VOLUME 79, NUMBER 25 PHYSICAL REVIEW LETTERS 22 DECEMBER 1997

Coherent Acoustic Phonons in a Semiconductor Quantum Dot

Todd D. Krauss and Frank W. Wise
Department of Applied Physics, Cornell University, Ithaca, New York 14853
(Received 10 July 1997)
Coherent acoustic phonons in PbS quantum dots are observed using femtosecond optical techniques.
This is the first observation of coherent acoustic phonons in a semiconductor quantum dot; the phonons
are generated through the deformation-potential coupling to the quantum-dot exciton. The acoustic
modes are weakly damped, and we also find extremely weak coupling (S , 0.01) to the optical modes.
These conclusions have important consequences for the vibronic nature of the exciton transition in the
quantum dot and its dephasing. [S0031-9007(97)04822-9]

PACS numbers: 73.20.Dx, 63.22. + m, 78.47. + p

The electronic and optical properties of semiconductor strong confinement of both charge carriers and phonons.
quantum dots (QD’s) have attracted much attention re- The vibrational modes of PbS QD’s in a polymer host
cently due to their interesting physical properties and po- are well characterized and understood theoretically [13].
tential utility in applications. Exciton-phonon coupling is Here we report the generation of coherent acoustic
a crucial issue in nanocrystal physics, and this has moti- phonons in 3-nm PbS QD’s. Short pulses resonant with
vated much of the previous work on the vibrational modes the lowest exciton transition drive a weakly damped
of quantum dots. The coupling to optical modes pro- spheroidal acoustic mode via the deformation-potential
duces phonon sidebands in the optical transitions. Acous- coupling to the exciton. Interestingly, in the femtosecond
tic modes will also produce sidebands, but the net result of experiments we observe only very weak coupling to an
the coupling to numerous, heavily damped acoustic modes optical mode that is prominent in the (time-integrated)
is an effective broadening of the zero-phonon line [1]. Raman spectrum. The implications of these findings for
Therefore, coupling of quantum-dot excitons to acoustic the exciton line shape will be discussed.
phonons, which generally occurs via the deformation- The vibrations of spherical elastic particles were first
potential interaction, sets a lower limit to the optical studied by Lamb [14]. Provided that the radius is not
linewidth in QD’s. In small dots this linewidth theo- too small, the acoustic modes of a QD are accurately
retically can be much larger than in the corresponding described by similar elastic continuum models [12].
bulk material [2]. Thus, there is ample motivation to Boundary conditions at the surface of the dot result in
develop a more thorough and direct understanding of discrete torsional and spheroidal eigenmodes. The tor-
the deformation-potential coupling of excitons to acoustic sional modes are completely transverse, while the sphe-
phonons in QD’s. The recent observation of ultranarrow roidal modes have mixed longitudinal and transverse
luminescence lines from single QD’s [3,4] adds to the in- character. Confined acoustic vibrations have been ob-
terest in this problem. served in the low-frequency Raman-scattering spectra of
Since the development of reliable sources of sub-100- MgCr2 O4 -MgAl2 O4 and Cd(S,Se) nanocrystals [11,12].
fs optical pulses, researchers have been using these pulses The mode frequencies and their dependence on size agree
to generate and monitor terahertz-frequency vibrations in well with calculations based on the continuum models.
materials. Coherent vibrations have been produced in The effects of confinement on the vibrational modes
molecules, semiconductors, insulators, and metals, as well of QD’s of IV-VI materials are dramatic owing to the
as in artificial structures such as superlattices. (For a com- dispersive phonons and unusual dielectric properties of
prehensive review of this area, see Ref. [5].) Coherent the bulk materials. The vibrational modes of PbS QD’s
optical phonons have been reported in QD’s of CdSe [6], in a polymer host were determined using resonant-Raman
InP [7], and PbS [8]. and far-infrared absorption spectroscopies [13]. The re-
The optical vibrational modes of QD’s are typically un- sults of those experiments agree with theoretical calcu-
derdamped, and couple strongly to the electronic states via lations of the optical vibrational modes that account for
the Fröhlich interaction [9,10]. The acoustic modes of the mechanical and electrostatic boundary conditions
QD’s are typically overdamped [11,12], and in some cases [15,16]. The Raman-scattering spectrum of PbS QD’s of
are not observed in the Raman spectrum [9,10]. Despite diameter 3 nm is shown in Fig. 1(a). The broad peak at
the fact that the acoustic phonon spectrum of a quantum ,215 cm21 is assigned to the lowest-order optical
dot is theoretically discrete, there has been no previous re- mode with angular momentum , ­ 0. This mode is
port of coherent acoustic phonons in a semiconductor QD. purely radial and has eigenfunctions similar to the , ­ 0
Because of their large Bohr radii and dispersive pho- longitudinal-optical (LO) vibrational modes obtained
nons, the lead salts offer unique opportunities for studying using dielectric theories [9,17]. The sharp peak at

5102 0031-9007y97y79(25)y5102(4)$10.00 © 1997 The American Physical Society

VOLUME 79, NUMBER 25 PHYSICAL REVIEW LETTERS 22 DECEMBER 1997

Time-domain data were obtained by measuring satu-

rated absorption in a standard pump-probe setup. Pulses
from a Ti:sapphire regenerative amplifier were used to
pump an optical parametric amplifier [20], the output of
which was frequency-doubled to produce 50-fs pulses,
wavelength-tunable across the lowest exciton transition at
600 nm. At the sample the excitation pulses had intensi-
ties of 5210 GWycm2 . The absorption of PbS nanocrys-
tals in poly(vinyl alcohol) can be permanently bleached
by strong illumination, so the experiments were done
with low pulse energies (,10 nJ) to avoid any bleach-
ing effects. We were particularly careful to ensure that
measurements were made on fresh material and that no
permanent photoinduced changes occurred during the ac-
quisition of each experimental trace.
The transient response of the QD’s is shown in Fig. 2.
The differential transmission of the probe pulse DT yT
is plotted versus the time delay between the pump and
probe pulses. The damped oscillation due to the , ­ 0
acoustic mode is clearly evident, superimposed on the
absorption bleaching due to saturation of the exciton.
The bleaching recovers as the exciton population decays,
probably into defect states. The Fourier transform of the
time-domain data is divided by the pulse spectrum to
account for the filtering of the finite spectral width and
the result is shown in Fig. 1(b) for comparison to the
FIG. 1. Comparison of Raman-scattering and time-domain re- Raman scattering spectrum. The lowest exciton transition
sults. (a) First-order Raman-scattering spectrum of 3-nm PbS of lead-salt QD’s is well approximated as a two-level
nanocrystals recorded at 4.2 K with excitation at 584 nm. system coupled to a vibrational mode, and in that case
Similar spectra were obtained for excitation wavelengths be- the Fourier transform of the time-domain data should be
tween 580 and 630 nm and temperatures between 4 and 300 K. identical to the Raman-scattering spectrum [5]. The 70-
(b) Fourier-transform amplitude of the time-domain trace shown
in Fig. 2. The spectrum has been divided by the power spec- cm21 frequency and the damping of the acoustic mode
trum of the laser pulse to account for the effect of the finite agree well with those measured by Raman scattering.
pulse duration. On the other hand, the , ­ 0 optical mode is barely
observable in the Fourier transform amplitude or through

,70 cm21 is due to the lowest-order spheroidal acoustic

mode with , ­ 0. Briefly, we reach this conclusion
based on the magnitude and polarization properties of
the peak and the agreement with the theoretical fre-
quency va ­ 68 cm21 [13]. There is also a small but
reproducible peak at 45 cm21 due to the lowest-order
spheroidal acoustic mode with , ­ 2.
Here we are primarily interested in the spheroidal
acoustic mode with , ­ 0. Based on the Raman spec-
trum, it should be possible to drive it impulsively with
pulses shorter than ,150 fs. With ,50-fs pulses the op-
tical mode will also be excited, so a superposition of the
two modes is expected.
Measurements were performed on PbS QD’s synthe- FIG. 2. The main figure shows a transient saturated-
sized in an aqueous solution of poly(vinyl alcohol) (PVA) absorption signal. The inset shows a three pulse photon echo
following the procedure in Ref. [18] and subsequently signal as a function of delay between the first and third pulses,
dried into thin films. X-ray scattering confirms the the delay for which the signal best exhibits the consequences
of exciton-phonon coupling. The solid line is experimental
sodium-chloride crystal structure with the bulk PbS lat- data and the dotted line is a theoretical fit using the standard
tice constant. The nanocrystal size is 3 nm, with a varia- analysis as in Ref. [6]. The exponential decay due to recovery
tion of 64% [19]. of the exciton population has been subtracted.

5103
VOLUME 79, NUMBER 25 PHYSICAL REVIEW LETTERS 22 DECEMBER 1997

its effect on the shape of the first few cycles of the time- vibration may be damped by radiating energy into the
domain trace; this will be explained below. host. The lack of host vibrational modes in the vicinity
In similar experiments on PbS QD’s, Machol et al. of the acoustic mode frequency may contribute to the
previously reported the observation of coherent oscilla- observed narrow linewidth, although there is no obvious
tions with a frequency of 90 cm21 [8]. This frequency reason why PbS should be fundamentally different from
matches the frequency of a coupled optical mode with other QD’s in polymer hosts. The density of acoustic
, ­ 1, which is not Raman active. We have not ob- modes in the PbS QD’s is somewhat lower than that in
served the 90-cm21 oscillation in an extensive series of QD’s studied previously, which contributes to isolating
nonlinear-absorption and four-wave-mixing experiments the observed spheroidal mode.
that includes the results described here. We cannot re- The relative couplings of the electronic states to
solve this discrepancy with certainty, but we did demon- acoustic and optical modes determines the intrinsic QD
strate that some photobleaching of the PbSyPVA samples lineshape and so deserves some discussion. The time-
is required to produce the induced-absorption signals ob- integrated Raman spectrum (fundamental and overtones)
served by Machol et al. Thus, the appearance of the cou- implies that the coupling to the , ­ 0 optical mode
pled mode may be peculiar to the photobleached sample. is large (Soptical , 0.7) and much larger than the cou-
The interaction of electrons with acoustic phonons pling to the , ­ 0 acoustic mode [21]. This would
arises primarily from deformation-potential and piezo- imply that the optical mode should appear as a large-
electric coupling. PbS is not piezoelectric. Takagahara amplitude oscillation in the first picosecond of the
has shown that even in piezoelectric materials the time-domain trace. However, it is barely observable in
deformation-potential coupling should dominate in small the time-resolved experiment. Briefly, we suggest the
QD’s [2]. We calculated the coupling strength using following explanation [21] for this discrepancy. Trapped
the exciton wave functions of Ref. [19] and the results charge builds up on the QD’s during the steady-state
are shown in Fig. 3 for several values of the difference Raman measurements, producing local electric fields. In
between the electron and hole deformation potentials the presence of trapped charge, the coupling of excitons to
sDe 2 Dh d. The Huang-Rhys parameter S (equal to the polar phonons increases by at least an order of magnitude
interaction energy in units of the phonon energy) is a [17,21]. A measurement that is performed before either
measure of the exciton-phonon coupling strength. For charge carrier is trapped (which occurs in picoseconds)
the 3-nm QD’s studied here, Sacoustic , 0.0120.1 for should detect the intrinsic coupling to optical phonons.
reasonable values of De 2 Dh . This is expected to be very weak in PbS QD’s (Soptical
The weak damping of the spheroidal acoustic mode in Fig. 3). Thus, the femtosecond time-resolved experi-
facilitates its observation. To our knowledge there is no ments are sensitive only to the weak coupling of the
treatment of the damping of the vibrational modes of a intrinsic wave function to the optical phonons. The cou-
QD, although it is plausible that there are fewer relaxation pling to acoustic modes is not essentially electromagnetic,
pathways in smaller QD’s. We find that the width of and so should not be distorted by the trapped charge and
the Raman peak increases by less than 5% between 4 associated fields. The trapped charge decays in less than
and 300 K. This suggests that the dominant damping the 1 ms between pulses, so each pair of pump and probe
mechanism does not involve other phonons, i.e., that the pulses encounters unexcited material.
The generation of coherent acoustic phonons and the
absence of coherent optical phonons contrast sharply with
previous studies of CdSe and InP QD’s [6,7], suggesting
major differences in the exciton-phonon coupling between
IV-VI and II-VI or III-V materials. Quantitative exciton-
phonon couplings cannot be inferred from saturated-
absorption data. To obtain the exciton- phonon couplings,
we performed three-pulse photon echo (3PPE) measure-
ments, in which the vibrations modulate the electronic
polarization. From those measurements (Fig. 2, inset)
we infer Sacoustic ­ 0.1 and Soptical ­ 0.01. The relative
strengths are consistent with the ratio of the intensities
in the Fourier transform of the saturated-absorption data
[Fig. 1(b)]. The magnitude of Sacoustic is also consistent
with the lack of a strong overtone of the acoustic mode in
FIG. 3. Calculated values of the Huang-Rhys factor S for the the Raman spectrum, and agrees with the calculated value
coupling of the lowest exciton to the spheroidal acoustic mode.
De 2 Dh has units of eV. S is the interaction energy in units of Sacoustic for reasonable values of (De 2 Dh ).
of the phonon energy. The polar coupling to the optical mode The occurrence of a small number of underdamped
with , ­ 0 is also shown. acoustic modes has profound implications for the

5104
VOLUME 79, NUMBER 25 PHYSICAL REVIEW LETTERS 22 DECEMBER 1997

homogeneous line shape of a QD. Underdamped optical contribute to irreversible dephasing of the exciton in
modes produce sidebands. In principle, acoustic modes these QD’s.
also produce sidebands that will be observed in the time The authors thank I. Kang for assistance with calcula-
domain as a coherent modulation or beat of the electronic tions. This work was supported by the National Science
polarization. However, a quasicontinuum of overdamped Foundation under Grant No. DMR-9321259, and made
acoustic modes will produce an irreversible decay use of Materials Science Center facilities supported by
of the dipole coherence and contribute to an effective Grant No. DMR-9121654. T. D. K. acknowledges the
broadening of the zero-phonon line. This is the case support of the New Focus Award of the Optical Society
in previous investigations of the exciton dephasing in of America.
CdSe and InP QD’s [6,7]. The acoustic modes were not
observed experimentally; the quasicontinuum of acoustic
modes (heat bath) was modeled as a single mode. It is [1] S. Schmitt-Rink, D. A. B. Miller, and D. S. Chemla, Phys.
interesting to try to connect the observation of coherent Rev. B 35, 8113 (1987).
acoustic phonons with the narrow (,100 meV) lumines- [2] T. Takagahara, Phys. Rev. Lett. 71, 3577 (1993).
cence linewidths of single CdSe QD’s [3]. Empedocles [3] S. A. Empedocles, D. J. Norris, and M. G. Bawendi, Phys.
et al. conclude that coupling to acoustic modes is negli- Rev. Lett. 77, 3873 (1996).
gible [3], but it is impossible to estimate the coupling [4] J.-Y. Marzin, J.-M. Gerard, A. Izrael, D. Barrier, and
with any accuracy without knowing the frequency of the G. Bastard, Phys. Rev. Lett. 73, 716 (1994).
mode. When the acoustic modes are underdamped, the [5] R. Merlin, Solid State Commun. 102, 207 (1997).
[6] R. W. Schoenlein, D. M. Mittleman, J. J. Shiang, A. P.
homogeneous linewidth of the zero-phonon line is not
Alivisatos, and C. V. Shank, Phys. Rev. Lett. 70, 1014
determined by the envelope of phonon sidebands, but by (1993).
energy fluctuations induced by multiphonon processes [7] U. Banin, G. Cerullo, A. A. Guzelian, C. J. Bardeen, A. P.
or eventually the decay of the exciton population. Thus Alivisatos, and C. V. Shank, Phys. Rev. B 55, 7059 (1997).
strong coupling to acoustic modes does not necessarily [8] J. L. Machol, F. W. Wise, R. C. Patel, and D. B. Tanner,
imply rapid dephasing and a broad homogeneous line. Phys. Rev. B 48, 2819 (1993).
In previous studies of the absorbing transition, strong [9] M. C. Klein, F. Hache, D. Ricard, and C. Flytzanis, Phys.
coupling to a few modes broadens the Franck-Condon Rev. B 42, 11 123 (1990).
envelope, and the zero-phonon line is lost. The results [10] J. J. Shiang, S. H. Risbud, and A. P. Alivisatos, J. Chem.
presented here indicate that in the strong-confinement Phys. 98, 8432 (1993).
limit the vibronic transitions are in fact recovered. [11] E. Duval, A. Boukenter, and B. Champagnon, Phys. Rev.
Lett. 56, 2052 (1986).
Measurements of the exciton dipole dephasing will be
[12] B. Champagnon, B. Andrianasolo, and E. Duval, J. Chem.
very interesting but require better sensitivity and time Phys. 94, 5237 (1991).
resolution than the experiments described here. [13] T. D. Krauss, F. W. Wise, and D. B. Tanner, Phys. Rev.
A systematic study of the acoustic modes of PbS QD’s Lett. 76, 1376 (1996).
with variable nanocrystal size will also be valuable. The [14] H. Lamb, Proc. London Math. Soc. 13, 187 (1882).
fabrication of monodisperse PbS quantum dots with vari- [15] E. Roca, C. Trallero-Giner, and M. Cardona, Phys. Rev. B
able particle size in a polymer host has not been reported. 49, 13 704 (1994).
Saturated absorption measurements of monodisperse PbS [16] M. P. Chamberlain, C. Trallero-Giner, and M. Cardona,
quantum dots with controllable particle size (diameter Phys. Rev. B 51, 1680 (1995).
$4 nm) in an oxide glass host show no evidence of co- [17] S. Nomura and T. Kobayashi, Phys. Rev. B 45, 1305
herent phonons. 3PPE measurements are in progress and (1992).
[18] M. T. Nenadovic, M. I. Comor, V. Vasic, and O. I. Micic,
should provide the magnitude of the exciton-phonon cou-
J. Phys. Chem. 94, 6930 (1990).
pling in these samples. [19] I. Kang and F. W. Wise, J. Opt. Soc. Am. B 14, 1632
In conclusion, we observe coherent spheroidal acoustic (1997).
phonons in PbS QD’s. The underdamped acoustic mode [20] K. Wilson and V. Yakovlev, J. Opt. Soc. Am. B 14, 444
appears as a phonon sideband, rather than contributing to (1997).
the width of the zero-phonon line of the exciton transition. [21] T. D. Krauss and F. W. Wise, Phys. Rev. B 55, 9860
Thus, first-order coupling to the acoustic mode does not (1997).

5105