1: %\documentclass[prb,preprint,showpacs]{revtex4}
2: %\documentclass[prb,showpacs]{revtex4}
3: \documentclass[prb,showpacs,twocolumn]{revtex4}
4:
5: \usepackage{amsmath,amssymb}
6:
7: \newcommand{\bu}[1]{{#1}^1B_u}
8: \newcommand{\but}[1]{{#1}^3B_u}
9: \newcommand{\buminus}[1]{{#1}^1B_u^-}
10: \newcommand{\ag}[1]{{#1}^1A_g}
11: \newcommand{\agtr}[1]{{#1}^3A_g}
12: \newcommand{\agplus}[1]{{#1}^1A_g^+}
13: \newcommand{\butplus}[1]{{#1}^3B_u^+}
14:
15: \bibliographystyle{apsrev}
16:
17: \usepackage{graphicx}
18:
19: \begin{document}
20:
21: \widetext
22:
23: \title{Computational Investigations of the Primary Excited States of Poly(\textit{para}-phenylene vinylene)}
24:
25: \author{Robert J. Bursill$^1$ and William Barford$^{2}$}
26:
27: \affiliation{
28: $^1$School of Physics, University of New South Wales, Sydney, New South Wales 2052, Australia\\
29: $^2$Department of Physics and Astronomy, University of Sheffield, Sheffield, S3 7RH, United Kingdom
30: }
31:
32: \begin{abstract}
33: The Pariser-Parr-Pople model of $\pi$-conjugated electrons is
34: solved by the density matrix renormalization group method for the
35: light emitting polymer, poly(\textit{para}-phenylene vinylene).
36: The energies of the primary excited states are calculated. When
37: solid state screening is incorporated into the model parameters there
38: is excellent agreement between theory and experiment, enabling
39: an identification of the origin of the key spectroscopic features.
40: \end{abstract}
41:
42: \pacs{71.10.-w, 71.20.Rv, 71.35.-y, 78.67.-n}
43:
44: \maketitle
45:
46: \section{Introduction}\label{Se:1}
47:
48: Since the discovery of electroluminescence in poly(para-phenylene
49: vinylene) (PPV) in 1990\cite{burroughes90}, a variety of
50: experimental and theoretical techniques have been deployed to
51: investigate the physics of the primary photoexcitations of the
52: phenyl-based light emitting polymers. Linear and non-linear
53: optical spectroscopies have revealed the energies and symmetries
54: of the dominant dipole-allowed transitions, as well as the
55: important dipole-forbidden transitions that participate in
56: non-linear optical processes.
57:
58: All phenyl-based light emitting polymers exhibit a characteristic
59: absorption spectra. These show two dominant excitations polarized
60: parallel to the chain axis (at $2.8$ eV and $6.1$ eV in
61: PPV\cite{martin99}). There are also two intermediate weaker
62: transitions (at $3.6$ eV and $4.8$ eV in PPV\cite{martin99}). The
63: higher of these intermediate transitions is predominantly
64: polarized perpendicular to the chain axis, whereas the
65: polarization of the lower transition is less well-defined. (In PPV
66: it is predominately polarized parallel to the
67: long-axis\cite{miller}.) Furthermore, the strength of this lower
68: transition is enhanced by chemical substitution.
69: Electroabsorption\cite{martin99} and two-photon
70: absorption\cite{frolov02} reveal a dipole-forbidden state at
71: approximately $0.7$ eV above the lowest dipole-allowed transition.
72: These two states are usually labelled the $\ag{m}$ and $\bu{1}$
73: states, respectively. Finally, a triplet state has been observed
74: at $0.7$ eV below the $\bu{1}$ excitation (see ref\cite{kohler04}
75: for references), with another dipole connected triplet state $1.4$
76: eV higher in energy\cite{monkman01}. (Notice that this higher
77: triplet state, labelled as the $\agtr{m}$ state, is virtually
78: degenerate with its singlet counterpart, namely the $\ag{m}$
79: state. These two states are often referred to as the singlet and
80: triplet charge-transfer states. Their degeneracy can be explained
81: by the fact that they are the lowest pseudomomentum branches of
82: the $n=2$ Mott-Wannier excitons, which have odd parity
83: electron-hole wavefunctions and therefore experience no exchange
84: interactions\cite{barford02}.)
85:
86: An important early insight into the nature of the primary
87: photoexcitations of the phenyl-based systems was provided by Rice
88: and Gartstein\cite{rice94,gartstein95}, who argued that they can
89: essentially be understood as arising from the delocalization of
90: the primitive benzene excitations. Kirova and Brazovskii, on the
91: other hand, have argued that a conventional semiconductor band
92: picture of bound particle-hole excitations is a more appropriate
93: description\cite{kirova99}. Other theoretical work on PPV
94: includes, a single configuration interaction (CI) calculation of
95: the Pariser-Parr-Pople model by Chandross and
96: Mazumdar\cite{chandross97}; density matrix renormalization group
97: calculations on reduced molecular-orbital models by Barford
98: \emph{et al.}\cite{barford97}; quantum chemistry calculations on
99: the INDO Hamiltonian by Beljonne \emph{et al.}\cite{beljonne99},
100: and Weibel and Yaron\cite{weibel02}; and \emph{ab initio}
101: Bethe-Salpeter equation (BSE) calculations by Rohlfing and
102: Louie\cite{rohlfing99}. These theoretical predictions will be
103: discussed more fully in Section \ref{Se:3}, when we discuss the
104: results of the calculations presented here and their comparisons
105: to experiment.
106:
107: In this paper we present density matrix renormalization group
108: (DMRG) calculations on the full Pariser-Parr-Pople model of PPV.
109: The DMRG method for calculating the full electronic spectra of
110: conjugated polymers has a number of advantages over its
111: competitors. It is more accurate than single or double CI
112: calculations, and does not suffer from finite-size consistency
113: errors; it makes no assumptions about the nature of the
114: excitations (unlike the BSE method, which assumes that the
115: excitations are particle-hole pairs), thus it is able to
116: accurately model highly correlated excited states; and unlike
117: standard quantum chemistry techniques, it is able to calculate the
118: excited spectra of large molecules. The DMRG method is most
119: readily suited to reduced basis Hamiltonians, such as the
120: $\pi$-electron Pariser-Parr-Pople model. This is not necessarily a
121: disadvantage over \emph{ab initio} methods, as when correctly
122: parameterized the Pariser-Parr-Pople model makes very accurate
123: predictions\cite{bursill98, chandross97}. Moreover, the
124: Pariser-Parr-Pople model possess particle-hole symmetry, which
125: means that spatial, particle-hole and spin-flip symmetries can be
126: employed to target high-lying excited states in the DMRG method
127: (see ref\cite{barford05}, for example).
128:
129:
130: DMRG calculations on the full Pariser-Parr-Pople model were
131: presented for poly(\emph{para}-phenylene) by Bursill and
132: Barford\cite{bursill02}. This paper extends that approach to PPV.
133: In the next section we define the Pariser-Parr-Pople model and
134: briefly describe our implementation of the DMRG method. In Section
135: \ref{Se:3} we describe and discuss our results, and
136: compare them to other approaches, concluding in section
137: \ref{Se:4}.
138:
139: \section{Model and Methodology}\label{Se:2}
140:
141: \subsection{The Pariser-Parr-Pople Model}
142:
143: The Pariser-Parr-Pople model is a $\pi$-electron model of conjugated polymers, defined by,
144: \begin{eqnarray}\label{Eq:1}
145: {\cal H}
146: & = &
147: - \sum_{<ij>\,\sigma} t_{ij}
148: \left[ c_{i\sigma}^{\dagger} c_{j\sigma} + c_{j\sigma}^{\dagger} c_{i\sigma} \right]
149: \nonumber
150: \\
151: & &
152: +\; U \sum_{i}
153: \left(n_{i\uparrow}-1/2\right)
154: \left(n_{i\downarrow}-1/2\right)
155: \nonumber
156: \\
157: & &
158: +\;
159: \frac{1}{2} \sum_{i\neq j} V_{ij} (n_i - 1)(n_j - 1),
160: \end{eqnarray}
161: where $<>$ represents nearest neighbors, $c_{i\sigma}$ destroys a $\pi$-electron on site $i$, $n_{i\sigma} =
162: c_{i\sigma}^{\dagger} c_{i \sigma}$, and $n_i = n_{i\uparrow} +
163: n_{i\downarrow}$.
164:
165: We use the Ohno parameterization for the Coulomb
166: interaction, defined by,
167: \begin{equation}\label{Eq:2}
168: V_{ij} = U / \sqrt{ 1 + ( U \epsilon r_{ij}/14.397)^2 },
169: \end{equation}
170: where $r_{ij}$ is the inter-atomic distance (in \AA), $U$ is the
171: on-site Coulomb interaction (in eV), and $\epsilon$ is the
172: dielectric constant. This interaction is an interpolation between
173: an on-site Coulomb repulsion, $U$, and a Coulomb potential,
174: $\textrm{e}^2/4\pi\epsilon\epsilon_0 r_{ij}$ as $r_{ij}
175: \rightarrow \infty$.
176:
177: \begin{figure}[tb]
178: \begin{center}
179: \includegraphics[scale=0.50]{FigPPV1.eps}
180: \end{center}
181: \caption{The band structure of PPV with all nearest neighbor bond integrals, $t$,
182: equal. The low-lying particle-hole transitions are labelled $1$, $2$
183: and $3$.} \label{Fi:1}
184: \end{figure}
185:
186: The band structure, obtained in the non-interacting limit ($U=0$),
187: is shown in Fig.\ \ref{Fi:1}. The pair of non-bonding bands are a
188: consequence of the $D_{2h}$ symmetry of the Hamiltonian in the
189: non-interacting limit with only nearest-neighbor bond integrals.
190: The low-lying particle-hole excitations are labelled $1$ (for
191: transitions involving $d_1$ and $d_1^*$), the degenerate pair
192: labelled $2$ (for transitions involving $d_1$ and $l^*$, and $l$
193: and $d_1^*$), and $3$ (for transitions involving $l$ and $l^*$).
194: Coulomb interactions lift the degeneracy of the intermediate pair
195: (which become the intermediate pair of transitions described in
196: Section \ref{Se:1}), while the transitions $1$ and $3$ become the
197: low and high energy transitions described in Section \ref{Se:1}.
198: Evidently, to obtain a more realistic description of the excited states it is
199: necessary to include Coulomb interactions.
200:
201: In the following we use two sets of parameters to model the
202: excited states. The first set, called the \emph{optimized}
203: parameters, were derived by fitting the predicted
204: Pariser-Parr-Pople excitation energies of stilbene to the
205: experimental spectrum of stilbene in vacuo\cite{castleton99}.
206: These are, $t_p = 2.539$ eV, $t_d = 2.684$ eV, $t_s = 2.22$ eV, $U
207: = 10.06$ eV, and $\epsilon = 1$ (where the phenyl, double and
208: single bond integrals are defined in Fig.\ \ref{Fi:2}). Although
209: giving instructive results, this parameter set fails to
210: quantitatively predict the excitation energies of polymers in the
211: solid state, as they fail to account for the solvation effects of
212: the surrounding dielectric\cite{moore98, barford04}. The other
213: parameter set, called the \emph{screened} parameters, were derived
214: by Chandross and Mazumdar\cite{chandross97} to account for solid
215: state solvation effects. These parameters are, $t_p = 2.4$ eV,
216: $t_d = 2.6$ eV, $t_s = 2.2$ eV, $U = 8$ eV, and $\epsilon = 2$.
217:
218: \subsection{The Density Matrix Renormalization Group (DMRG) Method}
219:
220: Eq.\ (\ref{Eq:1}) is solved by the DMRG method for chains of up to
221: $28$ phenyl rings (i.e.\ $222$ sites). The DMRG method is an
222: efficient truncation procedure for solving quantum lattice
223: Hamiltonians, especially in one-dimension\cite{white}. The details
224: of the DMRG implementation for this problem, particularly the
225: procedure of deriving an optimized reduced basis for the phenyl
226: ring\cite{zhang, bursill02, barford02b}, are described in detail
227: in ref\cite{bursill02}. Also shown in ref\cite{bursill02} are comprehensive DMRG convergence tests that are applicable to phenyl-based systems.
228:
229: As described in Section \ref{Se:1}, the Pariser-Parr-Pople model
230: possess spin-flip and particle-hole symmetries, which we exploit
231: in the DMRG method to target excited states. With only onsite and
232: nearest neighbor Coulomb interactions the Pariser-Parr-Pople model
233: applied to the PPV structure formally possess $D_{2h}$ symmetry.
234: In this limit this means that the eigenstates are labelled with
235: the spatial symmetry assignments, $A_g$, $B_{1u}$, $B_{2u}$, and
236: $B_{3g}$. Longer range Coulomb interactions with the PPV structure
237: reduce the $D_{2h}$ symmetry to $C_{2}$ symmetry, and thus the
238: true spatial symmetry assignments are $A_g$ and $B_{u}$. However,
239: it is computationally expedient, with very little loss of
240: accuracy, to retain $D_{2h}$ symmetry while incorporating long
241: range Coulomb interactions. This is achieved by taking the bond
242: angle between the single and double bonds in the vinylene unit to
243: be $180^0$, rather than $120^0$. To ensure that the overall
244: molecular size is consistent to the original structure the single
245: and double lengths are reduced to $1.283$ \AA~ and $1.194$ \AA,
246: respectively. The phenyl bond length is $1.40$ \AA. The next
247: section describes the calculated results.
248:
249: \section{Results and Discussions}\label{Se:3}
250:
251: \subsection{DMRG Calculations}
252:
253: \begin{figure}[tb]
254: \begin{center}
255: \includegraphics[scale=0.50]{FigPPV2.eps}
256: \end{center}
257: \caption{The DMRG calculated transition energies of
258: \textit{para}-phenylene vinylene oligomers as a function of
259: inverse chain length, calculated from the Pariser-Parr-Pople model
260: with optimized parameters: $U=10.06$ eV, $t_p = 2.539$ eV, $t_d =
261: 2.684$ eV, $t_s = 2.22$ eV, and dielectric constant, $\epsilon =
262: 1$. The excited states are identified with the spectroscopic
263: features shown in Fig.\ 3 of ref\protect\cite{martin99} and Fig.\
264: 1 of ref\protect\cite{frolov02}. The symmetry assignments of the
265: states shown in brackets are the symmetries the eigenstates would
266: have if PPV had $D_{2h}$ rather than $C_2$ symmetry. The inset
267: shows the oligo-phenylene vinylene structure with the bond
268: integrals $t_p$, $t_d$ and $t_s$. } \label{Fi:2}
269: \end{figure}
270:
271: Fig.\ \ref{Fi:2} shows the DMRG calculated excitation energies of
272: oligo(\textit{para}-phenylene vinylenes) using the
273: Pariser-Parr-Pople model with unscreened parameters ($U=10.06$ eV,
274: $t_p = 2.539$ eV, $t_d = 2.684$ eV, $t_s = 2.22$ eV, and $\epsilon
275: = 1$). ($N=2$ corresponds to stilbene.) The $\buminus{1}$ state is
276: the strong lowest energy dipole allowed transition (labelled I in
277: Fig.\ 3 of ref\cite{martin99}). The strong transition dipole
278: moment between this state and the $\agplus{2}$ state indicates
279: that the $\agplus{2}$ is the state labelled `$\ag{m}$' in
280: non-linear optical spectroscopies (see, for example, Fig.\ 1(a) in
281: ref\cite{frolov02}). The energies of the $\buminus{1}$ and
282: $\agplus{2}$ states initially reduce rapidly as a function of
283: chain length, indicating that these states reduce their kinetic
284: energy by readily delocalizing the exciton wavefunction along the
285: chain. This is primarily because these states are largely
286: constructed from particle-hole transitions between the bonding and
287: anti-bonding bands, labelled $d_1$ and $d_1^*$ in Fig.\
288: \ref{Fi:1}\cite{footnote1}. A description of these states as
289: strongly bound band exciton states is therefore appropriate in the
290: long-chain limit. Indeed, the $\buminus{1}$ state is the lowest
291: pseudomomentum branch of the $n=1$ Mott-Wannier exciton, while the
292: $\agplus{2}$ is the lowest pseudomomentum branch of the $n=2$
293: Mott-Wannier exciton\cite{barford02,barford05}. (Higher lying
294: pseudo-momentum branches of the $n=1$ and $n=2$ excitons for DMRG
295: calculations on poly(para-phenylene) are illusrated in Fig.\ 5 of
296: ref\cite{bursill02}.) Notice that this assignment places a
297: \emph{lower bound} on the binding energy of the $n=1$ exciton as
298: $E(mA_g) - E(1B_u)$.
299:
300: We next consider the state labelled II and shown by diamonds in
301: Fig.\ \ref{Fi:2}. This state has $^1B_{u}^+$ symmetry, but would
302: have $^1B_{2u}^+$ symmetry if PPV had $D_{2h}$ symmetry. From a
303: band picture analysis, this state is composed of an antisymmetric
304: combination of particle-hole transitions from $d_1$ to $l^*$ and
305: $l$ to $d_1^*$. At the Pariser-Parr-Pople model level, it has
306: positive particle-hole symmetry, and is therefore
307: dipole-forbidden. However, it is weakly dipole allowed in real
308: systems, and its oscillator strength is further enhanced by
309: chemical substitution. Its excitation energy is virtually
310: independent of chain length, because, first its wavefunction is
311: composed of non-bonding orbitals, and therefore there is no
312: hybridization via one-electron transfer terms, and second since it
313: has a very small oscillator strength, resonant exciton transfer
314: along the chain is virtually inoperative. These two features (weak
315: oscillator strength and a chain independent energy) strongly
316: indicate that the second spectroscopic feature (labelled II in
317: Fig.\ 3 of ref\cite{martin99}) originates from this state.
318:
319: The state labelled III and shown by triangles in Fig.\ \ref{Fi:2}
320: has $^1B_{u}^-$ symmetry, but would have $^1B_{2u}^-$ symmetry if
321: PPV had $D_{2h}$ symmetry. From a band picture analysis, this
322: state is composed of a symmetric combination of particle-hole
323: transitions from $d_1$ to $l^*$ and $l$ to $d_1^*$. It has
324: negative particle-hole symmetry, and therefore has an allowed
325: dipole transition from the ground state. This state is the third
326: spectroscopic feature of PPV (labelled III in Fig.\ 3 of
327: ref\cite{martin99}).
328:
329: The state labelled IV and shown by crosses in Fig.\ \ref{Fi:2} has
330: $^1B_{u}^-$ symmetry, but would have $^1B_{1u}^-$ symmetry if PPV
331: had $D_{2h}$ symmetry. From a band picture analysis, this state is
332: composed of particle-hole transitions from $l$ to $l^*$. It has negative particle-hole symmetry, has a strong
333: dipole transition from the ground state, and is polarized along the
334: chain axis. This state is the fourth spectroscopic feature of PPV
335: (labelled IV in Fig.\ 3 of ref\cite{martin99}), as is usually
336: referred to as the `Frenkel' exciton, as its particle-hole
337: wavefunction is typically confined to a single phenyl-ring.
338:
339: Finally, the two triplet states, $1^3B_u^+$ and $m^3A_g^-$, are
340: also shown in Fig.\ \ref{Fi:2}.
341:
342: The results shown in Fig.\ \ref{Fi:2} are obtained from the
343: Pariser-Parr-Pople model with optimized parameters that are not
344: parametrized to model solid state solvation effects. The
345: dielectric response of the environment is predicted to red-shift
346: the intra-molecular excitations of a single chain by up to $0.1$
347: eV for the $1^1B_u$ state, $0.7$ eV for the $m^1A_g$ state, and
348: $1.5$ eV for the charge gap\cite{moore98,barford04}. We would
349: therefore not expect these calculations on a single chain to
350: compare directly with experimental observations.
351:
352: \begin{figure}[tb]
353: \begin{center}
354: \includegraphics[scale=0.50]{FigPPV3.eps}
355: \end{center}
356: \caption{The DMRG calculated transition energies of
357: \textit{para}-phenylene vinylene oligomers as a function of
358: inverse chain length, calculated from the Pariser-Parr-Pople model
359: with screened parameters: $U=8$ eV, $t_p = 2.4$ eV, $t_d = 2.6$
360: eV, $t_s = 2.2$ eV, and $\epsilon = 2$. The symbols are defined in
361: the inset of Fig.\ \ref{Fi:2}. The excited states are identified
362: with the spectroscopic features shown in Fig.\ 3 of
363: ref\protect\cite{martin99} and Fig.\ 1 of
364: ref\protect\cite{frolov02}.} \label{Fi:3}
365: \end{figure}
366:
367: Chandross and Mazumdar attempted to model solvation affects by
368: adjusting the Pariser-Parr-Pople parameters and introducing a
369: static dielectric constant\cite{chandross97}. Although the
370: introduction of a \emph{static} dielectric constant is not a
371: faithful representation of the dielectric response (because the
372: timescale for the particle-hole motion is very similar to the
373: timescale of the dielectric response\cite{moore98}), these
374: screened parameters do lead to Pariser-Parr-Pople model
375: predictions that are remarkably close to the experimental
376: observations.
377:
378: Fig.\ \ref{Fi:3} shows the DMRG calculated results with the
379: screened parameters ($U=8$ eV, $t_p = 2.4$ eV, $t_d = 2.6$ eV,
380: $t_s = 2.2$ eV, and $\epsilon = 2$). Comparing this figure with
381: Fig.\ \ref{Fi:2} we see that the $\buminus{1}$ and $\agplus{2}$
382: states are red-shifted by ca.\ $0.1$ eV and $0.6$ eV,
383: respectively, as expected from solvation effects. The excited
384: states are identified with the spectroscopic features shown in
385: Fig.\ 3 of ref\protect\cite{martin99}. Their energies are
386: remarkably consistent with the experimental values of $2.8$,
387: $3.6$, $4.8$, and $6.1$ eV obtained in ref\cite{martin99}, lending
388: additional credence to our assignments of the excited state
389: origins of the features, discussed above.
390:
391: \subsection{Other Approaches}
392:
393: We now compare our DMRG predictions of the Pariser-Par-Pople model
394: with other approaches. Our results using the screened parameters
395: are consistent with those of Chandross and Mazumdar\cite{chandross97}. They
396: used SCI on the Pariser-Par-Pople model. For an eight-unit
397: oligomer they calculate the $\bu{1}$ state at $2.7$ eV, an
398: $m^1A_g$ state at $3.3$ eV, and the $\bu{n}$ state at 3.6 eV. The
399: $\bu{1}$ and $\ag{m}$ states are the $n=1$ and $n=2$ excitons,
400: while the $\bu{n}$ state coincides with the charge-gap and
401: therefore indicates the onset of the particle-hole continuum. They
402: also predict the $\but{1}$ state at $1.4$ eV.
403:
404: An \textit{ab initio} BSE calculation by Rohlfing and
405: Louie\cite{rohlfing99} on a PPV polymer predicts dipole allowed
406: and forbidden singlet excitons at $2.4$ eV and $2.8$ eV,
407: respectively, with the quasi-particle gap at $3.3$ eV. They also
408: predict triplet excitons at $1.5$ eV and $2.7$ eV. The $2.4$ eV
409: and $2.8$ eV singlet excitons are the $1^1B_u$ and $m^1A_g$
410: states, respectively, while the $1.5$ eV and $2.7$ eV triplet
411: excitons are the $1^3B_u$ and $m^3A_g$ states, respectively. The
412: $m^1A_g$ and $m^3A_g$ states are nearly degenerate, as predicted
413: by the Mott-Wannier exciton theory for odd parity particle-hole
414: wavefunctions. Using the same technique with a screened
415: electron-hole interaction van der Horst \textit{et
416: al.}\cite{horst01} predict a $1^1B_u$ binding energy in PPV of
417: $0.48$ eV.
418:
419: The origin of the higher-lying transitions has also been
420: investigated. Rohfling and Louie\cite{rohlfing99}, and Weibel and
421: Yaron\cite{weibel02} predict that peak II in PPV arises from an
422: exciton caused predominately by the antisymmetric combination of
423: particle-hole transitions from $d_1$ to $l^*$ and $l$ to $d_1^*$.
424: Weibel and Yaron\cite{weibel02} have also investigated the effects
425: of breaking particle-hole symmetry on the oscillator strength and
426: polarization of peak II. Using the semiempirical INDO Hamiltonian
427: on nonplanar di-hydroxy-PPV, their calculations indicate that
428: chemical substitution and mixing of the $\pi$ and $\sigma$
429: orbitals enhances the oscillator strength, as originally suggested
430: by Gartstein \emph{et al.}\cite{gartstein95} Moreover, as
431: illustrated in Fig.\ 5 of ref\cite{weibel02}, this peak becomes
432: predominately polarized along the chain axis, in agreement with
433: experiment\cite{miller}.
434:
435: \section{Conclusions}\label{Se:4}
436:
437: Gathering together the various theoretical predictions of the
438: origins of the primary excited states of PPV we now interpret the
439: key spectroscopic features of PPV.
440:
441: Peak I corresponds to the low-energy dipole active $1^1B_u^-$
442: state. This is the lowest pseudomomentum branch of the family of
443: $n=1$ Mott-Wannier singlet excitons resulting from the Coulomb
444: attraction between the particle-hole excitation from the valence
445: ($d_1$) to the conduction ($d_1^*$) bands. Approximately $0.7$ eV
446: higher in energy is the $m^1A_g$ state, identified by
447: electroabsorption\cite{martin99}, two-photon absorption and
448: photoinduced absorption\cite{frolov02}. The Pariser-Parr-Pople
449: model calculations described in this paper suggest that this state
450: is the $2^1A_g^+$ state, which is
451: the lowest pseudomomentum branch of the family of
452: $n=2$ Mott-Wannier excitons. This assignment places a lower bound
453: on the spectroscopically determined binding energy of the
454: $\bu{1}$ exciton of $0.7$ eV. Approximately $0.7$ eV below the
455: $1^1B_u^-$ exciton is the $1^3B_u^+$ triplet, indicating a large
456: exchange energy characteristic of correlated states. This state is
457: the lowest pseudomomentum branch of the family of $n=1$
458: Mott-Wannier triplet excitons. Photo-induced absorption from the
459: $1^3B_u^+$ triplet indicates another triplet, the $1^3A_g^-$
460: state, at approximately $1.4$ eV higher in energy, and essentially
461: degenerate with the $2^1A_g^+$ state. This triplet state is the
462: lowest pseudomomentum branch of the family of $n=2$ Mott-Wannier
463: triplet excitons. As expected from Mott-Wannier exciton theory in
464: one-dimension\cite{barford02, barford05}, the odd particle-hole
465: parity singlet and triplet (charge-transfer) excitons are
466: virtually degenerate. The $\bu{n}$ state at $0.1$ eV higher in
467: energy than the $\ag{m}$ state in PPV\cite{martin99} indicates
468: binding energies of $\sim 0.8$ eV and $0.1$ eV for the $n=1$ and
469: $n=2$ singlet excitons, respectively. Higher in energy are the
470: excitations associated with peaks II and III. These arise from
471: antisymmetric and symmetric combinations of particle-hole
472: transitions from $d_1$ to $l^*$ and $l$ to $d_1^*$, respectively.
473: Finally, peak IV is the intraphenyl, or Frenkel, exciton. This
474: identification of the primary excited states of PPV also applies
475: to other phenyl-based systems, e.g.\
476: PPP\cite{bursill02,barford05}.
477:
478: The calculations presented in this paper have entirely concerned
479: vertical transitions, where the excited states are calculated in
480: the geometry of the ground state. Of course, electron-lattice
481: relaxation is an important process in polymers that determines
482: Stokes shifts, self-trapping, and exciton and charge transfer
483: between polymers. The consequences of electron-lattice relaxation
484: on the excited states energies and structures have been considered
485: in refs\cite{el}.
486:
487:
488: \begin{acknowledgements}
489: R.\ J.\ B.\ acknowledges support from the Australian Research Council and the J.\ G.\ Russell Foundation.
490: W.\ B.\ thanks the Leverhulme Trust for financial support.
491: \end{acknowledgements}
492:
493:
494: \begin{references}
495:
496: \bibitem[]{email}
497: E.mail addresses: ph1rb@phys.unsw.edu.au, W.Barford@sheffield.ac.uk
498:
499: \bibitem{burroughes90} J. H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. Mackay, R. H. Friend,
500: P. L. Burn, and A. D. Holmes, \textit{Nature} \textbf{347}, 539 (1990)
501:
502: \bibitem{martin99} S. J. Martin, D. D. C. Bradley, P. A. Lane, H. Mellor, and P. L. Burn, \textit{Phys. Rev. B}
503: \textbf{59}, 15133 (1999)
504:
505: \bibitem{miller} E. K. Miller, D. Yoshida, C. Y. Yang, and A. J. Heeger, \emph{Phys. Rev. B}
506: \textbf{59}, 4661 (1999)
507:
508: \bibitem{frolov02} S. Frolov, Z. Bao, M. Wohlgenannt, and Z. V. Vardeny, \textit{Phys. Rev. B} \textbf{65},
509: 205209 (2002)
510:
511: \bibitem{kohler04} A. K\"ohler and D. Beljonne, \textit{Advanced Functional Materials} \textbf{14}, 11 (2004)
512:
513: \bibitem{monkman01} A. P. Monkman, H. D. Burrows, L. J. Hartwell, L. E. Horsburgh, I. Hamblett, and S. Navaratnam,
514: \textit{Phys. Rev. Lett.} \textbf{86}, 1358 (2001)
515:
516: \bibitem{barford02} W. Barford, R. J. Bursill, and R. W. Smith, \emph{Phys. Rev. B} {\bf 66}, 115205 (2002)
517:
518: \bibitem{rice94} M. J. Rice and Yu. N. Gartstein, \textit{Phys. Rev. Lett.} \textbf{73}, 2504 (1994)
519:
520: \bibitem{gartstein95} Yu. N. Gartstein, M. J. Rice, and E. M. Conwell, \textit{Phys.Rev.
521: B} \textbf{52}, 1683 (1995)
522:
523: \bibitem{kirova99} N. Kirova, S. Brazovskii, and A. R. Bishop, \textit{ Synth. Mets.} \textbf{100}, 29 (1999);
524: N. Kirova and S. Brazovskii, \textit{ Synth. Mets.} \textbf{141}, 139 (2004)
525:
526: \bibitem{chandross97} M. Chandross and S. Mazumdar, \textit{Phys. Rev. B} \textbf{55}, 1497 (1997)
527:
528: \bibitem{barford97} W. Barford and R. J. Bursill, \emph{Chem. Phys. Lett.}, \textbf{268}, 535 (1997);
529: M. Lavrentiev, W. Barford, S. Martin, H. Daly, and R. J. Bursill, \emph{Phys. Rev. B} \textbf{59}, 9987 (1999);
530: R. J. Bursill, W. Barford, and H. Daly, \emph{Chem. Phys.}, 243, 35 (1999)
531:
532: \bibitem{beljonne99} D. Beljonne, Z. Shuai, J. Cornil, D. A. dos Santos, and J. L. Br\'edas,
533: \textit{J. Chem. Phys.} \textbf{111}, 2829 (1999)
534:
535: \bibitem{weibel02} J. D. Weibel and D. Yaron, \emph{J. Chem. Phys.} \textbf{116}, 6846 (2002)
536:
537: \bibitem{rohlfing99} M. Rohlfing and S. G. Louie, \emph{Phys. Rev. Lett.} \textbf{82}, 1959 (1999)
538:
539: \bibitem{bursill98} R. J. Bursill, C. Castleton, and W. Barford, \textit{Chem. Phys. Lett.} \textbf{294}, 305 (1998)
540:
541: \bibitem{barford05} W. Barford, \emph{Electronic and Optical Properties of Conjugated Polymers},
542: Oxford University Press, Oxford (2005)
543:
544: \bibitem{bursill02} R. J. Bursill and W. Barford, \emph{Phys. Rev. B} {\bf 66}, 205112 (2002)
545:
546: \bibitem{castleton99} C. Castelton and W. Barford, \textit{Synth. Met.} \textbf{101}, 520 (1999); C. W. M. Castleton and W. Barford, \emph{J. Chem. Phys.},
547: \textbf{117}, 3570 (2002)
548:
549: \bibitem{moore98}
550: E. Moore, B. Gherman, and D. Yaron, \textit{J. Chem. Phys.}
551: \textbf{106} 4216 (1997); E. Moore and D. Yaron, \emph{J. Chem.
552: Phys.} {\bf 109}, 6147 (1998)
553:
554: \bibitem{barford04} W. Barford, R. J. Bursill, and D. Yaron, \emph{Phys. Rev. B} {\bf 69}, 155203 (2004)
555:
556: \bibitem{white} S. R. White, \emph{Phys. Rev. Lett.} {\bf 69}, 2863 (1992)
557:
558: \bibitem{zhang} C. Zhang, E. Jeckelmann, and S. R. White, \emph{Phys. Rev. Lett.} \textbf{80} 2661 (1998)
559:
560: \bibitem{barford02b} W. Barford, R. J. Bursill, and M. Lavrentiev, \textit{Phys. Rev. B} \textbf{65}, 075107 (2002);
561: W. Barford and R. J. Bursill, \textit{Phys. Rev. B} (in press)
562:
563: \bibitem{footnote1} However, as the chain length decreases below the particle-hole separation there
564: is also a `quantum confinement' effect, arising from the squeezing
565: of the particle-hole wavefunction. See, for example, Z. Shuai, J.
566: L. Bredas, S. K. Pati, and S. Ramasesha, \textit{Phys. Rev. B}
567: \textbf{56}, 9298 (1997); M. Y. Lavrentiev and W. Barford,
568: \textit{Phys. Rev. B} \textbf{59}, 15 048 (1999)
569:
570: \bibitem{horst01} J-W. van der Horst, P. A. Bobbert, M. A. J. Michels, and
571: H. B\"assler, \textit{J. Chem. Phys.} \textbf{114}, 6950 (2001)
572:
573: \bibitem{el} C. Ambrosch-Draxl, J. A. Majewski, P. Vogl, and G. Leising, \emph{Phys. Rev. B} {\bf 51}, 9668 (1995);
574: D. Beljonne, Z. Shuai, R. H. Friend, and J. L. Br\'edas, \textit{J. Chem. Phys.} \textbf{102}, 2042 (1995);
575: E. Zojer, N. Koch, P. Puschnig, F. Meghdadi, A. Niko, R. Resel, C. Ambrosch-Draxl, M. Knupfer, J. Fink,
576: J. L. Br\'edas, and G. Leising, \emph{Phys. Rev. B} \textbf{61}, 16538 (2000);
577: S. Tretiak, A. Saxena, R. L. Martin, and A. R. Bishop, \emph{Phys. Rev. Lett.}, \textbf{89}, 097402 (2002);
578: E. Artacho, M. Rohfling, M. C\^ote, P. D. Haynes, R. J. Needs, and C. Molteni,
579: \emph{Phys. Rev. Lett.} \textbf{93}, 116401 (2004);
580: E. Moore, W. Barford, and R. J. Bursill, \emph{Phys. Rev. B} \textbf{71}, 115107 (2005)
581:
582: \end{references}
583:
584: \end{document}
585: