NMR study of InP quantum dots: Surface structure and size effects

Article abstract of DOI:10.1063/1.478858

The results in this study are based upon the sensitivity of NMR to the local structure and electronic surroundings of the nuclear spins. A variety of surface components are identified and characterized using multinuclear polarization transfer and two-dimensional correlation techniques. It is expected that the application of recent developments in ab initio and density functional methods for the calculation of magnetic resonance parameters will complement and further elucidate the experimental observations in this new class of materials.

Full text of DOI:10.1063/1.478858

NMR study of InP quantum dots: Surface structure and size effects
M. Tomaselli, J. L. Yarger, M. Bruchez Jr., R. H. Havlin, D. deGraw, A. Pines, and A. P. Alivisatos
Citation: The Journal of Chemical Physics 110, 8861 (1999); doi: 10.1063/1.478858
View online: http://dx.doi.org/10.1063/1.478858
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/110/18?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Hydrogen and oxygen on InP nanowire surfaces
Appl. Phys. Lett. 89, 123117 (2006); 10.1063/1.2345599
Size-dependent Raman study of InP quantum dots
Appl. Phys. Lett. 82, 185 (2003); 10.1063/1.1535272
Effect of lattice mismatch on surface morphology of InAs quantum dots on (100) In 1−x Al x As/InP
Appl. Phys. Lett. 79, 4331 (2001); 10.1063/1.1428763
Selective positioning of InAs self-organized quantum dots on sub-250 nm GaAs facets
Appl. Phys. Lett. 71, 3254 (1997); 10.1063/1.120306
Reversible transition between InGaAs dot structure and InGaAsP flat surface
Appl. Phys. Lett. 71, 797 (1997); 10.1063/1.119649
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.88.90.140 On: Tue, 09 Dec 2014 11:56:40

JOURNAL OF CHEMICAL PHYSICS
VOLUME 110, NUMBER 18
8 MAY 1999
NMR study of InP quantum dots: Surface structure and size effects
M. Tomaselli, J. L. Yarger, M. Bruchez, Jr., R. H. Havlin, D. deGraw, A. Pines,
and A. P. Alivisatos
Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720
and Department of Chemistry, University of California, Berkeley, California, 94720
͑Received 22 December 1998; accepted 5 March 1999͒
We report the results of 31P NMR measurements on trioctylphosphine oxide ͑TOPO͒ passivated InP
quantum dots. The spectra show distinct surface-capping sites, implying a manifold of crystal–
ligand bonding configurations. Two In 31P surface components are resolved and related to different
electronic surroundings. With decreasing particle size the In 31P core resonance reveals an
increasing upfield chemical shift related to the overall size dependence of the InP electronic
structure. © 1999 American Institute of Physics. ͓S0021-9606͑99͒70718-X͔
were performed at
a ␯
31P Larmor frequency of
0
Semiconductor clusters with monodisperse diameters
ranging from 10–100 Å manifest quantum dot behavior.1
The surface composition of these colloidally prepared
particles2 has been shown to be important because of its
influence on the discrete electronic structure and quantum
confinement1,3–5 as well as its relation to electronic transport
properties,1,6 structural phase transitions, and thermodynamic
stability.7 In addition, the spectroscopic characterization of
the capping molecules can provide valuable information on
the morphology and faceting1 of the nanoparticles.
In this letter, we report an initial study on III–V semi-
conductor InP dots using one- and two-dimensional ͑1D and
2D͒ NMR. We find distinct capping and In 31P surface sites,
implying a variety of ligand–crystal bonding arrangements
and structural environments. The chemical shielding of the
In 31P resonance increases with decreasing dot size which
can be interpreted as a decrease in the 31P paramagnetic shift
with increasing electronic excitation energy of the quantum-
confined nanoparticles.
ϭ75.18 MHz (B0ϭ4.36 T) and ␯0ϭ161.99 MHz (B0
ϭ9.39 T) using Chemagnetics CMX spectrometers and 4
mm MAS probe assemblies from the same manufacturer.
The MAS frequency (␯mas) was stabilized within 3 Hz for all
experiments. The B1 nutation frequency was matched to ␯
1
ϭ140 kHz on all channels (1H,31P,13C). Spectrum 1͑A͒ was
obtained by a 31P single-pulse excitation with 1H decoupling
during data acquisition. The inhomogeneously broadened
resonance at Ϫ178 ppm ͑relative to 85% H3PO4) with a
linewidth of ␦1/2ϭ58 ppm (␯1/2ϭ4400 Hz) is assigned to the
In 31P core ͑interior͒ nuclei: For a spherical InP cluster with a
typical diameter of 45 Å, only ϳ20% of all atoms reside at
the surface. The lineshape is slightly asymmetric and upfield
shifted with respect to bulk In 31P ͓␦ϭϪ147 ppm, ␦1/2
ϭ43 ppm (␯1/2ϭ3200 Hz)͔, with the main source of broad-
ening being the indirect exchange interaction with the 115In
neighbors (Sϭ9/2),10 as well as a contribution from the
second-order
anisotropic
͑pseudo͒dipolar–quadrupolar
shift.11,12 The weak downfield segment of the spectrum
͑ϳ8%͒ in the range Ϫ20Ͻ␦Ͻ80 ppm corresponds to 31P
TOPO resonances at the crystal surface. From the relative
InP samples have been prepared under argon using the
dehalosilation reaction of InCl3 and P͑Si͑CH3͒3͒3 at 540–570
K in TOPO as coordinating solvent.8 Distinct size distribu-
tions of the particles were obtained by size-selective precipi-
tation resulting in InP clusters with average diameters in the
range of 20–50 Å and distributions of ϳ20%.8 For the sur-
face selective experiments, the separation and isolation were
carried out in a drybox. The precipitates were dried under
vacuum and the resulting powder samples sealed in pyrex
tubes ͑0.13 Pa͒. In a second synthesis, the identical precipi-
tation and isolation procedure was carried out in air, yielding
TOPO and oxide passivated InP dots.1,8 X-ray diffraction
spectra ͑Siemens D5000, Cu K␣ radiation͒ and transmission
electron microscopy images ͑TEM, TopCon EM002B͒
showed that the InP dots were highly crystalline ͑pure phase,
zinc-blende͒, roughly spherical in shape, with indications of
faceting. The cluster diameters ͑d͒ were inferred from UV/
vis absorption spectra ͑HP 8452͒ obtained immediately upon
separation.8
TABLE I. 31P chemical-shift parameters ͑75.18 MHz, 300 K͒; ␦aniso is the
anisotropy and
␩ the asymmetry of the chemical-shift anisotropy tensor
͑CSA͒ as defined in Ref. 18.
a
mas
1/2
a
␦
͑ppm͒
␩
␦
͑ppm͒
␦
͑ppm͒
iso
aniso
Free TOPO
InP-TOPO1
InP-TOPO2
InP-TOPO3
InP-TOPO4
47
2
118Ϯ3
96Ϯ5
0Ϯ0.05
0Ϯ0.1
71Ϯ2
36Ϯ2
6Ϯ2
10Ϯ2
12Ϯ2
10Ϯ2
12Ϯ2
b
b
¯
¯
78Ϯ6
1Ϯ0.1
b
b
Ϫ8Ϯ2
¯
¯
InP ͑bulk͒
Ϫ147c
43c
0 c
0 c
InP-core ͑45 Å͒
␣-InP ͑45 Å͒
␤-InP ͑45 Å͒
Ϫ178Ϯ4d
Ϫ118Ϯ8d
Ϫ199Ϯ6d
58Ϯ5d
63Ϯ7d
55Ϯ5d
aStatic CP and MAS rotary-resonance ͑Ref. 19͒ data.
bMotionally averaged CSA; ␦1st/a2ticϭ24Ϯ5 ppm.
cUndopted material; ͑Ref. 10͒.
Figures 1͑A͒ and 1͑B͒ show typical 1D 31P (Sϭ1/2)
NMR spectra of TOPO-InP (dӍ45 Å) recorded under con-
ditions of magic-angle spinning ͑MAS͒.9 The experiments
dLorentzian least-squares fit.
0021-9606/99/110(18)/8861/4/$15.00
8861
© 1999 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.88.90.140 On: Tue, 09 Dec 2014 11:56:40

8862
J. Chem. Phys., Vol. 110, No. 18, 8 May 1999
Communications
eters ͑see Table I͒ implies a manifold of ligand–crystal
bonding environments ͓i.e., unidentate and bridging oxygen
coordination14 to the cation ͑In͒ or perhaps anion ͑P͔͒.8 This
indicates that the capping molecules occupy distinct InP-
surface sites, conceivably due to the faceting of the nanopar-
ticles. Our finding contrasts with earlier studies on thiophe-
nol capped CdS or TOPO/TOP capped CdSe particles where
only one surfactant site was identified.15
The upfield part of the spectrum in Fig. 1͑B͒ shows two
partially overlapping In-bonded 31P resonances centered at
Ϫ118 ppm ͑␣-InP͒ and Ϫ199 ppm ͑␤-InP͒. Both lines are
shifted with respect to the core resonance ͓see Fig. 1͑A͔͒ and
they can be interpreted as two different structural
In 31P-surface environments. The assignment is corroborated
by a series of CP experiments with varying contact time ͑Fig.
2͒: We find a gradual downfield shift of the first moment for
␶cpϾ0.5 ms ͑dashed line in Fig. 2͒. This can be rationalized
by the presence of a slow polarization-transfer process from
TOPO protons to remote 31P core nuclei, shifted in resonance
frequency with respect to 31P surface spins. Neglecting
chemical-shift differences among the core nuclei, the full
series of CP spectra is fitted quite well with a minimum of
three Lorentzian components with varying intensities but
constant chemical shifts and linewidths ͑Fig. 2 and Table I͒.
The dependence of the extracted intensities as a function of
FIG. 1. 1D 31P NMR spectra of TOPO-InP dots (dӍ45 Å) recorded at
75.18 MHz and 300 K ͑ϳ5 mg sample sealed in a pyrex tube͒. ͑A͒ Single
pulse experiment; 300 s delay between experiments; 256 transients. ͑B͒
31
Surface selective 1H→ P cross-polarization experiment with ␶cpϭ300 ␮s;
1.5 s delay between experiments; 36 000 transients. Both experiments were
recorded under MAS (␯masϭ5 kHz) and on-resonance 1H decoupling during
data acquisition. Spinning sidebands are marked with asterisks. The labeled
resonances are described in the main text.
intensities of the two components in Fig. 1͑A͒ we infer an
average TOPO surface coverage of ϳ20%, consistent with
x-ray photoelectron spectroscopy ͑XPS͒ on similar samples.8
␶
is shown in Fig. 2, with the curves fitted by the form
cp
/T
Mqe(1ϪeϪ␶
) invoking the spin-temperature approxi-
cp IS
mation and neglecting T1␳ relaxation.9,16 It is evident that the
surface components ͑␣, ␤-InP, and TOPO͒ reach a state of
quasi-equilibrium for ␶cpϾ1.5 ms and are characterized by
buildup rates roughly one order-of-magnitude faster than that
of the core spins. For the present experimental conditions
͑i.e., high spinning speed where the IS zero-quantum line-
width is given by the effective IS dipolar coupling
frequency͒,17 T ϰ r3 . Here, r denotes a typical distance
1
31
Spectrum 1͑B͒ was obtained by H→ P cross polarization
͑CP͒13 with a contact time of ␶cpϭ300 ␮s and proton decou-
pling during data acquisition. The lineshape in Fig. 1͑B͒ dif-
fers considerably from that of Fig. 1͑A͒ since the CP experi-
ment probes 31P spins in close proximity to the TOPO
protons and is therefore surface selective. Four different 31P
capping resonances are resolved ͑TOPO 1–4͒ with isotropic
chemical shifts (␦iso) at 71, 36, 6, and Ϫ8 ppm and an aver-
age linewidth of 11 ppm ͑relative intensity ratios:
1.0:0.4:0.3:0.3͒. The distribution of chemical shift param-
͗
͘
IS
IS
IS
between 31P spins and surface protons. Assuming that the
TOPO 31P nuclei are efficiently polarized by the nearest-
31
FIG. 2. Left side: 1D 1H→ P CP spectra of TOPO-InP
dots with varying ␶cp : 2 s delay between scans; 18 000
transients. Other experimental parameters are as in Fig.
1͑B͒. Only the indium-bonded portion of the 31P spectra
is shown ͑dotted line: experiment, gray line: least-
squares Lorentzian fit described in the main text͒. The
vertical dashed curve connects the spectral maxima,
emphasizing the downfield shift of the first moment
with increasing ␶cp . Right side: Buildup of the 31P in-
tensities with increasing ␶cp . The extracted time con-
stants (TIS) are given in the figure.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.88.90.140 On: Tue, 09 Dec 2014 11:56:40

J. Chem. Phys., Vol. 110, No. 18, 8 May 1999
Communications
8863
fusion is almost exclusively confined to neighboring 31P spe-
cies. As expected, the spectra show no cross peaks due to
spin diffusion for short mixing times (␶mϭ0.5 ms). Never-
theless, the inhomogeneous character of the 31P spectrum is
clearly exposed by the elongated shape of the signal peaks
along the diagonal. At longer times of several tens of milli-
seconds, cross peaks between all surface sites and the In 31
P
core emerge. The asymmetry of the intensities across the
diagonal indicates a preferential surface→core directed spin-
diffusion process. No polarization transfer among the various
capping resonances could be detected on time scales of 0.1
Ͻ␶mϽ200 ms under MAS or static conditions. This indi-
cates no aggregation of capping molecules on the cluster
surface for dϾ30 Å. Furthermore, broadband21,22 as well as
frequency selective17 spin-diffusion experiments suggest that
the spectral features observed in the 1D data mainly result
from structural variations within the clusters, although con-
tributions of heterogeneities induced by the particle size dis-
tribution cannot be completely excluded. In particular, in
proton driven spin-diffusion experiments17 with 2Ͻ␶
m
Ͻ60 ms ͑data not shown͒ we find that the ␣-InP site exhibits
a rapid polarization transfer to the core and ␤ fraction, but
does not sample its own local environment. Consequently,
␣-31P surface sites share similar electronic surroundings but
are spatially remote from each other. Since XPS data indi-
cate that the large majority of TOPO ligands bind to indium,8
both In 31P surface species can be regarded as uncapped.
Based on the relative chemical-shift differences, we therefore
conclude that the ␤-InP fraction reflects a relatively ‘‘unper-
turbed’’ core-like nearest-neighbor configuration, whereas
␣-In 31P nuclei manifest significantly altered bonding envi-
ronments, possibly due to surface reconstruction to reduce
the number of dangling bonds.
The electronic quantum confinement of the InP particles
is reflected by the 31P chemical shift data: In analogy with
experiments on II–VI clusters,23 we find an increasing up-
field shift with decreasing dot size. Figure 4 illustrates this
behavior by correlating the chemical shift with the inverse
energy corresponding to the average size-dependent band
gap of the clusters. Since size variations in the electronic
structure arise through systematic transformations in the den-
sity of states1,3 ͑i.e., the electronic states shift to higher en-
ergy with the concentration of the oscillator strength into a
few transitions͒, it is surmised that the observed drift in the
chemical shift is largely dominated by the paramagnetic con-
tribution (␴p) to the chemical shielding interaction.24 The
FIG. 3. 2D 31P correlation spectra of TOPO/oxide passivated InP clusters
͑Ref. 20͒ (dӍ32 Å) recorded with ␶cpϭ5 ms (␯masϭ10 kHz). Similar spec-
tra were recorded for TOPO-InP with dӍ45 Å. Protons were decoupled
during the evolution (t1), detection (t2), and mixing (␶m) times; 1024 tran-
sients per t1 value. The data sets (100ϫ100 points͒ were zero filled to 256
points in each dimension and Fourier transformed to obtain pure absorption
mode spectra. 31P spin diffusion was driven by rotor-synchronized 180°
radio-frequency ͑rf͒ pulses during the mixing time ͑Ref. 21͒ and is mani-
fested by the occurrence of cross peaks (␯1Þ␯2) connecting the various
sites on the diagonal ͑thin lines͒. Contours are at 4%, 6%, 16%, 32%, 64%,
and 96% of the maximum signals.
neighbor protons only ͑i.e., r intraϭ2.5 Å), a lower bound
͗
͘
IS
for the 1H–31P distances can be estimated: We obtain
r
r
␣,␤Ӎ3.2 Å ͑a typical H–P van der Waals distance͒ and
͘
͘
͗
͗
IS
IS coreӍ6 Å, indicating that the capping octyl chains are in
closest contact with the semiconductor surface. This conclu-
sion is further corroborated by the observed upfield shift of 3
ppm for the InP-TOPO 13CH2 resonances compared to free
TOPO. Moreover, the core spins probed by the CP experi-
ment with 0.5Ͻ␶cpϽ4 ms reside within the first and second
shell from the crystal surface. For ␶cpϾ50 ms the CP spectra
converge to the single-pulse spectrum shown in Fig. 1͑A͒. In
fact, fitting spectrum 1͑A͒ with the parameters given in
Table I results in a relative InP-surface fraction ͑␣,␤-InP͒ of
24%, in agreement with the previous estimate of ϳ20%.
The spatial proximity of the various 31P sites can be
probed by 2D rf driven spin-diffusion experiments17,21 ͑Fig.
isotropic value of ␴ can be expressed in a perturbation treat-
p
ment as24
␮0e2ប2
8m2e␲
1
ץ
lץ␾
␴
isoϭϪ
0
͉
⌺
͉n
͗
͘
͚
p
ͫ
⌬En
nÞ0
l
ץ
1
ץ
1
ϫ n
͉
͉
⌺
⌺
͉
0 ϩ 0
͉
⌺
͉n
͘
͗
͘
͗
l
l
q3
q3
Ј
Ј
ץ␾
ץ␾
l
l
Ј
Ј
l
l
Ј
Ј
ץ
ϫ n
͗
͉
0
.
͑1͒
͘
lץ␾
ͬ
l
3͒. Since the rate of the dipole induced mutual spin flips
Ϫ3
strongly depends on the nuclear separation (r ),17 spin dif-
Here, me is the mass of the electron, ql and ␾ are the dis-
l
ij
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.88.90.140 On: Tue, 09 Dec 2014 11:56:40

8864
J. Chem. Phys., Vol. 110, No. 18, 8 May 1999
Communications
FIG. 4. Left side: 1D single pulse 31P MAS spectra of
bulk undoped InP and two TOPO/oxide passivated InP
clusters of different diameters; 60 s delay between ex-
periments; ␯masϭ10 kHz. Right side: Plot of the ex-
tracted 31P chemical shift values ͑circles͒ vs. inverse
energy. The nanoparticle UV/vis absorption curves
͑300 K toluene͒ are also displayed for comparison. The
chemical shift data is correlated with the inverse aver-
age band gap energy of the InP particles given by the
point of inflection of the absorption curves ͑vertical
dashed lines͒.
´ ´
tance from the origin and the azimuthal angle that define the
position of electron l in a spherical coordinate system cen-
tered at the nuclear spin with the z axis collinear with B0 .
Summing explicitly over all states n with electronic excita-
tion energies ⌬En is necessary for a rigorous calculation of
␴p . Qualitatively, Eq. ͑1͒ explains the experimental obser-
vation: With decreasing particle size, the electronic excita-
tion energies ⌬En increase and thereby reduce the downfield
paramagnetic chemical shift. The graph in Fig. 4 implies the
crude approximation that ⌬E1Ӷ⌬EnÞ1 . In this limit, the
fairly linear relationship between the 31P shift and the aver-
age inverse lattice band gap energy, extrapolated from the
bulk, suggests that a change from ⌬E1 towards a limiting
energy separation, more suitable for molecular systems, is
not yet reached for an InP cluster with dӍ25 Å.
The results in this study are based upon the sensitivity of
NMR to the local structure and electronic surroundings of
the nuclear spins. A variety of surface components are iden-
tified and characterized using multinuclear polarization
transfer and two-dimensional correlation techniques. We ex-
pect that the application of recent developments in ab initio
and density functional methods for the calculation of mag-
netic resonance parameters25 will complement and further
elucidate our experimental observations in this new class of
materials.
͑1996͒; O. I. Micic, H. M. Cheong, H. Fu, A. Zunger, J. R. Sprague, A.
Mascarenhas, and A. J. Nozik, J. Phys. Chem. 101, 4904 ͑1997͒.
5 H. Fu and A. Zunger, Phys. Rev. B 56, 1496 ͑1997͒.
6 D. L. Klein, P. L. McEuen, J. E. B. Katari, R. Roth, and A. P. Alivisatos,
Appl. Phys. Lett. 68, 2574 ͑1996͒.
7 C.-C. Chen, A. B. Herold, C. S. Johnson, and A. P. Alivisatos, Science
276, 398 ͑1997͒.
8 A. A. Guzelian, J. E. B. Katari, A. V. Kadavanich, U. Banin, K. Hamad,
E. Juban, P. Alivisatos, R. H. Wolters, C. C. Arnold, and J. R. Heath, J.
Phys. Chem. 100, 7212 ͑1996͒.
9 M. Mehring, Principles of High Resolution NMR in Solids ͑Springer, Ber-
lin, 1983͒.
10 M. Tomaselli, D. deGraw, J. L. Yarger, M. P. Augustine, and A. Pines,
Phys. Rev. B 58, 8627 ͑1998͒.
11 D. L. VanderHart, H. S. Gutowsky, and T. C. Farrar, J. Am. Chem. Soc.
89, 5056 ͑1967͒.
12 The 115In NMR signals of the investigated particles were broadened be-
yond detectability due to quadrupolar couplings, possibly caused by slight
bond strains and electric field gradients at or near the surface. Decoupling
at the bulk 115InP resonance frequency ͑Ref. 10͒ had no detectable nar-
rowing effect on the 31P lineshape.
13 A. Pines, M. G. Gibby, and J. S. Waugh, J. Chem. Phys. 59, 569 ͑1973͒.
14 T. S. Lobana, in The Chemistry of Organophosphorous Compounds Vol.
2, edited by F. R. Hartley ͑Wiley, New York, 1992͒.
15 J. R. Sachleben, E. W. Wooten, L. Emsley, A. Pines, V. L. Colvin, and A.
P. Alivisatos, Chem. Phys. Lett. 198, 431 ͑1992͒; L. R. Becerra, C. B.
Murray, R. G. Griffin, and M. G. Bawendi, J. Chem. Phys. 100, 3297
͑1994͒.
16 Since ␯
exceeds the strongest 31P–31P dipolar coupling frequency, the
mas
homonuclear 31P polarization transfer is negligible during the CP period.
17 B. H. Meier, Adv. Magn. Opt. Reson. 18, 1 ͑1994͒; S. M. De Paul, M.
Tomaselli, A. Pines, M. Ernst, and B. H. Meier, J. Chem. Phys. 108, 826
͑1998͒.
ACKNOWLEDGMENTS
18 U. Haeberlen, High Resolution NMR in Solids, Selective Averaging ͑Aca-
demic, New York, 1976͒.
This work was supported by the Director, Office of En-
ergy Research, Office of Basic Energy Sciences, Materials
Sciences Division, U.S. Department of Energy, under con-
tract No. DE-AC03-76SF00098. J.L.Y., M.B., and R.H.H.
acknowledge support from NSF research fellowships. M.T.
acknowledges support from the Swiss National Science
Foundation.
19 Z. Gan, D. M. Grant, and R. R. Ernst, Chem. Phys. Lett. 254, 349 ͑1996͒.
20 TOPO/oxide passivated InP particles showed an additional 31PO4-surface
resonance at 5 ppm (␦m1/2asϭ15 ppm). The assignment is based on
1H→ C→ P double cross-polarization experiments.
21 D. K. Sodickson, M. H. Levitt, S. Vega, and R. G. Griffin, J. Chem. Phys.
98, 6742 ͑1993͒.
13
13
22 M. Baldus, M. Tomaselli, B. H. Meier, and R. R. Ernst, Chem. Phys. Lett.
230, 329 ͑1994͒.
1 L. E. Brus, J. Phys. Chem. 90, 2555 ͑1986͒; A. P. Alivisatos, ibid. 100,
13226 ͑1996͒.
23 A. M. Thayer, M. L. Steigerwald, T. M. Duncan, and D. C. Douglass,
Phys. Rev. Lett. 60, 2673 ͑1988͒.
2 C. B. Murray, D. J. Norris, and M. G. Bawendi, J. Am. Chem. Soc. 115,
8706 ͑1993͒.
24 N. F. Ramsey, Phys. Rev. 78, 699 ͑1950͒.
25 M. Schindler and W. Kutzelnigg, J. Chem. Phys. 76, 1919 ͑1982͒; V. G.
Malkin, O. L. Malkina, and D. R. Salahub, Chem. Phys. Lett. 204, 80
͑1993͒; F. Mauri, B. G. Pfrommer, and S. G. Louie, Phys. Rev. Lett. 77,
5300 ͑1996͒.
3 V. L. Colvin, A. P. Alivisatos, and J. G. Tobin, Phys. Rev. Lett. 66, 2786
͑1991͒; M. G. Bawendi, P. J. Carroll, W. L. Wilson, and L. E. Brus, J.
Chem. Phys. 96, 946 ͑1992͒.
4 S.-H. Kim, R. H. Wolters, and J. R. Heath, J. Chem. Phys. 105, 7957
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.88.90.140 On: Tue, 09 Dec 2014 11:56:40