and
K-edge XANES spectra were investigated. A representative As
L3-edge spectrum for CrAs, resulting from excitation of As
2p electrons into either As 4s or 4d states, is shown in Fig. 5b.
Table 1 lists As L3-edge absorption energies for various MAs and
MAs1ꢀyPy compounds, and Fig. 4c shows a plot of these energies vs
0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75
ΔχP-M
CoP
2142.9
2142.7
2142.5
2142.3
2142.1
2141.9
2141.7
CoAs
FeAs
P
y
1-y
1
FeP
P
P
1-y
y
2
CrAs
1-y
y
3/4
7
MnP
8/9
CrP
difference in electronegativity (Dw ¼ wAsꢀwM). Like the P energies
10/11
in MP (Fig. 4a and b), the As L3-edge absorption energies decrease
linearly in the binary arsenides MAs through the series from CoAs
to VAs. The As L3-edge absorption energies in the mixed pnictides
MAs1ꢀyPy deviate from this linear trend and, opposite to the
behaviour of the P energies, they are higher than expected, with
deviations that increase with the increase in P concentration in a
given MAs1ꢀyPy series.
VP
The As K-edge absorption spectra are more complex. Repre-
sentative spectra for FeAs and some FeAs1ꢀyPy members collected
in transmission mode are shown in Fig. 6a. On the basis of
previous studies on other As-containing compounds, the low-
energy white-line intensity (labelled A) is assigned to be a pre-
edge peak [28]. By comparison of the spectra to the calculated
conduction states (Fig. 6b), this peak can be ascribed to a dipolar
transition in which As 1s electrons are excited into bound
As 4p states located below the continuum states (absorption
threshold). Because As 4p states are involved in bonding, the
intensity of this pre-edge peak can provide information on charge.
This peak intensifies going from FeAs to FeAs0.10P0.90, implying
that the As charge becomes less negative (more neutral) with
increasing y (greater P concentration) in FeAs1ꢀyPy. This trend is
also observed in the CoAs1ꢀyPy and CrAs1ꢀyPy series. Moreover,
this pre-edge peak becomes less intense through the sequence
CoAs, FeAs, and CrAs, consistent with the decrease in As L3-edge
energy seen earlier in Fig. 4c. In general, pre-edge intensities are
quite sensitive to charge, as has been shown in studies of the Au
L3-edge in a series of compounds where an increase in charge
from 0 to 1+ causes the pre-edge peak to intensify by as much as
ꢁ40% [29]. Likewise, the As charge does not have to change much
to produce substantial changes in the pre-edge peak intensity.
Peak A is much less intense in MAs and MAs1ꢀyPy (Fig. 6a), where
the As charge is near 1ꢀ, than in As2S3, where the As charge is
nominally 3+ [28]. (The P K-edge spectra could not be analysed
similarly because they were collected in fluorescence mode,
where self-absorption effects in thick samples may appreciably
affect peak intensities [30].)
TiP
0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75
ΔχP-M
4
1322.4
1322.3
1322.2
1322.1
1322.0
1321.9
1321.8
1321.7
1321.6
1321.5
CoAs
FeAs
P
y
1-y
3
P
P
1-y
y
CrAs
1-y
y
8/9
7
2
1
CoAs
FeAs
11
10
CrAs
VAs
0.20
0.25
0.30
0.35
ΔχAs-M
0.40
0.45
0.50
Fig. 4. Dependence of: (a) P 2p3/2 BE, (b) P K-edge absorption energy, and (c) As
L3-edge absorption energy on electronegativity difference for
binary pnictides MP or MAs (’) and ternary mixed pnictides CoAs1ꢀyPy
(1 ¼ CoAs0.75P0.25, 2 ¼ CoAs0.50P0.50, 3 ¼ CoAs0.25P0.75, 4 ¼ CoAs0.10P0.90), FeAs1ꢀyPy
(5 ¼ FeAs0.90P0.10 6 ¼ FeAs0.75P0.25 7 ¼ FeAs0.50P0.50 8 ¼ FeAs0.25P0.75
9 ¼ FeAs0.20P0.90), and CrAs1ꢀyPy ) (10 ¼ CrAs0.75P0.25, 11 ¼ CrAs0.10P0.90). The
MP P 2p3/2 BEs and P K-edge absorption energies are from Refs. [11,12].
(Dw ¼ wPn–wM)
(
)
(
)
,
,
,
,
Two higher-energy resonances (labelled B and C) are also
present in Fig. 6a. By comparison to the calculated conduction
(

ARTICLE IN PRESS
2554
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
A
FeAs
FeAs
P
0.90 0.10
FeAs
FeAs
P
0.50 0.50
P
0.10 0.90
B
C
CoAs0.10P0.90 P K-edge
11863 11868 11873 11878 11883 11888 11893 11898
Photon Energy (eV)
2136
2140
2144
2148
2152
2156
2160
Photon Energy (eV)
FeAs - As 4p
FeAs - Fe 4p
FeAs
FeAs - Fe 3d
FeAs
FeAs
P
- As 4p
0.50 0.50
FeAs
P
P
- P 3p
0.50 0.50
0.50 0.50
FeAs
P
- Fe 4p
0.50 0.50
CrAs L3-edge
0
5
10
15
20
25
30
35
1315
1320
1325
1330
1335
1340
1345
1350
Energy (eV)
Photon Energy (eV)
Fig. 6. (a) Normalized As K-edge absorption spectra, collected in transmission
mode, for FeAs and some FeAs1ꢀyPy compounds. Identification of peaks A–C is
discussed in the text. (b) Orbital projections of the calculated conduction states of
FeAs and FeAs0.50P0.50. The Fermi edge (EF) is at 0 eV.
Fig. 5. (a) P K-edge XANES spectrum for CoAs0.10P0.90 and (b) As L3-edge XANES
spectrum for CrAs. All spectra were measured in fluorescence mode.
states (Fig. 6b), they may be initially assigned as excitations from
As 1s to As 4p, P 3p, or Fe 4p states. However, these peaks
resemble those found in the As K-edge spectrum of As2S3, where
they are attributed to EXAFS phenomena [28]. This alternative
interpretation is supported by the observation that peaks B and C
shift to higher energy with greater y in FeAs1ꢀyPy (Fig. 6a),
consistent with the expectation that the energy is inversely
proportional to the scattering path length [31] and that the
average Fe–Pn bond distance shortens as the P concentration
increases. The presence of a strong pre-edge peak (A) and intense
EXAFS peaks (B and C) makes it difficult to track variations
in the threshold energy, which is buried beneath these features.
Nevertheless, there does appear to be a trend in the As K-edge
energies (Table 1). Similar to the As L3-edge absorption
energies, the As K-edge energies decrease going from CoAs to
CrAs. The absorption energies for the MAs1ꢀyPy compounds are, in
general, higher than those observed for MAs and increase with
increasing y. For example, in the FeAs1ꢀyPy series, the As K-edge
absorption energies are 11866.6 eV for FeAs to FeAs0.50P0.50, and
increase to 11866.8 eV for FeAs0.25P0.75 and to 11866.9 eV for

ARTICLE IN PRESS
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
2555
FeAs0.10P0.90. This agrees well with the trend in As L3-edge
energies (Fig. 4c, Table 1).
The Co 2p3/2 spectra also exhibit a high BE satellite (Fig. 7c),
which has been previously attributed to plasmon loss [11,12,27], in
which the kinetic energy of the photoelectron is reduced because
of oscillations of bound valence electrons, leading to an apparent
increase in BE [34]. This interpretation has been corroborated by
identifying the occurrence of the same phenomenon in most of
the first-row transition metals (Sc–Ni) [35–39]. Note that no such
plasmon loss peak is observed in the Fe or Cr 2p3/2 spectra, likely
because it is overlapped by the much more intense core line. The
plasmon loss peak weakens as Co becomes more positively
charged, as seen in the Co 2p3/2 spectra for the series Cox+Px
(x ¼ 0, 1, 3); this observation implies that the cross-section
for producing a plasmon oscillation decreases as the Co 3d
valence orbitals are depleted of electron density [27]. The
plasmon loss peak may also become less intense as the number
of spin-unpaired electrons decreases, as observed in the 1s XPS
3.4. M 2p XPS spectra
Fig. 7 shows M 2p XPS spectra for a representative member of
each MAs1ꢀyPy (M ¼ Cr, Fe, Co) series and for the parent binary
pnictides MP and MAs. Table 1 lists the M 2p3/2 BEs. All spectra
exhibit an asymmetric lineshape, which is characteristic of an
itinerant electronic structure, as first described by Doniach and
ˇ
´
Sunjic [32]. This type of lineshape arises when valence electrons,
interacting with the core hole (produced by photoionization), are
excited and scattered from filled states below the Fermi edge to
empty conduction states above. Because there is a continuum of
states above the Fermi edge in metallic compounds such as those
investigated here, an asymmetric tail comprising many closely
spaced states is observed instead of a few distinct satellite peaks
[33]. As in the binary and mixed-metal phosphides MP and
M1ꢀxM0xP investigated previously [11,12], this lineshape originates
from the electronic delocalization associated with the metal–
metal bonding network found in the MnP-type structure also
adopted by MAs and MAs1ꢀyPy.
spectra of Sc–Co measured with a Cu Ka X-ray source [39]. Fig. 8
shows a plot of the Co 2p3/2 plasmon loss intensity (normalized to
the core-line intensity) vs y for the CoAs1ꢀyPy series. As the P
concentration (y) increases, the intensity of the plasmon loss peak
diminishes, implying that Co becomes more positively charged as
we progress from CoAs to CoP.
Through the series from MP to MAs, the M 2p3/2 BE decreases
slightly for M ¼ Fe and Co, but hardly changes for M ¼ Cr (Fig. 7
and Table 1), when the precision of ꢁ70.10 eV is taken into
consideration. The smaller BE in the metal arsenides MAs is
consistent with the less positive M charge resulting from the
presence of less electronegative As atoms and more covalent
M–As bonds. The absence of a shift on going from CrP to CrAs
probably reflects the diminishing magnitude of the relative
electronegativity differences as M becomes less electronegative
3.5. M L- and K-edge XANES spectra
Further insight on the charges of M atoms in MAs1ꢀyPy can be
provided by the M L- and K-edge XANES spectra. In the L-edge
spectra, M 2p electrons are promoted into unoccupied M 3d and
possibly M 4s states, so the intensity of the L3 and L2 peaks reveals
information about the M charge. Fig. 9 shows normalized Fe
L-edge spectra for representative members of the FeAs1ꢀyPy series.
The two resolved L3 peaks can be assigned to the 3d t2g and eg
(cf.,
DwCr–P
/
DwCr–As ¼ 1.3 vs DwCo–P/DwCo–As ¼ 1.5) [26].
Fe 2p3/2
Co 2p3/2
Cr 2p3/2
FeP
CrP
CoP
CrAs0.75P0.25
FeAs0.50P0.50
CoAs0.50P0.50
CrAs
FeAs
CoAs
713
711
709
707
705
784
782
780
778
776
580
578
576
574
572
Binding Energy (eV)
Binding Energy (eV)
Binding Energy (eV)
Fig. 7. Representative metal 2p3/2 XPS spectra for (a) CrAs1ꢀyPy, (b) FeAs1ꢀyPy, and (c) CoAs1ꢀyPy, compared to the parent binary phosphides and arsenides. The spectra for
CrP, FeP, and CoP are from Ref. [11]. The Co 2p3/2 spectra were fitted with an asymmetric peak (thin solid line), revealing the superimposed high-BE satellite peak. The 2p3/2
BEs, which are listed in Table 1, are marked by vertical dashed lines.

ARTICLE IN PRESS
2556
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
0.22
0.21
0.20
0.19
0.18
0.17
0.16
0.15
0.14
0.13
FeAs
FeAs0.50P0.50
FeAs0.10P0.90
eg
t2g
L3
L2
0.0
0.2
0.4
0.6
0.8
1.0
y in CoAs1-yPy
CoAs
CoP
700
705
710
715
720
725
730
735
740
Fig. 8. Plot showing the variation in the normalized plasmon intensity vs y in the
Co 2p spectra for CoAs1ꢀyPy. The value for CoP is from Ref. [27].
Photon Energy (eV)
Fig. 9. Normalized total electron yield Fe L2,3-edge XANES spectra for FeAs and
some FeAs1ꢀyPy compounds. The small shoulders observed just above the L3 and L2
edges in the FeAs0.10P0.90 spectrum probably arise from some surface oxide formed.
states of the Fe atoms, which are surrounded by pnicogen atoms
in a distorted octahedral coordination. The L2 peak is not resolved
into distinct t2g and eg states because of spectral broadening
resulting from a decreased lifetime of the final state. Both L3 and
L2 peaks are more intense in FeAs0.10P0.90 and FeAs0.50P0.50 than in
FeAs; in particular, the eg portion of the L3 peak is considerably
enhanced in FeAs0.10P0.90. These intensity changes imply that the
3d states are depopulated and thus the Fe charge becomes more
positive as y increases in FeAs1ꢀyPy. Similar results were obtained
for the CoAs1ꢀyPy series, whereas difficulties in normalizing the
spectra made such an analysis impossible for the CrAs1ꢀyPy series.
Cr K-edge spectra for the CrAs1ꢀyPy series (including the end-
members CrP and CrAs) are shown in Fig. 10. The lineshape for the
spectrum of CrP is very similar to that observed in the Mn K-edge
spectrum of MnP [12,40]. The spectra contain three main
excitations, labelled A, B, and C. Peak A is a pre-edge peak in
which Cr 1s electrons are primarily excited into unoccupied Cr 3d
states, or possibly also into Cr 4p and Pn np states, from inspection
of the calculated conduction states for CrAs, which resemble those
for FeAs (Fig. 6b). These states overlap near the Fermi edge
because the distorted octahedral coordination around Cr atoms
entails some participation of Cr 4p orbitals in bonding [41]. Peak A
(Cr 1s-Cr 3d) is a quadrupolar transition and therefore it is much
weaker than peaks B (Cr 1s-Pn np) and C (Cr 1s-Cr 4p), which
are dipolar transitions. Similar transitions have been identified in
the Mn K-edge spectrum of MnP [40]. As in the argument
pertaining to the M L-edge spectra above, the intensity of the pre-
edge peak here reveals information about the M charge (when the
coordination environment of M is fixed). The pre-edge peak
intensifies on proceeding from CrAs, through CrAs1ꢀyPy, to CrP
(inset of Fig. 10), indicating that the 3d states are gradually
depopulated and the Cr charge becomes more positive as y
increases in CrAs1ꢀyPy. Also apparent from Fig. 10, the Cr K-edges
gradually shift to higher energy as we progress from CrAs to CrP,
consistent with an increase in Cr charge. Although the energy shift
of peak C (which affects the apparent intensity of peak B) may be
traced to the increase in energy of the M 4p states on going from
MAs through MAs1ꢀyPy (Fig. 6b), it can also be explained by a shift
CrP
CrAs
C
P
0.10 0.90
CrAs
P
0.75 0.25
CrAs
B
A
5988
5990
5992
5994
E (eV)
5985
5993
6001
6009
Photon Energy (eV)
6017
6025
6033
Fig. 10. Normalized Cr K-edge XANES spectra, collected in transmission mode, for
CrAs1ꢀyPy. Peak A primarily represents a pre-edge Cr 1s-Cr 3d transition. Based on
comparison to the calculated conduction states of FeAs and FeAs0.50P0.50 (Fig. 6b),
peak B may represent a Cr 1s-Pn np transition whereas peak C may represent a Cr
1s-Cr 4p transition or an EXAFS feature (see text). The inset highlights the
differences in pre-edge peak intensity among the four compounds.
of the EXAFS, as proposed in the discussion of the As K-edge
spectra (Section 3.3). Analysis of the FeAs1ꢀyPy and CoAs1ꢀyPy
K-edge spectra gave similar conclusions to those above.

ARTICLE IN PRESS
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
2557
4. Discussion
4.1. Shifts in P 2p3/2 BE and P K-edge absorption energies
The trend in P 2p3/2 BE within a given MAs1ꢀyPy series is
unusual: the BE gradually decreases as the MAs1ꢀyPy series
becomes more phosphorus rich (higher y) but the BE abruptly
increases when the end-member binary phosphide MP itself
is reached (Table 1 and Fig. 4a). Relaxation effects may be
considered first. Previous investigations in gas-phase molecules
have shown that, as long as the coordination geometry is retained,
substitution of more electropositive atoms in the coordination
environment enhances relaxation towards the nucleus of the
central atom under study, lowering its BE [42]. In the MAs1ꢀyPy
series, a central P atom is surrounded by M atoms in the first
coordination sphere and pnicogen (P or As) atoms in the second
coordination sphere. Substitution of more electronegative P atoms
in the coordination environment lowers the BE of the central
P atom, opposite to expectations and an observation that rules out
significant relaxation effects. Moreover, the trend in P K-edge
absorption energies (Fig. 4b) is consistent with the trend shown
by phosphorus BEs. Absorption energies are less influenced
by relaxation effects than are BEs, because of the greater screening
that is maintained in the absorption process. An estimate of
the ability of electrons to screen the nuclear charge can be
made from the values of Zeff, which can be calculated with Slater’s
rules [43]. A neutral P atom yields a Zeff value of 4.45, a
photoionized P atom with a 1s core hole (i.e., P1+) has a value of
5.45, and for the photo-absorption process, which can be
modelled by promotion of a 1s electron into an empty 3p orbital
(P*), a value of 5.10 is determined [43]. The nuclear charge is thus
better screened in P* than in P1+, implying that relaxation of
electrons towards the nucleus will be less important in the
absorption process compared to the photoemission process. The
trends in P BEs and absorption energies (Fig. 4a and b) are thus
attributed to ground-state effects as described by the charge
potential model (Eq. 1). Because these energy shifts are close to
the limit of precision of the measurements, we offer only a
qualitative analysis below.
The charges on M and As that are formed within a given
MAs1ꢀyPy series can be deduced. As y increases, the As atoms
become less negatively charged (from As K-edge spectra
(Section 3.3)), not only because they give up electron density to
the more electronegative P atoms via the weak Pn–Pn bonding
network, but also because they are less capable than the P atoms
in competing for electron density donated by the M atoms. As
y increases, the M atoms also become more positively charged
(as derived from Co plasmon loss peak intensities (Section 3.4)
and Fe L2,3-edge and Cr K-edge spectra (Section 3.5)), because the
M–Pn bond becomes more ionic, on average. Applying the
constraints that (i) the average Pn charge balances the M charge
in MAs1ꢀyPy (i.e., a concentration factor must be included) and
(ii) the P charge in MP is more negative than the As charge in MAs,
in agreement with relative electronegativities, we can deduce that
the P atoms become less negatively charged as y increases in
MAs1ꢀyPy (Fig. 11). A physical interpretation of this trend is that as
P atoms gradually substitute for As atoms in a MAs1ꢀyPy series,
the required total negative charge can be distributed over a
greater number of the more electronegative P atoms, such that
each P atom, on average, does not need to carry as high a negative
charge.
M
+
0
-
Slope = +X
As
Avg. Pn
Slope = -X
P
0.0
MAs
0.2
0.4
0.6
0.8
1.0
MP
y in MAs1-yPy
Fig. 11. Schematic representation of the dependence of M (’) and Pn (P, .; As,
)
charges on y in MAs1ꢀyPy. The average Pn charge (qPn ¼ yqP+(1ꢀy)qAs), represented
by the dashed line, balances the M charge in MAs1ꢀyPy.
shell containing the M atoms is considered. For very P-rich members
(high y), however, the second coordination shell now contains many
P atoms, which provide better screening to the photoemission
or absorption site than the As atoms because they are more
electronegative (more negative qj values) and they are closer
(smaller rij distances). Thus, next-nearest neighbour (interatomic)
effects can become quite important in overcoming and reversing
trends controlled by the atomic charge alone. In this present series,
the P 2p BE and P K-edge absorption energy values actually become
lower as y increases in MAs1ꢀyPy (0oyo1), with an abrupt increase
in the P 2p BE and K-edge absorption energy being observed for the
pure MP end member. This jump can be related to the reduced
screening of the P photoemission or absorption site provided by the
less negatively charged P atoms and the more positively charged
M atoms present in the pure end members compared to the values
found in the mixed MAs1ꢀyPy series.
4.2. Shifts in As L3- and K-edge absorption energies
In the binary arsenides MAs, the As L3- and K-edge absorption
energies decrease as the difference in electronegativity
(D
w
¼
wAs
ꢀ
w
M) increases, consistent with the development of
more negatively charged As atoms as the M–As bond becomes
more ionic and consistent with a more negative k qi term in
Eq. (1). This trend is confirmed by the attenuation of the pre-edge
peak in the As K-edge spectra through the series from CoAs to
CrAs (Section 3.3).
In the mixed pnictides MAs1ꢀyPy, the As L3- and K-edge
absorption energies increase with y, opposite to the trend shown
by the P energies. This occurs not only because the As atoms
D
become less negatively charged (less negative kDqi term), but also
According to Eq. (1), the P 2p3/2 BE and P K-edge absorption
energies should both increase with increasing proportion of
P atoms as we proceed from MAs1ꢀyPy (yo1) to MP, because
because the M atoms in the first coordination shell become more
positively charged (more positive DSj¼i qj/rij term). The next-
nearest neighbour effects associated with the Pn atoms in the
second coordination shell will not be as important because they
are further away from the larger As atom.
qi becomes less negative (in kDqi) and the Madelung potential
(
DSj¼i qj/rij) becomes more positive, if only the first coordination

ARTICLE IN PRESS
2558
A.P. Grosvenor et al. / Journal of Solid State Chemistry 181 (2008) 2549–2558
5. Conclusions
Appendix A. Supplementary data
Binary phosphides (MP) and arsenides (MAs) show a linear
decrease in the P or As BEs and absorption energies with greater
Supplementary data associated with this article can be found
in the online version at doi:10.1016/j.jssc.2008.06.034.
difference in electronegativity (Dw ¼ wPnꢀwM) because the pnico-
gen atoms become more negatively charged as the ionic character
in the M–Pn bond increases. However, mixed arsenide phosphides
MAs1ꢀyPy with the MnP-type structure do not show this
behaviour. As y increases, the P 2p3/2 XPS BEs and P K-edge
absorption energies decrease and are always lower than those for
the binary phosphides MP, whereas the As L3- and K-edge
absorption energies increase and are always higher than those
for the binary arsenides MAs. These trends can be rationalized
within the charge potential model, taking into consideration the
contribution of next-nearest neighbour effects arising from the
presence of a Pn–Pn bonding network and the competition
between the two different anions for electron density from the
more electropositive M atoms.
References
[1] P.A.W. van der Heide, Surf. Sci. 490 (2001) L619–L626.
[2] P.A.W. van der Heide, J. Electron Spectrosc. Relat. Phenom. 151 (2006) 79–91.
[3] A.R. Gerson, T. Bredow, Surf. Interface Anal. 29 (2000) 145–150.
[4] K. Siegbahn, C. Nordling, A. Fahlman, R. Nordberg, K. Hamrin, J. Hedman,
G. Johansson, T. Bergmark, S.-E. Karlsson, I. Lindgren, B. Lindberg, ESCA:
Atomic, Molecular and Solid State Structure Studied by Means of Electron
Spectroscopy, Almqvist and Wiksells Boktryckeri AB, Uppsala, 1967.
[5] D.T. Clark, D. Kilcast, J. Chem. Soc. B: Phys. Org. Chem. (1971) 2243–2247.
[6] D. Briggs, Surface Analysis of Polymers by XPS and Static SIMS, Cambridge
University Press, Cambridge, 1998.
[7] J.J. Rehr, R.C. Albers, Rev. Mod. Phys. 72 (2000) 621–654.
[8] J. Wong, F.W. Lytle, R.P. Messmer, D.H. Maylotte, Phys. Rev. B 30 (1984)
5596–5610.
With increasing y, the As atoms become less negatively
charged in MAs1ꢀyPy than in MAs, consistent with an observed
increase in the As L3- and K-edge absorption energy. The P atoms
are more negatively charged in MAs1ꢀyPy than in MP, consistent
with observed decreases in the P 2p3/2 BE and P K-edge absorption
energy. Lastly, the M atoms become more positively charged with
increasing y in MAs1ꢀyPy, consistent with the Co 2p XPS spectra
and the M L- and K-edge absorption spectra. Importantly, the
metal atoms in the first coordination shell and the pnicogen
atoms in the second coordination shell modify the Madelung
potential operating on the pnicogen centres and thus affect their
BE and absorption energies (particularly for P) indicating the
importance of near and next-nearest neighbour effects on these
energies. The variation of the M, As, and P charges in the MAs1ꢀyPy
series reflects a considerable tunability in the electronic structure
of MnP-type compounds, which could prove useful in catalytic
applications.
[9] A.H. de Vries, L. Hozoi, R. Broer, Int. J. Quantum Chem. 91 (2003) 57–61.
[10] A. Mottana, A. Marcelli, G. Biuli, Z. Wu, F. Seifert, E. Paris, Phys. Chem. Miner.
23 (1996) 227–228.
[11] A.P. Grosvenor, S.D. Wik, R.G. Cavell, A. Mar, Inorg. Chem. 44 (2005)
8988–8998.
[12] A.P. Grosvenor, R.G. Cavell, A. Mar, J. Solid State Chem. 180 (2007) 2702–2712.
[13] H. Fjellva˚ g, A. Kjekshus, A. Zieba, S. Foner, J. Phys. Chem. Solids 45 (1984)
709–718.
[14] S.T. Oyama, J. Catal. 216 (2003) 343–352.
[15] S. Rundqvist, P.C. Nawapong, Acta Chem. Scand. 19 (1965) 1006–1008.
[16] K. Selte, A. Kjekshus, T.A. Oftedal, A.F. Andresen, Acta Chem. Scand. A 28
(1974) 957–962.
[17] K. Kanaya, S. Abe, H. Yoshida, K. Kamigaki, T. Kaneko, J. Alloys Compd. 383
(2004) 189–194.
[18] A.R. Denton, N.W. Ashcroft, Phys. Rev. A 43 (1991) 3161–3164.
[19] N. Fairley, CasaXPS, version 2.3.9, Casa Software Ltd., Teighnmouth, Devon,
UK, 2003 /www.casaxps.comS.
[20] T. Regier, J. Paulsen, F. Wright, I. Coulthard, K. Tan, T.K. Sham, R.I.R. Blyth, AIP
Conf. Proc. 879 (2007) 473–476.
[21] C. Engemann, G. Kohring, A. Pantelouris, J. Hormes, S. Grimme, S.D.
Peyerimhoff, J. Clade, F. Frick, M. Jansen, Chem. Phys. 221 (1997) 189–198.
[22] B. Ravel, M. Newville, J. Synchrotron Radiat. 12 (2005) 537–541.
[23] O.K. Andersen, Phys. Rev. B 12 (1975) 3060–3083.
[24] J.E. Huheey, E.A. Keiter, R.L. Keiter, Inorganic Chemistry: Principles of
Structure and Reactivity, 4th ed, Harper Collins, New York, 1993.
[25] L. Pauling, The Nature of the Chemical Bond, 3rd ed, Cornell University Press,
Ithaca, NY, 1960.
Acknowledgments
[26] A.L. Allred, E.G. Rochow, J. Inorg. Nucl. Chem. 5 (1958) 264–268.
[27] A.P. Grosvenor, R.G. Cavell, A. Mar, Phys. Rev. B 74 (2006), 125102/1–125102/10.
[28] G. Pfeiffer, J.J. Rehr, D.E. Sayers, Phys. Rev. B 51 (1995) 804–810.
[29] A. Pantelouris, G. Kuper, J. Hormes, C. Feldmann, M. Jansen, J. Am. Chem. Soc.
117 (1995) 11749–11753.
[30] I. Arcon, J. Kolar, A. Kodre, D. Hanzel, M. Strlic, X-ray Spectrom. 36 (2007)
199–205.
[31] M.E. Fleet, Can. Miner. 43 (2005) 1811–1838.
The Natural Sciences and Engineering Research Council
(NSERC) of Canada supported this work through Discovery Grants
to R.G.C. and A.M. Access to the Kratos XPS was kindly provided by
the Alberta Centre for Surface Engineering and Science (ACSES) at
the University of Alberta. ACSES was established with capital
funding from the Canada Foundation for Innovation (CFI) and
Alberta Innovation and Science. CFI also provides interim operat-
ing support. Dr. Robert Blyth and Dr. J. Thompson of the Canadian
Light Source (CLS) are thanked for help in carrying out the M L-, As
L-, and P K-edge XANES measurements at 11-ID.1 at the CLS. The
CLS is supported by NSERC, NRC, CIHR, and the University of
Saskatchewan. Dr. Robert Gordon of PNC-XOR is thanked for help
in carrying out the M K-, and As K-edge XANES experiments at
PNC/XOR-CAT, sector 20 at the APS. PNC/XOR facilities at the
Advanced Photon Source (APS), and research at these facilities, are
supported by the US Department of Energy-Basic Energy Sciences,
a major research support grant (MRS) from NSERC, the University
of Washington, Simon Fraser University, and the APS. Use of the
APS is also supported by the US Department of Energy, Office of
Science, Office of Basic Energy Sciences, under Contract DE-AC02-
06CH11357. A.P.G. thanks NSERC, Alberta Ingenuity and the
University of Alberta for support.
ˇ
´
[32] S. Doniach, M. Sunjic, J. Phys. C: Solid State Phys. 3 (1970) 285–291.
[33] S. Hu¨fner, in: L. Ley, M. Cardona (Eds.), Photoemission in Solids II, Spinger,
Berlin, 1979.
[34] R.F. Egerton, M. Malac, J. Electron Spectrosc. Relat. Phenom. 143 (2005) 43–50.
[35] H.A.E. Hagelin-Weaver, G.B. Hoflund, D.M. Minahan, G.N. Salaita, Appl. Surf.
Sci. 235 (2004) 420–448.
[36] H.A.E. Hagelin-Weaver, J.F. Weaver, G.B. Hoflund, G.N. Salaita, J. Alloys Compd.
389 (2005) 34–41.
[37] A.P. Grosvenor, M.C. Biesinger, R.S.-C. Smart, N.S. McIntyre, Surf. Sci. 600
(2006) 1771–1779.
[38] N. Moslemzadeh, G. Beamson, P. Tsakiropoulos, J.F. Watts, Surf. Sci. 600
(2006) 265–277.
[39] N. Moslemzadeh, G. Beamson, P. Tsakiropoulos, J.F. Watts, S.R. Haines, P.
Weightman, J. Electron Spectrosc. Relat. Phenom. 152 (2006) 129–133.
[40] F.M.F. de Groot, S. Pizzini, A. Fontaine, K. Ha¨ma¨la¨inen, C.C. Kao, J.B. Hastins,
Phys. Rev. B 51 (1995) 1045–1050.
[41] P.G. Perkins, A.K. Marwaha, J.J.P. Stewart, Theor. Chim. Acta 59 (1981)
569–583.
[42] R.G. Cavell, R.N.S. Sodhi, J. Electron Spectrosc. Relat. Phenom. 41 (1986)
23–35.
[43] J.C. Slater, Phys. Rev. 36 (1930) 57–64.