Dynamics of Bulk vs. Nanoscale WS_2: Local Strain and Charging Effects

0.31 MB·English

by  R. D. Luttrell

Dynamics of Bulk vs. Nanoscale WS : Local Strain and Charging Effects
2
R.D. Luttrell1, S. Brown1, J. Cao1, J.L. Musfeldt1, R. Rosentsveig2, and R. Tenne2
1Department of Chemistry, University of Tennessee, Knoxville, Tennessee, USA 37996 and
2Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot, Israel 76100
WemeasuredtheinfraredvibrationalpropertiesofbulkandnanoparticleWS2inordertoinvesti-
6 gatethestructure-propertyrelationsinthesenovelmaterials. Inadditiontothesymmetry-breaking
0 effects of local strain, nanoparticle curvature modifies the local charging environment of the bulk
0 material. Performingachargeanalysisonthexy-polarizedE1uvibrationalmode,wefindanapprox-
2 imate1.5:1intralayerchargedifferencebetweenthelayered2Hmaterialandinorganicfullerene-like
(IF)nanoparticles. Thiseffectivechargedifferencemayimpactthesolid-statelubricationproperties
n
of nanoscale metal dichalcogenides.
a
J
PACSnumbers: 63.22.+m,78.20.Ci,61.46.+w,78.30.-j
5

]
i I. INTRODUCTION
c
s
-
l Inorganic fullerene-like (IF) nanostructures have re-
r
t centlyattractedattentionduetotheiruniqueclosedcage
m
structuresandoutstandingsolid-statelubricatingbehav-
. ior [1, 2, 3]. Just as carbon fullerenes are nanoscale
t
a analogs of layered graphite, IF nanoparticles and nan-
m
otubes are curved analogs of the corresponding quasi-
- two-dimensional material. Layered and nanoscale metal
d
dichalcogenides are prototypes in this regard, and the
n
o discovery of the WS2-based family of IF nanoparticles
c (Fig.1)providestheopportunitytoinvestigatestructure-
[ propertyrelationsinbulkvs. nanoscalematerials. Atthe

same time, the IF materials hold out the potential for
1
important applications. In addition to use in recharge-
v
FIG. 1: Transmission electron microscope images of (a) lay-
7 able batteries, optical devices, and in impact-resistant
0 nanocomposities, extensive mechanical properties test- ered 2H-WS2 and (b) a representative IF-WS2 nanoparticle
(b). Each nanoparticle consists of a hollow core and several
1 ing demonstrates that the friction coefficient of IF-WS2 W-S-W layers. The average particle diameter is ∼120 - 200
1 nanoparticles is reduced up to 50% compared with the
nm for the materials of interest here. The IF nanoparticle
0
2H-WS2 parent compound, maintaining excellent lubri- shown hereis slightly smaller than average.
6
cating behavior even under very high loads, ultra-high
0
vacuum, and in humid conditions [4, 5, 6, 7, 8, 9, 10].
/
at Because of these observations, major efforts have been motion,respectively. Thisdirectionalselectivityprovides
m directed at understanding the connection between bulk
a sensitive and microscopicprobe of chargeand bonding
and microscopic properties and exploiting the commer-
- interactions that we will employ in this work to assess
d cial promise of these novel nanomaterials.
effective total and local charge differences between the
n
o 2H-WS2 belongs to the space group P63/mmc (D64h) bulk and nanoscale materials [15, 16, 17, 18, 19]. X-ray
c and contains two formula units per unit cell [11]. The diffraction reveals that the local structure of 2H-WS2 is
: bonding is well-known to consist of strong covalent in- preservedwithin the unit cell of an IF nanoparticle with
v
tralayer forces and weak van der Waals interactions be- the exception of a 2% lattice expansion along the z-axis
i
X tween adjacent layers [12, 13]. Each MX2 layer (M = [20, 21]. This lattice expansion is attributed to strain
r group VIB metal, X = group VIA element) contains a in the curved WS2 layers [21, 22], each of which has
a
layerofmetalatoms,sandwichedbetweentwochalcogen a slightly different radius due to layer inhomogeneities
layers, with each metal atom bonded to six chalcogen (Fig. 1(b)). The lattice modes of 2H-WS2 were previ-
atomsinatrigonalprismaticarrangement. Agroupthe- ously investigated using combined Raman and inelastic
oretical analysis (Table A.1, Appendix) gives a total of neutron scattering, demonstrating that the two-phonon
18 normal modes [14]. The doubly degenerate E1u and resonance Raman effects are second-order processes in-
singly degenerate A2u vibrationalmodes are infraredac- volving the longitudinal acoustic mode at the K point
tive;the conjugategerademodesareRamanactive. Fig- of the Brillouin zone [23]. Previous studies also indicate
ure 2 displays the relevant displacement vectors. Note that the new Raman peaks in spectra of IF-WS2 nan-
thattheinfraredactivexy-polarizedE1u andz-polarized otubes andnanoparticlesshouldbe assignedasdisorder-
A2u modes are associated with intralayer and interlayer induced zone edge phonons [24]. More recent Raman
2
IR Active Raman Active
0.34
300 K 2H-WS2
Layer 1 0.30 IF-WS2
e
c
n
a 0.26
Layer 2 ct
efle 0.22 E1u(x,y) A2u(z)
1 2 R
E1u A2u E2g E2g E1g A1g
0.18
FIG. 2: Displacement vectors of the infrared and Raman
active modes of 2H-WS2. The dashed line between the two 0.14
layers represents theweak van der Waals force. 300 325 350 375 400 425 450 475
-1
Frequency (cm )
work demonstrates that the 33 cm−1 E2g shearing or
“rigid layer” mode is almost completely blocked in IF- FIG. 3: 300 K reflectance spectra of 2H- and IF-WS2. The
WS2 due tosurface strainhinderingintralayermotionin E1u and A2u modes are infrared active in these materials,
with xy-and z-directed polarizations, respectively.
the nanoparticles [25]. Extensive optical properties and
STMworkindicatesthattheindirectgapissmallerinthe
IFnanoparticlescomparedtothatinthebulk[26,27]. A
Elmer Lambda 900 grating spectrometer, covering the
andB excitonpositions arealso sensitiveto confinement frequency range from 25 - 52000 cm−1. A helium-cooled
and the number of layers in the nanoparticle [24, 26].
bolometer detector was employed in the far-infrared for
Inordertoinvestigatestructure-propertyrelationships added sensitivity. Both 0.5 and 2 cm−1 resolution were
in these chemically identical but morphologically differ-
usedintheinfrared,whereas3nmresolutionwasusedin
ent metal dichalcogenides, we measured the infrared re-
the optical regime. Variable temperature measurements
flectance spectra of both 2H- and IF-WS2 powders and were carried out with a continuous-flow helium cryostat
performedachargeanalysistoextracttotalandlocalef-
and temperature controller. A Kramers-Kronig analysis
fectivechargefromtheoscillatorparametersofthemajor
wasusedtocalculatetheopticalconstantsfromthemea-
infraredactivephononmodes. Wefoundanapproximate
sured reflectance, yielding information on the dispersive
1.5:1 intralayer charge difference between the 2H- bulk
and lossy response of each material [33, 34]. Standard
andIF-nanoparticlematerials,respectively. The trendis
peak-fitting techniques were employed, where appropri-
different in the interlayer direction, reflecting a slightly
ate.
enhanced z interaction in the nanoparticles. We discuss
how differences in both charge and strain may be con-
nected to the macroscopic properties of these materials. III. RESULTS AND DISCUSSION
A. Understanding Charge Localization Effects in
II. EXPERIMENTAL METHODS 2H- and IF-WS2
IF-WS2 was prepared from its oxide precursor, WO3, Figure 3 displays a close-up view of the far infrared
following previously published procedures [1, 21, 28, 29, reflectance spectra of 2H- and IF-WS2 at 300 K. We as-
30, 31, 32]. The IF nanoparticles of interest in this work signthemajorfeaturesinthereflectance(at356and437
range in size from ∼120 - 200 nm in diameter. The cm−1) as the E1u and A2u modes, respectively [14, 35].
particle size, shape and distribution have been studied These modes are strikingly different in the two materi-
by x-ray powder diffraction [21], scanning tunneling mi- als. The E1u mode appears damped and suppressed in
croscopy [27], and high-resolution transmission electron IF-WS2 compared to the 2H- analog, whereas the A2u
microscopy [21, 27, 28]. Pressed isotropic pellets were mode is slightly more pronounced in the IF compound
prepared to investigate the dynamical properties. Bulk compared to that in the bulk. These differences can be
2H-WS2 was also measured for comparison (Alfa Aesar, quantified using traditional dielectric oscillator models
99.8%). andfittingtechniqueswhich,incombinationwithappro-
Near-normal infrared reflectance was measured over a priate models, allow us to assess the charge characteris-
wide frequency range using a series of spectrometers in- tics of the nanomaterial as compared to the bulk.
cluding a Bruker 113V Fourier transform infrared spec- One well-established approach for quantifying charge
trometer, an Equinox 55 Fourier transform instrument in a material involves assessment of both total and local
(equipped with a microscope attachment), and a Perkin effective charge [15, 16, 17, 18, 19]. As discussed by
3
TABLE I:

Classical Oscillator Parameters and Optical Phonon
ODsaωcmLilωωOlpaTLi-tnωOOo3gTr0((O0csccotmmK(nrce−−sFmnt11rag))−entq1htu),,eγSnci2e33Hs00055..-f.6700WoE6..30r341115S70u22HM-Io33Fa00d055n-..e.6700Wd3..1119240SI78F2-W24H00S3..-2800WA.003224S2u2MIo4F00d3-..9e00W.00026S12 ()(Arb. Units) 0011111.......6802468 -2Im([1)/(T)]O LO )and) (Arb. Units)((12000156667.........048260482340 345 F35re0qu35e5n c3y6 0(cm36-15)370 375 0000000.......000000012233445050505 /()] (Arb. Units)
ǫ∞ 9.58 6.26 20.4 0.010 1
[
m
0.2 0.005
I
10 K -
0.0 0.000
Oscillator strength, S 0.040 0.017 0.005 0.004 350 355 360 365 370 375
Damping constant, γ 0.004 0.010 0.006 0.004 -1
ωTO (cm−1) 358.30 358.61 440.80 440.87 Frequency (cm )
ωLO (cm−1) 359.14 359.20
ωLO-ωTO (cm−1) 0.84 0.59 FIG. 4: Frequency dependence of the imaginary part of
ǫ∞ 8.92 6.46 the dielectric function, ǫ2(ω), and the energy loss function,
−Im(1/ǫ(ω)),ofIF-WS2 at300K.Longitudinaloptical(LO)
and transverse optical (TO) frequencies are indicated. The
Bursteinet al.[19], macroscopiceffectivecharge,e∗,is a inset displays an oscillator fit (white line) to the real (black)
T and imaginary (gray) parts of the dielectric function, ǫ1(ω),
measureoftheelectricmomentperunitcellandcontains
and ǫ2(ω),for IF-WS2 at 300 K.
contributions from both the localized charge on the ion
sites and the charge generated throughout the unit cell.
Therefore,e∗, can be separatedinto a localizedpart, e∗,
T l splitting is key to distinguishing local charge differences
and a nonlocalized part, e∗ ,
nl in 2H- and IF-WS2.
∗ ∗ ∗ Inordertoobtaintheparametersneededtoextractto-
e =e +e . (1)
T l nl tal effective and local charge for both 2H- and IF-WS2,
Here, e∗ is defined as the localized moment generated we carried out a Kramers-Kronig analysis of the mea-
l
per unit displacement of an ion. It induces a local field sured reflectance. LO and TO phonon frequencies are
through dipole-dipole interactions which contributes to obtaineddirectlyfromtheopticalconstants,forinstance,
the reduction of the transverse optical (TO) phonon fre- aspeaksintheenergylossfunction,−Im(1/ǫ(ω)),andin
quency. Although e∗ can be a good measure of bond the imaginarypartof the frequency dependent dielectric
T
ionicity or covalency in layered MX2 transition-metal function, ǫ2(ω), respectively. Figure 4 shows an example
dichalcogenides, e∗ gives a more appropriate representa- oftheseopticalconstantsandthestraightforwardextrac-
l
tion of bonding interactions because it quantifies charge tionofLOandTOphononfrequenciesforIF-WS2 at300
on the ionic sites [15]. K. Oscillator strength, S, is obtained by simultaneously
Total macroscopic effective charge, e∗, is given as fitting the real and imaginary parts of the complex di-
T
electric function, ǫ˜(ω) = ǫ1(ω) + iǫ2(ω), using the three
e∗T = ωTOc 4π2ǫ0m¯S. (2) parameter model
e e r N
S ω2
Here, ωTO is the TO phonon frequency (in cm−1), S is ǫ˜(ω)=ǫ∞+ ω2−ω2j−jiγ ω ω. (4)
the oscillator strength, N is the number of WS2 formula Xj j j j
units per unit volume, m¯ is the mode mass [36], e is
Here, the subscript j refers to the mode of interest, γ
the electronic charge, c is the speed of light (in cm/s),
is the damping constant, and ǫ∞ is the high frequency
and ǫ0 is the permittivity of free space. Note that for
dielectric constant [16, 18, 33, 38]. As an example, the
comparison of two similar materials, such as 2H- and
insetofFig.4showsanoscillatorfitoftherealandimag-
IF-WS2, oscillator strength and TO phonon frequency
willbethedistinguishingparameters. Localizedeffective inarypartsofthe complexdielectricfunctionfortheE1u
charge [15], e∗, is determined as modeofIF-WS2 at300K.Theclassicaloscillatorparam-
l etersofboth2H-andIF-WS2 aresummarizedinTableI.
e∗l =c (ω2 −ω2 ) m¯ǫ0 . (3) Tevhaeluseatpeatrhame cehtearrsg,eaclohnagrawctitehrisEtqicnss.. 2 and 3, allow us to
e r LO TO e2LN
The small LO-TO splitting in both 2H- and IF-WS2
Here, ω is the longitudinal optical (LO) phonon fre- indicates that the metal-chalcogen bond is highly cova-
LO
quency(incm−1)andListheLorentzfactorforahexag- lent within the layer, in agreement with previous results
onal lattice [37]. Precise measurement of the LO-TO forWS2 aswellasotherwell-knowncovalentcompounds
4
Table II also displays the total and local effective
TABLE II:
chargeof2H-andIF-WS2 atlowtemperature. Although
Macroscopic and Localized Effective Charge Values for 2H-
the exact values of total and local effective charge differ
and IF-WS2 slightly from their 300 K values (for instance, the total
E1u Mode A2u Mode charge extracted from analysis of the xy-polarized E1u
2H-WS2 IF-WS2 2H-WS2 IF-WS2 mode of 2H-WS2 is 0.51 at 10 K and 0.45 at 300 K),
the overalltrends between the layeredand nanomaterial
300 K e∗T/e 0.45 0.30 0.15 0.16 remainsimilartothose discussedabove. Macroscopicef-
e∗/e 0.20 0.14
l fective chargewithin the layerdecreasesfrom0.51inthe
bulk to 0.34 in the nanomaterial(again, an approximate
10 K e∗/e 0.51 0.34 0.22 0.20 1.5:1 charge difference), and local charge makes up ap-
eT∗/e 0.23 0.20 proximately 50% of the total charge. The total charge
l extracted from the interlayerA2u mode is slightly larger
in the 2H material than in the IF nanoparticles at low
ae∗l cannot beobtained for the z-polarized interlayer A2u temperature.
mode dueto thenegative Lorentz factor [40].
bError bars on thetotal and local charge valuesare ± 0.02.
B. Curvature-Induced Local Symmetry Breaking
including MoS2, MoSe2, and WSe2 [15, 16, 17, 18, 19]. in IF-WS2
In contrast, Uchida and Tanaka report large LO-TO
splittings for severalgroup IV transition-metal dichalco- Are there other manifestations of curvature in the
genides, including 1T-TiSe2, 1T-ZrSe2, and 1T-HfSe2, nanoparticles besides the aformentioned total and local
which are considered to be highly ionic materials [15]. effective charge differences? Certainly, strain and con-
Physically,the smallerLO-TOsplitting ofthe E1u mode finement have been of recent interest in both vanadium
inIF-WS2 (Table1),indicatesthattheIF-nanoparticles oxideinorganicnanotubes andsiliconnanowires[41,42].
areslightlymorecovalentthantheparent2H-compound. In both cases, strain broadens the vibrational modes.
Thisresultindicatesthatnanoparticlecurvaturechanges Another effect of curvature is that the local, short range
the charge-sharingenvironment within the layer. symmetryisformallylowerthan(andasubgroupof)the
TableIIdisplaysthetotalmacroscopicandlocalcharge unstrained bulk. The reduction of local symmetry can
for both 2H- and IF-WS2. Using the xy-polarized E1u change the selection rules, allowing formerly “infrared-
mode as a probe of charge within the layer,we find that silent”modestobecomeinfraredactive[43,44]. Further,
e∗T = 0.45 for 2H-WS2 and e∗T = 0.30 for IF-WS2 at 300 the curvature of each metal dichalcogenide layer within
K [39]. Thus, IF-WS2 has approximately two thirds the the nanoparticle is not uniform. This inhomogeneous
intralayer charge as 2H-WS2. This approximate 1.5:1 structure also results in mode dispersion. Evidence for
charge difference is replicated in the local charge num- these effects, while present in the 300 K spectrum of IF-
bers, with e∗ decreasing proportionally in the IF com- WS2 (Fig. 3), is best illustrated in the low temperature
l
pound. The intralayer charge differences summarized in spectral response.
Table II canbe tracedto differences inthe LO-TOsplit- Figure 5 displays the optical conductivity, σ1(ω), of
ting and oscillator strength of the E2u mode in 2H- and 2H- and IF-WS2 at 10 K [34]. The low temperature
IF-WS2. As already mentioned, these changes are easily spectral response of both materials is still dominated by
observedinthe spectraldata(Fig. 3). Theresultsimply the E1u and A2u modes, although because of the addi-
that there is in fact a significant difference in the local tional fine structure, the nanoparticle response is clearly
environmentandchemical bonding between the 2H- and much richer and more complicated than that of the bulk
IF- materials and that nanoparticle curvature changes material. We attribute the additional vibrational struc-
the charging environment within the plane. A blocked turetotheformally“silent”andcombinationmodes,ac-
“rigid layer” E2g Raman mode is also consistent with tivated (and dispersed) in the spectrum of IF-WS2 by
these observations [25]. thesymmetrybreakingthatresultsfromthecurvedcage
In the interlayer direction, the charge trend as char- structure. For instance, some Raman modes are well-
acterized by the behavior of the A2u mode is different known conjugates of infrared active features [45]. The
(Table II). At 300 K, e∗T = 0.16 for IF-WS2 compared E12g and E21u conjugate pair is an example. The symme-
withe∗T =0.15for2H-WS2. Thate∗T islargerinIF-WS2 try analysis and vector displacement diagrams of Verble
isindicativeofslightlystrongerinterlayerinteractionand andWieting[14](seealsoTableA.1,Appendix)alsopro-
enhancedchargeenvironmentinthecurvednanoparticles videseveralcandidatesforsilentmodeactivation,stating
comparedwith the bulk. As expected, the totaleffective thattheinactiveB1g,B2u,andE2u modesarenearlyde-
chargewithinthelayer(e∗T fromtheE1u mode)isalways generatewithseveralRaman-andinfrared-activemodes.
largerthanthatbetweenlayers(e∗T fromthe A2u mode), Thus, the optical conductivity of IF-WS2 likely contains
indicating that the majority of charge resides within the weakcontributionsfromallofthesefirst-orderstructures,
metal dichalcogenide layer in both materials. along with a substantial number of peaks that originate
5
tween the 2H- and IF- materials. The trend is different
in the interlayer direction (probed by the A2u mode) at
40 300 K. Here, effective charge increases slightly in the IF
10 K
materialcomparedto that in the bulk. We attribute the
) strongerinterlayerinteractionto nanoparticlecurvature.
-1 m 30 2H-WS2 Elucidatingtherelationshipbetweenstructureandeffec-
1 c IF-WS2 tive total and local charge in 2H- and IF-WS2 is only
- the first step in understanding the fundamental interac-
) ( 20 tions underlying the phenomenal mechanical and solid
( state lubricating properties of nanoscale metal dichalco-
1
genides. Tuning MX2 bond covalency may, for instance,
10 allow simultaneous exploration of macroscopic mechan-
ical properties, the charge environment, and potential
surface effects.
0
330 350 370 390 410 430 450
-1
Frequency (cm ) V. ACKNOWLEDGEMENTS
FIG. 5: Optical conductivity of 2H- and IF-WS2 at 10 K. Work at the University of Tennessee is supported by
Arrows highlight the low-temperature symmetry breaking in the Materials Science Division, Office of Basic Energy
the IF material. Ref. [46] lists the frequencies of these fea- Sciences at the U.S. Department of Energy under Grant
tures. All of these small modes are reproducible, although it No. (DE-FG02-01ER45885). Work at the Weizmann In-
should be noted that the exact details of this fine structure stitute of Science is supported by the Helen and Martin
will likely depend on particle size and distribution within a
KimmelCenterforNanoscaleScienceandbyNanoMate-
sample, thenumberof nanoparticle layers, and thenatureof
rials, Ltd. R.T. is the holder of the Drake Family Chair
thedefects.
inNanotechnology. We thankAlbertMiglioriandDavid
Toma´nek for interestingdiscussionsandRonitPopovitz-
from mode combinations and disorder effects. Biro for the transmission electron microscope images.
Group theory can also be used to investigate possible
second order infrared active combinations. When a di-
rect product of two fundamental modes contains A2u or APPENDIX: Group Theoretical Results for 2H-WS2
E1u symmetry in its character, it follows that a combi-
nation of those two modes will be infrared active. For In 1970, Verble and Wieting carried out a complete
example, E1u E2g = B1u + B2u + E1u, and A2u symmetryanalysisonhexagonallayeredcompoundswith
E1g = E1u. TNhese combinations are both infrared aNc- the goal of analyzing vibrational mode symmetries. Our
tive because their reducible representationcontains E1u. analysis differs from that in Ref. [14] in two ways: (1) a
(Taking the direct product of this state with the x, y, C′2′ correlation table accounts for the D3h 2(c) site sym-
and z dipole moment operators clearly contains the to- metryoftungsten,and(2)aσ correlationtableaccounts
d
tallysymmetricgroup.) Similarly,thesumanddifference for the C3v 4(f) site symmetry of sulfur [13, 47]. In Ref.
of other fundamental vibrational frequencies can lead to [47]refertoTable14,p. 50,andD6h correlationtable,p.
variousinfraredactivecombinationslikeA2u A1g,E1u 210. Table A.1 summarizes our results. Local strain in
A1g,E1u E1g,B2u E2g andB2u BN1g. Wenote the IF-WS2 breaksthe selectionrules ofthe 2H-“parent
NthatthelongNitudinalacoNusticmodeisinNtimatelyandex- compound”, as detailed in the text.
tensively involved with many combination modes that
appear in resonance Raman [23] and may be involved in
TABLE A.1:
theinfrared-activecombinationmodesofIF-WS2aswell.
Group Theoretical Analysis for 2H-WS2
Atom SiteSymmetry Irreducible Representation
IV. CONCLUSION W D3h 2(c) A2u + B1g + E2g + E1u
S C3v 4(f) A1g + A2u + B1g + B2u
We report the infrared vibrational properties of bulk + E1g + E1u + E2g + E2u
and nanoparticle WS2 in order to investigate the ΓTotal 2A2u + 2B1g + A1g + B2u
structure-property relations in these novel materials. In + E1g + 2E2g + 2E1u + E2u
additiontothesymmetry-breakingeffectsoflocalstrain, ΓInactive 2B1g + B2u + E2u
nanoparticle curvature modifies the well-known charge ΓAcoustical A2u + E1u
environment of the bulk material. Using the E1u mode ΓΓIRnfarmaarned 2AE2u2g++EA1u1g + E1g
as a local probe of charge behavior within the layer, we
findanapproximate1.5:1intralayerchargedifferencebe-
6
[1] R.Tenne,L.Margulis, M.Genut,andG.Hodes,Nature son,L.Margulis, andR.Tenne,J.Am.Chem.Soc.116,
360, 444 (1992). 1914 (1994).
[2] L. Margulis, G. Salitra, and R. Tenne, Nature 365, 113 [28] M.Homyonfer,B.Alperson,Y.Rosenberg,L.Sapir,S.R.
(1993). Cohen,G.Hodes,andR.Tenne,J.Am.Chem.Soc.119,
[3] P.A. Parilla, A.C. Dillon, K.M. Jones, G. Riker, D.L. 2693 (1997).
Schulz, D.S. Ginley, and M.J. Heben, Nature 397, 114 [29] Y. Feldman, A. Zak, R. Popovitz-Biro, and R. Tenne,
(1999). Solid State Sci. 2, 663 (2000).
[4] L.Rapoport,Yu.Bilik,Y.Feldman,M.Homyonfer,S.R. [30] R. Rosentsveig, A. Margolin, Y. Feldman, R. Popoviz-
Cohen, and R. Tenne,Nature 387, 791 (1997). Biro, and R.Tenne, Chem. Mater. 14, 471 (2002).
[5] L. Rapoport, Y. Feldman, M. Homyonfer, H. Cohen, J. [31] A.Margolin,R.Rosentsveig,A.Albu-Yaron,R.Popoviz-
Sloan, J.L. Hutchison, and R. Tenne, Wear 225, 975 Biro, and R.Tenne, J. Mater. Chem. 14, 617 (2004).
(1999). [32] R. Tenne, Angew. Chem. Int.Ed. 42, 5124 (2003).
[6] L. Rapoport, N. Fleischer, and R. Tenne, Adv. Mater. [33] F. Wooten, Optical Properties of Solids (Academic
15, 651 (2003). Press, New York,1972).
[7] W.X.Chen,J.P.Tu,Z.D.Xu,R.Tenne,R.Rosentsveig, [34] These optical constants include the real and imaginary
W.L. Chen, and H.Y. Gan, Adv. Eng. Mater. 4, 686 parts of the dielectric function, ǫ1(ω) and ǫ2(ω), the en-
(2002). ergylossfunction,−Im(1/ǫ(ω)),andtheopticalconduc-
[8] C. Schuffenhauer, G. Wildermuth, J. Felsche, and R. tivity,σ1(ω).Inouranalysis,thelowfrequencydatawas
Tenne,Phys. Chem. Chem. Phys. 6, 3991 (2004). extrapolated as a constant, and the high frequency data
[9] L. Rapoport, N. Fleischer, and R. Tenne, J. Mater. was extrapolated with an exponent of 2.4.
Chem. 15, 1782 (2005). [35] Intheopticalconductivity,σ1(ω),thesefeaturesarecen-
[10] Y.Q.Zhu,T.Sekine,Y.H.Li,W.X.Wang,M.W.Fay,H. teredat356.5and438.3cm−1 forthe2Hcompoundand
Edwards, P.D. Brown, N. Fleischer, and R.Tenne, Adv. at 357.0 and 439.0 cm−1 for the IFnanoparticles.
Mater. 17, 1500 (2005). [36] G. Herzberg, Infrared and Raman Spectra of Polyatomic
[11] Here,2HisbasedonRamsdellnotationwith2indicating Molecules (D. Van Nostrand Company, Inc., New York,
the number of S-W-S slabs within the unit cell and H 1945).
representing thelattice type(hexagonal). [37] H. Mu¨eller, Phys. Rev. 50, 547 (1936).
[12] J.A. Wilson and A.D.Yoffe, Adv.Phys. 18, 193 (1969). [38] F. Gervais, Infrared and Millimeter Waves, Vol. 8 (Aca-
[13] W.J.Schutte,J.L.DeBoer,andF.Jellinek,J.SolidState demic Press, New York,1983).
Chem. 70, 207 (1987). [39] WenotethatUchidaandTanaka[15]reporte∗ =1.5and
T
[14] J.L. Verble and T.J. Wieting, Phys. Rev. Lett. 25, 362 e∗l =0.4for2H-WS2,emphasizingthedifferencebetween
(1970). powder and single crystal measurements.
[15] S. Uchida and S. Tanaka, J. Phys. Soc. Jpn. 45, 153 [40] e∗l cannot be obtained for the z-polarized interlayer A2u
(1978). mode. The model for localized charge is developed for
[16] G. Lucovsky,R.M. White, J.A. Benda, and J.F. Revelli, studyingonlytheintralayerE1u modein thelayered 2H
Pys. Rev.B 7, 3859 (1973). bulkmaterial[15,19].SincetheLorentzfactorisnegative
[17] G. Lucovsky, R.M. Martin, and E. Burstein, Phys. Rev. for the A2u mode, no realistic charge values could be
B 4, 1367 (1971). determined.
[18] G. Lucovsky, J.C Mikkelsen, Jr., W.Y. Liang, R.M. [41] J.Cao,J.Choi,J.L.Musfeldt, S.Lutta,andM.S.Whit-
Whiteand R.M. Martin, Phys. Rev.B 14, 1663 (1976). tingham, Chem. Mater. 16, 731 (2004).
[19] E. Burstein, A. Pinczuk, and R.F. Wallis, in Proceed- [42] K.W. Adu,H.R. Gut´ierrez, U.J. Kim, G.U. Sumanasek-
ings of the Conference on the Physics of Semimetals era, and P.C. Eklund,Nano. Lett. 5, 409 (2005).
and Narrow-gap Semiconductors (Pergamon Press, New [43] Formally,thelocalsitesymmetryofacurvedsheetisC1,
York,1971), p. 251. andallAmodesareactive.Butpractically,thenanopar-
[20] Y. Feldman, E. Wasserman, D. J. Srolovitz, and R. ticle curvature is small, so it is useful to consider the
Tenne,Science 267, 222 (1995). additional spectral features as perturbations to the pla-
[21] Y. Feldman, G.L. Frey, M. Homyonfer, V. Lyakhovit- nar picture.
skaya,L.Margulis, H.Cohen,G.Hodes,J.L. Hutchison, [44] I. Loa, C. M¨oschel, A. Reich, W. Assenmacher, K.
and R.Tenne, J. Am. Chem. Soc. 118, 5362 (1996). Syassen, and M. Jansen, Phys. Status Solidi B 223, 293
[22] D.J.Srolovitz,S.A.Safran,M.Homyonfer,andR.Tenne, (2001).
Phys.Rev.Lett. 74, 1779 (1995). [45] T.Sekine,T.Nakashizu,K.Toyoda,K.Uchinokura,and
[23] C.Sourisseau,F.Cruege,andM.Fouassier,Chem.Phys. E. Matsuura, Solid State Commun. 35, 371 (1980).
150, 281 (1991). [46] Withintheclusterscenteredat∼347,379,and426cm−1,
[24] G.L. Frey, R. Tenne, M.J. Matthews, M.S. Dresselhaus, we observe vibrational fine structure at 338.7, 342.1,
and G. Dresselhaus, J. Mater. Res. 13, 2412 (1998). 345.0, 349.1, 355.09, 356.2; 362.6, 365.0, 368.1, 371.5,
[25] Unpublished data, P.C. Klipstein, Department of 376.0, 378.2, 381.9, 384.5, 388.0, 394.0, 396.8; and420.4,
Physics, Clarendon Laboratory, University of Oxford, 425.7, 430.1, 433.0 cm−1, respectively.
Oxford,England. [47] W.G. Fateley, F.R. Dollish, N.T. McDevitt, and F.F.
[26] G.L. Frey, S. Elani, M. Homyonfer, Y. Feldman, and R. Bentley, Infrared and Raman Selection Rules for Molec-
Tenne,Phys. Rev.B 57, 6666 (1998). ular and Lattice Vibrations: The Correlation Method
[27] M. Hershfinkel, L.A. Gheber, V. Volterra, J.L. Hutchi- (Wiley-Interscience, New York,1972).