Chemical Design and Example of Transparent Bipolar Semiconductors

Article abstract of DOI:10.1021/jacs.7b09806

Transparent bipolar semiconductors (TBSCs) are in demand for transparent electronics to serve as the basis for next generation optoelectronic devices. However, the poor carrier controllability in wide-bandgap materials makes the realization of a bipolar nature difficult. Only two materials, CuInO₂ and SnO, have been reported as TBSCs. To satisfy demand for the coexistence of transparency and bipolarity, we propose a design concept with three strategies; choice of early transition metals (eTM) such as Y³⁺ and Zr⁴⁺ for improving controllability of carrier doping, design of chemical bonds to obtain an appropriate band structure for bipolar doping, and use of a forbidden band-edge transition to retain transparency. This approach is verified through a practical examination of a candidate material, tetragonal ZrOS, which is chosen by following the criteria. ZrOS exhibits an excellent controllability of the electrical conductivity (10^-7-10^-2 S cm^-1), p- or n-type nature with ～10^-2 S cm^-1 by Y or F doping, respectively, and optically wide gap (below 10^-4 cm^-1 up to ～2.5 eV). This concept provides a new kind of TBSC based on eTM ionic compounds.

Full text of DOI:10.1021/jacs.7b09806

Subscriber access provided by READING UNIV
Article
Chemical Design and Example of Transparent Bipolar Semiconductors
Takeshi Arai, Soshi Iimura, Junghwan Kim, Yoshitake Toda, Shigenori Ueda, and Hideo Hosono
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.7b09806 • Publication Date (Web): 08 Nov 2017
Downloaded from http://pubs.acs.org on November 8, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of the American Chemical Society is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 8
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Chemical Design and Example of Transparent Bipolar Semiconduc-
tors
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Takeshi Arai1, Soshi Iimura1, *, Junghwan Kim2, Yoshitake Toda2, Shigenori Ueda3, 4, and Hideo Hoꢀ
sono1, 2, *
1Laboratory for Materials and Structures, Tokyo Institute of Technology, Yokohama 226ꢀ8503, Japan, 2Research Center for
Element Strategy, Tokyo Institute of Technology, Yokohama 226ꢀ8503, Japan, 3Synchrotron Xꢀray Station at SPringꢀ8, Naꢀ
tional Institute for Materials Science, Hyogo 679ꢀ5148, Japan, & 4Research Center for Advanced Measurement and Characꢀ
terization, National Institute for Materials Science, Tsukuba 305ꢀ0047, Japan
oxide semiconductors, bipolar semiconductors, material design, oxysulfide
ABSTRACT: Transparent bipolar semiconductors (TBSCs) are in demand for transparent electronics to serve as the basis for next
generation optoꢀelectronic devices. However, the poor carrier controllability in wideꢀbandgap materials makes the realization of a
bipolar nature difficult. Only two materials, CuInO2 and SnO, have been reported as TBSCs. To satisfy demand for the coexistence
of transparency and bipolarity, we propose a design concept with three strategies; choice of early transition metals (eTM) such as
Y3+ and Zr4+ for improving controllability of carrier doping, design of chemical bonds to obtain an appropriate band structure for
bipolar doping, and use of a forbidden bandꢀedge transition to retain transparency. This approach is verified through a practical
examination of a candidate material, tetragonal ZrOS, which is chosen by following the criteria. ZrOS exhibits an excellent controlꢀ
lability of the electrical conductivity (10−7–10−2 S cm−1), pꢀ or nꢀtype nature with ~10−2 S cm−1 by Y or F doping, respectively, and
optically wide gap (below 10−4 cm−1 up to ~2.5 eV). This concept provides a new kind of TBSC based on eTM ionic compounds.
Introduction
A transparent bipolar conductivity is a key functionality to
establish next generation optoꢀelectronic devices.(1–3) Howevꢀ
er, realization of a bipolar nature in wideꢀgap semiconductors,
such as oxides and chalcogenides, remains a challenging issue.
Wideꢀbandgap oxide semiconductors generally show nꢀtype
conduction, whereas pꢀtype materials are highly limited owing
to the high ionization potential arising from the high electroꢀ
negativity of oxygen.(4) Transparent pꢀtype semiconductors
reported to date are materials based on Cu1+ and Sn2+ such as
CuAlO2, CuI and SnO(5–8) except nanoꢀmaterials. Conversely,
obtaining nꢀtype conductivity in these materials is usually
difficult owing to Cu1+ or Sn2+ vacancy formation. The probꢀ
lematic bipolar conduction originates from the general features
of wideꢀbandgap semiconductors: easy generation of counter
doping through cation/anion vacancy formation and a fundaꢀ
mental dilemma between a wideꢀbandgap and carrier dopabilꢀ
ity. To overcome these obstacles, we propose a material deꢀ
sign concept with three strategies; [A] use of early transition
metals (eTM) to improve carrier doping controllability, [B]
chemical bond engineering appropriate for bipolar doping, and
[C] forbidden bandꢀedge transition ensuring optical transparꢀ
ency, as is summarized in Figure 1.
cations, leading to large band dispersion of the conduction
band minimum (CBM), that is, a small effective mass of the
electron.(9) However, the interaction of occupied Cu1+ 3d and
Sn2+ 5s orbitals with anion p orbitals increases the valence
band maximum (VBM), which also reduces the effective mass
of holes.(3,10,11) Until now, however, these pTMꢀbased semiꢀ
conductors have suffered from a poor doping efficiency origiꢀ
nating from the low formation energy of a metalꢀvacancy and
a facile valence change between two stable valence states, for
example, Sn2+ ↔ Sn4+ + 2e− and Cu1+ ↔ Cu2+ + e−. The cation
vacancy is generated at the lowꢀcoordination site derived from
the lowꢀoxidation state, such as a twoꢀfold coordination for
Cu1+ ion and asymmetric fourꢀfold coordination for a Sn2+ ion.
The formation of these defects has a low energy,(12–14) which
inhibits efficient electron doping by counter doping (i.e., Cu1+
or Sn2+ vacancy). However, the facile valence change is relatꢀ
ed to the competitive electronic configurations of the pTM
with (n−1)d9ns0 or (n−1)d10ns0 for groupꢀ11 elements, such as
Cu, and (n−1)d10ns2np0 or (n−1)d10ns0np0 for heavy groupꢀ13,
14, and 15 elements, such as In, Sn, and Pb. The two stable
electronic configurations also induce multiple stable phases,
for example, PbO (Pb2+) ↔ PbO2 (Pb4+), making carrier dopꢀ
ing into these compounds difficult. For example, when atꢀ
tempting to make a Pb2+ vacancy in PbO to attain pꢀtype conꢀ
duction, a mixture of PbO and PbO2 is obtained instead of Pb1–
xO.(15)
In a conventional material design, wideꢀbandgap semiconꢀ
ductors are based on postꢀtransition metal (pTM) ions such as
In3+, Cd2+, Cu1+, and Sn2+. The unoccupied, large spatial
spread of the 5s orbitals of In3+ and Cd2+ results in a large
overlap between these orbitals with those of the neighboring
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Material design concept for bipolar ionic semiconductor with optical transparency. (a) Comparative atomic features of
early and post transition metal (eTM and pTM). The valence state of eTM cations is stable and high relative to that of pTM cations,
stabilizing a single phase and inducing a high coordination number (CN). These chemical and structural properties are expected to
encourage an efficient carrier doping into eTM based compounds. (b) The schematic chemical bonding diagrams of conventional
ionic semiconductor and TBSCs. The CBM level below ~−4 eV and VBM level above ~−6 eV, relative to the vacuum level, are
empirical requirements for nꢀtype and pꢀtype doping, respectively, in the ionic semiconductors.(20) The material design concept conꢀ
sists of three requirements; [A] use of the eTM to improve the carrier doping efficiency, [B] nonꢀbonding between a cation and an
anion for nꢀtype doping and antiꢀbonding between anions for pꢀtype doping, [C] the forbidden bandꢀedge transition to attain a low
absorption coefficient in the energy range of visible light.
Design concept
Therefore, we chose early transition metal (eTM) cations,
such as Y3+, La3+, Zr4+, and Hf4+, as constituents for the funcꢀ
tional semiconductors. Contrary to the pTM ions, these ions
occupy a high coordination numbers (CN) site and form a
single stable ionic compound owing to their high and single
valence states, which prevents the formation of a cation vaꢀ
cancy and a valence change in their compounds. Furthermore,
they are environmentally benign, thus suitable for next generaꢀ
tion devices.
These eTM elements are yet to attract interest as a constituꢀ
ent of the nꢀ and pꢀtype semiconductors. Generally, electron
and hole doping into semiconductors require a material with a
large electronꢀaffinity (deep CBM) and small ionization potenꢀ
tial (shallow VBM),(16) respectively. The electrons are more
stabilized in a deeper CBM, while removing the electrons
from the VBM level (i.e., generating the holes) is easier in a
shallow VBM. However, unlike the pTM compounds, the relaꢀ
tively localized d orbital of eTM compounds forms a shallow
CBM level, and the eTM has no appropriate orbitals making
Figure 2. Crystal structures of tꢀZrOS (a) and cꢀZrOS (b).
the VBM level shallower, which contrasts with the 3d orbital
Blue, green, and grey spheres represent Zr, S, and O, respecꢀ
of Cu1+ or 5s orbital of Sn2+.
tively. Each local structure around Zr is shown below the crysꢀ
tal structure.
To solve the problematic band alignment of eTM comꢀ
pounds, the chemical bonding state is considered to identify an
electronic structure suitable for both nꢀ and pꢀdoping. To
deepen the CBM, introducing a nonꢀbonding state between
Table 1. Irreducible representations (Irrep) of the inꢀphase and
outꢀofꢀphase orbital configurations at the Γ point.
cation and anion would be effective, while an antiꢀbonding
state between anions will increase the VBM (strategy [B]).
Basically, the conduction band (CB) and the valence band
(VB) of conventional semiconductors are, respectively, comꢀ
posed of antiꢀbonding and bonding states between the unoccuꢀ
pied orbitals of the metal and the occupied orbital of the anion,
as shown in Figure 1b (left). If the nonbonding state is introꢀ
duced to the CB at a specific k point, the energy level at the k
point decreases, and the effective electron mass is reduced
owing to the large energy difference between the deep nonꢀ
bonding and the shallow antibonding states (see the right of
Figure 1b).(17,18) The antiꢀbonding between anions, which
increases the energy level of the VBM, is an analogy of the
ACS Paragon Plus Environment

Page 3 of 8
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Calculated electronic structures of tꢀZrOS and cꢀZrOS. (a,b) The band structures of tꢀZrOS (a) and cꢀZrOS (b). The vertiꢀ
cal axes are aligned respective to the energy level of Sꢀ3s as shown in Figure S3. Blue and green in (a) denote the contributions
from Zrꢀ4dx2−y2 and Sꢀ3px, y, respectively, while those in Figure 3(b) from Zrꢀ4d and Sꢀ3p, respectively. Here, the x and y axes are
chosen along a and b crystallographic axes of the tetragonal unit cell. (c,d) The kꢀdependences of transition matrix elements P2 beꢀ
tween the band edge states of tꢀZrOS (c) and cꢀZrOS (d). For the tꢀZrOS, the P2 are anisotropic because of the tetragonal symmetry.
cationꢀanion antibonding that is evident in Cu1+ and Sn2+ꢀ
based materials. Generally, as an anion has a larger ionic radiꢀ
us and are less ionic (e.g., from oxygen to tellurium in group
4), the energy levels of the p orbitals are increased, which, in
turn, results in a shallow VBM.(19) Therefore, chalcogenide
ions are more suitable for hole conduction than an oxide ion.
Moreover, a short separation between neighboring anions is
critical because the stronger interaction between these anions’
p orbitals raises the VBM derived from the anionꢀanion antiꢀ
bonding level and enhances the band dispersion.
ZrOS) in Figure 2b. The structure of cꢀZrOS is regarded as a
distorted fluorite structure, where the Zrꢀcentered polyhedra,
ZrO3S4, correspond to the Caꢀcentered cubes, CaF8, in fluorite.
The nearest neighbor ZrꢀZr bond length in the cꢀZrOS of 3.58
Å is comparable with that in tꢀZrOS (3.55 Å), whereas their
OꢀO and SꢀS lengths of 3.52 and 3.54 Å, respectively, are
significantly longer than those in the tetragonal phase.
Table 1 summarizes the irreducible representations (Irrep)
of tꢀZrOS at the Γ point. The P4/nmm space group of tꢀZrOS
belongs to the high symmetry point group D4h, and, furtherꢀ
more, the site symmetry of each constituent is high (the low
multiplicity), thus is suitable for forming a nonꢀbonding state.
Each atomic orbital has two configurations, inꢀphase and outꢀ
ofꢀphase, in the unit cell, and then the Irrep is assigned to each
configuration. In principle, the orbitals with the same Irrep can
interact with each other. Hence, according to Table 1, it is
found that the outꢀofꢀphase of Zr 4dx2ꢀy2 and both phases of Zr
4dxy form a nonꢀbonding state, that is, they cannot interact with
any p orbitals of S and O at the Γ point.
Finally, it is desirable to keep the visible transparency of the
semiconductor with a deep CBM and shallow VBM resulting
in a narrow bandgap, empirically, below ~2 eV.(20) Thus, semꢀ
iconductors that have a direct allowed type bandꢀtoꢀband tranꢀ
sition are eliminated as candidate materials. The low absorpꢀ
tion coefficient in the energy range of visible light is sufficient
for practical thin film applications. Thus, candidate materials
must have an indirect or direct forbiddenꢀtype bandꢀtoꢀband
transition (strategy [C]).(20,21)
Figure 3a and 3b show the band structures of tꢀZrOS and cꢀ
ZrOS, respectively. Here, the energy level of cꢀZrOS was
aligned to the S 3s level of tꢀZrOS located at E − EVBM ~−12
eV (see Figure S3 in the Supporting Information). These calꢀ
culated band structures reveal that the CBM and VBM of tꢀ
ZrOS are deeper by 2 eV and shallower by 0.7 eV than those
of cꢀZrOS, respectively, yielding the benefit of the emergence
of a bipolar nature. The deep CBM at the Γ point purely conꢀ
sists of the outꢀofꢀphase configuration of the Zr 4dx2ꢀy2 orbital.
This means that the Zr 4dx2ꢀy2 orbital forms a nonꢀbonding
state, which is consistent with the symmetry analysis based on
Table 1 (This chemical bonding picture is illustrated in Fig-
ure S4 in the Supporting Information). Away from the Γ
point, the Zr 4dx2−y2 orbital cancels the nonꢀbonding state and
eventually forms an antiꢀbonding state with O 2p orbitals at
Candidate material
Herein, we propose tetragonal ZrOS (tꢀZrOS) as a candidate
for TBSCs. The tꢀZrOS meets the requirements described
above; a high coordination number around Zr4+ cations preꢀ
venting the formation of a metal vacancy, a short SꢀS distance
suitable for shallow VBM, and a nonꢀbonding Zr 4dx2−y2 orꢀ
bital for a deep CBM.
Figure 2a shows the crystal structure of tꢀZrOS. Five S and
four O atoms are bonded to Zr (below the Figure 2a). The
nearest neighbor OꢀO and SꢀS bond lengths of 2.51 and 3.02
Å, respectively, are shorter than the summation of each ionic
radius at the corresponding sites. This fact suggests a strong
interaction between neighboring anion p orbitals. For compariꢀ
son, we also show the crystal structure of cubic ZrOS (cꢀ
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Electrical transport properties, hard Xꢀray photoemission spectra, and absorption coefficient of tꢀZrOS. (a) Doping deꢀ
pendence of conductivity of tꢀZrOS:F and tꢀZrOS:Y. Black filled star, open purple circle, and orange square represent the conducꢀ
tivity of nonꢀdoped tꢀZrOS, tꢀZrOS:F, and tꢀZrOS:Y, respectively. The solid lines are guides for the eye. (b) Doping dependence of
Seebeck coefficient, S, of tꢀZrOS:F and tꢀZrOS:Y. (c) Observed hard Xꢀray photoemission spectrum for tꢀZrOS:Y and calculated
atomꢀprojected density of states (PDOS). (d) Comparison of the spectra of tꢀZrOS:F and tꢀZrOS:Y. (e) Calculated absorption coefꢀ
ficient, α, of tꢀZrOS. Red dashed line is the α multiplied by 10 for clarity. (f) Absorption spectra of tꢀZrOS derived from diffuse
reflectance spectra at 300 K and the Kubelka–Munk relation, where α, s, and hν are the absorption coefficient, scattering factor, and
photon energy, respectively.(24) The vertical axis for direct allowed transition and direct forbidden transition are (hvα/s)2/3 and
(hvα/s)2, respectively.
another k point, such as the M point, which shapes a highly
dispersed band with a small effective mass (me* along the ΓꢀM
line is ~0.36m0, m0; a free electron mass). In contrast, the CB
of cꢀZrOS is flat and shallow owing to an antiꢀbonding nature
between Zr d and anion p orbitals at the whole reciprocal
space. As for the VB, S 3px/y orbitals of tꢀZrOS form a shalꢀ
lower and sharper VBM (a light hole effective mass; mh*(Γꢀ
M) ~ 0.24m0) relative to that of cꢀZrOS (a light hole effective
mass; mh*(ΓꢀM) ~ 0.37m0). This profound difference in the
VB of tꢀ and cꢀZrOS originates from the SꢀS separation in
both phases.
cm−1, and simultaneously, an inversion (plus to minus) of the
Seebeck coefficient sign was also confirmed upon changing
from tꢀZrOS:Y to tꢀZrOS:F. We performed Hall measurement
for the 3% tꢀZrOS:Y and tꢀZrOS:F by two different techꢀ
niques; DC mode at high magnetic field up to 9 T and AC
mode using a lockꢀin detection under AC field. However, no
reliable Hall voltage beyond noise level was obtained. Such
situation is almost common for polycrystalline bulk samples of
oxide semiconductors due mainly to grain boundary. It is reꢀ
quired to fabricate thin films or grow single crystals to obtain
reliable value of carrier concentration and mobility by Hall
effect measurements. Seebeck coefficient provides faithful
values even for disordered systems such as polycrystals and
amorphous semiconductors.(23) The formation of free carrier is
distinctly observed in the near infrared absorption for Yꢀ or Fꢀ
doped samples as shown in Figure S9. The low conductivity
in the undoped sample and the easy controllability of the carriꢀ
erꢀpolarity in the doped tꢀZrOS indicate the high formation
energy of an atomic vacancy, antiꢀsite, and/or interstitialꢀsite,
which is in contrast to the conventional wideꢀgap semiconducꢀ
tors based on a pTM cation. The donor and acceptor levels
estimated from the activation energy of the conductivity
measurement were 100 meV for the F 2.0 mol% doped sample
and 130 meV for the Y 2.2% doped sample (see Figure S10 in
the Supporting Information).
To examine the optical properties, we calculated the matrix
elements for direct interband transitions between band edge
states. Figure 3c and 3d show the k dependences of the transiꢀ
tion matrix element (P2) between the band edge states. They
reveal that direct transitions between CBM and VBM at the Γ
point are forbidden in tꢀZrOS because the Irrep of CBM and
VBM have the same parity (Γ4− for CBM and Γ5− for
VBM).(22) However, in the cꢀZrOS, the optical transitions at
the band edges are allowed at all k points (Figure 3d).
Experimental validation
Figure 4a and 4b show the doping dependences of the elecꢀ
trical conductivity σ and Seebeck coefficient S of yttrium and
fluorineꢀdoped tꢀZrOS (tꢀZrOS:Y and tꢀZrOS:F, respectively)
at room temperature (see Figures S5–S7 for structural analꢀ
yses and Figure S8 for transport measurements in the Supꢀ
porting Information). For the undoped tꢀZrOS, a much lower
conductivity of ~10−7 S cm−1 was achieved. Once the Y or F
Figure 4c shows the comparison between the hard Xꢀray
photoemission spectrum and the atomꢀprojected density of
state (PDOS) for tꢀZrOS. The observed spectra for the VB
agrees well with the calculated electronic structure. Moreover,
Figure 4d shows the shift of the band edges with doping, from
was doped, the conductivity was steeply enhanced to 10−2
S
ACS Paragon Plus Environment

Page 5 of 8
Journal of the American Chemical Society
be a new way to raise the VBM without pTM cations, such as
Cu1+ and Sn2+.
1
2
3
Conclusion
We proposed a novel design concept for transparent bipolar
semiconductors. The essential difference from the bipolar
semiconductors reported so far is to use an early transition
metal, rather than a postꢀtransition metal. The effectiveness of
this approach was demonstrated through experiments and calꢀ
culations using tetragonal ZrOS with a nonꢀbonding Zr d orꢀ
bital and a short SꢀS separation. This material has an appropriꢀ
ate direct forbiddenꢀtype bandgap (~2.5 eV) and exhibits clear
bipolarity upon substitutional doping. Since the present design
concept is based on three simple requirements, we expect it to
have applicability to various mixedꢀanion systems containing
an early transition metal.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Band alignment of tꢀZrOS, cꢀZrOS, and related semꢀ
iconductors. The ionization potentials and electron affinity for
tꢀZrOS and cꢀZrOS are determined by combining ultraviolet
photoemission spectroscopy, hard Xꢀray photoemission specꢀ
troscopy and diffuse reflection measurement (see Figures S11
and S12), while the data of other semiconductors except for
ZnS were taken from Ref. (20) and the CBM and VBM level
of ZnS were determined relative to those of ZnO, which is
attained from Ref. (24). The gray region in tꢀZrOS and SnO
expresses the energy difference of their fundamental bandgap
and the direct allowed bandgap. The energy window for bipoꢀ
lar doping is denoted as the yellow color area between ~−4 eV
and ~−6 eV.
Methodology
Sample preparation. Polycrystalline samples of tꢀZrOS
were synthesized by solidꢀstate reactions using starting mateꢀ
rials ZrO2, ZrS2, ZrS3, Yꢀsubstituted ZrO2, and ZrF4 under 5
GPa at 1200 °C for 1 hour. The ZrO2 was prepared by heating
a Zr foil (99.98%) at 1000 °C in air for 2 days, while the ZrS2
and ZrS3 were synthesized by heating the stoichiometric mixꢀ
ture of Zr foil and S grains (99.9999%) at 450 °C for 1 day
and 1000 °C for 2 days in an evacuated silica glass tube. For
the Yꢀsubstituted ZrO2, we mixed the ZrCl2O·8H2O and
Y(NO3)3·6H2O in water with a magnetic stirrer for 20 minutes
and neutralized it with ammonia, then, the resultant salt was
filtered by pure water. Finally, the final product was obtained
by annealing the salt at 600 °C for 10 hours. Yttriumꢀdoped
samples were prepared from the mixture: xZrS2 + xZrO2 +
(1−2x)/3Y2O3 + (1−2x)/3ZrS3 → Zr(4x+1)/3Y2(1−2x)/3OS (Indeed,
we used Y2O3ꢀstabilized ZrO2 instead of the mixture of ZrO2
and Y2O3.). Fluorineꢀdoped samples were prepared from the
mixture: (0.5−y)ZrS2 + 0.5ZrO2 + yZrF4 → ZrOS1−4yF4y + 2yS
↑. A polycrystalline sample of cꢀZrOS was synthesized by
heating the stoichiometric mixture of ZrO2 and ZrS2 at 1000
°C for 5 days in a silica glass tube filled argon.
tꢀZrOS:Y to tꢀZrOS:F, corresponding to the calculated
bandgap of 1.4 eV, shown in Figure 3a.
Figure 4e and 4f show the calculated and experimental abꢀ
sorption coefficients α of tꢀZrOS, respectively. Owing to the
direct forbidden transition between the band edge states, the α
of tꢀZrOS is as low as ~104 cm−1 below 2.5 eV, as shown in
Figure 4e. Above that, the α is sharply raised at hν = 2.5 eV
(Egdirect allowed), owing to the direct allowed transition from the
VBM to the second lowest conduction band at the Γ point. The
direct forbidden bandgap was experimentally estimated to be
1.5 eV (Egdirect fobbiden) from the (hvα/s)2/3 plot in Figure 4f,
which is consistent with the calculated bandgap and the energy
shift of 1.4 eV, observed in the hard Xꢀray photoemission
measurements.
Structural and chemical analyses. The crystal structure of
the samples was analyzed by powder Xꢀray diffraction
(D8ADVANCEꢀTXS Bruker AXS) using Cu Kα radiation
(power: 45 kV × 360 mA) and equipped with a position sensiꢀ
tive detector. The amount of Y in the tꢀZrOS:Y was measured
by an electron probe micro analyzer (EPMA), while the nomiꢀ
nal composition was adopted for the F content in tꢀZrOS:F
because of the difficulty to measure a reliable F content by the
EPMA.
Figure 5 shows the experimentally determined band alignꢀ
ment of tꢀZrOS, cꢀZrOS, and relevant semiconductors. Acꢀ
cording to an empirical rule, the lower limit of the electron
affinity (CBM position from the vacuum level (Evac.)) for sucꢀ
cessful electronꢀdoping is ~−4 eV, while the higher limit of
the ionization potential (VBM from the Evac.) for holeꢀdoping
is ~−6 eV. Since the CBM and VBM of tꢀZrOS are located at
−4.1 eV and −5.6 eV, respectively, it is evident that the band
alignment of tꢀZrOS is suitable for bipolar conductivity. Addiꢀ
tionally, the CBM and VBM levels of cꢀZrOS are out of the
conventional doping range, and thus cꢀZrOS is predicted to be
not promising for electron or hole doping. The transport and
electronic properties of cꢀZrOS are shown in Figure S13. Note
that even though the 3p orbitals of sulfur form shallower enerꢀ
gy levels compared with the 2p orbitals of O, the VBMs of cꢀ
ZrOS and ZnS are located below the pꢀtype doping limit,
which supports that the strong interaction between neighborꢀ
ing anions plays a vital role to increase the VBM level. This
result demonstrates that the short anionꢀanion separation will
Transport measurements. The electrical conductivity was
measured by the van der Pauw method with a sputtered Pt
electrode (for pꢀtype and unꢀdoped tꢀZrOS) or a Ag electrode
paste (for nꢀtype) using a physical property measurement sysꢀ
tem (PPMS Quantum Design). The Seebeck coefficient was
obtained from the gradient of the slope for the thermovoltage
versus temperature relationship at room temperature.
Ultraviolet photoemission spectroscopy (UPS). The enerꢀ
gy levels of the CBM and the VBM from the Evac., which corꢀ
respond to the electron affinity, χ, and ionization potential, Ip,
respectively, were measured by UPS (excited by He I and II
light sources). To prepare chemically pure surfaces, the dense
tꢀZrOS:F pellet was cleaved in a globe box that was directly
connected to the UPS chamber. The χ was calculated using the
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 8
measured optical bandgap by χ = Ip − Eg. The UPS measureꢀ
ment was performed at ~2×108 Pa.
1
ABBREVIATIONS
2
3
4
5
6
7
8
9
Hard X-ray photoemission spectroscopy. Hard Xꢀray
photoemission measurements at room temperature were perꢀ
formed at the BL15XU undulator beamline (the excitation Xꢀ
ray energy: hν = 5953.4 eV) of SPringꢀ8.(25) The binding enerꢀ
gy was calibrated with the Fermi level of an evaporated Au
thin film, and the total energy resolution was set to 240 meV,
which was confirmed by the Fermi cutꢀoff of the Au film. To
obtain a clean sample surface, samples were cleaved in a prepꢀ
aration chamber.
CCR2, CC chemokine receptor 2; CCL2, CC chemokine ligand 2;
CCR5, CC chemokine receptor 5; TLC, thin layer chromatogꢀ
raphy.
REFERENCES
(1)
383.
(2)
Yu, X.; Marks, T. J.; Facchetti, A. Nat Mater 2016, 15,
Fortunato, E.; Ginley, D.; Hosono, H.; Paine, D. C. MRS
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Bulletin 2007, 32, 242.
Diffuse reflectance measurement. Diffuse reflectance
spectra Robs were measured with powder samples, and calibratꢀ
ed by measuring a reference material (MgO powder) at the
same time. The absorption coefficient α was obtained by using
the Kubelka–Munk theory (1ꢀRobs)2/(2Robs) = α /s (s is the scatꢀ
tering coefficient).(26) The direct allowed, indirect, and direct
forbidden bandgaps were estimated by (hvα/s)2, (hvα/s)1/2, and
(hvα/s)2/3 versus hv, respectively.
(3)
24, 2945.
(4)
Physics 1998, 83, 3192.
(5)
Fortunato, E.; Barquinha, P.; Martins, R. Adv. Mater. 2012,
Zhang, S. B.; Wei, S.ꢀH.; Zunger, A. Journal of Applied
Kawazoe, H.; Yasukawa, M.; Hyodo, H.; Kurita, M.; Yanꢀ
agi, H.; Hosono, H. Nature 1997, 389, 939.
(6)
Tennakone, K.; Kumara, G. R. R. A.; Kumarasinghe, A. R.;
Wijayantha, K. G. U.; Sirimanne, P. M. Semicond. Sci. Technol. 1995,
10, 1689.
(7)
T.; Hirano, M.; Hosono, H. Appl. Phys. Lett. 2008, 93, 032113.
(8) Hosono, H.; Ogo, Y.; Yanagi, H.; Kamiya, T. Electrochem.
SolidꢀState Lett. 2011, 14, H13.
Ogo, Y.; Hiramatsu, H.; Nomura, K.; Yanagi, H.; Kamiya,
Theoretical calculations. The electronic structures of tꢀ
ZrOS and cꢀZrOS were derived by using the generalized graꢀ
dient approximation Perdew–Burke–Ernzerhof (GGAꢀPBEsol)
functional(27) and the fullꢀpotential linearized augmented plane
wave with WIEN2k code.(28) The mesh samplings in the Brilꢀ
louin zone (BZ) were 5 × 5 × 5 for the tꢀZrOS and cꢀZrOS. To
correct the error of the bandgap derived from the GGAꢀ
PBEsol method, the bandgap of tꢀZrOS was calculated by
HSE06(29) with the VASP code and the bandgap of cꢀZrOS
was taken from Ref. (30) where the bandgap was calculated by
GW method. The energy of the conduction band was shifted to
match the correlated bandgap value. The absorption coeffiꢀ
cient and the transition probability were calculated with
WIEN2k code.(31) The edge of the absorption spectrum was
shifted by the scissors operator to fit the bandgap estimated by
the diffuse reflectance. The broadening parameter Γ was set
zero in every absorption spectra. Maximally localized Wannier
functions as projected functions were constructed from Zr 4d,
S 3p, S 3s, and O 2p states within an energy window from −15
eV to approximately 7.5 eV.(32)
(9)
Kamiya, T.; Hosono, H. NPG Asia Mater 2010, 2, 15.
(10) Kawazoe, H.; Yanagi, H.; Ueda, K.; Hosono, H. MRS Bulꢀ
letin 2000, 25, 28.
(11) Walsh, A.; Watson, G. W. J. Phys. Chem. B 2005, 109,
18868.
(12) Pohl, J.; Albe, K. Phys. Rev. B 2013, 87, 245203.
(13) Raebiger, H.; Lany, S.; Zunger, A. Phys. Rev. B 2007, 76,
045209.
(14) Togo, A.; Oba, F.; Tanaka, I.; Tatsumi, K. Phys. Rev. B
2006, 74, 195128.
(15) Mizoguchi, H.; Kawazoe, H.; Hosono, H. Chem. Mater.
1996, 8, 2769.
(16) Zunger, A. Appl. Phys. Lett. 2003, 83, 57.
(17) Mizoguchi, H.; Kamiya, T.; Matsuishi, S.; Hosono, H. Naꢀ
ture Communications 2011, 2, 470.
(18) A nonꢀbonding state can be found in materials with the
space group belonging to highꢀsymmetry point groups: D2h, D4h, D6h,
and Oh, and low multiplicity of each constituent. Some related examꢀ
ples are shown in Table S1-S5.
(19) Kutzelnigg, W. Angew. Chem. Int. Ed. Engl. 1984, 23, 272.
(20) Hosono, H. Jpn. J. Appl. Phys. 2013, 52, 090001.
(21) The direct forbidden bandꢀtoꢀband transition frequently ocꢀ
curs in crystals belonging to high symmetry point groups. See Fig-
ures S1 and S2.
(22) Dresselhaus, M. S.; Dresselhaus, G.; Jorio, A. Group Theoꢀ
ry: Application to the Physics of Condensed Matter; Springer Science
& Business Media, 2007.
(23) Elliott, S. R. Physics of Amorphous Materials; Longman,
1983.
(24) Persson, C.; PlatzerꢀBjörkman, C.; Malmström, J.; Törndahl,
T.; Edoff, M. Phys. Rev. Lett. 2006, 97, 146403.
(25) Ueda, S.; Katsuya, Y.; Tanaka, M.; Yoshikawa, H.; Yamashiꢀ
ta, Y.; Ishimaru, S.; Matsushita, Y.; Kobayashi, K. AIP Conference
Proceedings 2010, 1234, 403.
(26) Kubelka, P.; Munk, F.; Z. Tech. Phys. 1931, 12, 593.
(27) Perdew, J. P.; Ruzsinszky, A.; Csonka, G. I.; Vydrov, O. A.;
Scuseria, G. E.; Constantin, L. A.; Zhou, X.; Burke, K. Phys. Rev.
Lett. 2008, 100, 136406.
(28) Blaha, P.; Schwarz, K.; Madsen, G.; Kvasnicka, D.; Luitz, J.
WIEN2k; Vienna University of Technology; Vienna, Austria, 2001.
(29) Heyd, J.; Scuseria, G. E.; Ernzerhof, M. The Journal of Chemꢀ
ical Physics 2003, 118, 8207.
(30) Hautier, G.; Miglio, A.; Ceder, G.; Rignanese, G.ꢀM.; Gonze,
X. Nature Communications 2013, 4, 2292.
ASSOCIATED CONTENT
Supporting Information. The Supporting Information is availaꢀ
ble free of charge on the ACS Publications website. The descripꢀ
tions of nonꢀbonding nature and selection rule, characterization
data, transport properties, HAXPES, and UPS data. (PDF file)
AUTHOR INFORMATION
Corresponding Author
*S. I, eꢀmail: s_iimura@mces.titech.ac.jp
*H. H, eꢀmail: hosono@msl.titech.ac.jp
Notes
Any additional relevant notes should be placed here.
ACKNOWLEDGMENT
The hard Xꢀray photoemission measurements were performed
under the approval of NIMS Synchrotron Xꢀray Station (Proposal
No. 2016B4703). S.U. would like to thank the staff of HiSOR,
Hitoshima University and JAEA at SPringꢀ8 for the development
of hard Xꢀray photoemission apparatus at BL15XU of SPringꢀ8.
This work was partly supported by the MEXT Elements Strategy
Initiative to Form Core Research Center.
ACS Paragon Plus Environment

Page 7 of 8
Journal of the American Chemical Society
(31) AmbroschꢀDraxl, C.; Sofo, J. O. Computer Physics Communiꢀ
cations 2006, 175, 1.
1
(32) Mostofi, A. A.; Yates, J. R.; Lee, Y.ꢀS.; Souza, I.; Vanderbilt,
D.; Marzari, N. Computer Physics Communications 2008, 178, 685.
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 8
Insert Table of Contents artwork here
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
8
ACS Paragon Plus Environment