TD-DFT study of the spin transition complex [Fe(pmea)(NCS)2]

Spectrochimica Acta Part A 94 (2012) 205–209

Contents lists available at SciVerse ScienceDirect

Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy journal homepage: www.elsevier.com/locate/saa

DFT/TD-DFT study of the spin transition complex [Fe(pmea)(NCS)2 ] Yuhui Qu ∗ School of Chemistry and Chemical Engineering, Shandong Institute of Light Industry, Shandong, Jinan 250353, PR China

a r t i c l e

i n f o

a b s t r a c t

Article history: Received 13 September 2011 Received in revised form 9 January 2012 Accepted 30 January 2012

The spin crossover (SCO) compound [Fe(pmea)(NCS)2 ] (where pmea symbolizes the ligand bis[(2pyridyl)methyl]-2-(2-pyridyl)ethylamine) has been studied by DFT/TD-DFT methods. Several density functionals and basis sets were used in the calculations to obtain optimized geometries of the compound in the low-(LS), intermediate-(IS) and high-spin (HS) states. The vibrational modes and IR spectra, spin splittings energies, excited states and UV/vis absorption spectra were calculated. From the TD-DFT calculations, it can be inferred that this complex may act as a reversible optical switch via the LIESST effect and its reverse process. © 2012 Published by Elsevier B.V.

Keywords: Spin crossover (SCO) compounds LS–HS splitting energies Vibrations and IR spectra Excited states and UV/vis absorption spectra Light-induced excited spin state trapping (LIESST)

1. Introduction Spin-crossover (SCO) complexes exhibit a change of the spin ground state as a function of external perturbations such as temperature, pressure, light irradiation, and pulsed magnetic fields. Hauser et al. [1–3] found that upon excitation with green light, crystalline and powder forms of [Fe(ptz)6 ](BF4 )2 (ptz = 1-propyltetrazole) make spin forbidden transitions from the low spin S = 0 state to a high spin S = 2 metastable state which becomes trapped below a critical temperature of about 50 K and later observed that a red light changes HS to LS [3,4]. This phenomenon is known as light-induced excited-state spin state trapping (LIESST). A mechanism for the reversible LIESST effect has been proposed as follows. For a d6 metal electronic occupation, the LS singlet ground state is first induced by photoexcitation to an excited singlet, then is converted to the metastable HS quintet state via the intermediate triplet. For the reverse LIESST effect, it consists of a first excitation from the lowest HS quintet state to an excited quintet state. Next, the system relaxes to an intermediate triplet state through an interconversion mechanism, and finally ends in the LS singlet ground state [5]. Spin configuration of the central Fe2+ valence shell which, in a cubic ligand field, can be described ↑ ↑↓ ↑↓ ↑↓ ↑ ↑ ↑ as d↑↓ xy dxz dyz dx2 −y2 dz 2 → dxy dxz dyz d 2 2 d 2 . Such ability of these x −y

z

complexes to control spin transitions, magnetic transition and color changes by means of optical excitation define this class of

∗ Corresponding author. E-mail address: [email protected] 1386-1425/$ – see front matter © 2012 Published by Elsevier B.V. doi:10.1016/j.saa.2012.01.078

compounds as especially promising materials for future applications in the fields of molecular electronics, data storage, nonlinear optics, and photomagnetism. The possibilities for the development of some novel properties such as photomagnetic behavior [6,7], superconductivity [8,9], spintronic property [10,11], and so on make spin-crossover (SCO) complexes an interesting field to probe theoretically. Present theoretical study and computational technique lead us to predict the structural, IR spectra; Excited states and UV/vis absorption spectra of this complex possible on the basis of density functional theory (DFT) calculations [12,13]. Recently, the synthesis, crystallographic structure, Mössbauer characterization, IR and Raman spectroscopy and DFT calculations of the complex were reported by Li and Brehm [14,15]. The goal of this paper is to characterize the new complex theoretically, and make a close contact to experimental data. We are interested in properties of the LS and HS, such as their structure and vibrational spectra and also focused on HS–LS energy splittings. We have estimated electronically excited singlet, triplet, and quintet states and UV/vis absorption spectra theoretically and predict the possibility of this complex acting as a candidate for reversible photo-switching via the LIESST effect and its reverse process.

2. Computational methods The DFT/TD-DFT calculations of the spin crossover compound [Fe(pmea)(NCS)2 ] have been performed using GAUSSIAN 09 package suite [16]. The complex geometries of the lowest energy LS and HS states were fully optimized with the restricted and unrestricted approach respectively. In the Fe(II) complex, the LS and

206

Y. Qu / Spectrochimica Acta Part A 94 (2012) 205–209

HS states correspond to the singlet and quintet, respectively. The key problem in DFT calculations is the choice of the exchangecorrelation functional. In this work, three different methods have been applied to this spin crossover complex: (i) Becke exchange and Lee, Yang, and Parr correlation functionals (BLYP) [17]. (ii) Becke’s nonlocal three-parameter hybrid functional combined with the Lee–Yang–Parr correlation functional with 20% exchange (B3LYP) [18]. (iii) Reparameterized hybrid functional, specially recommended for spin crossover systems, differing from B3LYP by the coefficient for the Hartree–Fock exchange (B3LYP*) [19,20]. To be more specific, in B3LYP, B3LYP* the exchange-correlation functional is given as: B3LYP(∗,∗∗ )

EXC

= (1 − a)EXLDA + aEXHF + bEXB88 + cECLYP + (1 − c)ECVWN (1)

EXHF

EXLDA

is the HF-type, ‘exact’ exchange, is the LDA where exchange energy, ECVWN is the LDA correlation energy in the Vosko–Wilk–Nusair [21] parameterization, EXB88 is the Becke 88 gradient-correction to the exchange energy [22] and ECLYP is the Lee–Yang–Parr gradient-corrected correlation energy [23]. The parameters are b = 0.72, c = 0.81, and a = 0.2 for B3LYP, a = 0.15 for B3LYP*. These functionals were combined with two basis sets. One is the all-electron 6−311G basis set implemented in GAUSSIAN 09 used for all atoms, which is in CPU time more expensive but also potentially more accurate. This basis set will be denoted as ‘A’ in the following. The other is the Los Alamos effective core potential plus double-␨ basis (LANL2DZ), which was employed for the Fe atom [24,25]. H, C, N and S are treated as all-electron atoms with the D95 basis set of Dunning/Huzinaga [26]. This basis will be called as ‘B’ below. The ‘model chemistries’ (functional/basis) were applied first of all for geometry optimizations in the S0 and Q1 ground states. In addition, the nature of the stationary points were determined in all cases by the analytical evaluation of the complete matrix of force constants and the associated harmonic vibrational frequencies at its corresponding level of theory. This enables us to verify that they correspond to true minima on the potential surface, and also to estimate the zero-point vibrational energy (ZPVE) correction. In any photoinduced process estimation of excitation energy is of extreme importance. In the case of the complex [Fe(pmea)(NCS)2 ] under investigation, time-dependent DFT (TDDFT) [27] was employed for excited states. The TD-DFT method to obtain excitation energies is based on the fact that the dynamic polarizability ˛(ω) of a system has poles at frequencies corresponding to its excitation energy. If one obtains the frequency dependent polarizability from TD-DFT calculations and substitutes in the sum-overstates relation [27] ˛(ω) =

 I

fI ωI2 − ω2

(2)

where ωI are the excitation energies and fI are corresponding oscillator strengths. The excitation energies ωI and oscillator strengths fI out of the respective ground states of a given multiplicity were calculated for singlet and quintet states in their optimized geometries following TD-DFT method described above. Finally, in view of a possible LIESST effect, the geometry of a triplet ground state T1 was determined with unrestricted DFT, and the triplet excitation energies and oscillator strengths were calculated by TD-DFT. 3. Results and discussion We optimized the structure of [Fe(pmea)(NCS)2 ] in full starting from the experimental structure without any symmetry’s constraint using restricted (for the singlet) and unrestricted (for the

Fig. 1. Optimized structures of the [Fe(pmea)(NCS)2 ] complex at B3LYP*/6−311G level of theory. Table 1 Obtained Fe-N distances (angstroms) of the [Fe(pmea)(NCS)2 ] complex at B3LYP*/6−311G level. Bond length

Fe Fe Fe Fe Fe Fe a

N1 N2 N3 N4 N5 N6

Singlet a

Triplet

Quintet

B3LYP*

Expt

B3LYP*

B3LYP*

Expta

1.989 1.968 2.034 2.098 1.933 1.948

1.985 1.973 2.006 2.042 1.938 1.971

1.984 2.195 2.005 2.341 2.059 1.984

2.210 2.225 2.210 2.379 2.002 2.056

2.192 2.212 2.189 2.266 2.060 2.125

Ref. [12].

quintet) B3LYP* method with 6−311G basis set. The optimized geometry of the complex is shown in Fig. 1 Fig. 1, and the Fe N bond lengths and N Fe N bond angles are listed in Tables 1 and 2. The calculated Fe N bond lengths fall in the range of 1.933–2.098 A˚ in the singlet state and 2.002–2.379 A˚ in the quintet state, being in good agreement with a general trend in the metal–ligand distances in low-spin and high-spin complexes. The computed singlet state lies 993 cm−1 below the quintet state. This result is consistent with the fact that only the singlet state is observed at low temperature [28]. We further optimized the complex using the B3LYP method with the same basis set. The B3LYP structures are similar to those of the B3LYP* calculations, but the singlet state is erroneously computed to lie 3424 cm−1 above the quintet state. Thus, the B3LYP* method, but not B3LYP, correctly predicts the energy ordering of the singlet and quintet states. Paulsen et al. [28] indicated that the B3LYP method tends to overestimate the stability of Table 2 Obtained N Fe N bond angles (degree) of [Fe(pmea)(NCS)2 ] complex at B3LYP*/6−311G level. Bond angle

N(1) N(1) N(1) N(1) N(2) N(2) N(2) N(3) N(3) N(3) N(5) a

Fe Fe Fe Fe Fe Fe Fe Fe Fe Fe Fe

N(2) N(4) N(5) N(6) N(3) N(4) N(5) N(4) N(5) N(6) N(6)

Ref. [12].

Singlet

Triplet

Quintet

B3LYP*

Expta

B3LYP*

B3LYP*

Expta

95.30 80.50 92.80 85.80 91.50 83.40 92.10 95.50 91.80 87.20 91.20

92.50 80.30 92.70

93.40 78.50 93.90 86.80 88.90 75.56 88.98 94.51 94.15 88.92 101.37

96.68 73.63 98.42 83.87 88.65 75.00 90.13 89.77 100.36 86.77 102.78

92.00 74.50 96.30

90.30 83.40 94.10 89.30 87.90

86.70 75.80 89.30 90.20 94.30

Y. Qu / Spectrochimica Acta Part A 94 (2012) 205–209

207

Table 3 Experimental and calculated mono-dentate ligand (mono), tetra-dentate ligand (tetra) and average (av.) Fe N bond lengths in A˚ of complex [Fe(pmea)(NCS)2 ] in the low-spin (LS) and high-spin (HS) states and average bond length change rHL . Method

LS

HS ωFeN a

Bond lengths

Experimentc B3LYP/A B3LYP/B BLYP/A BLYP/B B3LYP*/A B3LYP*/B a b c

ωHL a

0.188 0.190 0.190 0.183 0.195 0.185 0.193

519 548 506 545 520 539

EHL ◦

b

ωFeN a

Bond lengths

Mono

Tetra

Av.

Mono

Tetra

Av.

1.955 1.956 1.977 1.948 1.970 1.931 1.953

2.002 2.038 2.035 2.022 2.021 2.010 2.011

1.986 2.011 2.016 1.997 2.004 1.983 1.992

2.093 2.048 2.083 2.029 2.067 2.031 2.068

2.215 2.277 2.268 2.256 2.265 2.237 2.243

2.174 2.201 2.206 2.180 2.199 2.168 2.185

769 780 753 770 773 783

rHL

250 232 247 225 253 244

−3424 −2680 2742 2811 993 1265

Vibrational frequencies ωFeNX of the Fe NX (X = 5,6) symmetry stretch mode in cm−1 and their shift ωHL . 0 0 ZPE LS–HS splitting energies EHL for complex [Fe(pmea)(NCS)2 ] in cm−1 . EHL = EHL + EHL Ref. [12].

high-spin states in transition-metal complexes. Our result is fully consistent with theirs. We also show the X-ray structures of the experimental compounds in Tables 1 and 2, in comparison to fully geometry-optimized structures obtained for the SCO compounds [Fe(pmea)(NCS)2 ] using B3LYP*/A. We observe in Table 3 that with almost all model chemistries bond lengths involving Fe are rea˚ For example, the sonable, in many cases in the order of ≈0.02 A. average error in bond lengths is 0.003 A˚ for LS and 0.006 A˚ for HS at B3LYP*/A level. The Fe N bond lengths are somewhat less accurate for HS. The absolute errors in average Fe N bond lengths ˚ or 1.5% for B3LYP/B. The situation is better for are up to 0.03 A, BLYP/B, where the average error in Fe N bond lengths is reduced to 0.018 A˚ (0.9%). Generally we find for the geometries of these complexes that the accuracy in functionals was B3LYP < BLYP < B3LYP* for both LS and HS. Table 3 also contains Fe N bond length differences between HS and LS state, which are around 0.183–0.193 A˚ for this complex according to several functionals. The geometric distortion, i.e. rHL = rFeN (HS) − rFeN (LS), is qualitatively understood from the fact that in HS two electrons have been transferred from essentially non-bonding orbitals (t2g for the octahedral case) to Fe N antibonding ones (eg for the octahedral case). In summary, the calculated geometries are reasonable, given the fact that no solid state environment have been considered in the calculations. On the contrary to equilibrium geometries, the HS–LS energy splittings are very sensitive to the level of theory. Zero point energy 0 was calculated by taking the corrected HS–LS energy splitting EHL electronic energy difference between both spin states EHL , plus zero point corrections: 0 ZPE EHL = EHL + EHL

(3)

It can be seen from the Table 3 that the stability of HS vs. LS increase as the content of exact exchange in the exchangecorrelation functional increase. B3LYP functional with 20% exact 0 < 0 at room temperature, suggesting that exchange predict EHL HS is more stable than LS in the gas phase, which is different from experimental data. The functional B3LYP* with only 15% exchange, which was made to obtain correct HS–LS splittings for certain quasi-octahedral iron complexes perform better. The vibrational frequencies and intensities of the complex in their respective low-, intermediate- and high-spin states were calculated with B3LYP*/A methods (shown in Fig. 2 Fig. 2). The computed main IR spectra, intensity and their Vibrational Modes for complex LS and HS are listed in Table 4. Compared with B3LYP*/A case, the IR spectra obtained from other model chemistries look similar. The computed IR spectrum exhibit several distinct regions: (i) a high-energy region between about 3000 cm−1 containing C H stretching modes; (ii) an intermediate region ranging from about 2030 cm−1 to 800 cm−1 containing C NX C (X = 1, 2, 3), H C H bending modes and C N, C C, C S stretching modes;

Fig. 2. Calculated IR spectra of complex [Fe(pmea)(NCS)2 ], LS (low), IS (middle) and HS (up) obtained at B3LYP*/6−311G level of theory.

Table 4 Assignment of Vibrational Modes of complex [Fe(pmea)(NCS)2 ] in the low- and high spin states calculated with B3LYP*/6−311G. Frequency (int.)

Mode type

Low-spin

High-spin

2032(857) 2019(1128) 773(18) 424(4) 331(7) 317(1) 231(1) 212(4) 190(2) 177(10)

1987(888) 1971(1303) 253(30) 283(66) 213(1) 241(16) 178(1) 162(1) 125(5) 129(11)

N5 C, N6 C sym. str. N5 C, N6 C asym. str. FeN6 , FeN5 sym. str. FeN6 , FeN5 asym. str. FeN1 , FeN3 asym. str. FeN2 , FeN6 asym. str. N6 FeN5 bend FeN sym. str. FeNeq asym. str. FeNax sym. str.

208

Y. Qu / Spectrochimica Acta Part A 94 (2012) 205–209

(iii) a ‘fingerprint region’ below about 770 cm−1 which contains vibrations involving the heavier atoms Fe (Fe N stretching modes, the Fe N Fe bending modes) and other ligand vibrations (N C S bending mode). Of particular interest here is the symmetric ‘Fe NX (X = 5, 6) stretching mode’ in (quasi-) octahedral Fe N6 complex. For low-spin, we find ωFeNX = 773 cm−1 and a force constant of 0.5403 mdyn/A˚ at B3LYP*/A level. The frequency is similar to the one obtained with B3LYP and BLYP values (see Table 3). In the highstate, prominent change including clear frequency shifts relative to LS occur in the fingerprint region (see Table 4). In particular, all model chemistries correctly predict a significant softening of the Fe NX symmetrical stretching mode upon spin excitation LS → HS. The softening can easily be explained by the occupation of Fe N antibonding orbitals, in the high-spin case. We find ωFeNX = 253 cm−1 and a force constant of 0.5978 mdyn/A˚ in the HS state of the complex at B3LYP*/A level. The largest shift result in the Fe NX (X = 5, 6) symmetrical stretching mode that is blueshifted by 520 cm−1 at the B3LYP*/A level of theory. The intensity of these modes are predicted to be 1.7 times greater than that of the low-state. The frequency shifts between the levels of theory are consistently predicted in terms of magnitude and sign. The blueshift of frequencies is between about 519 and 548 wave numbers for the Fe NX symmetrical stretching mode (see Table 3). The Fe NX stretching vibration is not the most intense peak, because the dipole moment changes only weakly along this mode. We predict the C NX (X = 5, 6) asymmetrical stretching vibration mode in both LS ˚ and (ωCNX = 2019 cm−1 and a force constant of 31.3352 mdyn/A) ˚ HS (ωCNX = 1971 cm−1 and a force constant of 29.5232 mdyn/A) to be the most intense band. Therefore, this vibration is easy to see in the spectra of Fig. 2. For the LS complex, the C NX (X = 5, 6) symmetric and antisymmetric stretching vibrations are predicted at 2032/2019 cm−1 whereas the experimental values are 2117 and 2104 cm−1 [15]. For the HS complex, the corresponding pair of bands are predicted at 1987/1971 cm−1 whereas the experimental values are 2076 and 2065 cm−1 [15]. In conclusion, experimental and calculated wave numbers differ for these modes for the LS complex by about 85/85 (s) cm−1 and for the HS complex by about 89/94 (s) cm−1 . If error compensation is considered for B3LYP*/A, calculated wave numbers will increase in the right direction. Interestingly, we find for the HS complex a new, intense signal centered around 241 cm−1 (B3LYP*/A), since this signal is absent for LS. The signal consists of several subsignals which also involve Fe and ligand N atom motions. The observation of these vibrational modes in the laboratory might be used as a feature to distinguish between the two complexes. In any photoinduced process estimation of excitation energy is of extreme importance. In the case of the complex under investigation, the transition are of the d–d type, metal to ligand charge transfer (MLCT) type, ligand to metal charge transfer (LMCT) type, or ligand to ligand (LL) type (e.g. n–␲* and ␲–␲* transitions). By calculating the excitation energies through time-dependent density functional theory approach (TD-DFT), which is essentially a response theory method for LS, IS and HS isomers, excitation spectra for each species can be obtained. In Table 5, we show some of the excited singlet and excited quintet states of the LS and HS of complex [Fe(pmea)(NCS)2 ], when the BLYP/B method and the respective optimized structures were used. The most intense signal for LS takes place at 495 nm, which corresponds to the LMCT transition and by which a CT occurs from NCS to Fe, i.e. from the mono-dentate ligand. A less intense transition for LS occur at 460 nm from bonding-type to antibonding-type of ␲orbitals in the NCS moiety. For the HS, both MLCT states are shifted into the blue, with greatly reduced oscillator strengths. We note generally lower oscillator strengths for the HS. In addition, MLCT

Table 5 Selected electronic transitions for complex [Fe(pmea)(NCS)2 ], LS (upper half) and HS (lower half). ␭ (nm) Low-spin 619 589 587 555 495 460 430 379 342 335 High-spin 772 694 688 652 643 555 492 427 410 397 379 261

E (eV)

Osc. strength

Type

2.0021 2.1046 2.1137 2.2355 2.5069 2.6938 2.8838 3.2714 3.6244 3.7035

0.0141 0.0359 0.0120 0.0316 0.1212 0.0710 0.0113 0.0199 0.0138 0.0112

␲–␲* (pmea → pmea) ␲–␲* (pmea → pmea) MLCT(Fe → pmea) LMCT (NCS → Fe) LMCT (NCS → Fe) ␲–␲* (NCS → NCS) LMCT (NCS → Fe) LMCT (pmea → Fe) ␲–␲* (pmea → pmea) ␲–␲* (pmea → pmea)

1.6055 1.7855 1.8009 1.9009 1.9263 2.2338 2.5215 2.9034 3.0214 3.1233 3.2683 4.7431

0.0112 0.0110 0.0112 0.0162 0.0177 0.0105 0.0124 0.0139 0.0121 0.0281 0.0089 0.0249

LMCT(pmea → Fe) LMCT(pmea → Fe) d–d d–d d–d MLCT(Fe → pmea) MLCT(Fe → pmea) MLCT(Fe → pmea) MLCT(Fe → pmea) LMCT(pmea → Fe) LMCT(pmea → Fe) MLCT(Fe → pmea)

transitions play a role, for wavelengths up to and around 261 nm. The LMCT transition at 397 nm is the most intense transition for HS, which corresponds to a CT occurred from pmea ligand to Fe. A less intense CT Fe → pmea to the tetra-dentate ligand occurs at 261 nm. In addition to the visible transitions, there is a great number of dark states with vanishing oscillator strengths, both for LS and HS. In order to elucidate the possibility of LS → HS switching via the LIESST effect, we selected energy levels of singlet, triplet, and quintet multiplicity for the complex at B3LYP*/B and BLYP/B level (see Fig. 3 Fig. 3). Only those excited states are shown with an oscillator strength >0.01 from the respective spin ground state. The states are labeled Sn , Tn , and Qn , according to increasing energy. The electronic energy of the state concerned is plotted on the ordinate of Fig. 3, while the abscissa gives the averaged, optimized av for the respective ground states, S , T , Fe N bond length rFeN 0 1 and Q1 , out of which excited states have been calculated with TD-DFT. The quantum mechanical ground state is the lowest-lying singlet S0 with an average Fe N bond length of 1.992 A˚ for BLYP/B (see Table 3). Between the Sn states and the Qn states are several triplet states Tn , of which only the “non-dark” states are shown. In approximation, we consider the average Fe N distance only, and use the harmonic approximation obtained from normal mode analysis along this reaction coordinate. The triplet T1 is 0.6278 eV (5063 cm−1 ) above S0 , while Q1 is 0.3485 eV (2811 cm−1 ) above S0 at the BLYP/B level of theory (shown in Fig. 3). At the same av is 2.095 A ˚ for the T1 state, level, the average Fe N bond length rFeN ˚ and 2.185 A for Q1 (see also Fig. 3). This means that the r values for singlet to triplet are in between the r values for singlet to quintet transitions. The overall bond elongation of the triplet vs. singlet, is due to the t2g 5 eg 1 vs. t2g 6 eg 0 configuration, for an ideal octahedral complex. Therefore, we may speculate for LIESST in the complex that an optical transition transfers the system from S0 to one of the optical excited states Sn . The transition with the largest oscillator strength is the LMCT transition S0 → S2 , which is at 495 nm (2.5069 eV) at B3LYP*/B level (see solid arrow in Fig. 3). Then the excited Sn states will relax to lower-lying singlets by internal conversion (IC) and at the same time undergo transitions to the Tn by intersystem crossing (ISC). The ISC is accompanied by an elongation of the Fe N bonds. Similarly, IC will deexcite the triplet states and simultaneously ISC

Y. Qu / Spectrochimica Acta Part A 94 (2012) 205–209

209

both B3LYP*/B and BLYP/B for the complex qualifies it as possible candidates for reversible photo-switching. 4. Conclusions A theoretical study of the SCO complex [Fe(pmea)(NCS)2 ] has been carried out by using DFT/TD-DFT methods. We have made predictions of their geometries, IR and UV/vis spectra and LS–HS splitting energies. Calculated vibrational and electronic spectra could help to assign the experimental signals. From TD-DFT calculations, we predict that the complex might be functioning as optical switch. In the future, we wish to perform more study on spin–orbit coupling and from quantum dynamical point of view to give conclusive evidence. Acknowledgments This research has been supported by the Natural Science Foundation of Shandong Province, under grant no. Y2006B43. The authors also give grateful thanks for the support of the college outstanding young teacher domestic visiting scholar project of Shandong Province and Ministry of Education, PR China. References

Fig. 3. Energy levels of complex [Fe(pmea)(NCS)2 ] obtained with (1) BLYP/B and (2) B3LYP*/B. The ground states are the lowest-lying singlet state S0 , triplet state T1 and quintet state Q1 . Excited states are labeled with Sn , Tn , and Qn , respectively. The arrows indicate the most (solid) intense transitions for the LIESST effect, and its reverse process.

from Tn to Qn will occur, which is characterized by further Fe N bond stretching. Finally, the system ends up in the Q1 state, where it is trapped for long times before ISC to S0 occurs. The latter process must be slow as it requires the spin flip of two electrons rather than one. It can also be suggested from the energy diagrams in Fig. 3 that a reverse switching [5] should be possible for the complex. Firstly, initiated by an optical transition from Q1 to Q10 state (see solid arrow in Fig. 3). The quintet absorbs at different wavelengths (397 nm, 3.1233 eV) than the singlet and requires larger intensities because the oscillator strengths are typically at least an order of magnitude smaller (see Table 5). The results obtained using

[1] S. Decurtins, P. Gütlich, C.P. Köhler, H. Spiering, A. Hauser, Chem. Phys. Lett. 105 (1984) 1. [2] S. Decurtins, P. Gütlich, K.M. Hasselbach, A. Hauser, H. Spiering, Inorg. Chem. 24 (1985) 2174. [3] A. Hauser, J. Chem. Phys. 94 (1991) 2741. [4] A. Hauser, Chem. Phys. Lett. 124 (1986) 543. [5] J.J. McGarvey, I. Lawthers, J. Chem. Soc. Chem. Commun. 906 (1982). [6] K. Matsuda, M. Matsuo, M. Irie, J. Org. Chem. 66 (2001) 8799. [7] N. Tanifuji, M. Irie, K. Matsuda, J. Am. Chem. Soc. 127 (2005) 13344. [8] H. Kobayashi, A. Kobayashi, P. Cassoux, Chem. Soc. Res. 29 (2000) 325. [9] S. Uji, H. Shinagawa, T. Terashima, T. Yakabe, Y. Terai, M. Tokumoto, A. Kobayashi, H. Tanaka, H. Kobayashi, Nature (Lond.) 410 (2001) 908. [10] G.A. Prinz, Science 282 (1998) 1660. [11] E.G. Emberly, G. Kirczenow, Chem. Phys. 281 (2002) 311. [12] M. Dierksen, S. Grimme, J. Phys. Chem. A 108 (2004) 10225. [13] F. Santoro, R. Improta, A. Lami, J. Bloino, V. Barone, J. Chem. Phys. 126 (2007) 084509. [14] B. Li, R.J. Wei, J. Tao, R.B. Huang, L.S. Zheng, Inorg. Chem. 49 (2010) 745–751. [15] G. Brehm, M. Reiher, B. Guennic, M. Leibold, S. Schindler, F.W. Heinemann, S. Schneider, J. Raman Spectrosc. 37 (2006) 108–122. [16] M.J. Frisch, et al., Gaussian 09, Revision A.1, Gaussian, Inc., Wallingford, CT, 2009. [17] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785. [18] A.D. Becke, J. Chem. Phys. 98 (1993) 5648. [19] O. Salomon, M. Reiher, B.A. Hess, J. Chem. Phys. 117 (2002) 4729. [20] M. Reiher, O. Salomon, B.A. Hess, Theor. Chem. Acc. 107 (2001) 48. [21] S.H. Vosko, L. Wilk, M. Nusair, Can. J. Phys. 58 (1980) 1200. [22] A.D. Becke, Phys. Rev. A 38 (1988) 3098. [23] C. Lee, W. Wang, R.G. Parr, Phys. Rev. B 37 (1988) 785. [24] T.H. Dunning Jr., P.J. Hay, in: H.F. Schaefer (Ed.), Modern Theoretical Chemistry, vol. 3, Plenum, New York, 1976, p. 1. [25] P.J. Hay, W.R. Wadt, J. Chem. Phys. 82 (1985) 270. [26] T.H. Dunning Jr., P.J. Hay, Modern Theoretical Chemistry, vol. 3, Plenum, New York, 1976. [27] M.E. Casida, C. Jamorski, K.C. Casida, D.R. Salahub, J. Chem. Phys. 108 (1998) 4439. [28] H. Paulsen, L. Duelund, H. Winkler, H. Toftlund, A.X. Trautwein, Inorg. Chem. 40 (2001) 2201.