Molecular Germanium selenophosphate salts: Phase-change properties and strong second harmonic generation

Article abstract of DOI:10.1021/ja309386e

A new series of germanium chalcophosphates with the formula A ₄GeP₄Q₁₂ (A = K, Rb, Cs; Q = S, Se) have been synthesized. The selenium compounds are isostructural and crystallize in the polar orthorhombic space group Pca2

Full text of DOI:10.1021/ja309386e

Article
pubs.acs.org/JACS
Molecular Germanium Selenophosphate Salts: Phase-Change
Properties and Strong Second Harmonic Generation
†
†
‡
‡
‡
‡
Collin D. Morris, In Chung, Sungoh Park, Connor M. Harrison, Daniel J. Clark, Joon I. Jang,
,†
and Mercouri G. Kanatzidis*
†
Department of Chemistry, Northwestern University, Evanston, Illinois 60208, United States
‡
Department of Physics, Applied Physics and Astronomy, State University of New York (SUNY) at Binghamton, Binghamton, New
York 13902, United States
*
S Supporting Information
ABSTRACT: A new series of germanium chalcophosphates
with the formula A GeP Q (A = K, Rb, Cs; Q = S, Se) have
4
4
12
been synthesized. The selenium compounds are isostructural
and crystallize in the polar orthorhombic space group Pca21.
The sulfur analogues are isostructural to one another but
crystallize in the centrosymmetric monoclinic space group C2/
4
−
c. All structures contain the new molecular anion [GeP Q ] ;
4
12
however, the difference between the sulfides and selenides
arises from the change in crystal packing. Each discrete
molecule is comprised of two ethane-like P Q units that
2
6
4
+
chelate to a central tetrahedral Ge ion in a bidentate fashion.
The selenides were synthesized pure by stoichiometric
reaction of the starting materials, whereas the sulfides
contained second phases. The band gaps of the molecular salts are independent of the alkali metal counterions and have a
value of 2.0 eV for the selenides and 3.0−3.1 eV for the sulfides. All A GeP Se compounds melt congruently, and the potassium
4
4
12
analogue can be quenched to give a glassy phase that retains its short-range order as shown by Raman spectroscopy and powder
X-ray diffraction. Interestingly, K GeP Se is a phase-change material that reversibly converts between glassy and crystalline
4
4
12
states and passes through a metastable crystalline state upon heating just before crystallizing into its slow-cooled form. Initial
second harmonic generation (SHG) experiments showed crystalline K GeP Se outperforms the other alkali metal analogues
4
4
12
and exhibits the strongest second harmonic generation response among reported quaternary chalcophosphates, ∼30 times that of
AgGaSe at 730 nm. A more thorough investigation of the nonlinear optical (NLO) properties was performed across a range of
2
wavelengths that is almost triple that of previous reports (λ = 1200−2700 nm) and highlights the importance of broadband
measurements. Glassy K GeP Se also exhibits a measurable SHG response with no poling.
4
4
12
INTRODUCTION
or two-photon absorption (TPA) are problems that are
■
33−39
encountered as well.
Chalcophosphates, and chalcogenides
New materials are of primary importance as technologies evolve
and more stringent requirements are demanded. Chalcophos-
phates have proven to be promising candidates in areas that
in general, are promising for NLO applications in the mid-IR, a
region in which they are commonly transparent, and are far
superior to well-studied oxides that tend to have lower
1
−3
entail such properties as ferroelectricity,
photolumines-
40
4
−6
5,7,8
efficiencies and absorption bands in this area of the spectrum.
Ternary chalcophosphates such as K P Se and KPSe show
cence,
reversible crystal-to-glass phase transitions,
9,10
2
2
6
6
superionic conductors,
and second harmonic generation
The wide structural diversity exhibited by these
compounds also merits exploration in this area since new and
4
,7,11−13
exceptionally strong SHG response in the visible/near-IR
region as compared to benchmark chalcopyrite materials
AgGaQ (Q = S, Se). KPSe has the added benefit of exhibiting
(
SHG).
14−24
2
6
interesting phases continue to be discovered.
Infrared
excellent SHG response in melt-drawn optical glassy fibers as
well as being solution processable in N,N-dimethylformamide
to obtain highly nonlinear crystalline thin films with strong
SHG and DFG (difference frequency generation) efficien-
nonlinear optical (IR-NLO) materials that can generate a
coherent beam of radiation between 3 and 14 μm are highly
sought after for use in the environmental monitoring of toxic
2
5,26
gases,
high-precision surgery via laser ablation of
detection of breath biomarkers, telecommuni-
and standoff detection of explosives. Currently
7
,11,13
2
7,28
29
cies.
Chalcogenide glasses are also an important family of
biotissue,
cations,
3
0,31
32
materials and find use in areas such as phase-change random
used IR-NLO materials such as AgGaQ (Q = S, Se) and
2
ZnGeP exhibit several of the necessary properties required of
Received: September 21, 2012
Published: November 16, 2012
2
NLO materials; however, low laser-induced damage thresholds
©
2012 American Chemical Society
20733
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
access memory (PCRAM) and solar cells. 1,41−45 Because of
lack of long-range order and macroscopic center of symmetry,
glasses typically do not possess second-order NLO properties.
Poling, however, has been utilized to prepare glasses with
3
500 °C in 6 h, and held there for 24 h at which point it was removed
from the furnace and immediately quenched in a cool water bath. EDS
on several glassy samples from the red ingot gave an average
stoichiometry of K Ge P Se .
3.6
1.0 3.8 9.2
46−48
Rb
GeP Se12(2). A mixture of 0.125 g (0.5 mmol) of Rb Se, 0.018 g
4 4 2
significant SHG responses.
(
0.25 mmol) of Ge, 0.031 g (1.00 mmol) of P, and 0.197 g (2.50
Noncentrosymmetric chalcogenides have been discovered by
adding a main group or transition metal to a ternary flux that
can serve to link the anionic units in novel ways.
Recently, we found the incorporation of Ge into a
chalcophosphate structure to be a promising route for obtaining
new stuctures and herein report our findings of a new family
of germanium chalcophosphates, both of which contain the
novel molecular anion [GeP Q ] (Q = S, Se). The sulfide
analogue crystallizes in a centrosymmetric space group, whereas
mmol) of Se was added to a 9 mm fused silica tube inside a nitrogen
filled glovebox. The tube was sealed under vacuum and subjected to
the heating profile above. An ingot containing red, rod-like crystals of
Rb GeP Se was obtained in quantitative yield. EDS analysis on
4
,12,17,49−53
4
4
12
several crystals gave an average stoichiometry of Rb3.9Ge1.0
P4.3Se11.6.
5
4
Cs GeP Se (3). A mixture of 0.212 g (0.5 mmol) of Cs Se , 0.018
4
4
12
2
2
g (0.25 mmol) of Ge, 0.031 g (1.00 mmol) of P, and 0.158 g (2.00
mmol) of Se was added to a 9 mm fused silica tube inside a nitrogen
filled glovebox. The tube was sealed under vacuum and subjected to
the heating profile above. An ingot containing red, rod-like crystals of
Cs GeP Se was obtained in quantitative yield. EDS analysis on
4−
4
12
the selenide adopts the polar space group Pca2 and exhibits a
significant NLO response in the near-IR and visible region. Our
initial experiments revealed crystalline K GeP Se has a
1
4
4
12
several crystals gave an average stoichiometry of Cs Ge P Se12.0.
3.4
1.0 4.6
4
4
12
K GeP S (4). A mixture of 0.055 g (0.5 mmol) of K S, 0.018 g
4 4 12 2
maximum SHG response ∼30 times that of the benchmark
(0.25 mmol) of Ge, 0.031 g (1.00 mmol) of P, and 0.080 g (2.50
mmol) of S was added to a 9 mm fused silica tube inside a nitrogen
filled glovebox. The tube was sealed under vacuum and subjected to
the heating profile above. An ingot containing pale yellow, rod-like
crystals of K GeP S was obtained in ∼90% yield along with a small
material AgGaSe at 730 nm. We were also able to extend our
2
region of observation to a much broader wavelength range (λ/2
=
600−1350 nm) that is nearly triple that of previous SHG
4
4 12
studies on similar compounds. Additionally, K GeP Se melts
4
4
12
amount of a black polycrystalline impurity with a similar stoichiometry
by EDS. EDS analysis on several crystals gave an average stoichiometry
of K Ge P S11.7.
congruently and was found to be a phase-change material
whose melt-quenched amorphous phase undergoes an
exothermic glass to crystal transition on heating. The local
structure is preserved in the glassy phase, which leads to a
measurable SHG response with no poling.
3.8
1.0 4.0
Rb GeP S (5). A mixture of 0.102 g (0.5 mmol) of Rb S, 0.018 g
4
4
12
2
(0.25 mmol) of Ge, 0.031 g (1.00 mmol) of P, and 0.080 g (2.50
mmol) of S was added to a 9 mm fused silica tube inside a nitrogen
filled glovebox. The tube was sealed under vacuum and subjected to
the heating profile above. An ingot containing pale yellow, rod-like
crystals of Rb GeP S was obtained in ∼90% yield along with a small
EXPERIMENTAL SECTION
■
4
4 12
Reagents. All reagents were used as obtained from the specified
supplier: potassium metal (98%, Sigma Aldrich, St. Louis, MO);
rubidium metal (99.9+%, Strem Chemicals, Inc., Newburyport, MA);
cesium metal (99.9+%, Strem Chemicals, Inc., Newburyport, MA);
germanium powder (99.99%, Sigma Aldrich, St. Louis, MO); red
phosphorus powder (99%, Sigma Aldrich, St. Louis, MO); selenium
pellets (99.99%, Sigma Aldrich, St Louis, MO); sulfur powder
amount of a black polycrystalline impurity with a similar stoichiometry
by EDS. EDS analysis on several crystals gave an average stoichiometry
of Rb Ge P S11.3.
3.4
1.0 4.3
Cs GeP S (6). A mixture of 0.149 g (0.5 mmol) of Cs S, 0.018 g
4
4
12
2
(0.25 mmol) of Ge, 0.031 g (1.00 mmol) of P, and 0.080 g (2.50
mmol) of S was added to a 9 mm fused silica tube inside a nitrogen
filled glovebox. The tube was sealed under vacuum and subjected to
the heating profile above. An ingot containing pale yellow, rod-like
crystals of Cs GeP S was obtained in ∼90% yield along with a small
(
(
99.98%, Sigma Aldrich, St. Louis, MO); N,N-dimethylformamide
ACS grade, Mallinckrodt Chemical, Phillipsburg, NJ); and diethyl
4
4 12
ether (ACS grade, BDH Chemical, Leicestershire, U.K.). K Se was
amount of a black polycrystalline impurity with a similar stoichiometry
by EDS. EDS analysis on several crystals gave an average stoichiometry
of Cs Ge P S11.1.
2
prepared by reaction of stoichiometric amounts of potassium and
55
selenium in liquid ammonia as described elsewhere. K S, Rb S, Cs S,
2
2
2
3.6
1.0 3.9
Rb Se, and Cs Se were prepared using the same procedure. P Se was
Single-Crystal X-ray Crystallography. Data collections were
performed on a STOE IPDS II diffractometer using Mo Kα radiation
(λ = 0.71073 Å) operating at 50 kV and 40 mA at 293 K. Integration
and numerical absorption corrections were performed on each
2
2
2
2
5
prepared by stoichiometric reaction of the elements in a 13 mm fused
56
silica tube at 460 °C for 24 h as described elsewhere.
Synthesis. All compounds were synthesized by combining the
proper alkali metal chalcogenide, germanium, phosphorus, and
chalcogen in the correct stoichiometric ratios in a 9 mm fused silica
tube inside a nitrogen filled glovebox. The tubes were flame-sealed
5
7
structure using X-AREA, X-RED, and X-SHAPE. All structures
were solved using direct methods and refined by full-matrix least-
2
58
squares on F using the SHELXTL program package.
−4
under vacuum (10 mbar), heated to 500 °C in 6 h, held there for 24
h, cooled to 200 °C in 24 h and then to room temperature in 3 h. The
resulting products contained crystals of the target compounds. Pure
products were obtained for the selenides using this method; however,
unidentified impurities were observed in the sulfides even after several
attempts using modified synthesis conditions and washing with polar
organic solvents such as DMF.
Powder X-ray Diffraction. Powder X-ray diffraction patterns were
collected on ground crystalline samples of each product with a flat
sample geometry using a silicon-calibrated CPS 120 INEL powder X-
ray diffractometer (Cu Kα graphite-monochromatized radiation)
operating at 40 kV and 20 mA equipped with a position-sensitive
detector. Simulated patterns were generated using the CIF of each
refined structure and the Visualizer program within FindIt.
K GeP Se (1a). A mixture of 0.079 g (0.5 mmol) of K Se, 0.018 g
0.25 mmol) of Ge, 0.031 g (1.00 mmol) of P, and 0.197 g (2.50
Scanning Electron Microscopy. Images and semiquantitative
energy dispersive X-ray spectroscopy (EDS) analyses were obtained
using a Hitachi S-3400 scanning electron microscope equipped with a
PGT energy-dipsersive X-ray analyzer. Spectra were collected using an
accelerating voltage of 25 kV and a 60 s accumulation time.
Solid State UV−Vis Spectroscopy. Diffuse reflectance spectra
were collected in the range of 200−2500 nm using a Shimadzu UV-
3101 PC double-beam, double-monochromator spectrophotometer.
The instrument was equipped with an integrating sphere and
4
4
12
2
(
mmol) of Se was added to a 9 mm fused silica tube inside a nitrogen
filled glovebox. The tube was sealed under vacuum and subjected to
the heating profile above. An ingot containing red, rod-like crystals of
K GeP Se was obtained in quantitative yield. EDS analysis on several
4
4
12
crystals gave an average stoichiometry of K Ge P Se11.7.
4.1
1.0 4.2
Glassy K GeP Se (1b). A mixture of 0.079 g (0.5 mmol) of K Se,
4
4
12
2
0
(
.018 g (0.25 mmol) of Ge, 0.031 g (1.00 mmol) of P, and 0.197 g
2.50 mmol) of Se was added to a 9 mm fused silica tube inside a
nitrogen filled glovebox. The tube was sealed under vacuum, heated to
controlled by a personal computer. BaSO was used as a standard
4
and set to 100% reflectance. Samples were prepared by placing ground
2
0734
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
Figure 1. (a) SEM micrograph of a typical K GeP Se crystal (scale bar: 100 μm). (b) PXRD of simulated (blue), as-synthesized (red), and DTA
4
4
12
products (black) of K GeP Se . (c) DTA plot of K GeP Se showing a single, reproducible melting, and crystallization event over two cycles and
4
4
12
4
4
12
congruent melting behavior. (d) Electronic optical absorption spectra of K GeP Se (black) and K GeP S (blue) indicating band gaps of 2.0 and
4
4
12
4
4 12
3.0 eV, respectively.
crystalline products on a bed of BaSO . Collected reflectance data were
amplifier (OPA), which was synchronously pumped by a frequency-
tripled output (355 nm) of a Nd:YAG laser (EKSPLA PL 2143 series).
The SHG signals were collected and dispersed with a Spex Spec-One
4
converted to absorbance according to the Kubelka−Munk equation:
2
α/S = (1 − R) /2R, where R is the reflectance and α and S are the
5
5,56
absorption and scattering coefficients, respectively.
5
00 M spectrometer coupled to a nitrogen-cooled CCD camera.
A more thorough NLO study covering an exceptionally wide range
Differential Thermal Analysis. Experiments were performed
using a Shimadzu DTA-50 thermal analyzer. Ground crystalline
of wavelengths was carried out at room temperature using a Horiba
iHR320 spectroraph equipped with a CCD camera (Synapse) detector
as well as an IGA (Symphony) dectector. Incident laser light was
produced using an EKSPLA PL-2250 series diode-pumped picosecond
Nd:YAG laser with a pulse width of 30 ps and a repetition rate of 50
Hz. An EKSPLA Harmonics Unit H400 was utilized along with an
EKSPLA PG403-SH-DFG Optical Parametric Generator (OPG) to
obtain a variety of coherent fundamental wavelengths (λ) ranging from
−4
samples (∼40 mg) were sealed under vacuum (∼10 mbar) in a fused
silica ampule. A similar amount of Al O was sealed in a separate
2
3
ampule under vacuum and used as a reference. The samples were
heated and cooled at a rate of 5 °C/min to a maximum temperature of
500 °C. Two cycles of heating and cooling were performed to check
the reproducibility of the obtained values. The melting and
crystallization points were taken as the onset of the endothermic
and exothermic peaks, respectively.
1
200 to 4000 nm at increments of 100 nm. The CCD detector was
Raman Spectroscopy. Raman spectra of ground crystalline and
glassy samples were collected on a DeltaNu Advantage NIR
spectrometer equipped with a CCD detector using 785 nm radiation
from a diode laser. The samples were loaded into borosilicate glass
capillaries for the measurement. A max power of 60 mW and beam
diameter of 35 μm were used. The spectrum was collected using an
integration time of 10 s.
used to measure NLO signals in the range of 600−1050 nm, and the
IGA detector was employed in the 1100−2000 nm range.
Spectrometer gratings of 1800, 600, and 300 grooves/mm were used
for measured wavelengths in the regions of 400−850, 850−1400, and
1400−2000 nm, respectively. The output laser energy was tuned to 20
μJ using a linear polarizer before being focused onto powdered
samples in capillary tubes by a conventional focusing lens (f = 75
mm). The corresponding spot size was ∼0.5 mm in diameter. For the
fundamental wavelengths inaccessible with 20 μJ, the NLO counts
were properly scaled in accordance with the measured SHG power
dependence (see, for example, Figure 10 at λ = 1200 nm). Our
experimental range was limited up to 2700 nm due to significant
absorption of the incident beam at the focusing lens. Also, the data at λ
31P MAS Solid-State NMR. 31P magic angle spinning (MAS)
NMR spectra were collected using a Varian VNMRS 400 MHz NMR.
Chemical shifts are referenced to NH H PO (δ 0.8 ppm), and
4
2
4
coupling constants (J) are given in Hz. Pure samples of the crystalline
and glassy forms of K GeP Se were loaded into a 5 mm zirconia
4
4
12
rotor, and experiments were performed using a pulse width (pwx90) of
.5 μs and relaxation delay of 5 min. A total of 64 and 128 scans were
3
collected for the crystalline and glassy samples, respectively, with a spin
rate of 10 000 rpm for both samples.
=
2200 nm were omitted because the pulse energy was too low.
Radiated NLO signals passed through a combination of collection
lenses and were guided to the spectrograph through a fiber optic
bundle. SHG signals were recorded as a function of wavelength for one
sample of each material with similar particle sizes (53−63 μm). SHG
responses were then measured for all samples at λ = 1800 nm for
particle size dependence. The relative NLO signals in a broad
wavelength range were precisely calibrated with the known and
measured efficiencies of all optical components.
Nonlinear Optical (NLO) Measurements. Pure samples of each
selenide analogue were ground with a mortar and pestle to give a
powder with a wide range of particle sizes. The powders were then
mechanically sieved, resulting in eight different particle size ranges
between 20 and 150 μm. Initial experiments were performed at room
temperature using a reflection geometry by measuring the SHG
responses from our samples as a function of the fundamental
wavelength (idler beam: 1250−1800 nm) from an optical parametric
2
0735
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
Table 1. Crystallographic Refinement Details for Selenide Compounds at 293(2) K
K GeP Se
Rb GeP Se
Cs GeP Se
4 4 12
4
4
12
4
4
12
formula weight
crystal system
space group
1300.39
1485.87
1675.63
orthorhombic
Pca21
crystal color
red
unit cell dimensions
a = 13.9465(4) Å, α = 90.00°
b = 7.2435(2) Å, β = 90.00°
c = 24.0511(9) Å, γ = 90.00°
2429.68(17)
4
a = 14.1881(5) Å, α = 90.00°
b = 7.3566(3) Å, β = 90.00°
c = 24.2914(15) Å, γ = 90.00°
2535.4(2)
a = 14.3268(7) Å, α = 90.00°
b = 7.6202(3) Å, β = 90.00°
c = 24.6301(14) Å, γ = 90.00°
2688.9(2)
volume (Å3)
Z
3
ρ, g/cm
μ, mm−
3.555
3.893
4.139
1
20.174
26.305
22.947
F(000)
2304
2592
2880
θ range for data collection
reflections collected
independent reflections
refinement method
data/restraints/parameters
GOF
2.81−29.00°
22 093
2.87−29.17°
9700
2.67−29.20°
9471
5819 [Rint = 0.0727]
full-matrix least-squares on F2
5819/1/190
1.108
5645 [Rint = 0.0700]
5645 [Rint = 0.0682]
5645/1/190
1.035
5645/1/191
0.998
final R indices [>2σ(I)], R /wR
0.0349/0.0877
0.0443/0.1178
0.008(17)
0.0561/0.0627
0.0878/0.0689
0.022(19)
0.0540/0.0703
0.0853/0.0767
0.492(18)
1
2
a
R indices (all data), R1/wR2
Flack parameter
−
3
−3
−3
largest diff. peak and hole
0.874 and −1.525 e Å
1.248 and −1.081 e Å
1.091 and −1.042 e Å
a
2
2
2
4
1/2
2
2
o
2
2
2
c
R = ∑||F | − |F ||/∑|F |, wR = {∑[w(|F | − |F | ) ]/∑[w(|F | )]} and calc w = 1/[σ (F ) + (0.0207P) + 0.0000P], where P = (F + 2F )/3.
1
o
c
o
2
o
c
o
o
Figure 2. Discrete anionic unit (a) and extended unit cell viewed down the b-axis (b) of K GeP Se with thermal ellipsoids set at 90%. The polar
4
4
12
nature of the structure can be seen in the pointing of the tetrahedral units down the c-axis. Color code: K, dark blue; Ge, light blue; P, black; Se, red;
gray polyhedra, GeSe tetrahedral units.
4
RESULTS AND DISCUSSION
Supporting Information). The sulfides, however, all showed
evidence of an unknown impurity by PXRD, and several
attempts to modify the syntheses led to the same results
■
Synthesis and Crystal Structure Description. Red, rod-
like crystals of K GeP Se were first obtained from the reaction
4
4
12
(
Figure S3, Supporting Information). Washing the as-
of K Se, Ge, P Se , and Se in a 3:6:1:12 molar ratio at 850 °C
2
2
5
synthesized products with polar organic solvents such as
N,N-dimethylformamide, methanol, and acetonitrile also failed
to affect their purity. Energy-dispersive X-ray spectroscopy
for 24 h followed by cooling to 200 °C in 100 h. Once the
structure was solved and the elemental ratios determined,
stoichiometric reactions were performed for the K, Rb, and Cs
analogues, as well as for the corresponding sulfides, at a soaking
temperature of 500 °C followed by cooling to 200 °C in 24 h
and to room temperature in 3 h. Pure phases were obtained for
the selenides as confirmed by the excellent match between the
calculated and experimental powder X-ray diffraction (PXRD)
patterns (Figures 1b, S1, S2, Supporting Information).
Differential thermal analysis (DTA) experiments show a single
endo- and exothermic peak at 415 and 372 °C for K GeP Se ,
(
EDS) on multiple crystals of each compound gave
stoichiometries close to those obtained from the single-crystal
refinement.
The selenides are stable in air for several weeks, after which
slight decomposition is observed on the surface of the crystals.
Solubility tests were performed on the selenides under a
nitrogen atmosphere. They did not dissolve or decompose in
methanol or acetonitrile over the course of several days but
decomposed in several hours in N,N-dimethylformamide, N-
methylformamide, and formamide and immediately decom-
4
4
12
4
16 and 414 °C for Rb GeP Se , and 426 and 406 °C for
4 4 12
Cs GeP Se , verifying their phase purity (Figures 1c, S1, S2,
4
4
12
2
0736
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
posed in hydrazine. Interestingly, the selenides dissolve and
form a yellow solution in deionized water; however, the yellow
color fades to give a colorless solution within an hour.
Therefore, the compounds are water-soluble, but the
The polarity of the structure, which can be seen in the
arrangement of the apexes of the GeSe tetrahedra that point
4
along the c-axis at a ∼47° angle and is consistent with the polar
point group mm2, arises from the 2 screw axis along [001] and
1
4
−
[
GeP Se ] anion is most likely dismantled as the P−Se
the absence of a perpendicular mirror plane (Figure 2b). Each
of the discrete molecules stacks directly on top of another along
the b-axis with three short nonbonding intermolecular Se···Se
interactions of 3.358(2), 3.640(1), and 3.720(2) Å forming a
pseudo one-dimensional structure along the [010] direction
(Figure 3a, Table 3). These distances are less than 3.80 Å, that
4
12
bonds and Ge−Se bonds hydrolyze. The observed insolubility
of the selenides in the organic solvents may be due to the high
charge on each of the anions as observed previously in similar
2
3,24,59
systems.
Since the selenides are isostructural to one another,
K GeP Se will be discussed as a representative example.
4
4
12
The compound crystallizes in the polar orthorhombic space
group Pca2 (No. 29), which belongs to the polar point group
1
mm2 of the mmm Laue class, with axes of a = 13.9465(4) Å, b =
6
0,61
7.2435(2) Å, and c = 24.0511(9) Å (Table 1).
The
structure contains the new discrete molecular unit
4
−
4+
[
GeP Se ] that is built of a central Ge ion in a tetrahedral
4 12
4−
environment formed by two bidentate ethane-like [P Se ]
2
6
units (Figure 2a). All Ge−Se bonds are normal and fall in the
narrow range between 2.354(2) and 2.378(1) Å (Table 2). The
Table 2. Selected Bond Lengths (Å) and Angles (deg) for
K GeP Se at 293(2) K with Estimated Standard Deviations
4
4
12
in Parentheses
Se(1)−P(1)
Se(2)−P(1)
Se(3)−P(1)
Se(3)-Ge
2.121(3)
2.132(3)
2.316(3)
Se(10)−Ge−Se(7)
Se(6)−Ge−Se(7)
Se(3)−Ge−Se(7)
109.06(6)
115.52(5)
107.55(5)
118.03(13)
108.29(12)
106.68(13)
111.79(12)
107.17(11)
103.86(12)
117.22(12)
107.31(14)
112.49(13)
108.67(12)
108.38(12)
101.59(12)
2.3645(14) Se(1)−P(1)−Se(2)
Se(4)−P(2)
Se(5)−P(2)
Se(6)−P(2)
Se(6)-Ge
2.120(3)
2.144(3)
2.291(3)
Se(1)−P(1)−P(2)
Se(2)−P(1)−P(2)
Se(1)−P(1)−Se(3)
2.3631(14) Se(2)−P(1)−Se(3)
2.297(3) P(2)−P(1)−Se(3)
2.3778(14) Se(4)−P(2)−Se(5)
2.134(3)
2.135(3)
2.320(3)
Se(7)−P(4)
Se(7)−Ge
Se(8)−P(4)
Se(9)−P(4)
Se(10)−P(3)
Se(10)−Ge
Se(11)−P(3)
Se(12)−P(3)
Ge−Se(6)
Se(4)−P(2)−P(1)
Se(5)−P(2)−P(1)
Se(4)−P(2)−Se(6)
Figure 3. Intermolecular Se···Se interactions shown as dotted lines (a)
along the b-axis forming a pseudo one-dimensional chain and (b)
along the c-axis, with thermal ellipsoids set at 90%. K cations have been
omitted for clarity. Color code: Ge, light blue; P, black; Se, red.
2.3542(15) Se(5)−P(2)−Se(6)
2.116(3)
P(1)−P(2)−Se(6)
Se(11)−P(3)−Se(12) 118.90(12)
2.143(3)
Table 3. Selected Intermolecular Se···Se Interactions (Å) at
2.3631(14) Se(11)−P(3)−P(4)
107.53(13)
2
93(2) K with Estimated Standard Deviations in Parentheses
P(1)−P(2)
P(2)−Se(5)
P(2)−P(1)
P(3)−Se(11)
P(3)−P(4)
P(4)−Se(8)
P(4)−Se(9)
Se(10)−Ge−Se(6) 103.56(5)
Se(10)−Ge−Se(3) 114.97(5)
Se(6)−Ge−Se(3)
2.260(3)
2.144(3)
2.260(3)
2.116(3)
2.240(3)
2.134(3)
2.135(3)
Se(12)−P(3)−P(4)
Se(11)−P(3)−Se(10) 110.28(11)
Se(12)−P(3)−Se(10) 108.05(12)
P(4)−P(3)−Se(10)
Se(8)−P(4)−Se(9)
Se(8)−P(4)−P(3)
Se(9)−P(4)−P(3)
Se(8)−P(4)−Se(7)
Se(9)−P(4)−Se(7)
P(3)−P(4)−Se(7)
110.94(13)
a
for A GeP Se
4
4
12
label
K GeP Se
Rb GeP Se
Cs GeP Se
12
direction
4
4
12
4
4
12
4
4
99.39(11)
116.36(12)
111.40(13)
107.19(12)
109.81(11)
108.15(11)
103.05(12)
Se(9)−Se(11)
Se(7)−Se(10)
Se(1)−Se(2)
Se(1)−Se(11)
Se(4)−Se(11)
Se(1)−Se(8)
3.3582(15)
3.6404(14)
3.7198(16)
3.6709(16)
3.7070(17)
3.7120(16)
3.4037(18)
3.7027(18)
3.791(2)
3.720(2)
3.785(2)
3.835(2)
3.536(2)
3.896(2)
4.020(3)
3.868(2)
3.948(2)
4.046(2)
b
b
b
c
c
c
a
106.39(5)
Direction identifies the axis along which the interaction occurs.
is, twice the van der Waals radii of Se, and interactions such as
these have been previously observed in stabilizing the packing
P−Se bonds are also typical but can be divided into two groups,
terminal and bridging, with the expected observation of the P−
Se bonds being ∼0.18 Å longer than the P−Se bonds on
1
2,62,63
of molecular and polymeric selenophosphates.
Short
nonbonding Se···Se interactions are also found between
molecules along the c-axis, although their average distance is
∼0.12 Å longer than those along the b-axis. The distances
between Se atoms along [001] are 3.671(2), 3.707(2), and
3.712(2) Å (Table 3). Potassium cations fill the voids between
the discrete molecules to balance the negative charge with K−
Se distances ranging from 3.265(4) to 3.903(4) Å. The alkali
b
t
average. P−Se bonds range from 2.116(3) to 2.144(3) Å, while
t
those for P−Se are between 2.291(3) and 2.320(3) Å. The
b
angles around Ge and P are close to those expected for
tetrahedral coordination and are in the ranges of 103.56(5)−
1
15.52(5)° and 99.4(1)−118.0(1)° for the two atoms,
respectively.
2
0737
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
Table 4. Crystallographic Refinement Details for Sulfide Compounds at 293(2) K
K GeP S
Rb GeP S
Cs GeP S
4 4 12
4
4
12
4
4
12
formula weight
crystal system
space group
737.59
923.07
1112.83
monoclinic
C2/c
crystal color
yellow
unit cell dimensions
a = 12.1695(8) Å, α = 90.00°
b = 7.8377(5) Å, β = 100.189(6)°
c = 23.1846(17) Å, γ = 90.00°
2176.5(3)
a = 12.3217(6) Å, α = 90.00°
b = 8.0462(3) Å, β = 100.622(4)°
c = 23.5743(11) Å, γ = 90.00°
2297.17(18)
a = 12.6475(6) Å, α = 90.00°
b = 8.3020(3) Å, β = 102.012(4)°
c = 24.2428(13) Å, γ = 90.00°
2489.7(2)
volume (Å3)
Z
4
3
ρ, g/cm
μ, mm−
2.251
2.669
2.969
1
3.603
11.110
8.246
F(000)
1440
1728
2016
θ range for data collection
reflections collected
independent reflections
refinement method
data/restraints/parameters
GOF
1.78−25.00°
4569
1.76−29.22°
10 664
1.72−29.20°
11 630
1917 [Rint = 0.0565]
full-matrix least-squares on F2
1917/0/96
3083 [Rint = 0.0453]
3366 [Rint = 0.0396]
3083/0/96
1.136
3366/0/96
1.026
1.058
final R indices [>2σ(I)], R /wR
0.0482/0.0639
0.0771/0.0691
0.0403/0.0507
0.0609/0.0536
0.0368/0.0591
0.0606/0.0635
1
2
a
R indices (all data), R1/wR2
−3
−3
−3
largest diff. peak and hole
0.469 and −0.421 e Å
0.520 and −0.582 e Å
0.587 and −1.187 e Å
a
2
2
2
4
1/2
2
2
o
2
2
o
2
c
R = ∑||F | − |F ||/∑|F |, wR = {∑[w(|F | − |F | ) ]/∑[w(|F | )]} and calc w = 1/[σ (F ) + (0.0207P) + 0.0000P], where P = (F + 2F )/3.
1
o
c
o
2
o
c
o
Figure 4. Discrete anionic unit (a) and extended unit cell viewed down the b-axis (b) of K GeP S with thermal ellipsoids set at 70%. Color code: K,
4
4 12
dark blue; Ge, light blue; P, black; S, yellow.
metal cations are either 7 or 8 coordinate, and all have
somewhat irregular coordination geometries (Figure S4,
Supporting Information).
in the monoclinic space group C2/c (No. 15) with axes of a =
12.1695(8) Å, b = 7.8377(5) Å, c = 23.185(2) Å, and β =
100.189(6)° (Table 4). Although the molecular anion is the
same as that found in the selenium compounds, Ge sits on the
special position (0, y, 1/4) of the C2/c space group (Wyckoff
position 4e, site symmetry 2) in K GeP S (Figure 4). The
A closer look at the other selenide analogues reveals the
expected trend of increasing Se···Se distances along the b- and
c-axes as the size of the alkali metal increases. Although the
absolute changes are similar for the interactions along the two
different axes, the percent change is greater for the Se···Se
interactions along the b-axis. Also, the percent increase in the
length of the b-axis is much greater as compared to the other
two axes as the size of the alkali metal increases. The changes in
the b-axis are 1.6% and 3.6%, whereas those for the a-axis are
4
4 12
second coordinating P S ligand of the anion is generated by a
2
6
2-fold rotation axis that passes through the special position.
K GeP S exhibits types of trends in its bond lengths and
4
4 12
angles similar to its selenide analogue. The Ge−S bond lengths
are 2.210(1) and 2.232(1) Å, and the P−S bond lengths are
1.950(2), 1.964(2), 1.966(2), and 1.982(2) Å for P−S and
t
1
.7% and 1.0% and for the c-axis 1.0% and 1.4% in going from
2.147(2) and 2.161(2) Å for P−S (Table S1). The angles
b
+
+
+
+
the K to the Rb and the Rb to the Cs analogue, respectively.
The corresponding percent increase in the ionic radii of the
around Ge and P are also similar and range from 98.06(7)° to
122.42(8)°. The discrete molecules again stack directly on top
of one other along the b-axis; however, there are no
intermolecular S···S interactions less than twice the van der
Waals radii (i.e., 3.60 Å) between molecules along [010]. One
short intermolecular S···S interaction of 3.431(2) Å is found
along the [100] direction though.
64
alkali metal cations is 6.6% and 8.1%. Therefore, it seems as
though the pseudo one-dimensional chain along the b-axis
becomes more molecular in nature as the alkali metal gets
larger and decreases the intereaction between neighboring
molecules.
The sulfides are also isostructural to one another, and only
Thermal Properties and Glass Formation. Differential
thermal analysis (DTA) experiments were performed on the
the potassium analogue will be discussed. K GeP S crystallizes
4
4 12
2
0738
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
three selenide analogues and confirm their phase purity. Single
reproducible melting and crystallization points were observed
when performing multiple cycles of DTA, and analysis of the
products by PXRD reveals an excellent match between the
calculated and pristine patterns, indicating their congruently
melting behavior (Figures 1, S1, S2, Supporting Information).
This, along with the relatively low melting and crystallization
temperatures (415 and 372 °C for K GeP Se , 416 and 414 °C
4
4
12
for Rb GeP Se , and 426 and 406 °C for Cs GeP Se ),
4
4
12
4
4
12
suggests that large crystals for further optical characterization
can be grown using well-known growth techniques such as the
Bridgman method.
Given the congruently melting behavior of these phases, a
reaction for K GeP Se was quenched from 500 °C into a cool
4
4
12
water bath to investigate its glass forming ability. PXRD analysis
of the quenched ingot (1b) revealed only broad peaks whose
profile fits that of the crystalline compound, indicating that the
local structure is most likely maintained but the long-range
packing of the molecules has been lost (Figure 5a).
Figure 6. Differential thermal analysis (DTA) of glassy K GeP Se .
4
4
12
(a) A single cycle to 450 °C (1c). (b) Two cycles to 450 °C (1d). (c)
A single run to 325 °C (1e). (d) A single DTA run stopped just after
passing through the first exotherm on heating (1f).
Figure 7. (a) PXRD patterns of the simulated crystalline K GeP Se
12
4
4
(
(
pink), and DTA products 1c (green), 1d (blue), 1e (red), and 1f
black). The positions of the peaks remain unchanged for all of the
samples, although their relative intensities change and have a higher
background for 1e and 1f. (b) Raman spectra for pristine K GeP Se
4
4
12
(
1a, purple), glassy (1b, gray), 1c (green), 1e (red), and 1f (black).
Figure 5. (a) Experimental PXRD pattern of a melt-quenched glass
K GeP Se sample (1b, black) and simulated pattern for crystalline
The positions of the Raman shifts remain fairly constant over all
samples and do not broaden for the glassy phase.
4
4
12
K GeP Se (red). (b) Electronic absorption spectrum of glassy
4
4
12
K GeP Se . (c) SEM micrograph of glassy K GeP Se .
4
4
12
4
4
12
phase, demonstrating the ability of K GeP Se to convert from
4
4
12
Investigation of the red-orange glassy product by EDS showed
the presence of all elements with an average composition of
K Ge P Se10.9. Thermal analysis of the glassy phase gave
the glassy to the crystalline phase without melting and
41,65
establishing it as a phase-change material.
Interestingly,
the PXRD pattern of a fourth sample of glassy K GeP Se (1f)
3.5
1.0 3.8
4
4
12
unexpected results, and several different DTA experiments were
run to investigate their products (1c−1f) by PXRD (Figures 6
and 7a). A single cycle of DTA to 450 °C shows two exotherms
at 230 and 245 °C on heating, followed by endothermic
melting at 410 °C, and crystallization on cooling at 398 °C.
PXRD of the DTA product (1c) reveals the crystalline nature
of the product and matches very well with the calculated
pattern of K GeP Se . Two consecutive DTA cycles to 450 °C
heated through only the first exotherm in the DTA has peaks at
the same 2θ values but with different intensities and a more
amorphous background. It seems as though the glassy phase
goes through a metastable crystalline state at a temperature
slightly below that which results in full conversion to the most
thermodynamically stable crystalline state.
Optical Absorption Spectroscopy. Optical diffuse
reflectance spectra were recorded for the pure selenide
compounds and the sulfide analogues containing impurities.
Despite the size of the alkali metal ion increasing by ∼15% in
4
4
12
of a second sample of the glassy phase showed the same
behavior as before for the first cycle. The second cycle is similar
to that of the slow-cooled K GeP Se and shows only a single
+
+
going from K to Cs , the band gap remains at a constant value
of 2.0 eV for all three selenides (Figures 1, S1, S2, Supporting
Information). Therefore, the transitions between the valence
and conduction bands must be based solely on the molecular
anion. This is confirmed by the band gap of the glassy phase of
K GeP Se , which is slightly higher at 2.1 eV (Figure 5b).
4
4
12
melting and crystallization at 412 and 395 °C, respectively.
Again, the PXRD of the DTA product (1d) is crystalline and
matches with the calculated pattern. Heating a third sample of
the glassy material to 325 °C confirms the reproducibility of the
two exotherms on heating. PXRD of the DTA product (1e)
matches well with the calculated pattern of the crystalline
4
4
12
Typically, energy band gaps are reduced when going from a
2
0739
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
crystalline to a glassy phase as defect and midgap states are
305 Hz for δ 61.16, and J = 320 Hz for δ 56.81). The slight
differences in chemical shifts and coupling constants are
believed to arise from the different second nearest neighbor
environments (e.g., Se−K bonds and short nonbonding Se···Se
interactions) around each P atom, because the bond lengths
and angles of nearest neighbors are very similar.
The spectrum of the glassy phase shows a single broad
resonance at δ 64.50. Although the molecular unit was shown
to be intact in the amorphous phase by Raman spectroscopy,
the loss of long-range order causes the discrete chemical shifts
of the 31P spectrum for the crystalline phase to inhomoge-
neously broaden into one.
66
introduced near the band edges. The fact that the band gap
remains fairly constant as the long-range order is lost in the
glassy phase verifies that the discrete molecular salt structure is
retained. The band gaps of the sulfides remain fairly constant
for the different alkali metal analogues as well (3.0 eV for K, Rb;
3
.1 eV for Cs) (Figures 1, S5, Supporting Information).
Raman spectra were recorded for pristine K GeP Se (1a) as
4
4
12
well as for several DTA products from the experiments on
glassy K GeP Se (1b, 1c, 1e, and 1f) (Figure 7b). Relatively
4
4
12
sharp peaks at similar wavenumbers are observed in all of the
spectra, and no broadening occurs for the glassy phase, which
4
−
provides further evidence that the [GeP Se ] molecular
4
12
Nonlinear Optical (NLO) Properties and Second
Harmonic Generation. The mm2 crystal class to which the
anion remains intact in the glassy phase. The shifts in the
Raman spectra can be divided into three groups: the weak
space group Pca2 belongs prompted us to perform second
1
−
1
peaks at ∼156 and 175 cm are attributed to P Se and GeSe
2
6
4
harmonic generation (SHG) studies on the selenium-
−
1
6
0,61
bends, the strongest peaks at ∼206 and 215 cm are attributed
to the GeSe and P Se symmetric stretches, and those above
00 cm arise from various asymmetric stretches of GeSe and
4
8
containing compounds.
Structures of this point group are
4
2
6
pyroelectric, optically active, and piezoelectric (i.e., exhibit
SHG). A modified Kurtz and Perry powder method was first
used to perform a preliminary study of the SHG response as a
function of particle size for each alkali metal analogue in the
−1
3
,67−69
P Se .
2
6
3
1
P NMR Spectroscopy. Ground crystalline and glassy
samples of K GeP Se were investigated by P magic angle
3
1
70
4
4
12
range of λ/2 = 625−900 nm. Ground crystalline samples were
spinning (MAS) NMR. Since there are four crystallographically
unique P atoms with very similar environments in the structure
refinement, all of which are involved in a P−P bond, one would
expect four doublets with relatively similar chemical shift values.
At first glance, the spectrum of the crystalline phase appears to
have only three doublets centered at δ 65.78, 61.16, and 56.81;
however, the peak shape of the resonance at δ 64.97 appears to
be abnormal (Figure 8). When integrating the doublets at δ
mechanically sieved into eight different particle size ranges from
2
4
0 to 150 μm. For each compound, a peak in intensity for the
5−53 μm particles was observed followed by a steady decrease
in the SHG response as the particle sizes increased, indicating
that all three are type-I nonphase-matchable at 760 nm (Figure
S6, Supporting Information). The same trend was observed in
the range of 625−900 nm of the SHG wavelength. Nonphase-
matchable materials are characterized by a maximum response
when the particle size is close to the average coherence length
of the compound and a decrease in intensity as the particle sizes
70
become larger. Therefore, the coherence lengths, l , for the
c
three compounds are assumed to be in the range of 45−53 μm.
Because phase matchability is a wavelength-dependent
property, it is possible that these materials are phase matchable
71
at a different wavelength.
Wavelength-dependent SHG measurements from λ = 1250−
1
800 nm were performed on the 45−53 μm particles, and the
recorded intensities were compared to those collected for
similar size particles of the benchmark IR NLO material
AgGaSe (Figure 9). AgGaSe is also nonphase-matchable at
2
2
these wavelengths, with the well-established SHG coefficient
(
2)
72
χ
= 66 pm/V, and is a suitable reference for our materials.
The observed response of each chalcophosphate was
Figure 8. 31P solid-state magic angle spinning (MAS) NMR spectra for
crystalline (a) and glassy (b) K GeP Se .
4
4
12
6
5.78, 61.16, and 56.81, values of 2.37, 1.28, and 1.00 are
obtained, respectively. Therefore, the spectrum can be
explained as having two doublets that overlap almost perfectly
and two doublets that are shifted upfield by ∼4 and 9 ppm. It is
difficult to assign the four crystallographically unique P atoms
to the different doublets, but their chemical shifts are in the
Figure 9. Relative SHG intensity of A GeP Se (A = K, Rb, Cs) as
4
4
12
compared to AgGaSe as a function of wavelength. Intensities have
2
expected region when compared to the similar molecular anion
been normalized to AgGaSe at each collected wavelength. Code:
2
8
in Cs P Se . The coupling constants for the overlapped
K GeP Se ( ); Rb GeP Se ( ); Cs GeP Se (▲); AgGaSe (▼).
■
●
5
5
12
4 4 12 4 4 4 4 12 2
12
doublets are difficult to measure but are estimated to be similar
to those for the other two doublets (J ≈ 305 Hz for δ 65.78, J =
The dotted line represents the absorption edge of AgGaSe (∼1.7 eV),
2
below which a proper comparison of intensities cannot be made.
2
0740
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
normalized to that of AgGaSe2 at each wavelength for
comparison. K GeP Se exhibited the strongest signal over
this entire wavelength range, followed by the Rb and Cs
for the intensities at ω and 2ω considering the second-order
macroscopic NLO polarization of the material and TPA
simultaneously. This task usually requires numerical treatments
because the NLO effects are described in terms of the electric
field, whereas the TPA effects are described by the intensity.
However, it is possible to estimate β by fitting the observed
power dependence based on a simplified model with a modified
fundamental intensity by TPA:
4
4
12
analogues. At 730 nm the K, Rb, and Cs analogues are ∼30, 9,
and 5 times stronger than AgGaSe . To the best of our
2
knowledge, the SHG response of the K analogue is the
strongest reported among quaternary chalcophosphates in this
wavelength range. Although the relative intensities of the
reported compounds are much greater at shorter wavelengths,
comparisons below ∼720 nm cannot be made due to the strong
dI(ω)
dz
I(ω)
1 + βI(ω)d
2
=
−βI (ω); I(ω) →
3
3
absorption of AgGaSe (E ≈ 1.7 eV).
(2)
2
g
Because of the strong response of K GeP Se , a more
4
4
12
where d is the powder size in the range of 53−63 μm. The red
curve in Figure 10b corresponds to a least-squares fit to SHG
power dependence with a single fit parameter of β = 30.7 ± 2.3
cm/GW, where the uncertainty arises from the uncertainty in
the particle size. We found that this β value is reasonable
considering that values for other conventional chalcogenide
materials range from a few cm/GW to several hundred cm/
thorough NLO study was performed covering a much wider
wavelength range, that is, 1200−2700 nm for the fundamental
wavelength. A power-dependent SHG experiment was
conducted at λ = 1200 nm where two-photon absorption
(
TPA) of the fundamental beam can affect the observed SHG
efficiency of our sample (E = 2.0 eV). TPA is a nonparametric
g
NLO process and is characterized by the coefficient, β, which is
33
GW. For NLO applications involving beam mixing, signal
proportional to the imaginary part of the third-order NLO
(3)
amplification, and Kerr-effect-based optical switches, materials
with minimal TPA, that is, low β values, are desirable.
Broadband SHG experiments were then performed on the
A GeP Se compounds in the range of λ = 1200−2700 nm,
tensor, χ . The measured SHG responses obtained from
crystalline K GeP Se when the incident pulse energy was
4
4
12
varied from 0 to 30 μJ at a fundamental wavelength of 1200 nm
are shown in Figure 10a. The SHG responses as a function of
4
4
12
with a corresponding SHG wavelength, λ/2, of 600−1350 nm.
Our experimental setup is quite unique in that previous studies
have typically involved a fixed or limited wavelength range. We
have performed experiments over a range that is almost triple
that of reports on similar compounds. Although we did not use
any band-pass filter for eliminating the pump beam in our SHG
measurements, we confirmed that any NLO signal from the
capillary tube or any other optical component is negligible in
our experimental range. The NLO response of each
chalcophosphate was again directly compared to that from
AgGaSe2.
The series of blue traces in Figure 11 show the observed
SHG signals from crystalline K GeP Se powders as a function
of wavelength; the red traces correspond to the SHG signals
Figure 10. (a) SHG counts for 10, 20, and 30 μJ excitation pulse
energies for crystalline K GeP Se at λ = 1200 nm. (b) The
corresponding power-dependence plot. The black curve is the
predicted power dependence of the SHG when two-photon
absorption is absent. The red curve is a fit to the SHG power
dependence with β = 30.7 cm/GW.
4
4
12
4
4
12
from AgGaSe under the same experimental conditions. The
2
pulse energy are presented in Figure 10b. Under negligible
pump depletion, and in the slowly varying envelope
approximation, the expression for the SHG intensity, I(2ω, l),
with l being the interaction length, is given by
2
(2) 2 2
ω |χ | l
⎛Δkl ⎞
sinc2
I (ω) ∝ |χ | I (ω)
2
(2) 2 2
I(2ω, l) = 2
⎜
⎝
⎟
⎠
2
3
n n c ε
2
2
ω ω
0
(1)
where nω and n2ω are refractive indices at ω and 2ω,
respectively, c is the speed of light in vacuum, ε is the vacuum
0
dielectric constant, Δk is the phase mismatch, and I(ω) is the
pump intensity. In our powder experiments based on a
reflection geometry, the interaction length is the powder size,
d. The solid black curve in Figure 10b corresponds to eq 1. The
noticeable deviation from this simple prediction indicates that
there is significant pump depletion at λ = 1200 nm. This most
likely arises due to TPA of the pump beam because the
fundamental wavelength is slightly over the band-to-band two-
photon resonance. To rigorously explain the observed power
dependence, one needs to solve coupled differential equations
Figure 11. Wavelength-dependent SHG response for crystalline
K GeP Se (blue) and AgGaSe (red) from λ = 1200−2700 nm.
4
4
12
2
(b) and (c) are zoomed in regions of (a).
2
0741
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
first result that is noticed is the sharp decrease in the SHG
response when λ/2 approaches the optical band gaps of
K GeP Se and AgGaSe , ∼620 and 720 nm, respectively. This
that the coherence length, l , at this wavelength is similar to
what was found previously and is about 50 μm. We also
c
confirmed AgGaSe is type-I nonphase-matchable at 1800 nm
4
4
12
2
2
occurs due to strong absorption of the produced SHG light
near the band gap as well as TPA of the pump beam. Similarly,
the decrease in the SHG responses in the range of λ/2 = 1000−
(Figure 13b). Because of comparable SHG counts with
significant uncertainties in the 20−40 μm size range, we
could not determine the corresponding coherence length of
AgGaSe . However, we directly calculated l = λ/(4|n − n |)
1
200 nm can be attributed to higher-order multiphoton
2
c
2ω
ω
absorption effects. Although these experiments confirm our
≈ 5 μm using the measured birefringent refractive indices for
33
initial results that K GeP Se is a much stronger SHG material
AgGaSe available elsewhere. This small coherence length
4
4
12
2
than AgGaSe in the range of λ/2 = 600−900 nm, we see that
cannot be inferred from our powder method because the
2
this is not true at all longer wavelengths: AgGaSe outperforms
minimum accessible particle size is ∼20 μm. AgGaSe becomes
2
2
K GeP Se at 1000 and 1050 nm; however, our material
phase-matchable at λ > 3000 nm, which is beyond our
4
4
12
slightly outperforms AgGaSe at 1150 and 1200 nm before their
experimental range.
2
(
2)
responses are essentially equal at 1250 nm and AgGaSe2
becomes stronger at 1350 and 1400 nm. These results
emphasize the importance of performing broadband studies
because SHG responses vary widely with wavelength. Measure-
ment at λ/2 = 1100 nm was omitted due to low pulse energy of
the laser.
We estimated the absolute value of χ for Cs GeP Se
4 4 12
using eq 3 for the nonphase-matching case using χ( = 66 pm/
2)
ref
V:
1
/2
2
lc,Ref ⎛ I (2 ) ⎞
l
2
ω
(2) 2
c
(2)
(2)
Cs
I(2ω) ∝ |χ
|
;
|χ | = |χRef
|
⎜
⎟
Cs
d
lc,Cs ⎝ I (2ω)⎠
Ref
Figure 12 shows the relative SHG signal of all A GeP Se
4
4
12
(3)
compounds and AgGaSe on a semilog plot. Again, the decrease
2
(
2)
yielding |χ | ≈ 6.1 pm/V for Cs GeP Se . According to
4
4
12
(
2)
Figure 12, this implies that |χ | ≈ 7.1 and 11.1 pm/V for
Rb GeP Se and K GeP Se , respectively. We believe our
4
4
12
4
4
12
estimations are correct within a factor of 2, where the
uncertainty arises mainly from inhomogeneous distribution of
powders with different sizes within the specified range.
Since glassy K GeP Se was shown to retain its short-range
4
4
12
order, SHG measurements were performed from λ/2 = 600−
350 nm (Figure 14). Although the relative efficiency is about 2
1
Figure 12. Broadband SHG responses of AgGaSe (▼), K GeP Se
12
2
4
4
(
nm.
■), Rb GeP Se (●), and Cs GeP Se (▲) from λ = 1200−2700
4
4
12
4
4
12
in signal at wavelengths corresponding to the band gaps of the
materials and multiphoton absorptions are observed.
K GeP Se is confirmed to be the strongest material among
4
4
12
the three reported chalcophosphates. On the basis of the
measured wavelength-dependent NLO responses, we conclude
Figure 14. Wavelength-dependent SHG response from glassy
K GeP Se .
4
4
12
(2) (2)
(2) (2)
that |χ /χ | ≈ 1.56 and |χ /χ | ≈ 1.17 according to eq 1.
K
Rb
Rb
Cs
Additionally, the absolute values of the NLO tensors can be
obtained by comparing them with the NLO efficiencies of the
reference material, as explained below.
orders of magnitude smaller than crystalline K GeP Se , the
4
4
12
glassy phase exhibits a measurable SHG response (Figure 14).
Contrary to conventional glassy materials, the innate second-
order NLO properties of our glassy phase presumably arise
because the noncentrosymmetric building blocks remain7
partially aligned, similar to what was observed in KPSe6.
Interestingly, we found that the SHG efficiencies for λ > 2000
nm are strongly suppressed. This can be qualitatively explained
by the fact that the short-range order present in the glassy
phase will become averaged out for longer wavelengths, which
in turn cancels out the SHG dipole moments.
Figure 13 shows the particle size dependence of the
measured SHG responses from Cs GeP Se and AgGaSe at
4
4
12
2
a fundamental wavelength of 1800 nm. Figure 13a indicates
CONCLUSIONS
■
[
The new family of polar selenophosphates containing the novel
4−
GeP Se ] molecular anion exhibits a significant second
4
12
harmonic generation (SHG) response in the visible and near IR
wavelength range. The response is believed to arise from the
Figure 13. Particle size-dependent SHG responses from (a)
Cs GeP Se and (b) AgGaSe measured at λ = 1800 nm.
alignment of the GeSe tetrahedra along the c-axis of the
structure. Short nonbonding Se···Se interactions help stabilize
4
4
4
12
2
2
0742
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
the polar character of the structure because similar
intermolecular interactions are not observed in the centrosym-
metric sulfur analogues. The three selenides are type-I
nonphase-matchable over the investigated wavelength range,
(5) Chung, I.; Song, J.-H.; Kim, M. G.; Malliakas, C. D.; Karst, A. L.;
Freeman, A. J.; Weliky, D. P.; Kanatzidis, M. G. J. Am. Chem. Soc.
2
(
009, 131, 16303.
6) Banerjee, S.; Szarko, J. M.; Yuhas, B. D.; Malliakas, C. D.; Chen,
L. X.; Kanatzidis, M. G. J. Am. Chem. Soc. 2010, 132, 5348.
7) Chung, I.; Jang, J.-I.; Malliakas, C. D.; Ketterson, J. B.; Kanatzidis,
M. G. J. Am. Chem. Soc. 2010, 132, 384.
8) Chung, I.; Jang, J. I.; Gave, M. A.; Weliky, D. P.; Kanatzidis, M. G.
+
and the K analogue shows a second-order NLO response ∼30
(
times that of AgGaSe at λ/2 = 730 nm, which, to the best of
2
our knowledge, is the strongest among reported quaternary
(
chalcophosphates in the range of λ/2 = 600−900 nm. It is also
Chem. Commun. 2007, 4998.
∼
20 times stronger than AgGaSe at the telecommunications
2
(9) Kamaya, N.; Homma, K.; Yamakawa, Y.; Hirayama, M.; Kanno,
R.; Yonemura, M.; Kamiyama, T.; Kato, Y.; Hama, S.; Kawamoto, K.;
Mitsui, A. Nat. Mater. 2011, 10, 682.
relevant wavelength of λ = 1.55 μm. Our broadband optical
measurements from λ = 1200−2700 nm roughly triple the
range that has been studied in the past and demonstrate how
important investigation of materials over a significantly wide
range is because of their widely varying SHG efficiencies.
K GeP Se is also a phase-change material that can be
(10) Mo, Y. F.; Ong, S. P.; Ceder, G. Chem. Mater. 2012, 24, 15.
(
11) Chung, I.; Malliakas, C. D.; Jang, J. I.; Canlas, C. G.; Weliky, D.
P.; Kanatzidis, M. G. J. Am. Chem. Soc. 2007, 129, 14996.
12) Chung, I.; Song, J.-H.; Jang, J. I.; Freeman, A. J.; Ketterson, J. B.;
Kanatzidis, M. G. J. Am. Chem. Soc. 2009, 131, 2647.
13) Chung, I.; Kim, M. G.; Jang, J. I.; He, J. Q.; Ketterson, J. B.;
Kanatzidis, M. G. Angew. Chem., Int. Ed. 2011, 50, 10867.
(
4
4
12
quenched from the melt to obtain a glassy form of the
compound that, when heated below the melting point, passes
through a metastable semicrystalline state before converting to
the crystalline state obtained when slow cooling the melt. The
amorphous phase was shown to contain intact molecular anions
by Raman spectroscopy, and the metastable phase is thought to
be an alternate packing of the anions that is slightly different
from the slow-cooled sample. SHG measurements on glassy
K GeP Se also showed measurable responses that were ∼2
(
(
(
14) Kanatzidis, M. G. Curr. Opin. Solid State Mater. Sci. 1997, 2, 139.
15) Derstroff, V.; Ensling, J.; Ksenofontov, V.; Gutlich, P.; Tremel,
W. Z. Anorg. Allg. Chem. 2002, 628, 1346.
(16) Gieck, C.; Derstroff, V.; Block, T.; Felser, C.; Regelsky, G.;
Jepsen, O.; Ksenofontov, V.; Gutlich, P.; Eckert, H.; Tremel, W.
Chem.-Eur. J. 2004, 10, 382.
(17) Wu, Y. D.; Bensch, W. Inorg. Chem. 2007, 46, 6170.
(18) Loken, S.; Tremel, W. Eur. J. Inorg. Chem. 1998, 283.
(19) Milot, S.; Wu, Y.; Nather, C.; Bensch, W.; Klepp, K. O. Z. Anorg.
4
4
12
orders of magnitude smaller than the crystalline phase with no
poling. Congruently melting molecular materials such as those
reported here may prove to be beneficial as phase-change
materials in that once the molecule is formed only a small
amount of thermal energy is required for the reorganization of
the disordered molecules into an ordered crystalline state.
Allg. Chem. 2008, 634, 1575.
20) Komm, T.; Strobel, S.; Schleid, T. J. Alloys Compd. 2008, 451,
48.
21) Piccoli, P. M. B.; Abney, K. D.; Schoonover, J. D.; Dorhout, P.
K. Inorg. Chem. 2001, 40, 4871.
22) Hess, R. F.; Abney, K. D.; Burris, J. L.; Hochheimer, H. D.;
Dorhout, P. K. Inorg. Chem. 2001, 40, 2851.
23) Chung, I.; Holmes, D.; Weliky, D. P.; Kanatzidis, M. G. Inorg.
Chem. 2010, 49, 3092.
(
6
(
(
ASSOCIATED CONTENT
■
(
*
S
Supporting Information
PXRD, DTA, and UV−vis for Rb GeP Se and Cs GeP Se ,
4
4
12
4
4
12
(24) Morris, C. D.; Kanatzidis, M. G. Inorg. Chem. 2010, 49, 9049.
PXRD and UV−vis for the sulfides, coordination environments
(25) Sorokina, I. T.; Vodopyanov, K. L. Solid-state Mid-infrared Laser
for K in K GeP Se , particle size-dependent SHG measure-
4
4
12
Sources; Springer: Berlin; New York, 2003.
26) Partnership, S. Nat. Photonics 2010, 4, 576.
ments, and crystallographic information files (CIF). This
material is available free of charge via the Internet at http://
pubs.acs.org.
(
(27) Serebryakov, V. A.; Boiko, E. V.; Petrishchev, N. N.; Yan, A. V. J.
Opt. Technol. 2010, 77, 6.
(
28) Waynant, R. W.; Ilev, I. K.; Gannot, I. Philos. Trans. R. Soc., A
001, 359, 635.
29) Skeldon, K. D.; Gibson, G. M.; Wyse, C. A.; McMillan, L. C.;
Monk, S. D.; Longbottom, C.; Padgett, M. J. Appl. Opt. 2005, 44,
712.
30) Dorn, R.; Baums, D.; Kersten, P.; Regener, R. Adv. Mater. 1992,
, 464.
31) Eggleton, B. J.; Luther-Davies, B.; Richardson, K. Nat. Photonics
011, 5, 141.
2
(
AUTHOR INFORMATION
■
Corresponding Author
m-kanatzidis@northwestern.edu
4
(
4
(
Notes
The authors declare no competing financial interest.
2
ACKNOWLEDGMENTS
■
(
(32) Mukherjee, A.; Von der Porten, S.; Patel, C. K. N. Appl. Opt.
2010, 49, 2072.
(33) Nikogosyan, D. N. Nonlinear Optical Crystals: A Complete
Survey; Springer-Science: New York, 2005.
Financial support from the National Science Foundation
Grant DMR-1104965) is gratefully acknowledged. The
SEM/EDS work was performed in the EPIC facility of
NUANCE Center at Northwestern University. NUANCE
Center is supported by the NSF-NSEC, NSF-MRSEC, Keck
Foundation, the State of Illinois, and Northwestern University.
(34) Harasaki, A.; Kato, K. Jpn. J. Appl. Phys., Part 1 1997, 36, 700.
(35) Ziegler, B. C.; Schepler, K. L. Appl. Opt. 1991, 30, 5077.
(36) Peterson, R. D.; Schepler, K. L.; Brown, J. L.; Schunemann, P.
G. J. Opt. Soc. Am. B 1995, 12, 2142.
37) Pearl, S.; Fastig, S.; Ehrlich, Y.; Lavi, R. Appl. Opt. 2001, 40,
(
REFERENCES
1) Carpentier, C. D.; Nitsche, R. Mater. Res. Bull. 1974, 9, 1097.
2) Rogach, E. D.; Sviridov, E. V.; Arnautova, E. A.; Savchenko, E. A.;
■
2490.
(
(
(38) Petrov, V.; Rotermund, F.; Noack, F.; Schunemann, P. Opt. Lett.
1999, 24, 414.
Protsenko, N. P. Zh. Tekh. Fiz. 1991, 61, 201.
3) Scott, B.; Pressprich, M.; Willet, R. D.; Cleary, D. A. J. Solid State
Chem. 1992, 96, 294.
4) Banerjee, S.; Malliakas, C. D.; Jang, J. I.; Ketterson, J. B.;
Kanatzidis, M. G. J. Am. Chem. Soc. 2008, 130, 12270.
(39) Song, J.; Xia, J. F.; Zhang, Z.; Strickland, D. Opt. Lett. 2002, 27,
200.
(40) (a) Sasaki, T.; Mori, Y.; Yoshimura, M.; Yap, Y. K.; Kamimura,
T. Mater. Sci. Eng., R 2000, 30, 1. (b) Belokoneva, E. L.; Stefanovich, S.
Y. Journal of Solid State Chemistry 2012, 195, 79. (c) Smith, M. D.;
(
(
2
0743
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744

Journal of the American Chemical Society
Article
Blau, S. M.; Chang, K. B.; Tran, T. T.; Zeller, M.; Halasyamani, P. S.;
Schrier, J.; Norquist, A. J. Journal of Solid State Chemistry 2012, 195,
86. (d) West, J. P.; Hwu, S. J. Journal of Solid State Chemistry 2012,
195, 101. (e) Zhang, J. J.; Kong, F.; Xu, X.; Mao, J. G. Journal of Solid
State Chemistry 2012, 195, 63. (f) Zhang, J. J.; Zhang, Z. H.; Sun, Y. X.;
Zhang, C. Q.; Tao, X. T. Journal of Solid State Chemistry 2012, 195,
120. (g) Zhao, W. W.; Pan, S. L.; Wang, Y. J.; Yang, Z. H.; Wang, X.;
Han, J. Journal of Solid State Chemistry 2012, 195, 73. (h) Chung, I.;
Song, J. H.; Jang, J. I.; Freeman, A. J.; Kanatzidis, M. G.; Han, J. Journal
of Solid State Chemistry 2012, 195, 161. (i) Lin, X. S.; Guo, Y. F.; Ye,
N. Journal of Solid State Chemistry 2012, 195, 172. (j) Zhou, L. J.;
Chen, L.; Li, J. Q.; Wu, L. M. Journal of Solid State Chemistry 2012,
195, 166.
(
41) Raoux, S.; Welnic, W.; Ielmini, D. Chem. Rev. 2010, 110, 240.
(
42) Iovu, M. S.; Colomeico, E. P.; Benea, V. G.; Popescu, M.;
Lorinczi, A.; Velea, A. J. Optoelectron. Adv. Mater. 2011, 13, 1483.
43) Wang, R. Y.; Caldwell, M. A.; Jeyasingh, R. G. D.; Aloni, S.;
Shelby, R. M.; Wong, H. S. P.; Milliron, D. J. J. Appl. Phys. 2011, 109,
13506.
44) Eom, T.; Choi, S.; Choi, B. J.; Lee, M. H.; Gwon, T.; Rha, S. H.;
(
1
(
Lee, W.; Kim, M.-S.; Xiao, M.; Buchanan, I.; Cho, D.-Y.; Hwang, C. S.
Chem. Mater. 2012, 24, 2099.
(
5
(
1
45) Bogoslovskiy, N. A.; Tsendin, K. D. Semiconductors 2012, 46,
59.
46) Myers, R. A.; Mukherjee, N.; Brueck, S. R. J. Opt. Lett. 1991, 16,
732.
(
47) Qiu, J. R.; Si, J. H.; Hirao, K. Opt. Lett. 2001, 26, 914.
(
48) Liu, Q. M.; He, X. A.; Hu, T.; Gao, C.; Hou, Y. N.; Zhao, X. J.
Appl. Phys. A: Mater. Sci. Process. 2011, 102, 245.
49) Chen, M. C.; Wu, L. M.; Lin, H.; Zhou, L. J.; Chen, L. J. Am.
Chem. Soc. 2012, 134, 6058.
50) Zhao, H.-J.; Zhang, Y.-F.; Chen, L. J. Am. Chem. Soc. 2012, 134,
993.
51) Chen, M.-C.; Li, P.; Zhou, L.-J.; Li, L.-H.; Chen, L. Inorg. Chem.
011, 50, 12402.
52) Chen, M.-C.; Li, L.-H.; Chen, Y.-B.; Chen, L. J. Am. Chem. Soc.
011, 133, 4617.
53) Mei, D. J.; Yin, W. L.; Feng, K.; Lin, Z. S.; Bai, L.; Yao, J. Y.; Wu,
Y. C. Inorg. Chem. 2012, 51, 1035.
54) Morris, C. D.; Malliakas, C. D.; Kanatzidis, M. G. Inorg. Chem.
011, 50, 10241.
55) McCarthy, T. J.; Ngeyi, S. P.; Liao, J. H.; DeGroot, D. C.;
Hogan, T.; Kannewurf, C. R.; Kanatzidis, M. G. Chem. Mater. 1993, 5,
31.
(
(
1
(
2
(
2
(
(
2
(
3
(
(
56) Mccarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 1257.
57) Stoe & Cie. X-AREA, X-RED, and X-SHAPE; Darmstadt,
Germany, 1998.
(58) Sheldrick, G. M. Acta Crystallogr., Sect. A 2008, 64, 112.
(59) Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1998, 37, 2582.
(60) Nye, J. F. Physical Properties of Crystals: Their Representation by
Tensors and Matrices; Clarendon Press; Oxford University Press:
Oxford, Oxfordshire, NY, 1984.
(
61) Halasyamani, P. S.; Poeppelmeier, K. R. Chem. Mater. 1998, 10,
753.
62) Chung, I.; Do, J.; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G.
Inorg. Chem. 2004, 43, 2762.
2
(
(
(
(
(
(
63) Bondi, A. J. Phys. Chem. 1964, 68, 441.
64) Shannon, R. D. Acta Crystallogr., Sect. A 1976, 32, 751.
65) Wuttig, M.; Yamada, N. Nat. Mater. 2007, 6, 824.
66) Dhingra, S.; Kanatzidis, M. G. Science 1992, 258, 1769.
67) Aitken, J. A.; Evain, M.; Iordanidis, L.; Kanatzidis, M. G. Inorg.
Chem. 2002, 41, 180.
(68) Martin, B. R.; Dorhout, P. K. Inorg. Chem. 2004, 43, 385.
(69) Choudhury, A.; Dorhout, P. K. Inorg. Chem. 2006, 45, 5245.
(70) Kurtz, S. K.; Perry, T. T. J. Appl. Phys. 1968, 39, 3798.
(71) Boyd, R. W. Nonlinear Optics, 3rd ed.; Academic Press:
Amsterdam; Boston, 2008.
(
72) Bhar, G. C. Jpn. J. Appl. Phys., Part 1 1993, 32, 653.
2
0744
dx.doi.org/10.1021/ja309386e | J. Am. Chem. Soc. 2012, 134, 20733−20744