Tunable Light Emission through the Range 1.8-3.2 eV and p-Type Conductivity at Room Temperature for Nitride Semiconductors, Ca(Mg1- xZn x)2N2 (x = 0-1)

Article abstract of DOI:10.1021/acs.inorgchem.9b01811

The ternary nitride CaZn₂N₂, composed only of earth-abundant elements, is a novel semiconductor with a band gap of ～1.8 eV. First-principles calculations predict that continuous Mg substitution at the Zn site will change the optical band gap in a wide range from ～3.3-1.9 eV for Ca(Mg_1-xZn_x)₂N₂ (x = 0-1). In this study, we demonstrate that a solid-state reaction at ambient pressure and a high-pressure synthesis at 5 GPa produce x = 0 and 0.12 and 0.12 < x ≤ 1 polycrystalline samples, respectively. It is experimentally confirmed that the optical band gap can be continuously tuned from ～3.2 to ～1.8 eV, a range very close to that predicted by theory. Band to band photoluminescence is observed at room temperature in the ultraviolet-red region depending on x. A 2% Na doping at the Ca site of Ca(Mg_1-xZn_x)₂N₂ converts its highly resistive state to a p-type conducting state. Particularly, the x = 0.50 sample exhibits intense green emission with a peak at 2.45 eV (506 nm) without any other emission from deep-level defects. These features meet the demands of III-V group nitride and arsenide/phosphide light-emitting semiconductors.

Full text of DOI:10.1021/acs.inorgchem.9b01811

Article
Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX
pubs.acs.org/IC
Tunable Light Emission through the Range 1.8−3.2 eV and p‑Type
Conductivity at Room Temperature for Nitride Semiconductors,
Ca(Mg Zn ) N (x = 0−1)
1
−x
x 2 2
†
,†,‡
and Hideo Hosono‡
Masatake Tsuji, Hidenori Hiramatsu,*
†Laboratory for Materials and Structures, Institute of Innovative Research, Tokyo Institute of Technology, Mailbox R3-3, 4259
Nagatsuta-cho, Midori-ku, Yokohama 226-8503, Japan
‡
Materials Research Center for Element Strategy, Tokyo Institute of Technology, Mailbox SE-1, 4259 Nagatsuta-cho, Midori-ku,
Yokohama 226-8503, Japan
*
S Supporting Information
ABSTRACT: The ternary nitride CaZn N , composed only
2
2
of earth-abundant elements, is a novel semiconductor with a
band gap of ∼1.8 eV. First-principles calculations predict that
continuous Mg substitution at the Zn site will change the
optical band gap in a wide range from ∼3.3−1.9 eV for
Ca(Mg Zn ) N (x = 0−1). In this study, we demonstrate
1−x
x 2
2
that a solid-state reaction at ambient pressure and a high-
pressure synthesis at 5 GPa produce x = 0 and 0.12 and 0.12 <
x ≤ 1 polycrystalline samples, respectively. It is experimentally
confirmed that the optical band gap can be continuously tuned
from ∼3.2 to ∼1.8 eV, a range very close to that predicted by
theory. Band to band photoluminescence is observed at room temperature in the ultraviolet−red region depending on x. A 2%
Na doping at the Ca site of Ca(Mg1 Zn ) N converts its highly resistive state to a p-type conducting state. Particularly, the x =
−x
x 2
2
0
.50 sample exhibits intense green emission with a peak at 2.45 eV (506 nm) without any other emission from deep-level
defects. These features meet the demands of III−V group nitride and arsenide/phosphide light-emitting semiconductors.
INTRODUCTION
nitride semiconductors have mostly been limited to III−V
group nitrides, GaN, and its solid solutions, with appropriate
band gaps intentionally tuned by solid-solution alloying
techniques.
■
Rapid exploration of novel functional materials, composed only
of earth-abundant elements, has been urgently required for the
future realization of a sustainable society. Furthermore, such
functional materials will lead to highly efficient energy
conversion, storage, and optoelectronic functionalities, such
as electron transport and light emission, and will be able to
replace those materials currently being used, such as heavy,
rare, and toxic elements. To fulfill such a role, nitrides are
promising candidates because they are environmentally benign
and remain unexplored in comparison with other materials
systems, such as oxides, chalcogenides, and pnictides (e.g.,
arsenides and phosphides). The primary reason for such
exploration results from the very high stability of a nitrogen
molecule with a triple bond in comparison with an oxygen
A light-emitting semiconductor with electron current
injection is one of the key components in practical
optoelectronic devices. Although such components have been
commercially used, especially the III−V group semiconductors,
such as (Ga,Al,In)N- and (Ga,Al,In)(As,P)-based materials,
there remains a strong demand for next-generation materials to
exhibit higher performance in terms of light brightness,
quantum efficiency, and color accuracy (i.e., emission
1
bandwidth). However, irrespective of the rapid growth in
science and technology, light-emitting semiconductors have a
serious issue, the so-called green gap problem, where the
emission quantum efficiency drastically decreases to ca. 20% in
the particular wavelength region around green light, even for
molecule, chalcogen, or pnictogen. N is therefore inactive
2
even at high temperatures and under high pressures without
appropriate catalysts, which has led to the historical reluctance
for the wider exploration of nitrides in comparison with other
compounds, in addition to very successful research on III−V
group nitrides, because one may easily use solid reagent
sources in cases of oxides, chalcogenides, and pnictides. This is
the reason associated with the requirements of more active
nitrogen sources, such as ammonia and discharged plasma, for
novel nitride research. Thus, the currently commercialized
2
the practical III−V group based light-emitting diodes. This is
mainly because of the large in-plane lattice mismatch between
In-based III−V nitrides and single-crystal substrates and the
thermal instability of In-based III−V nitrides. Thus, no highly
efficient green-light-emitting semiconductor material has yet
Received: June 18, 2019
©
XXXX American Chemical Society
A
DOI: 10.1021/acs.inorgchem.9b01811
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
been realized for next-generation optoelectronic devices.
Therefore, it is necessary to explore and discover novel
semiconductor materials, which possibly have higher thermal
stability and more compatible in-plane lattice mismatches with
III−V group nitrides, with a high quantum efficiency of light
emission in the green wavelength region (band gap 2.3−2.5
eV, wavelength λ = 500−550 nm).
Recent progress of materials informatics, which combines
comprehensive first-principles calculations with experimental
validations, has led to an acceleration in the discoveries and/or
predictions of novel ternary nitride semiconductors, such as
3
4
5
Cu(Nb,Ta)N2, LaWN3, and CaZn N , which were all
2
2
previously unknown. Therefore, materials informatics provides
the possibility to overcome the aforementioned historical lack
of exploration research of novel functional nitride compounds.
CaZn N was predicted through such an approach and was
2
2
proposed with experimental validation as a novel promising
optoelectronic functional nitride semiconductor with a band
gap of ∼1.8 eV because of its small electron and hole effective
masses and direct transition-type energy band structure with a
wide tunability of its optical band gap of isomorphism nitrides
by an alloying technique similar to III−V group semi-
conductors. In ref 5, it was predicted that the band gaps
could be widely tuned from ∼3.3 eV (for CaMg N , i.e.,
2
2
ultraviolet region) to ∼1.6 eV (for SrZn N ) and ∼0.4 eV (for
Figure 1. Structure, crystalline phase, and composition of Ca-
2
2
(
Mg1−xZn ) N (x = 0−1) polycrystalline samples obtained. (a) XRD
CaCd N ) (i.e., red−infrared region) by solid-solution alloying
x
2
2
2
2
patterns. The three dashed rectangles denote areas of the diffraction
peak positions, which are assigned as 1011 and 0111 (2θ = 32.6−
3.3°), 1012 and 0112 (41.7−42.5°), and 0003 (44.6−45.1°)
at both the Ca and Zn sites. This prediction indicates that this
nitride is a promising host candidate as a light-emitting
semiconductor with high electron and hole transport proper-
ties in the widely tunable ultraviolet−infrared wavelength
region.
̅
̅
3
̅
̅
diffractions from the Ca(Mg Zn ) N main phase with a hexagonal
1
−x
x
2
2
structure of a trigonal system with P3
̅
m1. Diffraction peak positions of
Zn, Mg N , CaO, and MgO impurities are indicated at the bottom by
3
2
In this study, we focused on an earth-abundant Ca-
vertical bars. A small amount of BN, which is the material of the
capsules in the high-pressure cells, was observed at the vertical arrow
positiona (∼26.8°). (b) Lattice parameters a and c as a function of x.
Closed circles and triangles are a and c obtained in this study,
respectively. Open circles and triangles are the calculated results with
their structure relaxation reported in ref 5. Straight lines are shown on
the assumption of Vegard’s law between both end members. (c)
Volume fraction of the main phase and impurities as a function of x.
All of the data are taken from the results of the Rietveld analysis. The
fitted raw results are provided in Figure S1.
(
Mg1−xZn ) N (x = 0−1) complete solid-solution nitride
x 2
2
system because its band gap is controllable in the range of ∼1.9
to ∼3.3 eV, following the computational prediction. Here we
experimentally report the validity of the prediction; that is, its
band gap is tunable, the x = 0.50 sample exhibits an intense
green emission without deep defect emission even at room
temperature, and it can be doped to a p-type semiconductor by
2
% Na doping at the Ca site.
STRUCTURE AND PHASE PURITY OF
■
Ca(Mg1 Zn ) N (x = 0−1)
for Mg2+ and 60 pm for Zn in the case where the
2+
−x
x 2
2
7
Figure 1a shows XRD patterns of polycrystalline Ca-
coordination number is 4, this variation trend with changing
(
Mg1−xZn ) N (x = 0−1). The results of Rietveld analysis
x implied that the chemical bond between (Mg,Zn) and N was
more covalent than ionic because the covalent bond radius of
Zn is slightly smaller than that of Mg. Note that a small
amount of BN was observed at 2θ ≈ 26.8°, which is indicated
by vertical arrows in Figure 1a. It was very difficult to
completely remove this BN impurity in some samples because
a high pressure and a high temperature of 5.0 GPa and 1300
°C were applied to the mixed-powder disks tightly contacted
with capsules made of BN in the high-pressure cells. However,
this thin BN impurity located only around the outside wall of
the resulting disk samples, that is, the BN impurity, did not
diffuse inside the bulk region of the samples. It should be
noted that we observed the phase stability in air at room
temperature became strong with increasing x, that is, the x = 0
sample was unstable in air because it decomposed into the
x 2
2
are summarized in Figure S1. We confirmed that the main
crystalline phase of all the samples had the same crystal
structure as that of both end members, CaMg N and
CaZn N , which consisted of a hexagonal structure of a
6
2
2
5
2
2
trigonal system with P3
fabrication of complete solid-solution samples between x = 0
and 1. With an increase in x, the three diffraction peaks of 101
and 0111, 1012 and 0112, and 0003 of the Ca(Mg1−xZn ) N
main phase, which are surrounded by dotted squares in Figure
a, were continuously shifted to higher 2θ positions.
̅
m1 (No. 164): that is, the successful
̅
1
2
̅
̅
̅
x 2
1
Corresponding to this peak shift, both the a and c axis lattice
parameters shrank with increasing x [(see Figure 1b). In
comparison with the calculated results by first-principles
5
structure relaxation, the absolute values of both end members
(
x = 0 and 1) and their variation trend with a change in x
amorphous-like hydroxide Ae(OH) (Ae = Mg, Ca) just after
2
agreed well with each other, although the calculated values
were slightly smaller than the observed values. Because the
ionic radii of Mg and Zn are almost the same, that is, 57 pm
air exposure, which is similar to the result reported on the
8
binary nitride precursors Mg N and Ca N . In contrast, the x
3
2
3
2
2
+
2+
≥ 0.50 samples were stable at room temperature in air. In
B
DOI: 10.1021/acs.inorgchem.9b01811
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
addition, the sample purity strongly depended on the x
content. The volume fraction of the main phase and impurity
concentration for each x sample are summarized in Figure 1c.
The phase purity of the x = 0 sample was the best (∼90%)
among all the x contents. Addition of Zn decreased the volume
fraction of the main phase to 50−60% for the x = 0.29−1
samples. Accordingly, the Mg N impurity slightly increased up
the calculated results by the Heyd−Scuseria−Ernzerhof
5
(HSE06) hybrid functional, although all of them were slightly
lower than the calculated values (see Figure 2b). Accordingly,
band to band PL originating from each band gap was clearly
observed even at room temperature (see open circles and
vertical bars in Figure 2a). For the x = 0 sample, an additional
large and broad emission band that peaked at ∼2.6 eV was
observed. The decay curve of this emission band was
decomposed into two components, which were a relatively
short component of <∼30 ns and a longer component of ∼670
ns (see Figure S2). Although the exact origins of these two
lifetimes were unclear, we propose possible origins. The former
could be related to a small amount (7 mol %) of the Mg N
3
2
to ∼20% at x = 0.29, and the metal Zn impurity drastically
increased at x ≥ 0.69.
OPTICAL PROPERTIES OF Ca(Mg1−xZn ) N (x =
x 2
2
■ 0
−1)
3
2
Figure 2 shows the optical properties of polycrystalline
impurity because the energy position is close to the band gap
Ca(Mg1−xZn ) N (x = 0−1) at room temperature. With an
9
x 2
2
of Mg N (∼2.5 eV) or PL originating from recombination of
3
2
increase in x, the absorption edge (nearly equal to the optical
10
holes and electrons at in-gap state defects in Mg N . The
3
2
band gap energy in this case) continuously shifted to the lower
latter could originate from defects such as donor−acceptor
energy side from ∼3.2 to ∼1.8 eV (see black slope lines for
pairs in CaMg N because the sample has a defective
2
2
2
(
αhν/s) versus hν plots (data plotted by solid lines) in Figure
polycrystalline nature and the lifetime of ∼670 ns is close to
those of donor−acceptor pairs with a short donor−acceptor
2
a). This continuously controllable band gap agreed well with
1
1
distance ≪10 nm of conventional semiconductors. For the x
=
0.29 sample, we confirmed that the emission lifetime of an
emitting band that peaked at ∼2.8 eV was ∼7 ns (see Figure
S3), which is comparable to the pulse width of the excitation
laser we used: that is, the exact lifetime could not be estimated
by this experiment. However, the information on a short
lifetime was sufficient to assign it to a band to band transition.
Therefore, we concluded that the ∼2.8 eV PL band originates
from a band to band emission for x = 0.29. In addition, an
emission band that peaked at a ∼3.1 eV was observed. This PL
12
originated from the BN impurity, because we also observed
the same emission band from the x = 0.50 and 0.69 samples
when we set the observation wavelength region of the
spectrometer around this region (data not shown) and a
small amount of BN impurity was commonly observed in these
three samples in their XRD patterns (see vertical arrows at 2θ
=
∼26.8° in Figure 1a). Consequently, on the basis of the
above experimental optical properties at room temperature, we
continuously and widely tuned the optical band gaps and
emission bands from ∼3.2 eV (λ = ca. 390 nm, ultraviolet
region) to ∼1.8 eV (ca. 670 nm, red region) in complete solid-
solution samples of Ca(Mg Zn ) N (x = 0−1). In particular,
it should be noted that the x = 0.50 sample exhibited intense
green light emission that peaked at ∼506 nm (∼2.45 eV) even
at room temperature.
1−x
x 2
2
In this study, we focused on light-emitting semiconductors
in the green color region (i.e., band gap 2.3−2.5 eV, λ = 500−
5
50 nm). Thus, we examined the PL properties of the x = 0.50
sample in more detail because the PL band was located in the
appropriate wavelength region for green emission. Figure 3a
shows the temperature dependence of the PL spectra. We
observed three emission bands, A−C. Band A originated from
12
a small amount of a BN impurity, as discussed in the last
section. Band B was assigned to a band to band transition
because the emission peak position at 300 K was almost the
same as that of the absorption edge observed in Figure 2a.
Band C arose from deep-level defects and/or impurity phases
because the energy difference between B and C was ∼200 meV
at 20 K, although the origin was unclear because the emission
peak energy of band C did not depend on the measurement
temperature and almost disappeared at ≥150 K. Because a
similar temperature dependence has been reported for a Mg
Figure 2. Optical properties of Ca(Mg1−xZn ) N (x = 0−1)
x
2
2
polycrystalline samples at room temperature. (a) PL (open circles)
and R (=α/s, solid lines) spectra. The dashed lines are the results of
the deconvolution of PL spectra. The PL peak positions arising from
band to band transitions are shown by vertical bars. The black slope
2
lines for (αhν/s) versus hν plots are used to estimate the absorption
edges. A vertical arrow in the x = 0.29 spectrum indicates an emission
band from a BN impurity. (b) Optical band gaps determined on the
2
basis of the plots of (αhν/s) versus hν in (a) as a function of x. The
calculated values (open circles) by the HSE06 hybrid functional,
reported in ref 5, are shown for comparison.
13
acceptor in GaN:Mg, band C could have similar origins: that
C
DOI: 10.1021/acs.inorgchem.9b01811
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
16
to be on the sub-nanosecond order. However, these above
assignments are tentative and speculative possibilities.
ELECTRONIC TRANSPORT PROPERTIES OF Na 2%
■
Doped (Ca1−yNa )(Mg Zn ) N (x = 0.67 and y =
y
1−x
x 2
2
0.02)
Although Na-undoped (Ca1−yNa )(Mg Zn ) N (x = 0−1
y
1−x
x 2
2
and y = 0) samples were all too highly resistive to allow
measurement of the electrical conductivity (much higher than
6
1
0 Ω cm), 2% Na doping exhibited measurable conductivity.
̅
Figure 4a shows an enlargement of the strongest 1011 and
Figure 3. (a) Temperature dependence of PL spectra of the
Ca(Mg1−xZn ) N (x = 0.50) polycrystalline sample. The dashed
x
2
2
lines are deconvolution results of each PL spectrum. Note that the
spectrum at 300 K is the same as that for x = 0.50 in Figure 2a. (b) PL
peak energy (squares, left axis) and normalized PL intensity (circles,
right axis) of band B originating from band to band transition as a
function of measurement temperature. The vertical arrow indicates
the position of an anomalous blue shift between 75 and 100 K.
is, point defects trapped at doped-impurity sites (e.g., band to
impurity transitions). Owing to the band to band recombina-
tion, the energy of band B shifted to the lower energy side with
increasing temperature from 2.49 eV at 20 K to 2.45 eV at 300
K (see Figure 3b). At 75−100 K, an anomalous PL peak shift
from 2.48 to 2.49 eV (i.e., a blue shift with increasing
temperature) was observed. The possible origin of this
anomalous shift was a result of a transition from bound
excitons to free excitons, as observed in room-temperature-
exciton (i.e., electron−hole pairs stable at room temperature)
materials, such as ZnO and LaCuOS; whereas details of its
origin should be discussed using more high resolution and
continuous wave excited PL measurements on higher quality
samples, such as single crystals and epitaxial films. The PL
intensity was decreased by approximately 1 order of magnitude
from 20 to 300 K, owing to temperature deactivation (see
Figure 3b). The origin of this temperature quenching would be
defects originating from the polycrystalline nature of the
sample. However, the emission observed in the x = 0.50
sample shown in Figure 2a was also clearly observed at room
temperature. Next, we examined the emission lifetime of bands
B and C. The estimated lifetimes at 20 K were ∼2 ns for band
B and ∼250 ns for band C (see Figure S4). Because the pulse
width of the excitation Nd:YAG we used was ∼7 ns, the
lifetime of band B could not be exactly estimated. This result
suggested that the emission band B originated from bound/
free excitons because the lifetime was sufficiently shorter than
several nanoseconds. The relatively fast lifetime of ∼250 ns for
band C was roughly consistent with the origin proposed above,
that is, point defects trapped at doped-impurity sites, because
the lifetime of the Mg acceptor in GaN:Mg has been reported
Figure 4. Structure and electronic transport properties of the
Ca1−yNa )(Mg Zn ) N (x = 0.67 and y = 0.02) polycrystalline
sample. (a) XRD diffraction pattern around 1011 and 0111 (red
̅ ̅
(
y
1−x
x
2
2
pattern), which is the strongest peak of the main phase. The peak of
the undoped sample (black pattern) is plotted for comparison. (b)
Temperature dependence of electrical resistivity (ρ). The inset shows
thermoelectric power (ΔV) as a function of difference in temperature
(
ΔT) at room temperature (RT), revealing successful carrier doping
and its p-type electrical conductivity at room temperature.
14
15
̅
0111 peak region (2θ ≈ 33°) in the XRD pattern of the Na 2%
doped polycrystalline sample. The Rietveld analysis result is
shown in Figure S5. The main phase purity was low, 25%,
while those of Zn and Mg N impurities that might possibly
3
2
exhibit electrical conduction were also small, 8% and 15%,
respectively. Thus, it was considered that such small amounts
of possibly electrical conductive impurities should not
dominate the electrical conduction of the sample. Irrespective
2
+
+
of almost the same ionic radii of Ca and Na , the XRD peak
slightly shifted to the higher 2θ side by Na doping, which
indicated its lattice shrinkage. Although the undoped sample
with x ≥ 0.5 was chemically stable, the Na-doped sample with
x ≥ 0.5 was quite sensitive to air exposure. As observed in
Figure S5, Na substitution at the Ca site would be the assertive
doping process for the main crystalline phase because phase
separation of CaMg N is observed only in this Na-doped
2
2
sample and higher concentration doping did not work
effectively. The temperature dependence of ρ exhibited a
thermally activated type behavior, which implied that metallic
zinc did not contribute to the electronic conduction. The inset
of Figure 4b is the result of a thermoelectric power
D
DOI: 10.1021/acs.inorgchem.9b01811
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
measurement at room temperature. These data indicate p-type
electrical conduction with a positive Seebeck coefficient of
mixing and heating experiments for syntheses of Ca(Mg1−xZn ) N (x
x 2 2
=
0−1) and were completely used within 2 days, as it was difficult to
store them without degradation even in the glovebox for a long time,
such as ∼1 week.
+
2.3 μV/K. The absolute value of this Seebeck coefficient
seems to be too low for semiconductors; however, as discussed
above, the amounts of impurity phases observed are too small
to contribute electrical conduction of this sample by
considering percolation threshold (∼30% in three-dimensional
volume). Therefore, we concluded that this positive sign of
Seebeck coefficient comes from the Na-doped main crystalline
phase. However, the activation energy estimated using an
Arrhenius plot (see Figure S6) was ∼330 meV, which
suggested that relatively deep acceptor levels formed within
the energy band gap upon Na doping. We also tried n-type
carrier doping by adding La or Y to the Ca site, but no
enhancement of electrical conduction was observed. Finally we
tried to measure PL spectra of this Na-doped sample, but no
PL could be observed. This is due to easy degradation of the
surface arising from very high sensitivity to the ambient
atmosphere. These results indicated that we could successfully
obtain a p-type semiconducting sample with an appropriate
band gap near green emission by selecting proper x and Na
doping amounts.
To synthesize polycrystalline Ca(Mg1−xZn ) N (x = 0 and 0.12), a
x
2
2
conventional solid-state reaction was employed. Stoichiometric
mixtures of Ca N , Mg N , and Zn N precursor powders with
3
2
3
2
3
2
molar ratios of Ca N :Mg N :Zn N = 1:2(1−x):2x were pressed into
3
2
3
2
3
2
disks (size: diameter of ∼6 mm and height of ∼7 mm) in the
glovebox, which were sealed in Ar-filled stainless tubes, followed by a
heating process at 1050 °C for 24 h, as has previously been reported
6
for CaMg N .
2
2
In contrast, to synthesize Zn-substituted nitrides with higher x
Ca(Mg1−xZn ) N (0.12 < x ≤ 1) solid solutions, we used a belt-type
x
2
2
22
high-pressure apparatus to realize a high nitrogen chemical potential
during the chemical reaction between precursors in the high-pressure
cells. Stoichiometric mixtures of each binary precursor were pressed
into disks, which were put into BN capsules, followed by setting up in
high-pressure cells made of NaCl (+10 wt % ZrO ) with a graphite
2
heater. We also examined Na doping at the Ca site for hole-carrier
doping because the ionic radius of Na is close to that of Ca (100 pm
2
+
+
for Ca and 102 pm for Na for the case where their coordination
7
numbers are 6), whereas the atomic/covalent bond radius of Na is
slightly smaller than that of Ca. A commercially available NaN3
reagent powder (99.5%) was used as a sodium source. As-received
−
2
NaN was thermally annealed under a vacuum of ∼10 Pa at 250 °C
3
CONCLUSION
■
for 6 h to purify before mixing. We mixed the precursor powders so as
In summary, we synthesized Ca(Mg1 Zn ) N (x = 0−1) by a
to obtain (Ca1−yNa )(Mg Zn ) N (x = 0.67 and y = 0.02).
−x
x 2
2
y
1−x
x
2
2
conventional solid-state reaction for x = 0 and 0.12 and high-
pressure synthesis at 5 GPa for 0.12 < x ≤ 1 in
The pressure and heating temperature of the high-pressure cells,
including BN capsules and pressed disks in the glovebox for the
1
7
synthesis of undoped and Na-doped Ca(Mg1−xZn ) N (0.12 < x ≤
Ca(Mg Zn ) N . It was experimentally validated that the
x
2
2
1−x
x 2
2
1
), were commonly increased to 5.0 GPa and 1300 °C, respectively,
optical band gaps can be tuned continuously from ∼3.2 to
and they were then kept for 30 min. The final size of the high-
pressure-synthesized samples was a diameter of ∼5.5 mm and
thickness of ∼2 mm (after polishing). More details of the
experimental procedure and apparatus of the high-pressure synthesis
are reported in ref 5.
∼
1.8 eV. Owing to the direct transition type band structure,
ultraviolet−red photoluminescence was clearly observed in the
entire x = 0−1 region even at room temperature. p-type
electrical conduction was also demonstrated by 2% Na doping
at the Ca site. Especially, the sample with x = 0.50 exhibited an
intense green emission, which is strongly demanded in III−V
group light-emitting semiconductors. On the basis of this
achievement, the important issues of this new material series
for future application as light-emitting or photovoltaic devices
are growth of high-quality single-crystalline films by conven-
tional vapor phase epitaxy methods such as molecular beam
epitaxy and chemical vapor deposition and fabrication of a p−n
heterojunction using them.
Crystal structures of the samples, such as crystalline phases and
lattice parameters, were determined by powder X-ray diffraction
(XRD) with the Bragg−Brentano geometry with a Cu Kα radiation
source at room temperature, where an Ar-filled O-ring-sealed sample
holder was used because the Mg-rich samples (x < 0.5) were sensitive
to air. The lattice parameters of the main phase were determined by
the Pawley method using the TOPAS ver. 4.2 program (Karlsruhe,
Germany: Bruker AXS GmbH). Rietveld analysis, where the
fundamental parameter (FP) method is employed, was performed
to estimate the volume fraction of each crystalline phase observed and
x in Ca(Mg1−xZn ) N (i.e., crystallographic site occupancy of the
x
2
2
EXPERIMENTAL METHODS
Three polycrystalline binary nitrides, calcium nitride (Ca N ),
(
Mg,Zn) site) using the TOPAS program after initially determining
the lattice parameters by the Pawley method because it was difficult to
exactly distinguish the main phase of Ca(Mg1−xZn from some
impurity phases, such as Zn and Mg , by conventional
■
3
2
magnesium nitride (Mg N ), and zinc nitride (Zn N ), were used
) N
x 2 2
3
2
3
2
as starting precursors for undoped (Ca1−yNa )(Mg Zn ) N (x =
N
3 2
y
1−x
x
2
2
0
−1, y = 0; i.e., not hole carrier doped) polycrystalline samples. Ca N
spectroscopic quantitative techniques, such as X-ray fluorescence
spectroscopy and electron probe microanalysis.
3
2
powder was synthesized by heating Ca metal (purity 99.99%), which
was put on a rolled Fe foil in a silica-glass tube, at 900 °C for 10 h
under a N gas flow atmosphere (G1 grade, flow rate of N gas 100
mL/min). For Mg N powder, Mg metal (purity 99.5%) was reacted
with N2 gas in almost the same manner as that of the Ca N
Diffuse reflectance (R) spectra were measured at room temperature
with a spectrophotometer in the wavelength (λ) region of 200−2400
nm. The obtained R spectra were converted using the Kubelka−Munk
2
2
3
2
2
23
function (1 − R) /(2R) = α/s, where α and s denote the optical
absorption coefficient and scattering factor, respectively, to obtain the
quasi-optical absorption spectra. The optical band gap of each sample
3
2
precursor. All of these processes, including heating, were performed
in an Ar-filled glovebox (oxygen concentration ≤0.0 ppm, dew point
2
<
−100 °C) without exposure to air because both the binary nitrides
was estimated from plots of (αhν/s) versus hν, where h and ν are the
are sensitive to air; that is, the silica-glass tube furnace was directly
connected to the glovebox. Such a direct nitridization process of
Planck constant and frequency, respectively, because the band
structures of both end members (i.e., x = 0 and 1) have been
5
alkali-earth metals under an N atmosphere is similar to previous
reports.
product (purity 99%). Colors of the precursor powders were rubric
black for Ca N , dark chrome yellow for Mg N , and gray for Zn N ,
which were roughly consistent with their reported optical band gaps
of ∼1.5 eV for Ca N , ∼2.5 eV for Mg N , and ∼0.85 eV for
Zn N . Ca N and Mg N powders were synthesized just before
reported to be of the direct transition type. Photoluminescence (PL)
2
18,19
The Zn N powder we used was a commercially available
spectra and emission lifetimes at 20−300 K were attained with a CCD
3
2
camera and a spectrometer under excitation by 3ω of a Nd:YAG laser
2
(λ = 355 nm, pulse width ∼7 ns, excitation density ∼30 mJ/cm );
3
2
3
2
3
2
whereas only for the CaMg N (i.e., x = 0) sample, 4ω of the
2
2
20
9
2
Nd:YAG (λ = 266 nm, excitation density ∼10 mJ/cm ) was used
because of the widest band gap in Ca(Mg Zn ) N (x = 0−1). Note
3
2
3
2
21
3
2
3
2
3
2
1−x
x
2
2
E
DOI: 10.1021/acs.inorgchem.9b01811
Inorg. Chem. XXXX, XXX, XXX−XXX

Inorganic Chemistry
Article
that the sequence of performing experiments both of syntheses and
measurements was step by step; it took 2 years from x = 1 finally to x
Oba, F. Discovery of earth-abundant nitride semiconductors by
computational screening and high-pressure synthesis. Nat. Commun.
2016, 7, 11962.
=
0. If we obtained all the samples at the same time, we could employ
the shortest wavelength.
(6) Schultz-Coulon, V.; Schnick, W. CaMg N − ein gemischtes
2
2
Electronic transport properties were measured only for the
Erdalkalimetallnitrid mit anti-La O -Struktur. Z. Naturforsch., B: J.
2 3
(
Ca1−yNa )(Mg Zn ) N (x = 0.67 and y = 0.02) sample because
Chem. Sci. 1995, 50b, 619−622.
y
1−x
x
2
2
the electrical resistivity of the other undoped samples (i.e., x = 0−1
and y = 0) was too high to allow measurement of their transport
properties. We initially confirmed the electrical conduction with a
conventional electric tester in the glovebox because the Na-doped
sample was air sensitive. The temperature dependence of the electrical
resistivity and thermoelectric power (ΔV) as a function of difference
in temperature (ΔT) at room temperature (i.e., the Seebeck effect)
were measured with a physical property measurement system. For
these electrical measurements, we transferred the Na-doped sample
from the glovebox to the measuring system within ∼1 min together
with surface protection using commercially available Apiezon grease
to disturb the degradation.
(7) Shannon, R. D. Revised Effective Ionic Radii and Systematic
Studies of Interatomic Distances in Halides and Chalcogenides. Acta
Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976,
A32, 751−767.
(8) Heyns, A. M.; Prinsloo, L. C. The Vibrational Spectra and
Decomposition of α-Calcium Nitride (α-Ca N ) and Magnesium
3
2
Nitride (Mg N ). J. Solid State Chem. 1998, 137, 33−41.
3
2
(9) Wu, P.; Tiedje, T. Molecular beam epitaxy growth and optical
properties of Mg N films. Appl. Phys. Lett. 2018, 113, No. 082101.
3
2
(10) Uenaka, Y.; Uchino, T. Excitonic and Defect-Related
Photoluminescence in Mg N . J. Phys. Chem. C 2014, 118, 11895−
3
2
11901.
(
11) Nakashima, S.; Nakamura, A. Radiative recombination of
donor-acceptor pairs in ZnTe. Solid State Commun. 1981, 38, 1289−
292.
12) Berzina, B.; Korsaks, V.; Trinkler, L.; Sarakovskis, A.; Grube, J.;
ASSOCIATED CONTENT
Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.inorg-
chem.9b01811.
■
1
(
*
S
Bellucci, S. Defect-induced blue luminescence of hexagonal boron
nitride. Diamond Relat. Mater. 2016, 68, 131−137.
(
13) Reshchikov, M. A.; Ghimire, P.; Demchenko, D. O. Magnesium
acceptor in gallium nitride. I. Photoluminescence from Mg-doped
GaN. Phys. Rev. B: Condens. Matter Mater. Phys. 2018, 97, 205204.
XRD patterns and their Rietveld analysis results and
emission decay curves (PDF)
(14) Makino, T.; Tamura, K.; Chia, C. H.; Segawa, Y.; Kawasaki, M.;
Ohtomo, A.; Koinuma, H. Photoluminescence properties of ZnO
epitaxial layers grown on lattice-matched ScAlMgO4 substrates. J.
Appl. Phys. 2002, 92, 7157−7159.
AUTHOR INFORMATION
Corresponding Author
E-mail for H.H.: h-hirama@mces.titech.ac.jp.
■
*
(
15) Hiramatsu, H.; Ueda, K.; Takafuji, K.; Ohta, H.; Hirano, M.;
ORCID
Kamiya, T.; Hosono, H. Intrinsic excitonic photoluminescence and
band-gap engineering of wide-gap p-type oxychalcogenide epitaxial
films of LnCuOCh (Ln = La, Pr, and Nd; Ch = S or Se)
semiconductor alloys. J. Appl. Phys. 2003, 94, 5805−5808.
(16) Smith, M.; Chen, G. D.; Lin, J. Y.; Jiang, H. X.; Salvador, A.;
Sverdlov, B. N.; Botchkarev, A.; Morkoc, H.; Goldenberg, B.
Mechanisms of band-edge emission in Mg-doped p-type GaN. Appl.
Phys. Lett. 1996, 68, 1883−1885.
Hidenori Hiramatsu: 0000-0002-5664-5831
Hideo Hosono: 0000-0001-9260-6728
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
(17) Tsuji, M.; Hiramatsu, H.; Hosono, H. Tunable light-emission
This work was supported by JST CREST Grant No.
JPMJCR17J2, Japan. H. Hosono was supported by the
Ministry of Education, Culture, Sports, Science, and
Technology (MEXT) through the Element Strategy Initiative
to Form Core Research Center. H. Hiramatsu was also
supported by the Japan Society for the Promotion of Science
through the range 1.8−3.2 eV and p-type conductivity at room
temperature for nitride semiconductors, Ca(Mg − Zn ) N (x = 0
1
x
x
2
2
−
1). arXiv:1902.10495, https://arxiv.org/abs/1902.10495.
̈
(18) Hohn, P.; Hoffmann, S.; Hunger, J.; Leoni, S.; Nitsche, F.;
Schnelle, W.; Kniep, R. β-Ca N , a Metastable Nitride in the System
3
2
Ca−N. Chem. - Eur. J. 2009, 15, 3419−3425.
(19) Zong, F.; Meng, C.; Guo, Z.; Ji, F.; Xiao, H.; Zhang, X.; Ma, J.;
Ma, H. Synthesis and characterization of magnesium nitride powder
(
(
JSPS) through Grants-in-Aid for Scientific Research (A) and
B) (Grant Nos. 17H01318 and 18H01700) and Support for
formed by Mg direct reaction with N . J. Alloys Compd. 2010, 508,
Tokyotech Advanced Research (STAR).
2
1
(
72−176.
20) Reckeweg, O.; Lind, C.; Simon, A.; DiSalvo, F. J. Rietveld
Refinement of the Crystal Structure of α-Be N and the Experimental
REFERENCES
■
3
2
(
1) Tsao, J. Y.; Crawford, M. H.; Coltrin, M. E.; Fischer, A. J.;
Determination of Optical Band Gaps for Mg N , Ca N and
3
2
3
2
Koleske, D. D.; Subramania, G. S.; Wang, G. T.; Wierer, J. J.; Karlicek,
R. F., Jr. Toward Smart and Ultra-efficient Solid-State Lighting. Adv.
Opt. Mater. 2014, 2, 809−836.
CaMg N . Z. Naturforsch., B: J. Chem. Sci. 2003, 58b, 159−162.
2
2
(21) Kumagai, Y.; Harada, K.; Akamatsu, H.; Matsuzaki, K.; Oba, F.
Carrier-Induced Band-Gap Variation and Point Defects in Zn N
3
2
(
2) Auf der Maur, M.; Pecchia, A.; Penazzi, G.; Rodrigues, W.; Di
from First Principles. Phys. Rev. Appl. 2017, 8, No. 014015.
Carlo, A. Efficiency Drop in Green InGaN/GaN Light Emitting
Diodes: The Role of Random Alloy Fluctuations. Phys. Rev. Lett.
(22) Fukunaga, O.; Ko, Y. S.; Konoue, M.; Ohashi, N.; Tsurumi, T.
Pressure and temperature control in flat-belt type high pressure
apparatus for reproducible diamond synthesis. Diamond Relat. Mater.
2
(
016, 116, No. 027401.
3) Zakutayev, A.; Allen, A. J.; Zhang, X.; Vidal, J.; Cui, Z.; Lany, S.;
1
999, 8, 2036−2042.
23) Kubelka, P.; Munk, F. Ein Beitrag Zur Optik Der Farbanstriche.
Z. Technol. Phys. 1931, 12, 593−601.
Yang, M.; DiSalvo, F. J.; Ginley, D. S. Experimental Synthesis and
Properties of Metastable CuNbN2 and Theoretical Extension to
Other Ternary Copper Nitrides. Chem. Mater. 2014, 26, 4970−4977.
(
(
4) Sarmiento-Perez, R.; Cerqueira, T. F. T.; Korbel, S.; Botti, S.;
́
̈
Marques, M. A. L. Prediction of Stable Nitride Perovskites. Chem.
Mater. 2015, 27, 5957−5963.
(
5) Hinuma, Y.; Hatakeyama, T.; Kumagai, Y.; Burton, L. A.; Sato,
H.; Muraba, Y.; Iimura, S.; Hiramatsu, H.; Tanaka, I.; Hosono, H.;
F
DOI: 10.1021/acs.inorgchem.9b01811
Inorg. Chem. XXXX, XXX, XXX−XXX