ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/chuckDissertation/bulkmod.tex
Revision: 3496
Committed: Wed Apr 8 19:13:41 2009 UTC (16 years ago) by chuckv
Content type: application/x-tex
File size: 23252 byte(s)
Log Message:
Final Version

File Contents

# Content
1 %!TEX root = /Users/charles/Documents/chuckDissertation/dissertation.tex
2 \chapter{\label{chap:bulkmod}BREATHING MODE DYNAMICS AND ELASTIC PROPERTIES OF GOLD NANOPARTICLES}
3
4 In metallic nanoparticles, the relatively large surface area to volume ratio
5 induces a number of well-known size-dependent phenomena. Notable
6 among these are the depression of the bulk melting
7 temperature,\cite{Buffat:1976yq,el-sayed00,el-sayed01} surface melting
8 transitions, increased room-temperature alloying
9 rates,\cite{ShibataT._ja026764r} changes in the breathing mode
10 frequencies,\cite{delfatti99,henglein99,hartland02a,hartland02c} and
11 rapid heat transfer to the surrounding
12 medium.\cite{HuM._jp020581+,hartland02d}
13
14 In this chapter, atomic-level simulations of the transient
15 response of metallic nanoparticles to the nearly instantaneous heating
16 undergone when photons are absorbed during ultrafast laser excitation
17 experiments are discussed. The time scale for heating (determined by the {\it e-ph}
18 coupling constant) is faster than a single period of the breathing
19 mode for spherical nanoparticles.\cite{Simon2001,HartlandG.V._jp0276092}
20 Hot-electron pressure and direct lattice heating contribute to the
21 thermal excitation of the atomic degrees of freedom, and both
22 mechanisms are rapid enough to coherently excite the breathing mode of
23 the spherical particles.\cite{Hartland00}
24
25 There are questions posed by the experiments that may be easiest to
26 answer via computer simulation. For example, the dephasing seen
27 following coherent excitation of the breathing mode may be due to
28 inhomogeneous size distributions in the sample, but it may also be due
29 to softening of the breathing mode vibrational frequency following a
30 melting transition. Additionally, there are properties (such as the
31 bulk modulus) that may be nearly impossible to obtain experimentally,
32 but which are relatively easily obtained via simulation techniques.
33
34 We outline our simulation techniques in section \ref{bulkmod:sec:details}.
35 Results are presented in section \ref{bulkmod:sec:results}. We discuss our
36 results in terms of Lamb's classical theory of elastic spheres \cite{Lamb1882} in
37 section \ref{bulkmod:sec:discussion}.
38
39 \section{Computational Details}
40 \label{bulkmod:sec:details}
41
42 Spherical Au nanoparticles were created in a standard FCC lattice at
43 four different radii [20{\AA} (1926 atoms), 25{\AA} (3884 atoms),
44 30{\AA} (6602 atoms), and 35{\AA} (10606 atoms)]. To create spherical
45 nanoparticles, a large FCC lattice was built at the normal Au lattice
46 spacing (4.08 \AA) and any atoms outside the target radius were
47 excluded from the simulation.
48
49 \subsection{Simulation Methodology}
50
51 Potentials were calculated using the Voter-Chen
52 parameterization~\cite{Voter:87} of the Embedded Atom Method ({\sc
53 eam}) as described in Section \ref{introSec:eam} of this dissertation.
54 Before starting the molecular dynamics runs, a relatively short
55 steepest-descent minimization was performed to relax the lattice in
56 the initial configuration. To facilitate the study of the larger
57 particles, simulations were run in parallel over 16 processors using
58 Plimpton's force-decomposition method.\cite{plimpton93}
59
60 To mimic the events following the absorption of light in the ultrafast
61 laser heating experiments, we have used a simple two-step process to
62 prepare the simulations. Instantaneous heating of the lattice was
63 performed by sampling atomic velocities from a Maxwell-Boltzmann
64 distribution set to twice the target temperature for the simulation.
65 By equipartition, approximately half of the initial kinetic energy of
66 the system winds up in the potential energy of the system. The system
67 was a allowed a very short (10 fs) evolution period with the new
68 velocities.
69
70 Following this excitation step, the particles evolved under
71 Nos\'{e}-Hoover NVT dynamics~\cite{hoover85} for 40 ps. Given the
72 mass of the constituent metal atoms, time steps of 5 fs give excellent
73 energy conservation in standard NVE integrators, so the same time step
74 was used in the NVT simulations. Target temperatures for these
75 particles spanned the range from 300 K to 1350 K in 50 K intervals.
76 Five independent samples were run for each particle and temperature.
77 The results presented below are averaged properties for each of the
78 five independent samples.
79
80 \subsection{\label{bulkmod:sec:analysis}Analysis}
81
82 Of primary interest when comparing our simulations to experiments is
83 the dynamics of the low-frequency breathing mode of the particles. To
84 study this motion, access is needed to accurate measures of both the
85 volume and surface area as a function of time. Throughout the
86 simulations, both quantities were monitored using Barber {\it et al.}'s
87 very fast quickhull algorithm to obtain the convex hull for the
88 collection of 3-d coordinates of all of atoms at each point in
89 time.~\cite{barber96quickhull,qhull} The convex hull is the smallest convex
90 polyhedron which includes all of the atoms, so the volume of this
91 polyhedron is an excellent estimate of the volume of the nanoparticle.
92 The convex hull estimate of the volume will be problematic if the
93 nanoparticle breaks into pieces (i.e. if the bounding surface becomes
94 concave), but for the relatively short trajectories of this study, it
95 provides an excellent measure of particle volume as a function of
96 time.
97
98 The bulk modulus, which is the inverse of the compressibility,
99 \begin{equation}
100 K = \frac{1}{\kappa} = - V \left(\frac{\partial P}{\partial
101 V}\right)_{T}
102 \label{eq:Kdef}
103 \end{equation}
104 can be related to quantities that are relatively easily accessible
105 from our molecular dynamics simulations. Four
106 different approaches are presented here for estimating the bulk modulus directly from
107 basic or derived quantities:
108
109 \begin{itemize}
110 \item The traditional ``Energetic'' approach
111
112 Using basic thermodynamics and one of the Maxwell relations, it can be written that
113 \begin{equation}
114 P = T\left(\frac{\partial S}{\partial V}\right)_{T} -
115 \left(\frac{\partial U}{\partial V}\right)_{T}.
116 \label{eq:Pdef}
117 \end{equation}
118 It follows that
119 \begin{equation}
120 K = V \left(T \left(\frac{\partial^{2} S}{\partial V^{2}}\right)_{T} -
121 \left(\frac{\partial^{2} U}{\partial V^{2}}\right)_{T} \right).
122 \label{eq:Ksub}
123 \end{equation}
124 The standard practice in solid state physics is to assume the low
125 temperature limit ({\it i.e.} to neglect the entropic term), which
126 means the bulk modulus is usually expressed
127 \begin{equation}
128 K \approx V \left(\frac{\partial^{2} U }{\partial V^{2}}\right)_{T}.
129 \label{eq:Kuse}
130 \end{equation}
131 When the relationship between the total energy $U$ and volume $V$ of
132 the system are available at a fixed temperature (as it is in these
133 simulations), it is a simple matter to compute the bulk modulus from
134 the response of the system to the perturbation of the instantaneous
135 heating. Although this information would, in theory, be available
136 from longer constant temperature simulations, the ranges of volumes
137 and energies explored by a nanoparticle under equilibrium conditions
138 are actually quite small. Instantaneous heating, since it excites
139 coherent oscillations in the breathing mode, allows us to sample a
140 much wider range of volumes (and energies) for the particles. The
141 problem with this method is that it neglects the entropic term near
142 the melting transition, which gives spurious results (negative bulk
143 moduli) at higher temperatures.
144
145 \item The Linear Response Approach
146
147 Linear response theory gives us another approach to calculating the
148 bulk modulus. This method relates the low-wavelength fluctuations in
149 the density to the isothermal compressibility,\cite{BernePecora}
150 \begin{equation}
151 \underset{k \rightarrow 0}{\lim} \langle | \delta \rho(\vec{k}) |^2
152 \rangle = \frac{k_B T \rho^2 \kappa}{V}
153 \end{equation}
154 where the frequency dependent density fluctuations are the Fourier
155 transform of the spatial fluctuations,
156 \begin{equation}
157 \delta \rho(\vec{k}) = \underset{V}{\int} e^{i \vec{k}\cdot\vec{r}} \left( \rho(\vec{r}, t) - \langle \rho \rangle \right) dV
158 \end{equation}
159 This approach is essentially equivalent to using the volume
160 fluctuations directly to estimate the bulk modulus,
161 \begin{equation}
162 K = \frac{V k_B T}{\langle \delta V^2 \rangle} = k_B T \frac{\langle V \rangle}{\langle V^2 \rangle - \langle V \rangle^2}
163 \end{equation}
164 It should be noted that in these experiments, the particles are {\it
165 far from equilibrium}, and so a linear response approach will not be
166 the most suitable way to obtain estimates of the bulk modulus.
167
168 \item The Extended System Approach
169
170 Since these simulations are being performed in the NVT ensemble using
171 Nos\'e-Hoover thermostatting, the quantity conserved by our integrator
172 ($H_{NVT}$) can be expressed as:
173 \begin{equation}
174 H_{NVT} = U + f k_B T_{ext} \left( \frac{\tau_T^2 \chi(t)^2}{2} +
175 \int_0^t \chi(s) ds \right).
176 \end{equation}
177 Here $f$ is the number of degrees of freedom in the (real) system,
178 $T_{ext}$ is the temperature of the thermostat, $\tau_T$ is the time
179 constant for the thermostat, and $\chi(t)$ is the instantaneous value
180 of the extended system thermostat variable. The extended Hamiltonian
181 for the system, $H_{NVT}$ is, to within a constant, the Helmholtz free
182 energy.\cite{melchionna93} Since the pressure is a simple derivative
183 of the Helmholtz free energy,
184 \begin{equation}
185 P = -\left( \frac{\partial A}{\partial V} \right)_T ,
186 \end{equation}
187 the bulk modulus can be obtained (theoretically) by a quadratic fit of
188 the fluctuations in $H_{NVT}$ against fluctuations in the volume,
189 \begin{equation}
190 K = -V \left( \frac{\partial^2 H_{NVT}}{\partial V^2} \right)_T.
191 \end{equation}
192 However, $H_{NVT}$ is essentially conserved during these simulations,
193 so fitting fluctuations of this quantity to obtain meaningful physical
194 quantities is somewhat suspect. It should also noted that this method would
195 fail in periodic systems because the volume itself is fixed in
196 periodic NVT simulations.
197
198 \item The Direct Pressure Approach
199
200 Our preferred method for estimating the bulk modulus is to compute it
201 {\it directly} from the internal pressure in the nanoparticle. The
202 pressure is obtained via the trace of the pressure tensor,
203 \begin{equation}
204 P = \frac{1}{3}
205 \mathrm{Tr}\left[\overleftrightarrow{\mathsf{P}}\right],
206 \end{equation}
207 which has a kinetic contribution as well as a contribution from the
208 stress tensor ($\overleftrightarrow{\mathsf{W}}$):
209 \begin{equation}
210 \overleftrightarrow{\mathsf{P}} = \frac{1}{V} \left(
211 \sum_{i=1}^{N} m_i \vec{v}_i \otimes \vec{v}_i \right) +
212 \overleftrightarrow{\mathsf{W}}.
213 \end{equation}
214 Here the $\otimes$ symbol represents the {\it outer} product of the
215 velocity vector for atom $i$ to yield a $3 \times 3$ matrix. The
216 virial is computed during the simulation using forces between pairs of
217 particles,
218 \begin{equation}
219 \overleftrightarrow{\mathsf{W}} = \sum_{i} \sum_{j>i} \vec{r}_{ij} \otimes \vec{f}_{ij}
220 \end{equation}
221 During the simulation, the internal pressure, $P$, is recorded as well as
222 the total energy, $U$, the extended system's hamiltonian, $H_{NVT}$,
223 and the particle coordinates. Once the time
224 dependent volume of the nanoparticle has been calculated using the convex hull,
225 any of these four methods can be used to estimate the bulk modulus.
226 \end{itemize}
227
228 It is found, however, that only the fourth method (the direct pressure
229 approach) gives meaningful results. Bulk moduli for the 35 \AA\
230 particle were computed with the traditional (Energy vs. Volume)
231 approach as well as the direct pressure approach. A comparison of the
232 Bulk Modulus obtained via both methods and are shown in
233 Fig. \ref{fig:Methods} Note that the second derivative fits in the
234 traditional approach can give (in the liquid droplet region) negative
235 curvature, and this results in negative values for the bulk modulus.
236
237 \begin{figure}[htbp]
238 \centering
239 \includegraphics[height=3in]{images/Methods.pdf}
240 \caption{Comparison of two of the methods for estimating the bulk
241 modulus as a function of temperature for the 35\AA\ particle.}
242 \label{fig:Methods}
243 \end{figure}
244
245 The Bulk moduli reported in the rest of this paper were computed using
246 the direct pressure method.
247
248 To study the frequency of the breathing mode, the
249 power spectrum for volume ($V$) fluctuations have been calculated according to,
250 \begin{equation}
251 \rho_{\Delta V}(\omega) = \int_{-\infty}^{\infty} \langle \Delta V(t)
252 \Delta V(0) \rangle e^{-i \omega t} dt
253 \label{eq:volspect}
254 \end{equation}
255 where $\Delta V(t) = V(t) - \langle V \rangle$. Because the
256 instantaneous heating excites all of the vibrational modes of the
257 particle, the power spectrum will contain contributions from all modes
258 that perturb the total volume of the particle. The lowest frequency
259 peak in the power spectrum should give the frequency (and period) for
260 the breathing mode, and these quantities are most readily compared
261 with the Hartland group experiments.\cite{HartlandG.V._jp0276092} Further
262 analysis of the breathing dynamics follows in section
263 \ref{bulkmod:sec:discussion}.
264
265 The heat capacity for our simulations has also been computed to verify
266 the location of the melting transition. Calculations of the heat
267 capacity were performed on the non-equilibrium, instantaneous heating
268 simulations, as well as on simulations of nanoparticles that were at
269 equilibrium at the target temperature.
270
271 \section{Results}
272 \label{bulkmod:sec:results}
273
274 \subsection{The Bulk Modulus and Heat Capacity}
275
276 The upper panel in Fig. \ref{fig:BmCp} shows the temperature
277 dependence of the Bulk Modulus ($K$). In all samples, there is a
278 dramatic (size-dependent) drop in $K$ at temperatures well below the
279 melting temperature of bulk polycrystalline gold. This drop in $K$
280 coincides with the actual melting transition of the nanoparticles.
281 Surface melting has been confirmed at even lower temperatures using
282 the radial-dependent density, $\rho(r) / \rho$, which shows a merging
283 of the crystalline peaks in the outer layer of the nanoparticle.
284 However, the bulk modulus only has an appreciable drop when the
285 particle melts fully.
286
287 \begin{figure}[htbp]
288 \centering
289 \includegraphics[height=3in]{images/Stacked_Bulk_modulus_and_Cp.pdf}
290 \caption{The temperature dependence of the bulk modulus (upper panel)
291 and heat capacity (lower panel) for nanoparticles of four different
292 radii. Note that the peak in the heat capacity coincides with the {\em
293 start} of the peak in the bulk modulus.}
294 \label{fig:BmCp}
295 \end{figure}
296
297
298 Another feature of these transient (non-equilibrium) calculations is
299 the width of the peak in the heat capacity. Calculation of $C_{p}$
300 from longer equilibrium trajectories should indicate {\it sharper}
301 features in $C_{p}$ for the larger particles. Since the melting process itself is being initiated and observed in these calculations, the
302 smaller particles melt more rapidly, and thus exhibit sharper features
303 in $C_{p}$. Indeed, longer trajectories do show that $T_{m}$ occurs
304 at lower temperatures and with sharper transitions in larger particles
305 than can be observed from transient response calculations.
306 Fig. \ref{fig:Cp2} shows the results of 300 ps simulations which give
307 much sharper and lower temperature melting transitions than those
308 observed in the 40 ps simulations.
309
310 \begin{figure}[htbp]
311 \centering
312 \includegraphics[height=3in]{images/Cp_vs_T.pdf}
313 \caption{The dependence of the spike in the heat capacity on the
314 length of the simulation. Longer heating-response calculations result
315 in melting transitions that are sharper and lower in temperature than
316 the short-time transient response simulations. Shorter runs don't
317 allow the particles to melt completely.}
318 \label{fig:Cp2}
319 \end{figure}
320
321
322
323 \subsection{Breathing Mode Dynamics}
324
325 Fig.\ref{fig:VolTime} shows representative samples of the volume
326 vs. time traces for the 20 \AA\ and 35 \AA\ particles at a number of
327 different temperatures. It can clearly be seen that the period of the
328 breathing mode is dependent on temperature, and that the coherent
329 oscillations of the particles' volume are destroyed after only a few
330 ps in the smaller particles, while they live on for 10-20 ps in the
331 larger particles. The de-coherence is also strongly temperature
332 dependent, with the high temperature samples decohering much more
333 rapidly than lower temperatures.
334
335 \begin{figure}[htbp]
336 \centering
337 \includegraphics[height=3in]{images/Vol_vs_time.pdf}
338 \caption{Sample Volume traces for the 20 \AA\ and 35 \AA\ particles at a
339 range of temperatures. Note the relatively rapid ($<$ 10 ps)
340 decoherence due to melting in the 20 \AA\ particle as well as the
341 difference between the 1100 K and 1200 K traces in the 35 \AA\
342 particle.}
343 \label{fig:VolTime}
344 \end{figure}
345
346
347 Although $V$ vs. $t$ traces can say a great deal, it is more
348 instructive to compute the autocorrelation function for volume
349 fluctuations to give more accurate short-time information. Fig
350 \ref{fig:volcorr} shows representative autocorrelation functions for
351 volume fluctuations. Although many traces exhibit a single frequency
352 with decaying amplitude, a number of the samples show distinct beat
353 patterns indicating the presence of multiple frequency components in
354 the breathing motion of the nanoparticles. In particular, the 20 \AA\
355 particle shows a distinct beat in the volume fluctuations in the 800 K
356 trace.
357
358
359
360 \begin{figure}[htbp]
361 \centering
362 \includegraphics[height=3in]{images/volcorr.pdf}
363 \caption{Volume fluctuation autocorrelation functions for the 20 \AA\ (lower panel)
364 and 35 \AA\ (upper panel) particles at a range of temperatures. Successive
365 temperatures have been translated upwards by one unit. Note the beat
366 pattern in the 20 \AA\ particle at 800K.}
367 \label{fig:volcorr}
368 \end{figure}
369
370
371 When the power spectrum of the volume autocorrelation functions are
372 analyzed (Eq. (\ref{eq:volspect})), the samples which exhibit beat
373 patterns do indeed show multiple peaks in the power spectrum. The period corresponding to the two lowest frequency peaks is plotted in
374 Fig. \ref{fig:Period}. Smaller particles have the most evident
375 splitting, particularly as the temperature rises above the melting
376 points for these particles.
377
378 \begin{figure}[htbp]
379 \centering
380 \includegraphics[height=3in]{images/Period_vs_T.pdf}
381 \caption{The temperature dependence of the period of the breathing
382 mode for the four different nanoparticles studied in this
383 work.}
384 \label{fig:Period}
385 \end{figure}
386
387
388 \section{Discussion}
389 \label{bulkmod:sec:discussion}
390
391 Lamb's classical theory of elastic spheres~\cite{Lamb1882} provides
392 two possible explanations for the split peak in the vibrational
393 spectrum. The periods of the longitudinal and transverse vibrations
394 in an elastic sphere of radius $R$ are given by:
395 \begin{equation}
396 \tau_{t} = \frac{2 \pi R}{\theta c_{t}}
397 \end{equation}
398 and
399 \begin{equation}
400 \tau_{l} = \frac{2 \pi R}{\eta c_{l}}
401 \end{equation}
402 where $\theta$ and $n$ are obtained from the solutions to the
403 transcendental equations
404 \begin{equation}
405 \tan \theta = \frac{3 \theta}{3 - \theta^{2}}
406 \end{equation}
407 \begin{equation}
408 \tan \eta = \frac{4 \eta}{4 - \eta^{2}\frac{c_{l}^{2}}{c_{t}^{2}}}
409 \end{equation}.
410
411 $c_{l}$ and $c_{t}$ are the longitudinal and transverse speeds
412 of sound in the material. In an isotropic material, these speeds are
413 simply related to the elastic constants and the density ($\rho$),
414 \begin{equation}
415 c_{l} = \sqrt{c_{11}/\rho}
416 \end{equation}
417 \begin{equation}
418 c_{t} = \sqrt{c_{44}/\rho}.
419 \end{equation}
420
421 In crystalline materials, the speeds depend on the direction of
422 propagation of the wave relative to the crystal plane.\cite{Kittel:1996fk}
423 For the remainder of our analysis, it is assumed that the nanoparticles are
424 isotropic (which should be valid only above the melting transition).
425 A more detailed analysis of the lower temperature particles would take
426 the crystal lattice into account.
427
428 Using the experimental values for the elastic constants for 30
429 \AA\ Au particles at 300K, the low-frequency longitudinal (breathing)
430 mode should have a period of 2.19 ps while the low-frequency
431 transverse (toroidal) mode should have a period of 2.11 ps. Although
432 the actual calculated frequencies in the simulations are off of these
433 values, the difference in the periods (0.08 ps) is approximately half
434 of the splitting observed room-temperature simulations. This,
435 therefore, may be an explanation for the low-temperature splitting in
436 Fig. \ref{fig:Period}.
437
438 We note that Cerullo {\it et al.} used a similar treatment to obtain
439 the low frequency longitudinal frequencies for crystalline
440 semiconductor nanoparticles,\cite{Cerullo1999} and Simon and Geller
441 have investigated the effects of ensembles of particle size on the
442 Lamb mode using the classical Lamb theory results for isotropic
443 elastic spheres.\cite{Simon2001}
444
445 \subsection{Melted and Partially-Melted Particles}
446
447 Hartland {\it et al.} have extended the Lamb analysis to include
448 surface stress ($\gamma$).\cite{HartlandG.V._jp0276092} In this case, the
449 transcendental equation that must be solved to obtain the
450 low-frequency longitudinal mode is
451 \begin{equation}
452 \eta \cot \eta = 1 - \frac{\eta^{2} c_{l}^{2}}{4 c_{l}^{2} -
453 2 \gamma / (\rho R)}.
454 \end{equation}
455 In ideal liquids, inclusion of the surface stress is vital since the
456 transverse speed of sound ($c_{t}$) vanishes. Interested readers
457 should consult Hartland {\it et al.}'s paper for more details on the
458 extension to liquid-like particles, but the primary result is that the
459 vibrational period of the breathing mode for liquid droplets may be
460 written
461 \begin{equation}
462 \tau = \frac{2 R}{c_{l}(l)}
463 \end{equation}
464 where $c_{l}(l)$ is the longitudinal speed of sound in the liquid.
465 Iida and Guthrie list the speed of sound in liquid Au metal as
466 \begin{equation}
467 c_{l}(l) = 2560 - 0.55 (T - T_{m}) (\mbox{m s}^{-1})
468 \end{equation}
469 where $T_{m}$ is the melting temperature.\cite{Iida1988} A molten 35
470 \AA\ particle just above $T_{m}$ would therefore have a vibrational
471 period of 2.73 ps, and this would be markedly different from the
472 vibrational period just below $T_{m}$ if the melting transition were
473 sharp.
474
475 We know from our calculations of $C_{p}$ that the complete melting of
476 the particles is {\it not} sharp, and should take longer than the 40
477 ps observation time. There are therefore two explanations which are
478 commensurate with our observations.
479 \begin{enumerate}
480 \item The melting may occur at some time partway through observation
481 of the response to instantaneous heating. The early part
482 of the simulation would then show a higher-frequency breathing mode
483 than would be evident during the latter parts of the simulation.
484 \item The melting may take place by softening the outer layers of the
485 particle first, followed by a melting of the core at higher
486 temperatures. The liquid-like outer layer would then contribute a
487 lower frequency component than the interior of the particle.
488 \end{enumerate}
489
490 The second of these explanations is consistent with the core-shell
491 melting hypothesis advanced by Hartland {\it et al.} to explain their
492 laser heating experiments.\cite{HartlandG.V._jp0276092} At this stage, our
493 simulations cannot distinguish between the two hypotheses. One
494 possible avenue for future work would be the computation of a
495 radial-dependent order parameter to help evaluate whether the
496 solid-core/liquid-shell structure exists in our simulation.