We demonstrate generation and characterization of crystalline bismuth thin film from triphenyl bismuth in methanol. Upon ultraviolet (267 nm) femtosecond laser irradiation of the solution, a thin film of elemental bismuth forms on the inner side of the sample cuvette, confirmed by detection of the coherent A1g optical phonon mode of crystalline bismuth at ∼90 cm−1. Probe pulses at 267 and 400 nm are used to elucidate the excited state potential energy surface and photochemical reaction coordinate of triphenyl bismuth in solution with femtosecond resolution. The observed phonon mode blueshifts with increasing irradiation time, likely due to the gradual thickening of nascent bismuth thin film to ∼80 nm in 90 min. From transient absorption with the 400 nm probe, we observe a dominant ∼4 ps decay time constant of the excited-state absorption signal, which is attributed to a characteristic metal-ligand bond-weakening/breaking intermediate enroute to crystalline metallic thin film from the solution precursor molecules. Our versatile optical setup thus opens an appealing avenue to characterize the laser-induced crystallization process in situ and prepare high-quality thin films and nanopatterns directly from solution phase.

Among heavy metals and semimetals, bismuth (Bi) is considered to be a relatively green and environmentally benign element. With growing environmental concerns and the need for greener reagents, interest in Bi and its compounds has steadily increased in the past decade for wide applications in semiconductors, medicine, photocatalysis, gas sensing, optical coating, radiation detection, and thermoelectrics.1–3 In particular, triphenyl bismuth (Ph3Bi) has been used as an X-ray contrast agent in orthopedic bone cements,4 additives for plastics in dentistry,5 and in medical implants and sutures.6 From the molecular perspective, this compound contains photolabile Bi–C bonds and is highly soluble in many organic solvents. As a result, the photochemistry of Ph3Bi can be studied in solution as a precursor to Bi-centered radicals and Bi-containing solids.7,8 Such photochemistry is also fundamentally relevant to the emerging area of metal oxide photoresists, which now exhibit the highest patterning performance among all known extreme ultraviolet (EUV) materials.9 

It is known that Ph3Bi has strong absorption in the UV and experiences homolytic bond cleavage under UV irradiation.7,10,11 The timescale for relevant radical formation, however, remains unclear. Such studies are challenging because chemical bond rupture could occur on the molecular timescale down to the femtosecond (fs) regime. Therefore, ultrafast spectroscopic methods are needed to study excited state dynamics of such metal-organic complexes in solution,12–14 particularly regarding the photoinduced metal-to-ligand charge transfer (MLCT) process or metal-ligand bond dissociation. In this work, we study the photochemistry of Ph3Bi as a model system for metal-carbon bond breaking. We generate a thin film of Bi directly from Ph3Bi in methanol solution via fs-laser-induced photolysis and crystallization without use of a chemical reductant. We simultaneously characterize the nascent Bi thin film by analysing a dominant optical phonon mode at ∼90 cm−1 from spectral oscillations of fs transient absorption. The spectral data with characteristic picosecond (ps) time constants shed light on the photochemical scission of Ph3Bi in solution, revealing the formation of an intermediate electronic state on the ∼4 ps timescale. This in-situ methodology thus provides unique insights into Bi–C bond weakening/breaking events and the transition from molecule to condensed solid.

Our fs laser system consists of a mode-locked Ti:sapphire oscillator (Mantis-5) and regenerative amplifier (Legend Elite-USP-1K-HE, Coherent), providing the fundamental pulse (FP) at 800 nm with 35 fs pulse duration, 4.1 mJ pulse energy at 1 kHz repetition rate. Part of the FP passes through a 1-mm-thick type-I beta barium borate (BBO) crystal for second harmonic generation (SHG) to obtain the 400 nm pulse. A calcite plate provides compensation for group velocity delay, and a zero-order dual waveplate (λ/2 at 800 nm and λ at 400 nm) is subsequently used to achieve parallel polarization of the 800 and 400 nm pulses. The FP residual and SHG pulses are directed through another 1-mm-thick type-I BBO to generate the third-harmonic pulse at 267 nm, which is selected with a dichroic mirror to act as the UV pump with ∼0.5 μJ pulse energy [Fig. 1].

FIG. 1.

Schematic of the experimental setup. In typical noncollinear pump-probe geometry, both laser pulses are focused onto the sample cell, while the pump pulse is being chopped. Depending on the probe wavelength, two different detection schemes are used: a lock-in amplifier with Si photodiode for the 267 nm probe and a dispersive spectrograph with CCD array for the 400 nm probe. Bismuth nanocrystal thin film forms on the inner surface of sample cuvette under femtosecond UV-laser irradiation with ∼5-mm beam diameter (bottom left). The characteristic phonon mode is revealed after applying fast Fourier transform (FFT) of the observed spectral oscillations in femtosecond transient absorption (bottom right).

FIG. 1.

Schematic of the experimental setup. In typical noncollinear pump-probe geometry, both laser pulses are focused onto the sample cell, while the pump pulse is being chopped. Depending on the probe wavelength, two different detection schemes are used: a lock-in amplifier with Si photodiode for the 267 nm probe and a dispersive spectrograph with CCD array for the 400 nm probe. Bismuth nanocrystal thin film forms on the inner surface of sample cuvette under femtosecond UV-laser irradiation with ∼5-mm beam diameter (bottom left). The characteristic phonon mode is revealed after applying fast Fourier transform (FFT) of the observed spectral oscillations in femtosecond transient absorption (bottom right).

Close modal

Another portion of FP is used to generate either a 400-nm pulse via SHG in BBO or a 267-nm pulse via the broadband up-converted multicolor array technology in a 1-mm-thick calcite crystal plate.15,16 The probe pulse energy is set at ∼100 nJ. A high-resolution motor-controlled delay stage (07EAS504, CVI Melles Griot) is used to tune the time delay between the s-polarized pump and probe pulses, which are focused onto the sample solution using an f = 10 cm lens or UV-enhanced Al-coated concave mirror with ca. 0.2-mm beam diameter at the focal point. The 267-nm probe past the sample is selected through a UV bandpass filter (FGUV11, Thorlabs), then focused and measured by a Si-biased photodiode detector. The time-dependent signal is collected via a lock-in amplifier synchronized to the optical chopper at 500 Hz. For the 400-nm probe, we recollimate the beam and focus it into the spectrograph (MS127i, Oriel) with a 600-grooves/mm, 400-nm blaze grating. The dispersed signal is collected by a CCD array camera (Princeton Instruments, PIXIS 100F) that is synchronized with the laser source to achieve shot-to-shot spectral acquisition. A typical pump-probe scan with enough time delay points and signal averaging takes ∼10 min. Ph3Bi was purchased from Alfa Aesar and used without further purification. High performance liquid chromatography (HPLC)-grade methanol (EMD Millipore) was dried by standard procedures and freshly distilled prior to use. All experiments are carried out at atmospheric pressure (1 atm) and room temperature (72 °F). To aid the ground-state Raman mode assignment, Ph3Bi is computed in Gaussian at the restricted Hartree-Fock (RHF)/LANL2DZ level with IEFPCM (integral equation formalism polarizable continuum model)-methanol to include the solution effect on the vibrational normal modes (with the frequency scaling factor of 0.96).17 

We first irradiate 0.25M Ph3Bi in methanol solution in a 1-mm-pathlength quartz cell (Spectrosil 1-Q-1, Starna Cells) with 267-nm pump (ca. 2 μJ/pulse, 5-mm beam diameter) for 20 min. A black thin film forms on the inner side of the front surface (facing incident light) of the sample cuvette, while the solution remains clear and colorless [Fig. 1 bottom]. Interestingly, after ∼30 min, if we stop the UV irradiation, this black film becomes invisible. Alternatively, if we open the cell to air, the film rapidly becomes visibly transparent, making it difficult to further characterize the thin film. Given that the black film likely consists of Bi metal as the dominant species (see below), the observed film “disappearance” can be explained as Bi being oxidized to form bismuth oxide (Bi2O3).7 Notably, in comparison to pulsed-laser deposition of bismuth thin film in vacuum chamber,18,19 our solution-based crystallization method is much simpler with less energy footprint.16 For instance, our pump-laser fluence at ∼1 × 10−5 J/cm2 is much lower than previously used values18–20 above ∼1 mJ/cm2 mainly because we use a pump pulse with fs time duration.

To substantiate the merit of fs-laser-induced Bi crystallization, we performed an airtight control experiment. In a 50-ml Schlenk flask, 300-mg Ph3Bi was dissolved in 20-ml dry methanol (sample concentration of ≈34 mM), stirred for 5 min under an atmosphere of Argon, and irradiated by a mercury UV lamp (Porta-Cure 1000, American Ultraviolet, output wavelength of 185–400 nm, non-focused) at 125-W/in. setting for 75 min. After 15 min, the colorless solution changed to light yellow; after 60 min, it became dark brown. The solvent was removed in vacuo to yield a black powder, which mainly contains graphitic carbon and an insignificant amount of elemental Bi as revealed by XRD analysis.21 Decreasing the Ph3Bi concentration to 1 mM produces a light yellow solution without formation of black powder even after 120 min under the same condition. Changing the flask to the 1-mm-thick quartz cuvette makes no clear difference. It shows that the UV lamp cannot effectively generate Bi from Ph3Bi solution, which could be due to much less excitation energy density, more heat-induced degradation, and/or additional photochemistry driven by wavelengths in the UV lamp spectrum that are not in the ultrafast laser bandwidth.

In the UV pump-probe setup with 267 nm pulses, the time-resolved probe transmittance on 1-mM Ph3Bi in methanol solution as a function of irradiation time is presented in Fig. 2. The initial dip in transmittance [Fig. 2(a)] arises from two-photon absorption of methanol, reflecting the cross correlation between the pump and probe pulses (i.e., ∼100 fs instrument response time). This negative peak diminishes with time, which is consistent with the production and crystallization of Bi along the beampath; a control experiment with pure methanol shows the negative peak unchanged with irradiation time. The Bi film formed on the cuvette absorbs at 267 nm, so later datasets develop a positive feature after ∼100 fs that is indicative of electronic ground-state bleaching of Bi crystals.22 After ∼30 min, a clear exponential decay of the positive ΔT with oscillations appears with the time constant of ∼1 ps [Fig. 2(b)]. By subtracting the fitted curve from the time-resolved ΔT trace, the oscillatory residual is obtained [Fig. 2(c)]. Using fast Fourier transform (FFT), the underlying coherent modulation mode is retrieved at ∼90 cm−1, which matches well with the A1g phonon mode of crystalline Bi [Fig. 2(d)].20,23,24 The phonon lifetime is ∼500 fs by least-squares fitting the oscillatory trace to a damped sine wave.24,25 This result indicates that fs-UV pump pulse generates crystalline Bi on the cuvette inner surface, while the transient coherent optical phonon mode is impulsively excited and detected by the pump and probe pulses, respectively.26–28 We conjecture that the focused fs-pump has sufficient peak density to trigger photolysis of solvated Ph3Bi to generate crystalline Bi. After a certain amount of film is formed, the pump pulse induces A1g coherent phonons in the Bi nanocrystal film that modulate its ground-state bleaching signal. The aforementioned phonon decay on the sub-ps timescale is mainly due to scattering with the incoherent population elements including thermal phonons, lattice defects, and electrons.20,25 The observed ∼1-ps exponential decay in Fig. 2(b) corresponds to the excited-state population relaxation time of Bi crystal.20 

FIG. 2.

Pump-probe spectroscopy on 1-mM Ph3Bi in methanol solution with 267 nm pulses. (a) Spectral oscillations after ∼100 fs become prominent in the time-resolved probe transmittance after ∼30 min. The dashed-dotted line represents the zero trace. (b) The ΔT plot over a ∼2 ps time window (black solid line) shows a ∼1 ps exponential decay process (red dashed line). (c) The residual ΔT plot after removing the exponential fit in (b). (d) FFT analysis of the observed coherent oscillations in (c) yields a dominant 90 cm−1 mode, attributed to the A1g phonon motion in a rhombohedral unit cell of crystalline Bi (inset).

FIG. 2.

Pump-probe spectroscopy on 1-mM Ph3Bi in methanol solution with 267 nm pulses. (a) Spectral oscillations after ∼100 fs become prominent in the time-resolved probe transmittance after ∼30 min. The dashed-dotted line represents the zero trace. (b) The ΔT plot over a ∼2 ps time window (black solid line) shows a ∼1 ps exponential decay process (red dashed line). (c) The residual ΔT plot after removing the exponential fit in (b). (d) FFT analysis of the observed coherent oscillations in (c) yields a dominant 90 cm−1 mode, attributed to the A1g phonon motion in a rhombohedral unit cell of crystalline Bi (inset).

Close modal

In order to resolve the phonon dynamics with higher frequency resolution, we further analyze the transient absorption data using discrete Fourier transform (DiFT) without zero padding. Fig. 3(b) shows that the phonon mode blueshifts within ∼30 min, i.e., from ∼86 to 98 cm−1. This result is consistent with the formation and growth of the thin film, because it is observed that the Raman-active A1g mode undergoes a frequency blueshift with increasing thickness and stiffening of the nanocrystal film.29,30 Meanwhile, because Ph3Bi absorbs in the UV, we implement tunable femtosecond stimulated Raman spectroscopy (FSRS) using the ps 400-nm Raman pump via a home-built second harmonic bandwidth compressor31 and the first sideband of sum-frequency-generation-based broadband up-converted multicolor array (SFG-BUMA)32,33 as Raman probe to characterize the Ph3Bi precursor. Fig. 3(a) presents the anti-Stokes FSRS data with pre-resonance Raman enhancement.33,34 A number of Raman modes are clearly resolved within a ∼1600 cm−1 spectral window,35 suggesting that the Franck-Condon region of photoexcited Ph3Bi involves significant benzene ring motions and some degree of Bi–C stretching on the sub-ps timescale.36 Notably, we did not observe a 90 cm−1 mode in the ground-state FSRS data of Ph3Bi solution [Fig. 3(a)], corroborating that the phonon mode is associated with the Bi thin film [Fig. 3(b)].

FIG. 3.

Vibrational characterization of the reactant and product species during crystalline Bi thin film formation. (a) Ground-state FSRS spectrum with 400-nm Raman pump of 0.25M Ph3Bi in methanol. Major atomic motions are labeled. (b) DiFT result of transient absorption data [e.g., Fig. 2(c)] after different time of UV irradiation. The ∼90 cm−1 mode blueshift is indicated by the horizontal arrow.

FIG. 3.

Vibrational characterization of the reactant and product species during crystalline Bi thin film formation. (a) Ground-state FSRS spectrum with 400-nm Raman pump of 0.25M Ph3Bi in methanol. Major atomic motions are labeled. (b) DiFT result of transient absorption data [e.g., Fig. 2(c)] after different time of UV irradiation. The ∼90 cm−1 mode blueshift is indicated by the horizontal arrow.

Close modal

To provide deeper insights into a different region of the multidimensional potential energy surface of Ph3Bi in the photoexcited state, we use a 400-nm probe with the 267-nm pump. The transient electronic dynamics differ between datasets with increasing irradiation time, except that a dominant excited-state absorption feature, i.e., positive ΔOD, with a 4.0 ± 0.6 ps exponential decay constant is observed [Fig. 4 and inset (b)]. The time-resolved data trace after ∼30 min UV irradiation exhibits an oscillation of ∼350 fs period superimposed on the 4-ps decay [Fig. 4 inset (a)]. The phonon lifetime from least-squares fit of the oscillatory component is ∼1 ps, which is consistent with the characteristic damping time of impulsively excited coherent phonons to return to unperturbed positions.20,27 After the initial decay, the first dataset collected immediately following irradiation (0 min) shows an additional ∼165 ps decay (30 wt. %) that could be related to further bond breaking events. The subsequent negative ΔOD signal recovers on the nanosecond time scale (beyond our current experimental time window of ∼600 ps), which is indicative of stimulated emission from an intermediate state. In contrast, following 30 min of UV irradiation, the transient data exhibit complex dynamics, wherein after reaching the minimum at ∼20 ps, the negative ΔOD experiences a slow rise with nanosecond time constant [see Fig. 4].

FIG. 4.

Femtosecond transient absorption spectroscopy of 1-mM Ph3Bi in methanol solution with 267-nm pump and 400-nm probe pulses. Experimental data at 0 min (black, hollow squares) and 30 min (green, hollow circles) are fitted with multiple exponentials, respectively. Inset (a): Time-resolved spectral oscillations within the first ∼2.5 ps after photoexcitation exhibit a dominant ∼90 cm−1 modulation mode. Inset (b): The ∼4 ps initial decay constant from the least-squares fit (solid line) of the excited-state absorption peak kinetic trace up to ∼10 ps. The chemical structure of Ph3Bi is depicted in the inset (below, blue). The proposed main energy level diagram of solvated Ph3Bi is depicted in the right panel. The energy gap in wavelength and characteristic time constants are noted.

FIG. 4.

Femtosecond transient absorption spectroscopy of 1-mM Ph3Bi in methanol solution with 267-nm pump and 400-nm probe pulses. Experimental data at 0 min (black, hollow squares) and 30 min (green, hollow circles) are fitted with multiple exponentials, respectively. Inset (a): Time-resolved spectral oscillations within the first ∼2.5 ps after photoexcitation exhibit a dominant ∼90 cm−1 modulation mode. Inset (b): The ∼4 ps initial decay constant from the least-squares fit (solid line) of the excited-state absorption peak kinetic trace up to ∼10 ps. The chemical structure of Ph3Bi is depicted in the inset (below, blue). The proposed main energy level diagram of solvated Ph3Bi is depicted in the right panel. The energy gap in wavelength and characteristic time constants are noted.

Close modal

This unique series of transient absorption data collected in situ during Bi formation provides mechanistic insights into the Ph3Bi photochemistry in solution. At 0 min, the fs-UV pump pulse excites Ph3Bi molecules from ground state (S0) to first singlet excited state (S1, MLCT), which can further go to a higher lying excited state, e.g., S2 or Sn. The 4-ps exponential decay of Ph3Bi excited-state absorption signal reflects the S1 state relaxation. If all the S1 population returns to S0 on the ps timescale, the transient absorption signal would diminish to zero, but we observe an evolving negative signal. The least-squares fitting reveals a ∼2 ns recovery process that likely arises from stimulated emission of an intermediate state (I1) after photoexcited Ph3Bi navigates out of the Franck-Condon region toward the photolytic reaction state [Fig. 4, right panel]. Before all three Bi–C bonds dissociate to form Bi, some reactive Ph2Bi· or PhBi: radical species may form in a multiquantum process without dominant intramolecular electron transfer.12,37 After ∼30 min of UV irradiation, the nascent crystalline Bi absorbs the 400 nm probe to some degree22 and contributes the ∼1 ps decay (∼45 wt. %) of ground-state bleaching [Fig. 2(b)], which is coupled to stimulated emission of the I1 state (∼2 ns decay, 55 wt. %), leading to the observed complex dynamics. Notably, the observation that the ∼90 cm−1 oscillation associated with impulsively excited A1g coherent phonon motions from the Bi film is pronounced in the 30 min trace [Fig. 4 inset (a)] further substantiates this mechanism. Furthermore, after checking the 400-nm probe counts on CCD camera for each pump-probe scan, we note that at ∼30, 60, and 90 min the transmitted probe intensity decreases by ∼74%, 90%, and 93%, respectively. Since the penetration depth of crystalline Bi upon absorbing 400-nm light (∼3.1 eV) is ∼30 nm (Ref. 24) while Ph3Bi in methanol solution does not absorb at 400 nm, we estimate that the thickness of the nascent crystalline Bi film is ∼40, 70, and 80 nm after 30, 60, and 90 min of UV (267 nm) irradiation, respectively, corroborating the observed phonon frequency blueshift in Fig. 3(b).

To elucidate the ultrafast structural dynamics of this photochemical reaction, tunable FSRS with a UV excitation pulse is needed. The ground-state Raman measurement using SFG-BUMA as Raman probe pulse was performed,33 and three peaks of solvated Ph3Bi stand out in Fig. 3(a): 206 cm−1, Bi–C stretching and Ph3Bi symmetric stretching; 648 cm−1, benzene ring asymmetric in-plane deformation; and 1002 cm−1, benzene ring breathing motion.38 Following photoexcitation, the time-resolved FSRS data of Ph3Bi in solution will expose transient atomic motions by tracking multiple vibrational mode frequencies and intensities35,36 which reveal molecular structural evolution at the chemical bond level beyond the electronic domain [Figs. 2 and 4]. We are in the process of tuning the Raman pump to enhance excited-state FSRS features31,33 particularly within the first ∼10 ps [see Fig. 4, vertical dashed line] and unravel the initial photolysis reaction coordinate that leads to the metal thin film.

In conclusion, we use fs 267-nm pump pulse to generate a crystalline Bi thin film directly from Ph3Bi in methanol solution and simultaneously monitor the non-equilibrium reactant consumption and product formation with a time-delayed fs-probe pulse either at 267 or at 400 nm. We reveal a ∼4 ps time constant associated with the excited-state relaxation of Ph3Bi, which likely involves an intermediate state with weakened and/or broken Bi–C bonds. After a certain thickness (∼40 nm) of Bi film is generated, the coherent A1g phonon mode at ∼90 cm−1 of crystalline Bi is detected in the fs transient absorption and transmittance traces, inferring the dominant electron-lattice coupling in the functional crystalline material. As the nascent Bi film thickens (up to ∼80 nm) with time, a blueshift of the totally symmetric phonon mode is observed. Our setup offers a convenient and viable approach to use fs-laser pulses to generate crystalline metallic thin films from inexpensive solution precursors while offering the opportunity to simultaneously characterize in real time photochemical reaction pathways in electronic and vibrational domains.

This work was supported by the Oregon State University Faculty Startup Research Grant and Research Equipment Reserve Fund (C.F.) and the National Science Foundation (NSF) CAREER award (CHE-1455353, C.F.). We also acknowledge funding from the NSF Center for Sustainable Materials Chemistry (CHE-1102637, L.Z., S.S., and D.A.K.).

1.
G. G.
Briand
and
N.
Burford
,
Chem. Rev.
99
,
2601
(
1999
).
2.
N. M.
Leonard
,
L. C.
Wieland
, and
R. S.
Mohan
,
Tetrahedron
58
,
8373
(
2002
).
3.
D. G.
Cahill
,
P. V.
Braun
,
G.
Chen
,
D. R.
Clarke
,
S.
Fan
,
K. E.
Goodson
,
P.
Keblinski
,
W. P.
King
,
G. D.
Mahan
,
A.
Majumdar
 et al,
Appl. Phys. Rev.
1
,
011305
(
2014
).
4.
H. R.
Rawls
,
R. J.
Granier
,
J.
Smid
, and
I.
Cabasso
,
J. Biomed. Mater. Res.
31
,
339
(
1996
).
5.
L. O.
Reis
,
M. R.
Kaizer
,
F. A.
Ogliari
,
F. M.
Collares
, and
R. R.
Moraes
,
Int. J. Adhes. Adhes.
48
,
80
(
2014
).
6.
J.
Smid
,
I.
Cabasso
,
A.
Obligin
, and
H. R.
Rawls
, U.S. patent No. 4882392 (21 November
1989
).
7.
D. H.
Webber
and
R. L.
Brutchey
,
Inorg. Chem.
50
,
723
(
2011
).
8.
D.
Breitwieser
,
M.
Kriechbaum
,
H. M. A.
Ehmann
,
U.
Monkowius
,
S.
Coseri
,
L.
Sacarescu
, and
S.
Spirk
,
Carbohydr. Polym.
116
,
261
(
2015
).
9.
A.
Grenville
,
J. T.
Anderson
,
B. L.
Clark
,
P.
De Schepper
,
J.
Edson
,
M.
Greer
,
K.
Jiang
,
M.
Kocsis
,
S. T.
Meyers
,
J. K.
Stowers
 et al,
Proc. SPIE
9425
,
94250S
(
2015
).
10.
D. H.
Hey
,
D. A.
Shingleton
, and
G. H.
Williams
,
J. Chem. Soc.
1963
,
5612
.
11.
P.
Weiss
,
Pure Appl. Chem.
15
,
587
(
1967
).
12.
J.
Chen
,
H.
Zhang
,
I. V.
Tomov
,
X.
Ding
, and
P. M.
Rentzepis
,
Proc. Natl. Acad. Sci. U. S. A.
105
,
15235
(
2008
).
13.
A. S.
Mereshchenko
,
S. K.
Pal
,
K. E.
Karabaeva
,
P. Z.
El-Khoury
, and
A. N.
Tarnovsky
,
J. Phys. Chem. A
116
,
2791
(
2012
).
14.
L.
Hua
,
M.
Iwamura
,
S.
Takeuchi
, and
T.
Tahara
,
Phys. Chem. Chem. Phys.
17
,
2067
(
2015
).
15.
W.
Liu
,
L.
Wang
,
F.
Han
, and
C.
Fang
,
Opt. Lett.
38
,
3304
(
2013
).
16.
W.
Liu
,
L.
Wang
,
F.
Han
, and
C.
Fang
,
Appl. Phys. Lett.
103
,
201116
(
2013
).
17.
M. J.
Frisch
,
G. W.
Trucks
,
H. B.
Schlegel
,
G. E.
Scuseria
,
M. A.
Robb
,
J. R.
Cheeseman
,
G.
Scalmani
,
V.
Barone
,
B.
Mennucci
,
G. A.
Petersson
 et al,
Gaussian 09, Revision B.1
(
Gaussian, Inc.
,
Wallingford, CT
,
2009
).
18.
J. C. G.
de Sande
,
T.
Missana
, and
C. N.
Afonso
,
J. Appl. Phys.
80
,
7023
(
1996
).
19.
K.-S.
Wu
and
M.-Y.
Chern
,
Thin Solid Films
516
,
3808
(
2008
).
20.
A. A.
Melnikov
,
O. V.
Misochko
, and
S. V.
Chekalin
,
J. Appl. Phys.
114
,
033502
(
2013
).
21.
S. S.
Djokić
,
N. S.
Djokić
, and
T.
Thundat
,
J. Electrochem. Soc.
159
,
D587
(
2012
).
22.
Y. W.
Wang
,
B. H.
Hong
, and
K. S.
Kim
,
J. Phys. Chem. B
109
,
7067
(
2005
).
23.
J. S.
Lannin
,
J. M.
Calleja
, and
M.
Cardona
,
Phys. Rev. B
12
,
585
(
1975
).
24.
A. A.
Melnikov
,
O. V.
Misochko
, and
S. V.
Chekalin
,
Phys. Lett. A
375
,
2017
(
2011
).
25.
L.
Wang
,
W.
Liu
, and
C.
Fang
,
Phys. Chem. Chem. Phys.
17
,
17034
(
2015
).
26.
T. K.
Cheng
,
S. D.
Brorson
,
A. S.
Kazeroonian
,
J. S.
Moodera
,
G.
Dresselhaus
,
M. S.
Dresselhaus
, and
E. P.
Ippen
,
Appl. Phys. Lett.
57
,
1004
(
1990
).
27.
M. F.
DeCamp
,
D. A.
Reis
,
P. H.
Bucksbaum
, and
R.
Merlin
,
Phys. Rev. B
64
,
092301
(
2001
).
28.
M.
Hase
,
M.
Kitajima
,
S.-i.
Nakashima
, and
K.
Mizoguchi
,
Phys. Rev. Lett.
88
,
067401
(
2002
).
29.
C.
Lee
,
H.
Yan
,
L. E.
Brus
,
T. F.
Heinz
,
J.
Hone
, and
S.
Ryu
,
ACS Nano
4
,
2695
(
2010
).
30.
X.
Luo
,
Y.
Zhao
,
J.
Zhang
,
Q.
Xiong
, and
S. Y.
Quek
,
Phys. Rev. B
88
,
075320
(
2013
).
31.
L.
Zhu
,
W.
Liu
, and
C.
Fang
,
Appl. Phys. Lett.
105
,
041106
(
2014
).
32.
W.
Liu
,
L.
Zhu
, and
C.
Fang
,
Opt. Lett.
37
,
3783
(
2012
).
33.
L.
Zhu
,
W.
Liu
,
Y.
Wang
, and
C.
Fang
,
Appl. Sci.
5
,
48
(
2015
).
34.
A. B.
Myers
and
R. A.
Mathies
, in
Biological Applications of Raman Spectroscopy
, edited by
T. G.
Spiro
(
John Wiley & Sons, Inc.
,
New York
,
1987
), Vol. 2, p.
1
.
35.
W.
Wang
,
W.
Liu
,
I.-Y.
Chang
,
L. A.
Wills
,
L. N.
Zakharov
,
S. W.
Boettcher
,
P. H.-Y.
Cheong
,
C.
Fang
, and
D. A.
Keszler
,
Proc. Natl. Acad. Sci. U. S. A.
110
,
18397
(
2013
).
36.
R. R.
Frontiera
,
C.
Fang
,
J.
Dasgupta
, and
R. A.
Mathies
,
Phys. Chem. Chem. Phys.
14
,
405
(
2012
).
37.
O. P.
Strausz
,
J.
Connor
, and
P. J.
Young
,
J. Am. Chem. Soc.
93
,
822
(
1971
).
38.
C.
Ludwig
,
M.
Dolny
, and
H.-J.
Götze
,
Spectrochim. Acta, Part A
53
,
2363
(
1997
).