DOI: 10.1002/chem.201304375
Full Paper
& Photophysics
A Theoretical Investigation into the Luminescent Properties of d8Transition-Metal Complexes with Tetradentate Schiff Base Ligands Glenna So Ming Tong,*[a] Pui Keong Chow,[a] Wai-Pong To,[a] Wai-Ming Kwok,[b] and ChiMing Che*[a]
Abstract: A theoretical investigation on the luminescence efficiency of a series of d8 transition-metal Schiff base complexes was undertaken. The aim was to understand the different photophysics of [M-salen]n complexes (salen = N,N’bis(salicylidene)ethylenediamine; M = Pt, Pd (n = 0); Au (n = + 1)) in acetonitrile solutions at room temperature: [Pt-salen] is phosphorescent and [Au-salen] + is fluorescent, but [Pdsalen] is nonemissive. Based on the calculation results, it was proposed that incorporation of electron-withdrawing groups at the 4-position of the Schiff base ligand should widen the 3 MLCT–3MC gap (MLCT = metal-to-ligand charge transfer and MC = metal centered, that is, the dd excited state); thus permitting phosphorescence of the corresponding PdII Schiff base complex. Although it is experimentally proven that [Pdsalph-4E] (salph = N,N’-bis(salicylidene)-1,2-phenylenediamine; 4E means an electron-withdrawing substituent at the 4-position of the salicylidene) displays triplet emission, its quantum yield is low at room temperature. The correspond-
Introduction There is renewed interest in understanding the spectroscopic properties of transition-metal complexes. This is partly driven by the increased realization of their practical uses in diverse areas such as material sciences and solar chemistry. In particular, these applications involve processes with a change in spin state, and thus, the rates of these formally spin-forbidden transitions are essential in designing systems with desired functionalities. For example, molecules with fast triplet radiative
[a] Dr. G. S. M. Tong, Dr. P. K. Chow, Dr. W.-P. To, Prof. C.-M. Che State Key Laboratory of Synthetic Chemistry Institute of Molecular Functional Materials Department of Chemistry, The University of Hong Kong Pokfulam Road, Hong Kong (S.A.R. China) Fax: (+ 852) 2857-1586 E-mail: [email protected] [email protected] [b] Prof. W.-M. Kwok Department of Applied Biology and Chemical Technology The Hong Kong Polytechnic University Hung Hom, Kowloon, Hong Kong (S.A.R. China) Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/chem.201304375. Chem. Eur. J. 2014, 20, 6433 – 6443
ing PtII Schiff base complex, [Pt-salph-4E], is also much less emissive than the unsubstituted analogue, [Pt-salph]. Thus, a detailed theoretical analysis of how the substituent and central metal affected the photophysics of [M-salph-X] (X is a substituent on the salph ligand, M = Pt or Pd) was performed. Temperature effects were also investigated. The simple energy gap law underestimated the nonradiative decay rates and was insufficient to account for the temperature dependence of the nonradiative decay rates of the complexes studied herein. On the other hand, the present analysis demonstrates that inclusions of low-frequency modes and the associated frequency shifts are decisive in providing better quantitative estimates of the nonradiative decay rates and the experimentally observed temperature effects. Moreover, spin–orbit coupling, which is often considered only in the context of radiative decay rate, has a significant role in determining the nonradiative rate as well.
decay rates (kr) could serve as good triplet emitters in organic light-emitting diodes (OLEDs), whereas molecules with slow nonradiative decay rates (knr) and long emission lifetimes (tem) could act as sensitizers. In nonrelativistic quantum theory, the triplet-to-singlet transition, termed phosphorescence, is not allowed, but this restriction is partially lifted by the virtue of spin–orbit coupling (SOC). SOC scales approximately to the fourth power of the nuclear charge, Z (the so-called “heavy-atom effect”),[1] and therefore, many of the highly emissive materials come from third-row transition metals, such as [Ir(ppy)3] (ppy = C-deprotonated 2-phenylpyridines).[2] However, not all of the third-row transition-metal complexes are highly emissive. Prototypical examples are gold(III) complexes, of which only a few examples reported to date have emission quantum yields over 10 % in solution at room temperature.[3] Some derivatives of [Ir(ppy)3] are also only weakly emissive.[4] In addition to the heavy-atom effect, the rates of electronic transitions depend on the energies, electronic configurations, and nuclear structures of the triplet excited states.[5] Factors that are important in determining the phosphorescence emission lifetimes of transition-metal complexes include 1) the triplet excited-state energy (the so-called “energy gap law” (EGL)):[6] smaller energy gaps between the excited and ground
6433
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper states (DE) result in faster knr ; 2) the differences between the triplet excited- and ground-state coordinates: larger structural distortions of the excited state relative to the ground state result in faster knr ;[7] 3) the proximity of deactivating metal-centered (MC) dd excited state(s);[8] 4) the energy of singlet excited state(s): smaller energy gaps between the singlet and triplet excited states result in faster kr ;[9] 5) the transition dipole moments (m, or the oscillator strength, f) of the singlet excited states from which the triplet excited state can borrow intensity.[9] Although recent advances in femtosecond techniques provide valuable information in understanding ultrafast excitedstate dynamics, proper assignments of absorption and emission spectra are indispensable. In this regard, quantum mechanical calculations are powerful tools for both interpretative and predictive purposes, in particular, with the development in DFT and its time-dependent extension (TDDFT). The latter allows accurate and balanced treatments of both ground and excited states of medium to large systems that are computationally intractable with high-level multi-reference ab initio methods. Herein, DFT and TDDFT approaches are employed to study d8 transition-metal complexes with Schiff base ligands. Schiff bases are well known to have diverse and useful applications in chemistry. The remarkable features of Schiff bases include 1) easy preparation and modification from inexpensive starting materials, and 2) ready coordination to metal ions and the formation of stable transition-metal complexes. Recently, we have reported luminescent platinum(II) Schiff base complexes with phosphorescence quantum yields of up to 0.27 and with lifetimes less than 10 ms in acetonitrile at room temperature.[10] On the other hand, palladium(II) and gold(III) Schiff base complexes with the ligand N,N’-bis(salicylidene)ethylenediamine (salen), are nonemissive and fluorescent, respectively.[11] In addition, the 5-tert-butyl-substituted PtII–salph complex (salph = N,N’-bis(salicylidene)phenylenediamine; hereafter we refer to this complex as Pt-salph-5tB) has a lower emission energy (lmax = 625 nm) than the unsubstituted analogue, Pt-salph (lmax = 611 nm); the former has a slower nonradiative decay rate (knr = 1.58 105 s1 for Pt-salph-5tB and 2.24 105 s1 for Pt-salph),[10] which is contrary to the EGL prediction.
We first present the theoretical background for the radiative decay and nonradiative rate calculations. In the Results and Discussion section, we give an account of the different photophysics of [M-salen] complexes (M = Pt, Pd, or Au). Thereafter, a possibly emissive PdII Schiff base complex was proposed that was subsequently validated experimentally. Then, a detailed Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
theoretical account of how substitutions at the Schiff base ligand of salph affect the emission lifetimes and quantum yields at 77 and 298 K is presented.
Theoretical Background Spin–orbit coupling In principle, both radiative and nonradiative decay rates of triplet-to-singlet transitions depend on their SOC matrix element, HSOC : for radiative decay, kr , depends on SOC between the emitting triplet excited state (usually T1, according to the Kasha rule) and singlet excited states, Sm (m > 0), < Sm j HSOC j T1 > ;[12] on the other hand, the nonradiative decay rate, knr, is related to SOC between T1 and the singlet ground state S0, < S0 j HSOC j T1 > , if direct SOC is the dominant nonradiative decay path.[13] In the present context, pure-spin Born–Oppenheimer states are taken as the zero-order basis and the electronic-state wave functions, 1, 3Y, are taken as linear combinations of configuration-state functions (CSFs), fj [Eq (1)]: 1;3
Y¼
X
a1;3 j j
j
ð1Þ
in which aj is the configuration interaction (CI) coefficient of the CSF fj contributing to the electronic-state wave function, Y; fj is represented as a one-electron excitation from an occupied orbital to an unoccupied orbital that could be extracted from TDDFT calculations. These CSFs are generally classified as intraligand (IL(pp*)), metal-to-ligand charge transfer (MLCT(dp*)), ligand-to-ligand charge transfer (LLCT), ligand-to-metal charge transfer (LMCT), and MC (dd), according to the nature of the molecular orbitals (MOs; fj) involved in the CSFs. The MOs are usually expressed as linear combinations of atomic orbitals (AOs) qk [Eq. (2)]: j ¼
X j
cjk qk
ð2Þ
in which cjk are the mixing coefficients of the AO (qk) to the MO (fj). We have used an approach similar to that of Nozaki et al. in estimating the SOC matrix elements.[14] To simplify the SOC calculations, only one-center one-electron direct SOCs are considered and SOC involving atoms other than the transition metal is neglected (i.e., only qk = d orbitals of the transition metals are considered). Hence, HSOC ¼ 6 0 only when the coupling singlet and triplet excited states have 1) the same occupied or unoccupied orbitals; 2) different d orbitals; and 3) the same symmetry of the spin–orbit wave functions of the triplet substate, T1a, and the singlet state because the SOC operator is totally symmetric (there are three triplet substrates for the T1 excited state because S = 1; see the Supporting Information for definitions). Therefore, the HSOC matrix elements simplify to products of CI wave function coefficients (aj), mixing coefficients of the metal d orbital (cdj) and its SOC constant, z. The
6434
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper matrix elements between different d orbitals are tabulated in the Supporting Information.
Radiative decay rate (kr) The triplet radiative decay rate constant could be calculated from the Einstein coefficient of spontaneous emission [Eq. (3)]: 2 ~ðT Þ 8p h u kra ðT1 ! S0 Þ ¼ 3e0 h 1 MaT 2 3
3
rations between the three T1a sub-states (i.e., the ZFSs); kB is the Boltzmann constant. Usually, the ZFSs for d8 metal complexes are less than 100 cm1, and thus, at room temperature, the average triplet radiative decay rate could be approximated as Equation (8):
RT kr;avg ðT1 ! S0 Þ ¼
1X a k 3 a r
ð8Þ
ð3Þ Nonradiative decay rate (knr)
in which h is the refractive index of the solvent (herein, we used acetonitrile; h = 1.344), n˜ (T1) is the transition energy of the T1 excited state (in cm1), and MTa is the transition dipole moment of the T1a !S0 transition. Because phosphorescence is a spin-forbidden transition, MTa should be zero between pure spin states. The T1!S0 transition could steal its intensity by mixing the T1 excited state with singlet excited states through SOC. According to first-order perturbation theory, MTa is given by Equation (4): MaT ¼
X hTa1 jHSOC jSm i EðSm ÞEðT1 Þ MSm
ð4Þ
m
E(Sm) and E(T1) are the energies of the mth singlet excited state (Sm) and the lowest triplet excited state (T1), respectively; the bra-ket is the SOC matrix elements between the zeroth order T1 a-sub-state and Sm excited states (“zeroth order” means that these basis sets are the eigenstates before SOC); MSm is the transition dipole moment of the Sm !S0 transition. Therefore, Equation (3) could be rewritten as Equation (5):
a r
k ðT1 ! S0 Þ ¼
~ðT1 Þ3 8p2 h3 u 3e0 h
2 X X hT1a jHSOC jSm i EðSm ÞEðT1 Þ MSm ;j j2x;y;z m
ð5Þ
in which MSm,j is the j-axis projection of the Sm !S0 transition dipole moment. Expressing n˜ (T1) in cm1 and the transition 8p2 dipole moment in atomic units, ea0, the prefactor 3e0 h = 6 2.0261 10 . The energies of the triplet sub-states are given by Equation (6): a 1
EðT Þ ¼ EðT1 Þ þ
ð6Þ
EðSm ÞEðT1 Þ m
The zero-field splitting (ZFS) would be given by the energy difference between the triplet sub-states. The average radiative decay rate, kr , is thus given by Equation (7):
kr;avg ðT1 ! S0 Þ ¼
kI þkII e
1þe
DEI;II kB T
DEI;II kB T
e þkIII
þe
DEI;III kB T
DEI;III kB T
ð7Þ
hT1 jHSOC jS0 i2 h2
Z
i½Ev ðT1 ÞEv ðS0 Þt f ðtÞ exp dt h T
s
ð9Þ
in which uT and uS designate, respectively, the collective vibrational quantum numbers of the T1 and S0 states (i.e., j uT > = j u1T, u2T, u3T, …, ujT, …, uNT > ; the index j is the jth normal mode; N is the total number of normal modes; uS has a similar meaning); therefore, EuT(T1)EuS(S0) is the energy difference between the T1 and S0 vibronic states; f(t) relates to the thermally averaged Franck–Condon factors (FCFs). Two further assumptions are usually made to reduce the formal result into a useful and tractable expression: 1) the vibrational wave functions of the electronic states are independent harmonic oscillators, and 2) the potential energy surfaces of T1 and S0 states are parallel (i.e., wjS = wjT for all j = 1, 2, …, N; wjS(T) is the frequency of the jth normal mode of the S0 (T1) state). In the weak coupling limit (i.e., S < 1 or hwM @ kBT; S is the Huang–Rhys factor) with two-mode approximation, the nonradiative decay rate reduced to the EGL expression [Eq. (10)]:[16] 1=2 2p 1 knr ðEGLÞ ¼ h hT1 jHSOC jS0 i2 2phwM DE h i 2 DE gþ1 expðSM Þ exp ghwM þ hwM llf kB T DE g ¼ ln lM 1 1m w Sj ¼ 2 hj j DQ2j X SM ¼ Sj 0
0
j2hf
lM ¼
X
ð10Þ
Sj hwj
j2hf
l hwM ¼ M SM X llf ¼ Sj hwj j2lf 0
DE ¼ DE00 llf
kI,II,III are the radiative decay rates of the individual sub-states I, II, and III, respectively (I denotes the lowest energy T1 substate and III is the highest). DEI,II and DEI,III are the energy sepaChem. Eur. J. 2014, 20, 6433 – 6443
knr ðT1 ! S0 Þ ¼
0
X hTa1 jHSOC jSm i2
The nonradiative decay rate could be calculated by Fermi’s Golden Rule in the limit of Condon approximation by using the generating function method [Eq. (9)]:[15]
www.chemeurj.org
in which hf and lf designate the high-frequency ligand skeletal modes (1700 > whf > 1000 cm1) and low-frequency modes
6435
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper (wlf 1000 cm1), respectively; mj is the reduced mass of the jth normal mode; DQj represents the equilibrium displacement along the jth normal mode coordinate; and DE00 is the zeropoint energy difference between the T1 excited state and the S0 ground state. In Equation (10), it is assumed that there is effectively only one dominant accepting mode (designated with the subscript “M”). This EGL expression works well if the nuclear tunneling effect is large and a weak temperature dependence on knr should be observed. Herein, we would also apply another approach to obtain the nonradiative decay rates by considering that 1) the MLCT transition should cause a change in electron density on both the metal ion and ligand and the assumption of no frequency change between the T1 and S0 states should be removed; and 2) as described below, the Huang–Rhys factor for the low-frequency modes, Slf, are greater than one, which means that the lf modes should be treated in the strong coupling limit. In this limit, knr has a stronger temperature dependence because the lf modes with large values of Slf become better acceptors as the temperature increases. As such, knr is calculated by the convolution method in which the lf modes are treated in the strong coupling limit and the hf modes in the weak coupling limit. In addition, as demonstrated below, there are large frequency changes for the lf modes between the two electronic states, and thus, frequency change for the lf modes also need to be taken into account [Eq. (11)]:[17]
Results and Discussion In this section, an account of the different photophysics of [Msalen] (M = Pd, Au, and Pt) is presented, followed by a proposed way to get an emissive PdII Schiff base complex, and then a detailed theoretical study on the effect of substitution at the salph ligand on the photophysics of M-salph-X (M = Pt or Pd; X = a substituent at the salph ligand).
M-salen In the literature, it is reported that 1) [Pd-salen] is nonemissive,[11a] 2) [Pt-salen] is moderately emissive (fem = 0.19),[10] and 3) [Au-salen] + is fluorescent with fem = 0.15.[11b] All the measurements were performed in acetonitrile at 298 K. The following refers to our theoretical investigations.
Pd-salen At the ground-state geometry, the major contributions of the first three lowest-lying singlet excited states are HOMO! L + 1 (S1), HOMO!LUMO (S2), and HOMO!L + 2 (S3) (see Table 1). The HOMO is composed of ligand p and Pd(dp) orbitals, whereas the LUMO and L + 1 are mainly of ligand p* in character and L + 1 also have some metal dp character (MO diagrams in the Supporting Information). L + 2 is mainly Pd(ds*)
1=2 2p knr ðT1 ! S0 Þ ¼ hT jH jS i2 2ph2 ðD21 þ P2 Þ h 1 SOC 0 ðDE00 nM hwM l1 mÞ2 Sn exp expðSM ÞnMM ! 2 2 2 2p h ðD1 þ P Þ X Sj hwsj l1 ¼ bj j2lf
Table 1. Calculated low-lying singlet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions the optimized ground-state (S0) geometry of [Pd-salen] in CH3CN.
M
m¼
2 hwTj 1 X s 1 bj hw j coth bj 2kB T 2 j2lf
X h wsj 2 hwTj Sj coth bj 2kB T j2lf " #2 ð1 b2j Þ hwTj 1X 2 2 s hP ¼ hw j coth 2kB T bj 2 j2lf
ð11Þ
wTj wSj
nM ¼
DE00 l1 m hw M
Because nM is the number of vibrational quanta of hwM , the quotient is corrected to the smaller integer. From Equation (11), it can be seen that the new temperature-dependent term (m) arising from frequency changes could increase (if m < 0) or decrease (if m > 0) the effective energy gap (DE00l1m), and hence, increase or decrease the number of quanta (nM) of the hf mode wM needed for the T1!S0 transition, and correspondingly, lead to a decrease or increase of knr and the effect would be enhanced as the temperature increases. Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
l [nm]
f
Major contributions[a]
lexp [nm] (e [103 mol1 dm3 cm1])[b]
Assignment[c]
S1 S2 S3
387 384 379
0.0569 0.1665 0.0124
H!L + 1 H!L H!L + 2
420 (2.9) 392 (5.3)
MLCT/IL (14) MLCT/IL (14) MC dd
[a] H = HOMO; L = LUMO. [b] Ref. [11a]. [c] The values in parentheses indicate the metal character of the occupied orbital in the transition. IL = intraligand.
2 D21 ¼ h
bj ¼
State
in character. Thus, S1 and S2 are mainly 1MLCT/IL in nature, whereas S3 is a MC 1dd excited state. These three singlet excited states are very close in energy and the 1dd excited state is only about 290–520 cm1 above the two 1MLCT excited states. Thus, excitation close to the first peak maximum (S2 state from the calculation) can easily populate the 1dd excited state, and hence, fast nonradiative decay prior to intersystem crossing (ISC) takes place. Thus, this complex is expected to be nonemissive. [Au-salen] + At the ground-state geometry, the major contributions of the first four lowest-lying singlet excited states are HOMO!LUMO (S1), H1!LUMO (S2), HOMO!L + 1 and L + 2 (S3 and S4) (see
6436
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 2. Calculated low-lying singlet and triplet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions at the optimized ground-state (S0) geometry of [Au-salen] + in CH3CN. State
l [nm]
f
Nature
S1 T6
437 389
0.0037 0
S2 S3
375 364
0.0037 0.1685
S4
362
0.0083
T7
347
0
H!L H2!L + 2 H3!L + 1 H1!L H!L + 1 H!L + 2 H!L + 2 H!L + 1 H-1!L + 1 H!L + 2
lexp [nm] (e [103 mol1 dm3 cm1])[a]
Table 3. Calculated low-lying singlet and triplet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions at the optimized ground-state (S0) geometry of [Pt-salen] in CH3CN. State
l [nm]
f
Nature
lexp [nm] (e103 mol1 dm3 cm1)[a]
Assignment[b]
S1 T3
420 403
0.0977 0
417 (5.9)
LMCT IL/MLCT (3)
T4
403
0
IL/MLCT (3)
S2
400
0.1178
H!L H1!L H4!L + 1 H3!L H1!L + 1 H4!L H3!L + 1 H!L + 1
MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL
Assignment[b]
LMCT LMCT
375 (3.9)
(24) (10) (17) (34) (10) (17) (34) (24)
[a] Ref.[10]. [b] The values in the parentheses indicate the metal character of the occupied orbital in the transition.
IL/MLCT (2) IL/MLCT (4)
[a] Ref. [11b]. [b] The values in parentheses indicate the metal character of the occupied orbital in the transition.
Table 2). The HOMO is composed of ligand p and Au(dp) orbitals, similar to those of [Pd-salen], but with a smaller metal character in [Au-salen] + ( 3 %) than that in Pd-salen ( 14 %). On the other hand, the LUMO is mainly composed of Au(ds*) and is about 0.2 eV lower lying than the L + 1 and L + 2 orbitals. The latter two orbitals are predominantly of ligand p* character with an insignificant amount of Au(dp) character of 1 and 2 %, respectively. Thus, both S1 and S2 could be assigned as 1LMCT excited states, whereas S3 and S4 are 1IL/MLCT excited states. Upon excitation at l = 380 nm, the S3 excited state would be populated. Now S2 is only about 820 cm1 below the S3 excited state and more than 1700 cm1 above the closest triplet excited state (T6), so it would undergo internal conversion from S3 to S2 instead of ISC from S3 to T6. Because S2 is an LMCT excited state, fast nonradiative decay should take place. Based on calculations, [Au-salen] + should also be nonemissive in CH3CN. We thus repeated the experiment and found that [Au-salen] + was, at most, weakly emissive in CH3CN at room temperature (fem 1.6 104 with a time-resolved fluorescence decay lifetime, tf, of 2.7 ps; see experimental details in the Supporting Information). Pt-salen At the ground-state geometry, the first two lowest-lying singlet excited states are HOMO!LUMO (S1) and HOMO!L + 1 (S2) (Table 3). The HOMO, LUMO, and L + 1 orbitals have the same parentage as those of the PdII analogue, [Pd-salen]. The only difference is that the metal contributions are larger in the case of Pt-salen: the HOMO and L + 1 have 24 and 7 % Pt(dp) character, respectively. Therefore, S1 and S2 are both 1MLCT/IL excited states. The Pt(ds*) orbital is now L + 4 and lies about 2.0 eV above the LUMO. Thus, the MC 1dd excited state of Pt-salen is more than 9000 cm1 above the two 1MLCT/IL excited states, S1 and S2. Because the S1 and S2 excited states have similar osChem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
cillator strengths, both states may be populated upon excitation at l = 390 nm. For [Pt-salen], the closest-lying triplet excited states to S2 are T3 and T4, both are about 190–210 cm1 below the S2 excited state. Given the small singlet–triplet energy separation and the large metal contributions in both the S2 and T4 excited states, fast ISC could take place, even at the Franck–Condon geometry. After ISC from S2 to the T4 excited state, [Pt-salen] could undergo internal conversion to the T1 excited state, which is also predominantly of HOMO!LUMO character, that is, a 3MLCT excited state. The change in bond lengths between the T1 and S0 states does not exceed 0.1 and Sj is smaller than 0.5 (see the Supporting Information). Thus, structural distortion of T1 relative to the ground state in the case of [Pt-salen] is not significant and nonradiative decay is expected to be the slowest for [Pt-salen] among the three [M-salen] complexes presented in this section. Hence, phosphorescence could be observed with [Pt-salen]. The calculated value of kr is 5.03 104 s1, which is in good agreement with experimentally determined kr = fem/ tem = 5.43 104 s1;[10] this lends support that the method is adequate for d8 transition-metal complexes with Schiff base ligands. Is it possible to obtain emissive PdII Schiff base complexes? From the previous section, the dominant nonradiative decay path(s) comes from the MC dd and LMCT excited states at energies close to the MLCT/IL excited states; these low-lying MC dd and LMCT excited states would lead to large structural distortions and facile nonradiative decay. To alleviate this problem, the common strategy is to widen the MLCT–MC/LMCT energy gap. This could be done by introducing stronger sdonor atoms coordinated to the metal ion. However, since we want to keep the Schiff base O^N^N^O framework, we sought to increase this energy gap by introducing substituent(s) the Schiff base ligand. For [Pd-salen], the PdII ds* orbital is higher lying than the salen ligand p* orbital. On the other hand, for [Au-salen] + , the LUMO is ds* and is about 0.2 eV below the
6437
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper ligand p* orbitals. Thus, it may be more stringent for AuIII Schiff base complexes to push the ds* orbital sufficiently high in energy such that the LMCT excited state(s) would not be a viable deactivating path, if the O^N^N^O framework is to be kept. Given the higher emission quantum yield of the Pt-salph complex (fem = 0.23) relative to its [Pt-salen] counterpart (fem = 0.19),[10] we sought to modify the [Pd-salph] complex instead of [Pd-salen]. With the phenylene bridge replacing the ethylene bridge, conjugation across the ligand would be more extended, so that the ligand p* orbitals would become more low lying, but without significant effect on the character of the ds* orbital, since the energy of the latter would be less affected than that of the ligand p* orbitals. Thus, the ds* and p* orbitals using a phenylene bridge are expected to be pushed further apart in energy compared with the case using an ethylene bridge. To further increase the energy separation between the ligand p* orbital and the ds* orbital, while keeping the O^N^N^O framework for coordination to the metal ion, electron-withdrawing substituent(s) could be installed at the 4and/or 6-positions of the salicylidene moiety (i.e., meta to the phenoxide O). This is because there are significant electron densities at these positions in the ligand p* orbitals, but not in the ds* orbital (Figure 1). It is expected that such modifications
Figure 1. Lowest-lying unoccupied orbitals in the [M-salph] complexes (M = Pt or Pd). The subscripts o and i in the p* orbitals indicate that the combinations of the two salicylidene moieties are out-of-phase and inphase, respectively. The diagrams shown are those of [Pt-salph]; [Pd-salph] has analogous MOs.
would increase the MLCT–MC energy gap, thereby suppressing the nonradiative decay path through direct access to the dd excited states. We have chosen an ester, COOMe, as the electron-withdrawing group, resulting in the complex [Pd-salph-4E]. The ligand p* orbitals (LUMO and L + 1) are 1.63 and 1.17 eV, respectively, below the ds* orbital (L + 2). At its optimized S0 geometry, the MC 1dd excited state is more than 6500 cm1 above the lowest 1MLCT excited state [i.e., S1, and is derived from a HOMO!LUMO transition with the HOMO composed of ligand p and Pd(dp) fragment orbitals at the optimized ground-state geometry; see the Supporting Information for the MO surfaces]. From the results given in Table 4, it can be seen that the closest-lying triplet excited state that can have effective SOC with the S1 excited state is T3, which is about 310 cm1 below the S1 excited state. ISC should thus be rapid. Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
Table 4. Calculated low-lying singlet and triplet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions at the optimized ground-state (S0) geometry of [Pd-salph-4E] in CH3CN. State
l [nm]
f
Nature
Assignment[a]
T3
497
0
S1 T4
489 464
0.2432 0
S2
452
0.1856
H2!L H4!L + 1 H!L H!L + 2 H7!L + 2 H9!L + 2 H1!L H!L + 1
MLCT/IL MLCT/IL MLCT/IL dd dd dd MLCT/IL MLCT/IL
(12) (5) (14)
(2) (13)
[a] The values in parentheses indicate the metal character of the occupied orbital in the transition.
This T3 excited state can then decay to the T1 excited state through internal conversion. The optimized T1 excited state of [Pd-salph-4E] is mainly derived from the HOMO!LUMO transition with some contributions from the H1!LUMO transition, in which H1 is mainly composed of ligand p character (see the Supporting Information for the MO surfaces). Therefore, T1 is assigned to be a 3MLCT/IL excited state. At the optimized T1 geometry, the 3 dd excited state is more than 8200 cm1 above T1 (see the Supporting Information), and so it is, at first sight, envisioned that the 3dd excited state is not an effective deactivation channel for [Pd-salph-4E] phosphorescence. In addition, the differences in bond lengths between the optimized T1 and S0 states are less than 0.04 and individual Sj less than 0.4; thus, structural distortions should also not be large and we propose that [Pd-salph-4E] should display phosphorescence. Subsequently, we have prepared [Pd-salph-4E], which displays phosphorescence, albeit very weakly at room temperature (lem = 613 nm, fem = 1.7 104 at lex = 510 nm, t < 0.2 ms).[18] At 77 K in a glassy medium (MeOH/EtOH (1:4)), there is a substantial increase in emission lifetime: tem 143 ms. Such a large temperature dependence in tem is often attributed to the presence of thermally accessible 3dd excited state(s) close to the 3MLCT excited state. We thus performed a TDDFT optimization of the 3dd excited state and found that its potential energy minimum was only about 1530 cm1 above the T1 potential energy minimum. Thus, the activation energy for 3 MLCT!3dd is only about 2530 cm1. Comparable activation energies have also been reported for very weakly emissive cis[Ru(bpy)2L2]2 + (bpy = 2,2’-bipyridine; L is a nonchromophoric ligand).[6] Therefore, at 298 K, this 3dd excited state is thermally accessible and contributes to fast nonradiative decay (knr > 106 s1). At 77 K, the thermal energy is not sufficient to overcome the activation barrier and the population of 3dd excited state is no longer feasible; hence, the measured lifetime is that originating from the lowest 3MLCT excited state. Our value of knr calculated by using Equation (11) is about 1.07 104 s1, which is comparable to the experimentally estimated knr value of about 6.14 103 s1 at 77 K (see below).
6438
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Substituent effect on [Pt-salph-X]
stituted [Pt-salph]. From these results, it can be seen that [Ptsalph-4E] has the slowest radiative decay rate and the longest If the emitting triplet excited state of [Pt-salph-4E] has similar emission wavelengths of the three PtII Schiff base complexes. structural distortions to that of [Pd-salph-4E], the FCFs of the By the EGL, knr / exp(gDE/hwM), a smaller value of DE (or the two complexes would be similar and the estimated knr of [Ptlonger the emission wavelength, lmax) results in a faster nonrasalph-4E] would be larger than that of [Pd-salph-4E] by about diative decay rate. Based on this rationale, it is reasonable for an order of magnitude (because the SOC effect on knr would [Pt-salph-4E] to have a lower fem value than that of [Pt-salph]. become the determining factor: knr / HSOC2 / z2 ; z(Pt) = However, if the EGL is the sole reason, one would expect that [Pt-salph-4E] should have a knr value of more than 100-fold 4368 cm1 and z(Pd) = 1412 cm1). The 3dd excited state is not expected to be the deactivating path because, from previous faster than that of [Pt-salph] (see below). In addition, although theoretical analysis on [Pt-salen], there is no thermally accessi[Pt-salph-5tB] has a lower emission energy than that of [Ptble 3dd excited state close to the emitting 3MLCT excited state. salph], its knr is the slowest among the three PtII complexes Since the kr value of [Pt-salph] is experimentally found to be listed in Table 5. Moreover, within the applicability of the EGL, knr should display a weak temperature dependence. However, 6.71 104 s1,[10] we conceived that an emission quantum yield of approximately 0.45 for [Pt-salph-4E] could be feasible (asas seen from Table 5, the knr value of [Pt-salph] is more than ten times slower upon cooling from 298 to 77 K. These feasuming kr 6 104 s1 and knr 7 104 s1). Thus, [Pt-salph-4E] tures prompted us to undertake a detailed theoretical analysis was prepared and its photophysical properties examined. Its on the effect of substitution on the photophysical properties emission data together with those of [Pt-salph], [Pt-salph-5tB], of these PtII Schiff base complexes. and its PdII analogue, [Pd-salph-4E], are listed in Table 5. As depicted in Table 5, [Pt-salph-4E] is only weakly emissive In the following, we first present the results for calculated and its emission quantum yield is even less than that of unsubkr to justify that the electronic structure of their respective optimized T1 excited state is a good representation. Then, we list the calculated knr values within the context of the EGL Table 5. Photophysical parameters of [Pt-salph], [Pt-salph-5tB], [Pt-salph-4E], and [Pd-salph-4E] derived from [Eq. (10)] and the convoluted apexperimental data. proach [Eq. (11)], in which the tem knr tem[c] knr[e] lmax fem kr kr[d] latter takes into account the fre(298 K) (298 K) (298 K) (298 K) (298 K) (77 K) (77 K) (77 K) quency change between their [nm] [ms] [104 s1] [104 s1] [ms] [104 s1] [104 s1] respective optimized T1 and S0 [Pt-salph][a] 611 0.23 3.43 6.71 22.4 11.0 8.50 0.601 states and the low-frequency modes treated in the strong coupling limit.
[Pt-salph-5tB][a]
625
0.27
4.62
5.84
15.8
10.6
6.109
3.34
Radiative decay rates (kr)
[Pt-salph-4E][b]
661
0.033
[Pd-salph-4E][b]
613
1.7 104
1.6
< 0.2
2.06
0.085
60.4
> 500
3.70
143
2.42
0.085
24.6
0.614
[a] Measured in CH3CN at 298 K.[10] [b] Measured in dichloromethane because the ester-substituted complexes do not dissolve well in CH3CN (this work). [c] Recorded in MeOH/EtOH (1:4). [d] kr is estimated by assuming that there is a linear correlation between the calculated and experimental kr values; kr is calculated by using 1 Equation (7). [e] tem ¼ kr þ knr . Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
6439
Table 6 depicts the DE00, ZFS, and kr,avg values at 298 and 77 K for [Pt-salph], [Pt-salph-5tB], [Ptsalph-4E], and [Pd-salph-4E]. The calculated zero-point energy differences compare well with the corresponding experimental emission maxima. Theoretical values of kr at 298 K compare satisfactorily with those found from experiments (kr, exp = fem/tem), except for [Pt-salph5tB], which is predicted to have a slower radiative decay rate than that of [Pt-salph-4E], contrary to the experimental findings. The total emission rate should be written as Equation (12):[19]
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 6. Photophysical parameters of [Pt-salph], [Pt-salph-5tB], [Pt-salph-4E], and [Pdsalph-4E] obtained at their respective optimized T1 geometries. DE00 [cm1] ([nm]) [Pt-salph] [Pt-salph-5tB] [Pt-salph-4E] [Pd-salph-4E]
15 222 14 692 13 758 14 715
(657) (681) (727) (680)
kr,avg (298 K)[a] 104 s1]
ZFS (DEI,II ; DEI,III) [cm1]
kr,avg (77 K)[b] [104 s1]
5.35 (5.89) 3.23 (3.32) 4.86 (5.26) 0.0258 (0.0258)
13.7; 46.7 11.1; 41.53 0.6; 52.8
Full Paper
& Photophysics
A Theoretical Investigation into the Luminescent Properties of d8Transition-Metal Complexes with Tetradentate Schiff Base Ligands Glenna So Ming Tong,*[a] Pui Keong Chow,[a] Wai-Pong To,[a] Wai-Ming Kwok,[b] and ChiMing Che*[a]
Abstract: A theoretical investigation on the luminescence efficiency of a series of d8 transition-metal Schiff base complexes was undertaken. The aim was to understand the different photophysics of [M-salen]n complexes (salen = N,N’bis(salicylidene)ethylenediamine; M = Pt, Pd (n = 0); Au (n = + 1)) in acetonitrile solutions at room temperature: [Pt-salen] is phosphorescent and [Au-salen] + is fluorescent, but [Pdsalen] is nonemissive. Based on the calculation results, it was proposed that incorporation of electron-withdrawing groups at the 4-position of the Schiff base ligand should widen the 3 MLCT–3MC gap (MLCT = metal-to-ligand charge transfer and MC = metal centered, that is, the dd excited state); thus permitting phosphorescence of the corresponding PdII Schiff base complex. Although it is experimentally proven that [Pdsalph-4E] (salph = N,N’-bis(salicylidene)-1,2-phenylenediamine; 4E means an electron-withdrawing substituent at the 4-position of the salicylidene) displays triplet emission, its quantum yield is low at room temperature. The correspond-
Introduction There is renewed interest in understanding the spectroscopic properties of transition-metal complexes. This is partly driven by the increased realization of their practical uses in diverse areas such as material sciences and solar chemistry. In particular, these applications involve processes with a change in spin state, and thus, the rates of these formally spin-forbidden transitions are essential in designing systems with desired functionalities. For example, molecules with fast triplet radiative
[a] Dr. G. S. M. Tong, Dr. P. K. Chow, Dr. W.-P. To, Prof. C.-M. Che State Key Laboratory of Synthetic Chemistry Institute of Molecular Functional Materials Department of Chemistry, The University of Hong Kong Pokfulam Road, Hong Kong (S.A.R. China) Fax: (+ 852) 2857-1586 E-mail: [email protected] [email protected] [b] Prof. W.-M. Kwok Department of Applied Biology and Chemical Technology The Hong Kong Polytechnic University Hung Hom, Kowloon, Hong Kong (S.A.R. China) Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/chem.201304375. Chem. Eur. J. 2014, 20, 6433 – 6443
ing PtII Schiff base complex, [Pt-salph-4E], is also much less emissive than the unsubstituted analogue, [Pt-salph]. Thus, a detailed theoretical analysis of how the substituent and central metal affected the photophysics of [M-salph-X] (X is a substituent on the salph ligand, M = Pt or Pd) was performed. Temperature effects were also investigated. The simple energy gap law underestimated the nonradiative decay rates and was insufficient to account for the temperature dependence of the nonradiative decay rates of the complexes studied herein. On the other hand, the present analysis demonstrates that inclusions of low-frequency modes and the associated frequency shifts are decisive in providing better quantitative estimates of the nonradiative decay rates and the experimentally observed temperature effects. Moreover, spin–orbit coupling, which is often considered only in the context of radiative decay rate, has a significant role in determining the nonradiative rate as well.
decay rates (kr) could serve as good triplet emitters in organic light-emitting diodes (OLEDs), whereas molecules with slow nonradiative decay rates (knr) and long emission lifetimes (tem) could act as sensitizers. In nonrelativistic quantum theory, the triplet-to-singlet transition, termed phosphorescence, is not allowed, but this restriction is partially lifted by the virtue of spin–orbit coupling (SOC). SOC scales approximately to the fourth power of the nuclear charge, Z (the so-called “heavy-atom effect”),[1] and therefore, many of the highly emissive materials come from third-row transition metals, such as [Ir(ppy)3] (ppy = C-deprotonated 2-phenylpyridines).[2] However, not all of the third-row transition-metal complexes are highly emissive. Prototypical examples are gold(III) complexes, of which only a few examples reported to date have emission quantum yields over 10 % in solution at room temperature.[3] Some derivatives of [Ir(ppy)3] are also only weakly emissive.[4] In addition to the heavy-atom effect, the rates of electronic transitions depend on the energies, electronic configurations, and nuclear structures of the triplet excited states.[5] Factors that are important in determining the phosphorescence emission lifetimes of transition-metal complexes include 1) the triplet excited-state energy (the so-called “energy gap law” (EGL)):[6] smaller energy gaps between the excited and ground
6433
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper states (DE) result in faster knr ; 2) the differences between the triplet excited- and ground-state coordinates: larger structural distortions of the excited state relative to the ground state result in faster knr ;[7] 3) the proximity of deactivating metal-centered (MC) dd excited state(s);[8] 4) the energy of singlet excited state(s): smaller energy gaps between the singlet and triplet excited states result in faster kr ;[9] 5) the transition dipole moments (m, or the oscillator strength, f) of the singlet excited states from which the triplet excited state can borrow intensity.[9] Although recent advances in femtosecond techniques provide valuable information in understanding ultrafast excitedstate dynamics, proper assignments of absorption and emission spectra are indispensable. In this regard, quantum mechanical calculations are powerful tools for both interpretative and predictive purposes, in particular, with the development in DFT and its time-dependent extension (TDDFT). The latter allows accurate and balanced treatments of both ground and excited states of medium to large systems that are computationally intractable with high-level multi-reference ab initio methods. Herein, DFT and TDDFT approaches are employed to study d8 transition-metal complexes with Schiff base ligands. Schiff bases are well known to have diverse and useful applications in chemistry. The remarkable features of Schiff bases include 1) easy preparation and modification from inexpensive starting materials, and 2) ready coordination to metal ions and the formation of stable transition-metal complexes. Recently, we have reported luminescent platinum(II) Schiff base complexes with phosphorescence quantum yields of up to 0.27 and with lifetimes less than 10 ms in acetonitrile at room temperature.[10] On the other hand, palladium(II) and gold(III) Schiff base complexes with the ligand N,N’-bis(salicylidene)ethylenediamine (salen), are nonemissive and fluorescent, respectively.[11] In addition, the 5-tert-butyl-substituted PtII–salph complex (salph = N,N’-bis(salicylidene)phenylenediamine; hereafter we refer to this complex as Pt-salph-5tB) has a lower emission energy (lmax = 625 nm) than the unsubstituted analogue, Pt-salph (lmax = 611 nm); the former has a slower nonradiative decay rate (knr = 1.58 105 s1 for Pt-salph-5tB and 2.24 105 s1 for Pt-salph),[10] which is contrary to the EGL prediction.
We first present the theoretical background for the radiative decay and nonradiative rate calculations. In the Results and Discussion section, we give an account of the different photophysics of [M-salen] complexes (M = Pt, Pd, or Au). Thereafter, a possibly emissive PdII Schiff base complex was proposed that was subsequently validated experimentally. Then, a detailed Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
theoretical account of how substitutions at the Schiff base ligand of salph affect the emission lifetimes and quantum yields at 77 and 298 K is presented.
Theoretical Background Spin–orbit coupling In principle, both radiative and nonradiative decay rates of triplet-to-singlet transitions depend on their SOC matrix element, HSOC : for radiative decay, kr , depends on SOC between the emitting triplet excited state (usually T1, according to the Kasha rule) and singlet excited states, Sm (m > 0), < Sm j HSOC j T1 > ;[12] on the other hand, the nonradiative decay rate, knr, is related to SOC between T1 and the singlet ground state S0, < S0 j HSOC j T1 > , if direct SOC is the dominant nonradiative decay path.[13] In the present context, pure-spin Born–Oppenheimer states are taken as the zero-order basis and the electronic-state wave functions, 1, 3Y, are taken as linear combinations of configuration-state functions (CSFs), fj [Eq (1)]: 1;3
Y¼
X
a1;3 j j
j
ð1Þ
in which aj is the configuration interaction (CI) coefficient of the CSF fj contributing to the electronic-state wave function, Y; fj is represented as a one-electron excitation from an occupied orbital to an unoccupied orbital that could be extracted from TDDFT calculations. These CSFs are generally classified as intraligand (IL(pp*)), metal-to-ligand charge transfer (MLCT(dp*)), ligand-to-ligand charge transfer (LLCT), ligand-to-metal charge transfer (LMCT), and MC (dd), according to the nature of the molecular orbitals (MOs; fj) involved in the CSFs. The MOs are usually expressed as linear combinations of atomic orbitals (AOs) qk [Eq. (2)]: j ¼
X j
cjk qk
ð2Þ
in which cjk are the mixing coefficients of the AO (qk) to the MO (fj). We have used an approach similar to that of Nozaki et al. in estimating the SOC matrix elements.[14] To simplify the SOC calculations, only one-center one-electron direct SOCs are considered and SOC involving atoms other than the transition metal is neglected (i.e., only qk = d orbitals of the transition metals are considered). Hence, HSOC ¼ 6 0 only when the coupling singlet and triplet excited states have 1) the same occupied or unoccupied orbitals; 2) different d orbitals; and 3) the same symmetry of the spin–orbit wave functions of the triplet substate, T1a, and the singlet state because the SOC operator is totally symmetric (there are three triplet substrates for the T1 excited state because S = 1; see the Supporting Information for definitions). Therefore, the HSOC matrix elements simplify to products of CI wave function coefficients (aj), mixing coefficients of the metal d orbital (cdj) and its SOC constant, z. The
6434
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper matrix elements between different d orbitals are tabulated in the Supporting Information.
Radiative decay rate (kr) The triplet radiative decay rate constant could be calculated from the Einstein coefficient of spontaneous emission [Eq. (3)]: 2 ~ðT Þ 8p h u kra ðT1 ! S0 Þ ¼ 3e0 h 1 MaT 2 3
3
rations between the three T1a sub-states (i.e., the ZFSs); kB is the Boltzmann constant. Usually, the ZFSs for d8 metal complexes are less than 100 cm1, and thus, at room temperature, the average triplet radiative decay rate could be approximated as Equation (8):
RT kr;avg ðT1 ! S0 Þ ¼
1X a k 3 a r
ð8Þ
ð3Þ Nonradiative decay rate (knr)
in which h is the refractive index of the solvent (herein, we used acetonitrile; h = 1.344), n˜ (T1) is the transition energy of the T1 excited state (in cm1), and MTa is the transition dipole moment of the T1a !S0 transition. Because phosphorescence is a spin-forbidden transition, MTa should be zero between pure spin states. The T1!S0 transition could steal its intensity by mixing the T1 excited state with singlet excited states through SOC. According to first-order perturbation theory, MTa is given by Equation (4): MaT ¼
X hTa1 jHSOC jSm i EðSm ÞEðT1 Þ MSm
ð4Þ
m
E(Sm) and E(T1) are the energies of the mth singlet excited state (Sm) and the lowest triplet excited state (T1), respectively; the bra-ket is the SOC matrix elements between the zeroth order T1 a-sub-state and Sm excited states (“zeroth order” means that these basis sets are the eigenstates before SOC); MSm is the transition dipole moment of the Sm !S0 transition. Therefore, Equation (3) could be rewritten as Equation (5):
a r
k ðT1 ! S0 Þ ¼
~ðT1 Þ3 8p2 h3 u 3e0 h
2 X X hT1a jHSOC jSm i EðSm ÞEðT1 Þ MSm ;j j2x;y;z m
ð5Þ
in which MSm,j is the j-axis projection of the Sm !S0 transition dipole moment. Expressing n˜ (T1) in cm1 and the transition 8p2 dipole moment in atomic units, ea0, the prefactor 3e0 h = 6 2.0261 10 . The energies of the triplet sub-states are given by Equation (6): a 1
EðT Þ ¼ EðT1 Þ þ
ð6Þ
EðSm ÞEðT1 Þ m
The zero-field splitting (ZFS) would be given by the energy difference between the triplet sub-states. The average radiative decay rate, kr , is thus given by Equation (7):
kr;avg ðT1 ! S0 Þ ¼
kI þkII e
1þe
DEI;II kB T
DEI;II kB T
e þkIII
þe
DEI;III kB T
DEI;III kB T
ð7Þ
hT1 jHSOC jS0 i2 h2
Z
i½Ev ðT1 ÞEv ðS0 Þt f ðtÞ exp dt h T
s
ð9Þ
in which uT and uS designate, respectively, the collective vibrational quantum numbers of the T1 and S0 states (i.e., j uT > = j u1T, u2T, u3T, …, ujT, …, uNT > ; the index j is the jth normal mode; N is the total number of normal modes; uS has a similar meaning); therefore, EuT(T1)EuS(S0) is the energy difference between the T1 and S0 vibronic states; f(t) relates to the thermally averaged Franck–Condon factors (FCFs). Two further assumptions are usually made to reduce the formal result into a useful and tractable expression: 1) the vibrational wave functions of the electronic states are independent harmonic oscillators, and 2) the potential energy surfaces of T1 and S0 states are parallel (i.e., wjS = wjT for all j = 1, 2, …, N; wjS(T) is the frequency of the jth normal mode of the S0 (T1) state). In the weak coupling limit (i.e., S < 1 or hwM @ kBT; S is the Huang–Rhys factor) with two-mode approximation, the nonradiative decay rate reduced to the EGL expression [Eq. (10)]:[16] 1=2 2p 1 knr ðEGLÞ ¼ h hT1 jHSOC jS0 i2 2phwM DE h i 2 DE gþ1 expðSM Þ exp ghwM þ hwM llf kB T DE g ¼ ln lM 1 1m w Sj ¼ 2 hj j DQ2j X SM ¼ Sj 0
0
j2hf
lM ¼
X
ð10Þ
Sj hwj
j2hf
l hwM ¼ M SM X llf ¼ Sj hwj j2lf 0
DE ¼ DE00 llf
kI,II,III are the radiative decay rates of the individual sub-states I, II, and III, respectively (I denotes the lowest energy T1 substate and III is the highest). DEI,II and DEI,III are the energy sepaChem. Eur. J. 2014, 20, 6433 – 6443
knr ðT1 ! S0 Þ ¼
0
X hTa1 jHSOC jSm i2
The nonradiative decay rate could be calculated by Fermi’s Golden Rule in the limit of Condon approximation by using the generating function method [Eq. (9)]:[15]
www.chemeurj.org
in which hf and lf designate the high-frequency ligand skeletal modes (1700 > whf > 1000 cm1) and low-frequency modes
6435
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper (wlf 1000 cm1), respectively; mj is the reduced mass of the jth normal mode; DQj represents the equilibrium displacement along the jth normal mode coordinate; and DE00 is the zeropoint energy difference between the T1 excited state and the S0 ground state. In Equation (10), it is assumed that there is effectively only one dominant accepting mode (designated with the subscript “M”). This EGL expression works well if the nuclear tunneling effect is large and a weak temperature dependence on knr should be observed. Herein, we would also apply another approach to obtain the nonradiative decay rates by considering that 1) the MLCT transition should cause a change in electron density on both the metal ion and ligand and the assumption of no frequency change between the T1 and S0 states should be removed; and 2) as described below, the Huang–Rhys factor for the low-frequency modes, Slf, are greater than one, which means that the lf modes should be treated in the strong coupling limit. In this limit, knr has a stronger temperature dependence because the lf modes with large values of Slf become better acceptors as the temperature increases. As such, knr is calculated by the convolution method in which the lf modes are treated in the strong coupling limit and the hf modes in the weak coupling limit. In addition, as demonstrated below, there are large frequency changes for the lf modes between the two electronic states, and thus, frequency change for the lf modes also need to be taken into account [Eq. (11)]:[17]
Results and Discussion In this section, an account of the different photophysics of [Msalen] (M = Pd, Au, and Pt) is presented, followed by a proposed way to get an emissive PdII Schiff base complex, and then a detailed theoretical study on the effect of substitution at the salph ligand on the photophysics of M-salph-X (M = Pt or Pd; X = a substituent at the salph ligand).
M-salen In the literature, it is reported that 1) [Pd-salen] is nonemissive,[11a] 2) [Pt-salen] is moderately emissive (fem = 0.19),[10] and 3) [Au-salen] + is fluorescent with fem = 0.15.[11b] All the measurements were performed in acetonitrile at 298 K. The following refers to our theoretical investigations.
Pd-salen At the ground-state geometry, the major contributions of the first three lowest-lying singlet excited states are HOMO! L + 1 (S1), HOMO!LUMO (S2), and HOMO!L + 2 (S3) (see Table 1). The HOMO is composed of ligand p and Pd(dp) orbitals, whereas the LUMO and L + 1 are mainly of ligand p* in character and L + 1 also have some metal dp character (MO diagrams in the Supporting Information). L + 2 is mainly Pd(ds*)
1=2 2p knr ðT1 ! S0 Þ ¼ hT jH jS i2 2ph2 ðD21 þ P2 Þ h 1 SOC 0 ðDE00 nM hwM l1 mÞ2 Sn exp expðSM ÞnMM ! 2 2 2 2p h ðD1 þ P Þ X Sj hwsj l1 ¼ bj j2lf
Table 1. Calculated low-lying singlet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions the optimized ground-state (S0) geometry of [Pd-salen] in CH3CN.
M
m¼
2 hwTj 1 X s 1 bj hw j coth bj 2kB T 2 j2lf
X h wsj 2 hwTj Sj coth bj 2kB T j2lf " #2 ð1 b2j Þ hwTj 1X 2 2 s hP ¼ hw j coth 2kB T bj 2 j2lf
ð11Þ
wTj wSj
nM ¼
DE00 l1 m hw M
Because nM is the number of vibrational quanta of hwM , the quotient is corrected to the smaller integer. From Equation (11), it can be seen that the new temperature-dependent term (m) arising from frequency changes could increase (if m < 0) or decrease (if m > 0) the effective energy gap (DE00l1m), and hence, increase or decrease the number of quanta (nM) of the hf mode wM needed for the T1!S0 transition, and correspondingly, lead to a decrease or increase of knr and the effect would be enhanced as the temperature increases. Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
l [nm]
f
Major contributions[a]
lexp [nm] (e [103 mol1 dm3 cm1])[b]
Assignment[c]
S1 S2 S3
387 384 379
0.0569 0.1665 0.0124
H!L + 1 H!L H!L + 2
420 (2.9) 392 (5.3)
MLCT/IL (14) MLCT/IL (14) MC dd
[a] H = HOMO; L = LUMO. [b] Ref. [11a]. [c] The values in parentheses indicate the metal character of the occupied orbital in the transition. IL = intraligand.
2 D21 ¼ h
bj ¼
State
in character. Thus, S1 and S2 are mainly 1MLCT/IL in nature, whereas S3 is a MC 1dd excited state. These three singlet excited states are very close in energy and the 1dd excited state is only about 290–520 cm1 above the two 1MLCT excited states. Thus, excitation close to the first peak maximum (S2 state from the calculation) can easily populate the 1dd excited state, and hence, fast nonradiative decay prior to intersystem crossing (ISC) takes place. Thus, this complex is expected to be nonemissive. [Au-salen] + At the ground-state geometry, the major contributions of the first four lowest-lying singlet excited states are HOMO!LUMO (S1), H1!LUMO (S2), HOMO!L + 1 and L + 2 (S3 and S4) (see
6436
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 2. Calculated low-lying singlet and triplet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions at the optimized ground-state (S0) geometry of [Au-salen] + in CH3CN. State
l [nm]
f
Nature
S1 T6
437 389
0.0037 0
S2 S3
375 364
0.0037 0.1685
S4
362
0.0083
T7
347
0
H!L H2!L + 2 H3!L + 1 H1!L H!L + 1 H!L + 2 H!L + 2 H!L + 1 H-1!L + 1 H!L + 2
lexp [nm] (e [103 mol1 dm3 cm1])[a]
Table 3. Calculated low-lying singlet and triplet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions at the optimized ground-state (S0) geometry of [Pt-salen] in CH3CN. State
l [nm]
f
Nature
lexp [nm] (e103 mol1 dm3 cm1)[a]
Assignment[b]
S1 T3
420 403
0.0977 0
417 (5.9)
LMCT IL/MLCT (3)
T4
403
0
IL/MLCT (3)
S2
400
0.1178
H!L H1!L H4!L + 1 H3!L H1!L + 1 H4!L H3!L + 1 H!L + 1
MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL MLCT/IL
Assignment[b]
LMCT LMCT
375 (3.9)
(24) (10) (17) (34) (10) (17) (34) (24)
[a] Ref.[10]. [b] The values in the parentheses indicate the metal character of the occupied orbital in the transition.
IL/MLCT (2) IL/MLCT (4)
[a] Ref. [11b]. [b] The values in parentheses indicate the metal character of the occupied orbital in the transition.
Table 2). The HOMO is composed of ligand p and Au(dp) orbitals, similar to those of [Pd-salen], but with a smaller metal character in [Au-salen] + ( 3 %) than that in Pd-salen ( 14 %). On the other hand, the LUMO is mainly composed of Au(ds*) and is about 0.2 eV lower lying than the L + 1 and L + 2 orbitals. The latter two orbitals are predominantly of ligand p* character with an insignificant amount of Au(dp) character of 1 and 2 %, respectively. Thus, both S1 and S2 could be assigned as 1LMCT excited states, whereas S3 and S4 are 1IL/MLCT excited states. Upon excitation at l = 380 nm, the S3 excited state would be populated. Now S2 is only about 820 cm1 below the S3 excited state and more than 1700 cm1 above the closest triplet excited state (T6), so it would undergo internal conversion from S3 to S2 instead of ISC from S3 to T6. Because S2 is an LMCT excited state, fast nonradiative decay should take place. Based on calculations, [Au-salen] + should also be nonemissive in CH3CN. We thus repeated the experiment and found that [Au-salen] + was, at most, weakly emissive in CH3CN at room temperature (fem 1.6 104 with a time-resolved fluorescence decay lifetime, tf, of 2.7 ps; see experimental details in the Supporting Information). Pt-salen At the ground-state geometry, the first two lowest-lying singlet excited states are HOMO!LUMO (S1) and HOMO!L + 1 (S2) (Table 3). The HOMO, LUMO, and L + 1 orbitals have the same parentage as those of the PdII analogue, [Pd-salen]. The only difference is that the metal contributions are larger in the case of Pt-salen: the HOMO and L + 1 have 24 and 7 % Pt(dp) character, respectively. Therefore, S1 and S2 are both 1MLCT/IL excited states. The Pt(ds*) orbital is now L + 4 and lies about 2.0 eV above the LUMO. Thus, the MC 1dd excited state of Pt-salen is more than 9000 cm1 above the two 1MLCT/IL excited states, S1 and S2. Because the S1 and S2 excited states have similar osChem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
cillator strengths, both states may be populated upon excitation at l = 390 nm. For [Pt-salen], the closest-lying triplet excited states to S2 are T3 and T4, both are about 190–210 cm1 below the S2 excited state. Given the small singlet–triplet energy separation and the large metal contributions in both the S2 and T4 excited states, fast ISC could take place, even at the Franck–Condon geometry. After ISC from S2 to the T4 excited state, [Pt-salen] could undergo internal conversion to the T1 excited state, which is also predominantly of HOMO!LUMO character, that is, a 3MLCT excited state. The change in bond lengths between the T1 and S0 states does not exceed 0.1 and Sj is smaller than 0.5 (see the Supporting Information). Thus, structural distortion of T1 relative to the ground state in the case of [Pt-salen] is not significant and nonradiative decay is expected to be the slowest for [Pt-salen] among the three [M-salen] complexes presented in this section. Hence, phosphorescence could be observed with [Pt-salen]. The calculated value of kr is 5.03 104 s1, which is in good agreement with experimentally determined kr = fem/ tem = 5.43 104 s1;[10] this lends support that the method is adequate for d8 transition-metal complexes with Schiff base ligands. Is it possible to obtain emissive PdII Schiff base complexes? From the previous section, the dominant nonradiative decay path(s) comes from the MC dd and LMCT excited states at energies close to the MLCT/IL excited states; these low-lying MC dd and LMCT excited states would lead to large structural distortions and facile nonradiative decay. To alleviate this problem, the common strategy is to widen the MLCT–MC/LMCT energy gap. This could be done by introducing stronger sdonor atoms coordinated to the metal ion. However, since we want to keep the Schiff base O^N^N^O framework, we sought to increase this energy gap by introducing substituent(s) the Schiff base ligand. For [Pd-salen], the PdII ds* orbital is higher lying than the salen ligand p* orbital. On the other hand, for [Au-salen] + , the LUMO is ds* and is about 0.2 eV below the
6437
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper ligand p* orbitals. Thus, it may be more stringent for AuIII Schiff base complexes to push the ds* orbital sufficiently high in energy such that the LMCT excited state(s) would not be a viable deactivating path, if the O^N^N^O framework is to be kept. Given the higher emission quantum yield of the Pt-salph complex (fem = 0.23) relative to its [Pt-salen] counterpart (fem = 0.19),[10] we sought to modify the [Pd-salph] complex instead of [Pd-salen]. With the phenylene bridge replacing the ethylene bridge, conjugation across the ligand would be more extended, so that the ligand p* orbitals would become more low lying, but without significant effect on the character of the ds* orbital, since the energy of the latter would be less affected than that of the ligand p* orbitals. Thus, the ds* and p* orbitals using a phenylene bridge are expected to be pushed further apart in energy compared with the case using an ethylene bridge. To further increase the energy separation between the ligand p* orbital and the ds* orbital, while keeping the O^N^N^O framework for coordination to the metal ion, electron-withdrawing substituent(s) could be installed at the 4and/or 6-positions of the salicylidene moiety (i.e., meta to the phenoxide O). This is because there are significant electron densities at these positions in the ligand p* orbitals, but not in the ds* orbital (Figure 1). It is expected that such modifications
Figure 1. Lowest-lying unoccupied orbitals in the [M-salph] complexes (M = Pt or Pd). The subscripts o and i in the p* orbitals indicate that the combinations of the two salicylidene moieties are out-of-phase and inphase, respectively. The diagrams shown are those of [Pt-salph]; [Pd-salph] has analogous MOs.
would increase the MLCT–MC energy gap, thereby suppressing the nonradiative decay path through direct access to the dd excited states. We have chosen an ester, COOMe, as the electron-withdrawing group, resulting in the complex [Pd-salph-4E]. The ligand p* orbitals (LUMO and L + 1) are 1.63 and 1.17 eV, respectively, below the ds* orbital (L + 2). At its optimized S0 geometry, the MC 1dd excited state is more than 6500 cm1 above the lowest 1MLCT excited state [i.e., S1, and is derived from a HOMO!LUMO transition with the HOMO composed of ligand p and Pd(dp) fragment orbitals at the optimized ground-state geometry; see the Supporting Information for the MO surfaces]. From the results given in Table 4, it can be seen that the closest-lying triplet excited state that can have effective SOC with the S1 excited state is T3, which is about 310 cm1 below the S1 excited state. ISC should thus be rapid. Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
Table 4. Calculated low-lying singlet and triplet excited-state energies (l in nm), the associated oscillator strengths (f), and the nature of the transitions at the optimized ground-state (S0) geometry of [Pd-salph-4E] in CH3CN. State
l [nm]
f
Nature
Assignment[a]
T3
497
0
S1 T4
489 464
0.2432 0
S2
452
0.1856
H2!L H4!L + 1 H!L H!L + 2 H7!L + 2 H9!L + 2 H1!L H!L + 1
MLCT/IL MLCT/IL MLCT/IL dd dd dd MLCT/IL MLCT/IL
(12) (5) (14)
(2) (13)
[a] The values in parentheses indicate the metal character of the occupied orbital in the transition.
This T3 excited state can then decay to the T1 excited state through internal conversion. The optimized T1 excited state of [Pd-salph-4E] is mainly derived from the HOMO!LUMO transition with some contributions from the H1!LUMO transition, in which H1 is mainly composed of ligand p character (see the Supporting Information for the MO surfaces). Therefore, T1 is assigned to be a 3MLCT/IL excited state. At the optimized T1 geometry, the 3 dd excited state is more than 8200 cm1 above T1 (see the Supporting Information), and so it is, at first sight, envisioned that the 3dd excited state is not an effective deactivation channel for [Pd-salph-4E] phosphorescence. In addition, the differences in bond lengths between the optimized T1 and S0 states are less than 0.04 and individual Sj less than 0.4; thus, structural distortions should also not be large and we propose that [Pd-salph-4E] should display phosphorescence. Subsequently, we have prepared [Pd-salph-4E], which displays phosphorescence, albeit very weakly at room temperature (lem = 613 nm, fem = 1.7 104 at lex = 510 nm, t < 0.2 ms).[18] At 77 K in a glassy medium (MeOH/EtOH (1:4)), there is a substantial increase in emission lifetime: tem 143 ms. Such a large temperature dependence in tem is often attributed to the presence of thermally accessible 3dd excited state(s) close to the 3MLCT excited state. We thus performed a TDDFT optimization of the 3dd excited state and found that its potential energy minimum was only about 1530 cm1 above the T1 potential energy minimum. Thus, the activation energy for 3 MLCT!3dd is only about 2530 cm1. Comparable activation energies have also been reported for very weakly emissive cis[Ru(bpy)2L2]2 + (bpy = 2,2’-bipyridine; L is a nonchromophoric ligand).[6] Therefore, at 298 K, this 3dd excited state is thermally accessible and contributes to fast nonradiative decay (knr > 106 s1). At 77 K, the thermal energy is not sufficient to overcome the activation barrier and the population of 3dd excited state is no longer feasible; hence, the measured lifetime is that originating from the lowest 3MLCT excited state. Our value of knr calculated by using Equation (11) is about 1.07 104 s1, which is comparable to the experimentally estimated knr value of about 6.14 103 s1 at 77 K (see below).
6438
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Substituent effect on [Pt-salph-X]
stituted [Pt-salph]. From these results, it can be seen that [Ptsalph-4E] has the slowest radiative decay rate and the longest If the emitting triplet excited state of [Pt-salph-4E] has similar emission wavelengths of the three PtII Schiff base complexes. structural distortions to that of [Pd-salph-4E], the FCFs of the By the EGL, knr / exp(gDE/hwM), a smaller value of DE (or the two complexes would be similar and the estimated knr of [Ptlonger the emission wavelength, lmax) results in a faster nonrasalph-4E] would be larger than that of [Pd-salph-4E] by about diative decay rate. Based on this rationale, it is reasonable for an order of magnitude (because the SOC effect on knr would [Pt-salph-4E] to have a lower fem value than that of [Pt-salph]. become the determining factor: knr / HSOC2 / z2 ; z(Pt) = However, if the EGL is the sole reason, one would expect that [Pt-salph-4E] should have a knr value of more than 100-fold 4368 cm1 and z(Pd) = 1412 cm1). The 3dd excited state is not expected to be the deactivating path because, from previous faster than that of [Pt-salph] (see below). In addition, although theoretical analysis on [Pt-salen], there is no thermally accessi[Pt-salph-5tB] has a lower emission energy than that of [Ptble 3dd excited state close to the emitting 3MLCT excited state. salph], its knr is the slowest among the three PtII complexes Since the kr value of [Pt-salph] is experimentally found to be listed in Table 5. Moreover, within the applicability of the EGL, knr should display a weak temperature dependence. However, 6.71 104 s1,[10] we conceived that an emission quantum yield of approximately 0.45 for [Pt-salph-4E] could be feasible (asas seen from Table 5, the knr value of [Pt-salph] is more than ten times slower upon cooling from 298 to 77 K. These feasuming kr 6 104 s1 and knr 7 104 s1). Thus, [Pt-salph-4E] tures prompted us to undertake a detailed theoretical analysis was prepared and its photophysical properties examined. Its on the effect of substitution on the photophysical properties emission data together with those of [Pt-salph], [Pt-salph-5tB], of these PtII Schiff base complexes. and its PdII analogue, [Pd-salph-4E], are listed in Table 5. As depicted in Table 5, [Pt-salph-4E] is only weakly emissive In the following, we first present the results for calculated and its emission quantum yield is even less than that of unsubkr to justify that the electronic structure of their respective optimized T1 excited state is a good representation. Then, we list the calculated knr values within the context of the EGL Table 5. Photophysical parameters of [Pt-salph], [Pt-salph-5tB], [Pt-salph-4E], and [Pd-salph-4E] derived from [Eq. (10)] and the convoluted apexperimental data. proach [Eq. (11)], in which the tem knr tem[c] knr[e] lmax fem kr kr[d] latter takes into account the fre(298 K) (298 K) (298 K) (298 K) (298 K) (77 K) (77 K) (77 K) quency change between their [nm] [ms] [104 s1] [104 s1] [ms] [104 s1] [104 s1] respective optimized T1 and S0 [Pt-salph][a] 611 0.23 3.43 6.71 22.4 11.0 8.50 0.601 states and the low-frequency modes treated in the strong coupling limit.
[Pt-salph-5tB][a]
625
0.27
4.62
5.84
15.8
10.6
6.109
3.34
Radiative decay rates (kr)
[Pt-salph-4E][b]
661
0.033
[Pd-salph-4E][b]
613
1.7 104
1.6
< 0.2
2.06
0.085
60.4
> 500
3.70
143
2.42
0.085
24.6
0.614
[a] Measured in CH3CN at 298 K.[10] [b] Measured in dichloromethane because the ester-substituted complexes do not dissolve well in CH3CN (this work). [c] Recorded in MeOH/EtOH (1:4). [d] kr is estimated by assuming that there is a linear correlation between the calculated and experimental kr values; kr is calculated by using 1 Equation (7). [e] tem ¼ kr þ knr . Chem. Eur. J. 2014, 20, 6433 – 6443
www.chemeurj.org
6439
Table 6 depicts the DE00, ZFS, and kr,avg values at 298 and 77 K for [Pt-salph], [Pt-salph-5tB], [Ptsalph-4E], and [Pd-salph-4E]. The calculated zero-point energy differences compare well with the corresponding experimental emission maxima. Theoretical values of kr at 298 K compare satisfactorily with those found from experiments (kr, exp = fem/tem), except for [Pt-salph5tB], which is predicted to have a slower radiative decay rate than that of [Pt-salph-4E], contrary to the experimental findings. The total emission rate should be written as Equation (12):[19]
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper Table 6. Photophysical parameters of [Pt-salph], [Pt-salph-5tB], [Pt-salph-4E], and [Pdsalph-4E] obtained at their respective optimized T1 geometries. DE00 [cm1] ([nm]) [Pt-salph] [Pt-salph-5tB] [Pt-salph-4E] [Pd-salph-4E]
15 222 14 692 13 758 14 715
(657) (681) (727) (680)
kr,avg (298 K)[a] 104 s1]
ZFS (DEI,II ; DEI,III) [cm1]
kr,avg (77 K)[b] [104 s1]
5.35 (5.89) 3.23 (3.32) 4.86 (5.26) 0.0258 (0.0258)
13.7; 46.7 11.1; 41.53 0.6; 52.8
