Comment on "Triplet-to-Singlet Exciton Formation in poly(p-phenylene-vinylene) Light-Emitting Diodes"

1

CO

o

O

CN

>

O

O
o

CN

i

CO

-4-*
c3

I

o
o

>

in

o

cn
o

c3

c

o
o

X

Comment on "Triplet-to-Singlet Exciton For-
mation in poly(p-phenylene-vinylene) Light-
Emitting Diodes"

Lin et al. claim to have determined the ratio 7 of
triplet to singlet excitons in poly(p-phenylene-vinylenc)
(PPV) light emitting diodes by measuring the singlet and
triplet absorption spectra under conditions of optical and
electrical excitation^. The authors find that (a) 7 >>
3 at weak electric fields, and (b) 7 < 3 at moderate
electric fields. This steep electric field dependence and
7 < 3 at moderate fields is explained assuming : (i) ini-
tial formation of a high energy triplet T2 from the triplet
electron-hole (e-h) pair, which lies slightly below the e-h
continuum and which is dipole allowed from the lowest
triplet T\ , (ii) a phonon bottleneck^ in the nonradiative
relaxation from T2 to T\ , due to the large energy gap be-
tween them, and (iii) the competing field-induced disso-
ciation of the Ti exciton, and the consequent suppression
of 7. The authors quote reference to state that since
the higher energy singlet exciton S% is only 0.3 eV above
Si, phonon bottleneck and field induced dissociation are
absent in the singlet channel.

The lowest excitations in PPV can be simulated within
the Pariser-Parr-Pople model for linear polyenes with ar-
tificially large effective bond alternation^,

ff=-5^t(i±<j)(4c i+1)<r +/i.c) +

XVtf(n,-l)(nj-l) (f)

i i<j

where all terms have their usual meanings. Using the
standard U and t, Ohno parametrization for Vij and 5 =
0.24 we have calculated the exact singlet and triplet ener-
gies for the chain with N = 12 atoms. In Fig. 1, we show
the exact triplet energy spectrum between Ti, hereafter
1 3 B~, and the dipole-coupled T 2 , hereafter 1 3 A+. We
have also included the optical singlet exciton Si (1 1 B^)
and the singlet exciton m 1 A+, which has the largest
transition dipole with the 1 within (1) (the quan-
tum number m is N-dependent and is 8 in N = 12)£,
and is closer to the e-h continuum^ than the 1 3 A+. The
m 1 A+ is observable in nonlinear spectroscopy^ which
places this state above the S2 state calculated in refer-
ence and cited by Lin et al. The S2 state in reference
3 is the lowest two-photon state 2 X A+ of PPV and not
the m 1 A+. The energy gap between the m 1 A+ and the
1 1 B~ is at least as dense as that between the 1 3 A+ and
the 1 3 B~, and hence there exist singlet excitons nearly
degenerate with 1 3 B~. The large energy of the 1 1 B~ and
the large energy gaps in Fig. 1 are finite size effects. With

increasing N, absolute energies as well as energy differ-
ences decrease rapidly^ and additional levels appear in
the energy gaps. Different theoretical treatments have
confirmed this picture of the long chain limi^Si.

Two conclusions emerge from Fig. 1. First, the triplet

-8 1 A + „

1 n B

FIG. 1: The triplet energy spectrum between the 1 3 B U and
the 1 3 A+ in a N = 12 chain relative to the singlet ground
state. Different symmetry subspaces are shown separately.
State marked by asterisk is dipole-coupled to the 1 3 B^. The
l 1 Bu and the m Ajj~ (m = 8 in N = 12) are also included.

spectrum between Ti and the lowest dipole connected
triplet, the so called T2 in reference^ is very dense in long
chains. Absorption measurements miss all the dipole-
forbidden states in between. Hence the phonon bottle-
neck in the nonradiative relaxation^ will not occur. The
observed field dependencei of 7 therefore cannot be ex-
plained within Lin et al.'s model. Second, the forma-
tion of the 1 3 B~ triplet exciton via the m 3 A+ would im-
ply similar formation of the 1 1 B~ singlet exciton via the
m 1 A+. If indeed the energy spectra were sparse, contra-
dicting our calculations, then exciton dissociation would
be larger in the singlet channel due to the closer prox-
imity of the m 1 A+ to the e-h continuum. This would
have caused a field dependence of 7 opposite to what
is observed by the Lin et aim . Our theoretical results
therefore cast severe doubts on Lin et al. 's model, and
by implication on the experimental results themselves.

This work was partially supported by NSF and DST.

S. Mazumdar

Department of Physics

University of Arizona, Tucson, AZ 85721

Mousumi Das and S. Ramasesha

Solid State and Structural Chemistry Unit

Indian Institute of Science, Bangalore 560012, India.

1 L.C. Lin et al, Phys. Rev. Lett. 90, 036601 (2003). 3 M. Rohlfing and S. Louie, Phys. Rev. Lett. 82, 1959 (1999).

2 T.M. Hong and H.F. Meng, Phys. Rev. B 63, 075206 (2001). 4 Z.G. Soos et al, Chem. Phys. Lett. 194, 341 (1992).

D. Guo et al., Phys. Rev. B 48, 1433 (1993). 8 S. Abe et al. Phys. Rev. B 45, 9432 (1992).

Y. Shimoi and S. Mazumdar, Synth. Metals 85, 1027 9 A. Race et al. Phys. Rev. B 64, 035208 (2001).
(1997).

M. Liess et al. Phys. Rev. B 56, 15712 (1997).