Dalton Transactions View Article Online
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
PAPER
Cite this: Dalton Trans., 2014, 43, 10040
View Journal | View Issue
Tuning of the ground state in electron doped anthracene Quynh T. N. Phan,a Satoshi Heguri,*b Yoichi Tanabe,a Hidekazu Shimotani,a Takehito Nakano,c Yasuo Nozuec and Katsumi Tanigaki*a,b High quality bulk samples of anthracene (AN) doped with potassium (K) in 1 : 1 and 2 : 1 stoichiometries were successfully prepared by a method involving a room temperature solid-state mechanical diffusion process prior to intercalation reactions during heat treatment, and their physical properties were studied using both magnetic and optical measurements. The transfer of almost one electron from K to AN in K1(AN) was confirmed by SQUID and ESR measurements. A pronounced magnetic hump centered at
Received 8th January 2014, Accepted 10th April 2014
150 K associated with antiferromagnetic interactions was observed, which can most likely be interpreted in terms of on-site Coulomb repulsions of the Mott insulating states. Optical spectra of K1(AN) clearly
DOI: 10.1039/c4dt00071d
showed the insulating states, as well as the electron occupation of the LUMO-derived band of AN. Our
www.rsc.org/dalton
results demonstrated tuning of the ground state of a typical bulk hydrocarbon by alkali metal intercalation.
1.
Introduction
A semiconducting state showing very high conductivity, 1 × 10−3 S cm−1, was observed in 1954 for perylene doped with halogens;1 this inspired research searching for possible metallic states composed only of simple polycyclic aromatic hydrocarbons. Unfortunately, true metallic properties were not observed in perylene. Afterwards, intrinsic metallic states showing metallic conduction bands were realized in donor– acceptor type organic charge transfer complexes, such as TTF-TCNQ,2 as well as in a single-component organic molecule.3 A recent report described a metallic state of polyaniline in a system of conducting polymers showing conductivity exceeding 1000 S cm−1, as well as electrical resistivity that decreases monotonically with temperature,4 which was an intriguing topic in this research area. However, the debate still continues as to whether true metallic states can be created in simple polycyclic aromatic hydrocarbons ( polyacenes), such as anthracene (AN), tetracene (TN) and pentacene (PN), where carriers are introduced by incorporating other elements, such as iodine, alkali metals, and so forth.
a
Department of Physics, Graduate School of Science, Tohoku University, 6-3 Aoba, Aramaki, Aoba, Sendai, Japan b WPI-Advanced Institute for Materials Research (WPI-AIMR), Electronic Materials Physics, Tohoku University, 1-1 2 Katahira, Aoba, Sendai, Japan. E-mail: [email protected], [email protected]; Fax: +81-22-217-6331; Tel: +81-22-217-6173 c Department of Physics, Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka, Japan
10040 | Dalton Trans., 2014, 43, 10040–10045
Looking back through the history of this research, many experimental approaches have been taken with these simple polyacenes. Hole doped simple polycyclic aromatic hydrocarbons, such as TN5 and PN,6 were studied by modifying the density of states at the Fermi level using chemical doping, but a metallic state has yet to be observed in the doped phases. The crystal structure and electronic properties of iodine doped PN,6 with single crystal, powder and thin film morphologies, were reported over a broad range of iodine concentrations (x) of 0.2 ≤ x ≤ 1.0. A thermally accessible excited triplet magnetic state ascribed to the presence of a dimer (PN+)2 was suggested by the results of experiments on powder samples containing intercalated iodine. From the point of view of electric transport, organic thin films have been widely investigated for PN,7 fullerene8–10 and phthalocyanine.11 Very recently, a metal– insulator transition in alkali metal doped PN thin films was discussed,7 and K1(PN) was reported to be a Mott insulator based on the electrical transport measurements, as well as theoretical calculations. In general, considering that the transfer integral, t, (below 0.5 eV) is much smaller than the on-site Coulomb repulsion energy (above 1 eV) due to the large molecular separation in van der Waals crystals of polyaromatic hydrocarbons, it is thought that a metallic state is not realized, but instead a Mott insulating state is generated. However, this controversy has not yet been settled, and the debate still continues in this intriguing area of condensed matter physics and chemistry. From a synthetic point of view, the quality of doped polyacenes is not high enough because various metastable phases with the equivalent ground state energies coexist, and this is known to be a serious problem for preparing a high quality
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
Dalton Transactions
Paper
single-phase sample by employing conventional synthetic techniques. In this paper, we report the synthesis of K-intercalated AN, the smallest molecule among polycyclic aromatic hydrocarbons, and investigate its physical properties. In order to improve the quality of the doped samples, a slightly different synthetic route, involving a solid phase diffusion process with careful mechanical grinding at room temperature prior to pelletization and annealing, is employed to promote better intercalation reactions. We show that K1(AN) prepared by this modified process is most likely to be in a Mott insulating state, and this is confirmed by both magnetic and optical measurements.
2. Experimental section A mixture of a K ingot and AN with the target stoichiometry was pelletized and ground inside a glove-box (O2: 0.1 ppm and H2O: 0.5 ppm). The process was repeated twice, and a homogeneous black powder was obtained. The resulting material was sealed in a tube filled with helium (He) after evacuation, and annealed at 413 K for two weeks. Powder X-ray diffraction (XRD) of the pristine and intercalated samples was measured at the BL02B2 beam port at SPring-8. XRD samples were sealed in soda-glass capillaries under an argon atmosphere. Raman spectroscopy was performed using a RENISHAW inVia Reflex spectrometer equipped with a single monochromator and a CCD detector at an excitation energy of 785 nm. The same capillaries were used for both XRD measurements and Raman measurements. Diffuse reflectance spectra were measured for samples sealed in quartz tubes at room temperature using a UV-vis-NIR spectrometer and an FTIR spectrometer. The magnetic susceptibilities were measured using a Quantum design MPMS-XL7 superconducting quantum interference device (SQUID) magnetometer at temperatures between 2 and 300 K under an applied magnetic field of up to 7 T. Electron spin resonance (ESR) measurements at temperatures between 4 and 300 K were performed using a Bruker E500 X-band ESR spectrometer with an Oxford ESR900 He flow cryostat. The samples for SQUID and ESR measurements were prepared in He gas in handmade partitioned quartz tubes and commercial ESR quartz tubes, respectively. Chromium powder was used as a standard g-factor reference, and the spin susceptibility was calibrated using a CuSO4·5H2O single crystal.
3. Results and discussion 3.1.
Preparation of K doped AN
We carefully prepared both K1(AN) and K2(AN) as nominal compositions by mechanical mixing at room temperature, followed by pelletization and annealing at higher temperatures that were still below the melting point of AN, as described in the Experimental section. It is difficult to prepare Kx(AN) with x exceeding 2, even when using our synthesis method. The syntheses of the target K1(AN) and K2(AN) phases were
This journal is © The Royal Society of Chemistry 2014
Fig. 1 XRD profiles of pristine AN (blue), K1(AN) (red), and K2(AN) (green).
repeated, and the high quality samples were subjected to XRD. The diffraction peaks of pristine AN could be indexed well using the crystal structure in ref. 12. Several sharp new diffraction peaks appeared, leaving a small trace of pristine AN, for the sample prepared from a 1 : 1 K to AN stoichiometry, as shown in Fig. 1. A low angle diffraction peak with a larger d-value than that of AN appeared in both the K1(AN) and the K2(AN) XRD profiles. The K1(AN) unit cell was slightly expanded compared with that of pristine AN, due to the K intercalation. When the K to AN stoichiometry of 2 : 1 was employed, the diffraction peaks of pristine AN disappeared almost entirely. In the diffraction profile of K2(AN), not only were new peaks at different positions from those of K1(AN) obtained, but other small broad peaks also appeared. The number of observed diffraction peaks was less than that of K1(AN). The appearance of broad peaks and the decrease in the number of peaks for K2(AN) may indicate that the crystal structure starts to decompose when the stoichiometry exceeds one to one. For K2(AN), new peaks different from those of K1(AN) were present, although some peaks still remained in the same positions as those of K1(AN). These diffraction profiles indicate that K1(AN) and K2(AN) are the most stable phases in the K-intercalated AN system. It is noted that when the temperature dependence for K1(AN) was studied, no distinct structural transition was detected above 90 K. 3.2.
Optical measurements
Optical spectra of pristine AN and K1(AN) are shown in Fig. 2(a). For pristine AN, absorptions due to the exciton bands were observed above 3 eV. Below this energy, there was no absorption except for a sharp peak at 0.38 eV, which was due to the C–H stretching vibrations. For K1(AN), since the regular reflection signals were dominant in the diffuse reflectivity above 0.8 eV, the reflectance spectrum was obtained by utilizing a transformation described in the literature.13 On the other hand, below 0.8 eV, the absorption spectrum was obtained using the Kubelka–Munk function (KMF) because the transmitted signals were dominant. In the reflectance spectrum,
Dalton Trans., 2014, 43, 10040–10045 | 10041
View Article Online
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
Paper
Dalton Transactions
Fig. 2 Optical absorption and reflectance spectra of (a) pristine AN and K1(AN), and (b) free solvated AN mono- and dianions. The data for the anions are from ref. 14.
several peaks were clearly observed, even below 3 eV, in contrast to pristine AN. For comparison, the absorption spectra of free solvated AN mono- and dianions14 are plotted in Fig. 2(b). For the free solvated anions, three sets of absorptions were observed at approximately 1.1–1.5 eV, 2.3–2.8 eV and 3.3 eV. They have been assigned to electron excitations from the LUMO to the LUMO+1, from the HOMO to the LUMO+1, and from the HOMO−1 to the LUMO, respectively. Note that the reflectance peaks of K1(AN) were observed at very similar energies to those of the free solvated anions, and thus, the same assignments are applicable. The appearance of the LUMO to LUMO+1 reflection band at 1–2 eV for K1(AN) indicates that the LUMO band is occupied by electrons in the ground state. This is evidence of charge transfer from K to AN in K1(AN), namely, successful electron doping. An absorption tail is observed in the low energy region in Fig. 2(a), and there appears to be no contribution from the Drude term for free carriers. In the case of K2(AN), the Drude term was not obtained in the low energy region. Hence, we can conclude that K1(AN) and K2(AN) are insulators. 3.3.
Raman spectra of AN, K1(AN), and K2(AN)
The room temperature Raman spectra of pristine AN, K1(AN) and K2(AN) are shown in Fig. 3. The peak widths became wider with increasing K intercalation. A similar increase in the peak width was reported for alkali doped fullerides, and was attributed to electron–phonon interactions.15 The AN molecule belongs to the D2h point group, and a total of thirty-three Raman-active vibration modes can be expected based on the symmetry considerations.16 However, only Raman-active modes with strong intensities associated with the C–C stretching vibrations were observed in our measurements for pristine AN. The Raman shift of AN observed in our present studies is consistent with that in other reported experiments.16 All observed vibration modes for both pristine AN and Kx(AN) were assigned and are listed in Table 1, together with those obtained in other experimental reports.16,17 Almost all the
10042 | Dalton Trans., 2014, 43, 10040–10045
Fig. 3 Raman spectra of AN (blue), K1(AN) (red), and K2(AN) (green) with λ0 = 785 nm.
peaks of Kx(AN) (x = 1, 2) were shifted to lower wavenumbers than those of pristine AN. This phonon softening is associated with the electron transfer from K to AN molecules, and accordingly the anti-bonding orbitals of AN are occupied, which weakens the vibrational force constant. A similar effect caused by charge transfer was reported for graphite intercalation compounds.18 As shown in Fig. 3 the peak at 1236 cm−1 in the spectrum of K1(AN) almost disappeared in the case of K2(AN) and the peak at 609 cm−1 for K1(AN) shifted to 596 cm−1 for K2(AN). However, low intensity peaks at 1236 cm−1 and 609 cm−1 were observed for K2(AN), and furthermore, a small trace of the peak at 596 cm−1 was also observed for K1(AN). These facts suggest that a small amount of the K2(AN) phase was present in K1(AN), and vice versa. From the intensities of the Raman spectra, the amount of K2(AN) contaminant in the K1(AN) sample was estimated to be 8%. Similarly, the amount of the contaminating K1(AN) phase in K2(AN) was estimated to be approximately 9%.
3.4.
Magnetic properties of K doped AN
Fig. 4 shows the temperature dependences of the magnetic susceptibilities (χ) of the KxAN phases (x = 0, 1, and 2). Pristine AN showed a temperature independent diamagnetism of −124 × 10−6 emu mol−1, and Curie behaviour was not observed at low temperatures. The diamagnetic value is consistent with the one observed by NMR.19 When K was intercalated, a large paramagnetic moment was detected in K1(AN), compared to pristine AN. However, when a larger amount of elemental K was intercalated, the paramagnetic susceptibility decreased again, as observed for K2(AN). This clearly indicates that electron transfer from K to AN occurs, and that the LUMO-derived band is gradually filled with electrons in proportion to the amount of K. The very small magnetic susceptibility of K2(AN) can be understood by picturing a band insulator, in which the LUMO-derived conduction band is completely filled by electrons. A weak increase in the temperature dependent χ observed at low temperatures for K2(AN) can be understood using the Curie term, and the spin number was estimated to
This journal is © The Royal Society of Chemistry 2014
View Article Online
Dalton Transactions Table 1
Paper
Vibration modes and Raman shifts of AN, K1(AN), and K2(AN), together with the theoretical and experimental results from ref. 15 and 16
K1(AN)
AN Vibration mode
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
ag
b3g
ω δ ω ω δ ω ω ω ω δ
Ref. 16
This work
Δν
Δν
1557 1482 1403 1260 1163 1007 753 622 395 1187
1561 1484 1406 1264 1168 1012 758 526 — 1191
K2(AN)
Ref. 17
This work
Intensity
Δν
Δν
3186 543 7199 989 967 883 6035 669 — 1087
1543 1466 1360 1230 1152 1024 738 608 390 1180
1534 1478 1352 1236 1152 1020 738 609 390 —
Ref. 17
This work
Intensity
Δν
Δν
Intensity
812 200 3056 1971 121 336 275 2621 144 —
1540 1470 1350 — 1150 1020 723 599 390 1175
1536 1467 1351 — 1143 1019 726 595 390 —
800 251 10 350 — 250 300 298 1246 306
the previous reports. Schematic models of the ground states are shown in Fig. 5. In principle, the thermally activated triplet state results from the formation of dimers of a band insulator in a localized electron system with a singlet ground state, while the Mott insulating state is in principle categorized as an itinerant electron system, in which electrons are localized due to the on-site Coulomb repulsion energy, U. First, we tentatively employed a similar singlet–triplet model by excluding the Pauli term, and found that eqn (1) gave a fairly good fit to the data.
Fig. 4 Temperature dependences of the magnetic susceptibilities of pristine AN (blue), K1(AN) (red), and K2(AN) (green).
be 0.05 spin per molecule. This can reasonably be attributed to the defects in the lattice created during the K intercalation processes. The magnetic susceptibilities of K1(AN) and K2(AN) are consistent with the XRD data discussed earlier. It should be noted that the increase in χ observed for K1(AN) at low temperatures is much larger than that for K2(AN) and does not follow the general Curie-type 1/T temperature dependence. This result will be discussed in the next paragraph. 3.5.
Detailed discussion of the magnetic properties of K1(AN)
The most striking and intriguing phenomenon in the magnetic susceptibility of K1(AN) is the fact that a broad hump centered at 150 K was observed, which was most evident between 50 K and room temperature (Fig. 4). It is also very important that K1(AN) exhibited a large spin number of more than 0.5 spins per AN, and therefore this cannot be ascribed to any defect spins. Brinkmann et al. previously reported a similar magnetic behaviour for iodine-doped I1(PN),6 and interpreted the result in terms of a thermally accessible singlet–triplet equilibrium state. This result should be discussed and compared to the observations for K1(PN) reported by Craciun et al.,7 where the observed electronic states are interpreted based on the Mott insulating state. The ground states of both electron (K doped) and hole (I doped) PNs are insulating, but their ground states are different according to
This journal is © The Royal Society of Chemistry 2014
χ¼
C1 4C2 þ T T ð3 þ eT 0 =T Þ
ð1Þ
In eqn (1), C1 is the Curie constant arising from the defects, C2 is related to the number of localized spins, and T0 (= Δ/kB) is the characteristic temperature associated with a thermally accessible singlet–triplet energy gap. The parameters in eqn (1) were optimized by the least squares method: C1 = 0.0199 emu mol−1 K, C2 = 0.246 emu mol−1 K, and T0 = 240 K. The number of spins per AN molecule (n: estimated from C2) was approximately 0.66. This model supposes that a biradical electronic state is formed as the ground state (G0), and that the next thermally accessible triplet state (T1) is located slightly above G0 on a kBT0 energy scale, as shown in Fig. 5a. In order to clarify the ground state of K1(AN), we performed ESR spectroscopy. In the case of the singlet–triplet model, the thermally accessible triplet species should be detected by ESR, as its spectrum shape (S = 1) is very different from that of a
Fig. 5 Schematic models of (a) thermal activation and (b) the Mott insulator in K1(AN).
Dalton Trans., 2014, 43, 10040–10045 | 10043
View Article Online
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
Paper
doublet (S = 1/2) due to the D and E zero-field splitting parameters in the traceless S·D·S spin–spin Hamiltonian.20 This has been very widely studied in biradical systems, as well as in photo-excited triplet states. Our experimental data shown in Fig. 6, however, displayed no such typical triplet spectral features, and instead a symmetrical signal that could be fitted using a single Lorentzian function was observed above 4 K. It should be noted that, strictly speaking, a small asymmetric feature was observed at very low temperatures, but the spectrum was described well by introducing a small anisotropic g tensor in the analysis, by integrating over all azimuth angles in g-space in the fashion of a general powder ESR simulation. A Pauli paramagnetic contribution associated with conduction electrons was also not observed. When these experimental facts are taken into account, the most reasonable explanation of the K1(AN) ground state is a Mott insulating state. This interpretation is consistent with the optical measurements described earlier. The singlet–triplet model, therefore, should be ruled out as an explanation of the electronic state of K1(AN). Consequently, according to these discussions, we temporarily employed an Ising model for an antiferromagnetic spin chain, described by Bonner and Fisher21 within the framework of a finite chain length, for the next stage of interpretation, which will be discussed in the next paragraph. In this model, the magnetic susceptibility decreases due to antiferromagnetic interactions between adjacent spins at temperatures below J/kB, and at the same time the increase in magnetic susceptibility is steeper than conventional Curie-type temperature dependence, which can be explained when the finite chain length is taken into consideration. It is noted here that
Fig. 6 First derivatives of the ESR signals of K1(AN) at 4 K, 40 K, and 280 K, and the curves simulated using Lorentzian functions. The blue empty circles are the first derivatives of the ESR signals, the red lines are the curves fitted using Lorentzian functions, and the green lines are the differences between the experimental data and the fitting results.
10044 | Dalton Trans., 2014, 43, 10040–10045
Dalton Transactions
Fig. 7
Schematic model of the finite spin chain in K1(AN).
although the influence of other phases may generally be considered in order to understand the experimental phenomena, other phases are unlikely to be involved in the present case because an ESR spectrum with only a single Lorentzian lineshape was observed, as shown in Fig. 6. It may be reasonable to imagine that the prepared intercalated phase does not have long-range ordering due to the low dimensional instability. We therefore employed a modified Bonner–Fisher model as an alternative choice by taking into account the different finite numbers of spin chains consisting of even (N) and odd (M) numbers of spins with the equivalent interaction energy ( J), whereas previously the dispersion of the chain length was neglected. The magnetic susceptibility is expressed by " # C1 J=kB T 1 ð tanhð J=2kB TÞÞN χ¼ e T 1 þ ð tanhð J=2kB TÞÞN " # C2 J=kB T 1 ð tanhð J=2kB TÞÞM þ e T 1 þ ð tanhð J=2kB TÞÞM
ð2Þ
where C1 and C2 correspond to the Curie constants of the chains with even and odd numbers of spins, respectively. A schematic model of a finite spin chain in K1(AN) is shown in Fig. 7. In this model, the antiferromagnetic interactions between two adjacent spins cause a broad hump at intermediate temperatures, while an odd number of spins increases the magnetic susceptibility at low temperatures, as shown in Fig. 8. The parameters in eqn (2) were optimized using the least squares method, and were evaluated as N = 12, M = 11, C1 = 0.168 emu mol−1 K, C2 = 0.183 emu mol−1 K, and J = 12 meV ( J/kB = 140 K). The number of spins per AN molecule (n, esti-
Fig. 8 Temperature dependence of the static magnetic susceptibility of K1(AN) and the fitting using the Bonner–Fisher model.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Dalton Transactions
mated as the summation of C1 and C2) was sufficiently large (approximately 0.93), and importantly, the estimated spin numbers were consistent, irrespective of the choice of model.
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
4.
Conclusion
We successfully synthesized K-intercalated AN, using an improved solid-state mechanical mixing and thermal diffusion methodology, and observed its optical, magnetic and vibrational properties. All the properties of K1(AN) can be well described based on a Mott insulating state, while K2(AN) is a band insulating state. Both magnetic and optical experiments provided evidence for the Mott insulating state of K1(AN), with the transfer of one electron from K to AN. This conclusion seems reasonable when the on-site Coulomb repulsion energy, U, (1 eV) and the band width energy, W, (0.6 eV) are considered in the itinerant band limit. We did not observe thermally accessible triplet states via biradical formation in the electron localization limit. Our results demonstrate tuning of the ground state in a typical molecular organic semiconductor, and suggest that a unique magnetically ordered state with a finite number of spins can be created in K-intercalated aromatic hydrocarbons. In the future, it would be interesting to see whether a metallic system can be produced by controlling molecular alignment at various degrees of charge transfer under elevated pressure.
Acknowledgements We acknowledge Prof. K. Akiyama for his fruitful discussions. This work was supported by the Scientific Research Fund (24651128, 24684023 and 25610084) and the World Premier International Research Center Initiative (WPI) from the MEXT of Japan and the Tohoku University GCOE program. This work was also supported by the Joint Studies Program (2011–2012) of the Institute for Molecular Science.
References 1 H. Akamatsu, H. Inokuchi and Y. Matsunaga, Nature, 1954, 173, 168–169.
This journal is © The Royal Society of Chemistry 2014
Paper
2 P. W. Anderson, P. A. Lee and M. Saitoh, Solid State Commun., 1973, 13(5), 595–598. 3 H. Tanaka, Y. Okano, H. Kobayashi, W. Suzuki and A. Kobayashi, Science, 2001, 291, 285–287. 4 K. Lee, S. Cho, S. H. Park, A. J. Heeger, C. W. Lee and S. H. Lee, Nature, 2006, 441, 65–68. 5 P. A. Stunzhas, V. A. Benderskii and L. A. Blyumenfel’d, J. Struct. Chem., 1968, 8, 393–398. 6 M. Brinkmann, V. S. Videva, A. Bieber, J. J. Andre, P. Turek, L. Zuppiroli and R. Humphry-Baker, J. Phys. Chem. A, 2004, 108, 8170–8179. 7 M. F. Craciun, G. Giovannetti, S. Rogge, G. Brocks, A. F. Morpurgo and J. van den Brink, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 79(12), 125116–125123. 8 T. T. M. Palstra, R. C. Haddon, A. F. Hebard and J. Zaanen, Phys. Rev. Lett., 1992, 68(7), 1054–1057. 9 R. C. Haddon, G. P. Kochanski, A. F. Hebard, A. T. Fiory and R. C. Morris, Science, 1992, 258, 1636–1638. 10 S. Heguri and M. Kobayashi, Solid State Commun., 2010, 150, 1489–1492. 11 M. F. Craciun, S. Rogge, M. J. den Boer, S. Margadonna, K. Prassides, Y. Iwasa and A. F. Morpurgo, Adv. Mater., 2006, 18, 320–324. 12 A. Charlesby, G. Finch and H. Willman, Proc. Phys. Soc., 1939, 51, 479–528. 13 T. Kodaira, Y. Nozue, S. Ohwashi, T. Goto and O. Terasaki, Phys. Rev. B: Condens. Matter, 1993, 48, 12245–12252. 14 K. H. J. Buschow and G. J. Hoijtink, J. Chem. Phys., 1964, 40, 2501–2504. 15 M. G. Mitch, S. J. Chase and J. S. Lannin, Phys. Rev. B: Condens. Matter, 1992, 46, 3696–3699. 16 J. Räsänen, F. Stenman and E. Penttinen, Spectrochim. Acta, Part A, 1973, 29, 395–403. 17 C. Takahashi and S. Maeda, Chem. Phys. Lett., 1973, 22, 364–367. 18 D. E. Nixon and G. S. Parry, J. Phys. C: Solid State Phys., 1969, 2, 1732–1741. 19 H. J. Dauben Jr., J. D. Wilson and J. L. Laity, J. Am. Chem. Soc., 1969, 91, 1991–1998. 20 E. Wasserman, L. C. Snyder and W. A. Yager, J. Chem. Phys., 1964, 41, 1763–1772. 21 J. C. Bonner and M. E. Fisher, Phys. Rev., 1964, 135, A640– A658.
Dalton Trans., 2014, 43, 10040–10045 | 10045
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
PAPER
Cite this: Dalton Trans., 2014, 43, 10040
View Journal | View Issue
Tuning of the ground state in electron doped anthracene Quynh T. N. Phan,a Satoshi Heguri,*b Yoichi Tanabe,a Hidekazu Shimotani,a Takehito Nakano,c Yasuo Nozuec and Katsumi Tanigaki*a,b High quality bulk samples of anthracene (AN) doped with potassium (K) in 1 : 1 and 2 : 1 stoichiometries were successfully prepared by a method involving a room temperature solid-state mechanical diffusion process prior to intercalation reactions during heat treatment, and their physical properties were studied using both magnetic and optical measurements. The transfer of almost one electron from K to AN in K1(AN) was confirmed by SQUID and ESR measurements. A pronounced magnetic hump centered at
Received 8th January 2014, Accepted 10th April 2014
150 K associated with antiferromagnetic interactions was observed, which can most likely be interpreted in terms of on-site Coulomb repulsions of the Mott insulating states. Optical spectra of K1(AN) clearly
DOI: 10.1039/c4dt00071d
showed the insulating states, as well as the electron occupation of the LUMO-derived band of AN. Our
www.rsc.org/dalton
results demonstrated tuning of the ground state of a typical bulk hydrocarbon by alkali metal intercalation.
1.
Introduction
A semiconducting state showing very high conductivity, 1 × 10−3 S cm−1, was observed in 1954 for perylene doped with halogens;1 this inspired research searching for possible metallic states composed only of simple polycyclic aromatic hydrocarbons. Unfortunately, true metallic properties were not observed in perylene. Afterwards, intrinsic metallic states showing metallic conduction bands were realized in donor– acceptor type organic charge transfer complexes, such as TTF-TCNQ,2 as well as in a single-component organic molecule.3 A recent report described a metallic state of polyaniline in a system of conducting polymers showing conductivity exceeding 1000 S cm−1, as well as electrical resistivity that decreases monotonically with temperature,4 which was an intriguing topic in this research area. However, the debate still continues as to whether true metallic states can be created in simple polycyclic aromatic hydrocarbons ( polyacenes), such as anthracene (AN), tetracene (TN) and pentacene (PN), where carriers are introduced by incorporating other elements, such as iodine, alkali metals, and so forth.
a
Department of Physics, Graduate School of Science, Tohoku University, 6-3 Aoba, Aramaki, Aoba, Sendai, Japan b WPI-Advanced Institute for Materials Research (WPI-AIMR), Electronic Materials Physics, Tohoku University, 1-1 2 Katahira, Aoba, Sendai, Japan. E-mail: [email protected], [email protected]; Fax: +81-22-217-6331; Tel: +81-22-217-6173 c Department of Physics, Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka, Japan
10040 | Dalton Trans., 2014, 43, 10040–10045
Looking back through the history of this research, many experimental approaches have been taken with these simple polyacenes. Hole doped simple polycyclic aromatic hydrocarbons, such as TN5 and PN,6 were studied by modifying the density of states at the Fermi level using chemical doping, but a metallic state has yet to be observed in the doped phases. The crystal structure and electronic properties of iodine doped PN,6 with single crystal, powder and thin film morphologies, were reported over a broad range of iodine concentrations (x) of 0.2 ≤ x ≤ 1.0. A thermally accessible excited triplet magnetic state ascribed to the presence of a dimer (PN+)2 was suggested by the results of experiments on powder samples containing intercalated iodine. From the point of view of electric transport, organic thin films have been widely investigated for PN,7 fullerene8–10 and phthalocyanine.11 Very recently, a metal– insulator transition in alkali metal doped PN thin films was discussed,7 and K1(PN) was reported to be a Mott insulator based on the electrical transport measurements, as well as theoretical calculations. In general, considering that the transfer integral, t, (below 0.5 eV) is much smaller than the on-site Coulomb repulsion energy (above 1 eV) due to the large molecular separation in van der Waals crystals of polyaromatic hydrocarbons, it is thought that a metallic state is not realized, but instead a Mott insulating state is generated. However, this controversy has not yet been settled, and the debate still continues in this intriguing area of condensed matter physics and chemistry. From a synthetic point of view, the quality of doped polyacenes is not high enough because various metastable phases with the equivalent ground state energies coexist, and this is known to be a serious problem for preparing a high quality
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
Dalton Transactions
Paper
single-phase sample by employing conventional synthetic techniques. In this paper, we report the synthesis of K-intercalated AN, the smallest molecule among polycyclic aromatic hydrocarbons, and investigate its physical properties. In order to improve the quality of the doped samples, a slightly different synthetic route, involving a solid phase diffusion process with careful mechanical grinding at room temperature prior to pelletization and annealing, is employed to promote better intercalation reactions. We show that K1(AN) prepared by this modified process is most likely to be in a Mott insulating state, and this is confirmed by both magnetic and optical measurements.
2. Experimental section A mixture of a K ingot and AN with the target stoichiometry was pelletized and ground inside a glove-box (O2: 0.1 ppm and H2O: 0.5 ppm). The process was repeated twice, and a homogeneous black powder was obtained. The resulting material was sealed in a tube filled with helium (He) after evacuation, and annealed at 413 K for two weeks. Powder X-ray diffraction (XRD) of the pristine and intercalated samples was measured at the BL02B2 beam port at SPring-8. XRD samples were sealed in soda-glass capillaries under an argon atmosphere. Raman spectroscopy was performed using a RENISHAW inVia Reflex spectrometer equipped with a single monochromator and a CCD detector at an excitation energy of 785 nm. The same capillaries were used for both XRD measurements and Raman measurements. Diffuse reflectance spectra were measured for samples sealed in quartz tubes at room temperature using a UV-vis-NIR spectrometer and an FTIR spectrometer. The magnetic susceptibilities were measured using a Quantum design MPMS-XL7 superconducting quantum interference device (SQUID) magnetometer at temperatures between 2 and 300 K under an applied magnetic field of up to 7 T. Electron spin resonance (ESR) measurements at temperatures between 4 and 300 K were performed using a Bruker E500 X-band ESR spectrometer with an Oxford ESR900 He flow cryostat. The samples for SQUID and ESR measurements were prepared in He gas in handmade partitioned quartz tubes and commercial ESR quartz tubes, respectively. Chromium powder was used as a standard g-factor reference, and the spin susceptibility was calibrated using a CuSO4·5H2O single crystal.
3. Results and discussion 3.1.
Preparation of K doped AN
We carefully prepared both K1(AN) and K2(AN) as nominal compositions by mechanical mixing at room temperature, followed by pelletization and annealing at higher temperatures that were still below the melting point of AN, as described in the Experimental section. It is difficult to prepare Kx(AN) with x exceeding 2, even when using our synthesis method. The syntheses of the target K1(AN) and K2(AN) phases were
This journal is © The Royal Society of Chemistry 2014
Fig. 1 XRD profiles of pristine AN (blue), K1(AN) (red), and K2(AN) (green).
repeated, and the high quality samples were subjected to XRD. The diffraction peaks of pristine AN could be indexed well using the crystal structure in ref. 12. Several sharp new diffraction peaks appeared, leaving a small trace of pristine AN, for the sample prepared from a 1 : 1 K to AN stoichiometry, as shown in Fig. 1. A low angle diffraction peak with a larger d-value than that of AN appeared in both the K1(AN) and the K2(AN) XRD profiles. The K1(AN) unit cell was slightly expanded compared with that of pristine AN, due to the K intercalation. When the K to AN stoichiometry of 2 : 1 was employed, the diffraction peaks of pristine AN disappeared almost entirely. In the diffraction profile of K2(AN), not only were new peaks at different positions from those of K1(AN) obtained, but other small broad peaks also appeared. The number of observed diffraction peaks was less than that of K1(AN). The appearance of broad peaks and the decrease in the number of peaks for K2(AN) may indicate that the crystal structure starts to decompose when the stoichiometry exceeds one to one. For K2(AN), new peaks different from those of K1(AN) were present, although some peaks still remained in the same positions as those of K1(AN). These diffraction profiles indicate that K1(AN) and K2(AN) are the most stable phases in the K-intercalated AN system. It is noted that when the temperature dependence for K1(AN) was studied, no distinct structural transition was detected above 90 K. 3.2.
Optical measurements
Optical spectra of pristine AN and K1(AN) are shown in Fig. 2(a). For pristine AN, absorptions due to the exciton bands were observed above 3 eV. Below this energy, there was no absorption except for a sharp peak at 0.38 eV, which was due to the C–H stretching vibrations. For K1(AN), since the regular reflection signals were dominant in the diffuse reflectivity above 0.8 eV, the reflectance spectrum was obtained by utilizing a transformation described in the literature.13 On the other hand, below 0.8 eV, the absorption spectrum was obtained using the Kubelka–Munk function (KMF) because the transmitted signals were dominant. In the reflectance spectrum,
Dalton Trans., 2014, 43, 10040–10045 | 10041
View Article Online
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
Paper
Dalton Transactions
Fig. 2 Optical absorption and reflectance spectra of (a) pristine AN and K1(AN), and (b) free solvated AN mono- and dianions. The data for the anions are from ref. 14.
several peaks were clearly observed, even below 3 eV, in contrast to pristine AN. For comparison, the absorption spectra of free solvated AN mono- and dianions14 are plotted in Fig. 2(b). For the free solvated anions, three sets of absorptions were observed at approximately 1.1–1.5 eV, 2.3–2.8 eV and 3.3 eV. They have been assigned to electron excitations from the LUMO to the LUMO+1, from the HOMO to the LUMO+1, and from the HOMO−1 to the LUMO, respectively. Note that the reflectance peaks of K1(AN) were observed at very similar energies to those of the free solvated anions, and thus, the same assignments are applicable. The appearance of the LUMO to LUMO+1 reflection band at 1–2 eV for K1(AN) indicates that the LUMO band is occupied by electrons in the ground state. This is evidence of charge transfer from K to AN in K1(AN), namely, successful electron doping. An absorption tail is observed in the low energy region in Fig. 2(a), and there appears to be no contribution from the Drude term for free carriers. In the case of K2(AN), the Drude term was not obtained in the low energy region. Hence, we can conclude that K1(AN) and K2(AN) are insulators. 3.3.
Raman spectra of AN, K1(AN), and K2(AN)
The room temperature Raman spectra of pristine AN, K1(AN) and K2(AN) are shown in Fig. 3. The peak widths became wider with increasing K intercalation. A similar increase in the peak width was reported for alkali doped fullerides, and was attributed to electron–phonon interactions.15 The AN molecule belongs to the D2h point group, and a total of thirty-three Raman-active vibration modes can be expected based on the symmetry considerations.16 However, only Raman-active modes with strong intensities associated with the C–C stretching vibrations were observed in our measurements for pristine AN. The Raman shift of AN observed in our present studies is consistent with that in other reported experiments.16 All observed vibration modes for both pristine AN and Kx(AN) were assigned and are listed in Table 1, together with those obtained in other experimental reports.16,17 Almost all the
10042 | Dalton Trans., 2014, 43, 10040–10045
Fig. 3 Raman spectra of AN (blue), K1(AN) (red), and K2(AN) (green) with λ0 = 785 nm.
peaks of Kx(AN) (x = 1, 2) were shifted to lower wavenumbers than those of pristine AN. This phonon softening is associated with the electron transfer from K to AN molecules, and accordingly the anti-bonding orbitals of AN are occupied, which weakens the vibrational force constant. A similar effect caused by charge transfer was reported for graphite intercalation compounds.18 As shown in Fig. 3 the peak at 1236 cm−1 in the spectrum of K1(AN) almost disappeared in the case of K2(AN) and the peak at 609 cm−1 for K1(AN) shifted to 596 cm−1 for K2(AN). However, low intensity peaks at 1236 cm−1 and 609 cm−1 were observed for K2(AN), and furthermore, a small trace of the peak at 596 cm−1 was also observed for K1(AN). These facts suggest that a small amount of the K2(AN) phase was present in K1(AN), and vice versa. From the intensities of the Raman spectra, the amount of K2(AN) contaminant in the K1(AN) sample was estimated to be 8%. Similarly, the amount of the contaminating K1(AN) phase in K2(AN) was estimated to be approximately 9%.
3.4.
Magnetic properties of K doped AN
Fig. 4 shows the temperature dependences of the magnetic susceptibilities (χ) of the KxAN phases (x = 0, 1, and 2). Pristine AN showed a temperature independent diamagnetism of −124 × 10−6 emu mol−1, and Curie behaviour was not observed at low temperatures. The diamagnetic value is consistent with the one observed by NMR.19 When K was intercalated, a large paramagnetic moment was detected in K1(AN), compared to pristine AN. However, when a larger amount of elemental K was intercalated, the paramagnetic susceptibility decreased again, as observed for K2(AN). This clearly indicates that electron transfer from K to AN occurs, and that the LUMO-derived band is gradually filled with electrons in proportion to the amount of K. The very small magnetic susceptibility of K2(AN) can be understood by picturing a band insulator, in which the LUMO-derived conduction band is completely filled by electrons. A weak increase in the temperature dependent χ observed at low temperatures for K2(AN) can be understood using the Curie term, and the spin number was estimated to
This journal is © The Royal Society of Chemistry 2014
View Article Online
Dalton Transactions Table 1
Paper
Vibration modes and Raman shifts of AN, K1(AN), and K2(AN), together with the theoretical and experimental results from ref. 15 and 16
K1(AN)
AN Vibration mode
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
ag
b3g
ω δ ω ω δ ω ω ω ω δ
Ref. 16
This work
Δν
Δν
1557 1482 1403 1260 1163 1007 753 622 395 1187
1561 1484 1406 1264 1168 1012 758 526 — 1191
K2(AN)
Ref. 17
This work
Intensity
Δν
Δν
3186 543 7199 989 967 883 6035 669 — 1087
1543 1466 1360 1230 1152 1024 738 608 390 1180
1534 1478 1352 1236 1152 1020 738 609 390 —
Ref. 17
This work
Intensity
Δν
Δν
Intensity
812 200 3056 1971 121 336 275 2621 144 —
1540 1470 1350 — 1150 1020 723 599 390 1175
1536 1467 1351 — 1143 1019 726 595 390 —
800 251 10 350 — 250 300 298 1246 306
the previous reports. Schematic models of the ground states are shown in Fig. 5. In principle, the thermally activated triplet state results from the formation of dimers of a band insulator in a localized electron system with a singlet ground state, while the Mott insulating state is in principle categorized as an itinerant electron system, in which electrons are localized due to the on-site Coulomb repulsion energy, U. First, we tentatively employed a similar singlet–triplet model by excluding the Pauli term, and found that eqn (1) gave a fairly good fit to the data.
Fig. 4 Temperature dependences of the magnetic susceptibilities of pristine AN (blue), K1(AN) (red), and K2(AN) (green).
be 0.05 spin per molecule. This can reasonably be attributed to the defects in the lattice created during the K intercalation processes. The magnetic susceptibilities of K1(AN) and K2(AN) are consistent with the XRD data discussed earlier. It should be noted that the increase in χ observed for K1(AN) at low temperatures is much larger than that for K2(AN) and does not follow the general Curie-type 1/T temperature dependence. This result will be discussed in the next paragraph. 3.5.
Detailed discussion of the magnetic properties of K1(AN)
The most striking and intriguing phenomenon in the magnetic susceptibility of K1(AN) is the fact that a broad hump centered at 150 K was observed, which was most evident between 50 K and room temperature (Fig. 4). It is also very important that K1(AN) exhibited a large spin number of more than 0.5 spins per AN, and therefore this cannot be ascribed to any defect spins. Brinkmann et al. previously reported a similar magnetic behaviour for iodine-doped I1(PN),6 and interpreted the result in terms of a thermally accessible singlet–triplet equilibrium state. This result should be discussed and compared to the observations for K1(PN) reported by Craciun et al.,7 where the observed electronic states are interpreted based on the Mott insulating state. The ground states of both electron (K doped) and hole (I doped) PNs are insulating, but their ground states are different according to
This journal is © The Royal Society of Chemistry 2014
χ¼
C1 4C2 þ T T ð3 þ eT 0 =T Þ
ð1Þ
In eqn (1), C1 is the Curie constant arising from the defects, C2 is related to the number of localized spins, and T0 (= Δ/kB) is the characteristic temperature associated with a thermally accessible singlet–triplet energy gap. The parameters in eqn (1) were optimized by the least squares method: C1 = 0.0199 emu mol−1 K, C2 = 0.246 emu mol−1 K, and T0 = 240 K. The number of spins per AN molecule (n: estimated from C2) was approximately 0.66. This model supposes that a biradical electronic state is formed as the ground state (G0), and that the next thermally accessible triplet state (T1) is located slightly above G0 on a kBT0 energy scale, as shown in Fig. 5a. In order to clarify the ground state of K1(AN), we performed ESR spectroscopy. In the case of the singlet–triplet model, the thermally accessible triplet species should be detected by ESR, as its spectrum shape (S = 1) is very different from that of a
Fig. 5 Schematic models of (a) thermal activation and (b) the Mott insulator in K1(AN).
Dalton Trans., 2014, 43, 10040–10045 | 10043
View Article Online
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
Paper
doublet (S = 1/2) due to the D and E zero-field splitting parameters in the traceless S·D·S spin–spin Hamiltonian.20 This has been very widely studied in biradical systems, as well as in photo-excited triplet states. Our experimental data shown in Fig. 6, however, displayed no such typical triplet spectral features, and instead a symmetrical signal that could be fitted using a single Lorentzian function was observed above 4 K. It should be noted that, strictly speaking, a small asymmetric feature was observed at very low temperatures, but the spectrum was described well by introducing a small anisotropic g tensor in the analysis, by integrating over all azimuth angles in g-space in the fashion of a general powder ESR simulation. A Pauli paramagnetic contribution associated with conduction electrons was also not observed. When these experimental facts are taken into account, the most reasonable explanation of the K1(AN) ground state is a Mott insulating state. This interpretation is consistent with the optical measurements described earlier. The singlet–triplet model, therefore, should be ruled out as an explanation of the electronic state of K1(AN). Consequently, according to these discussions, we temporarily employed an Ising model for an antiferromagnetic spin chain, described by Bonner and Fisher21 within the framework of a finite chain length, for the next stage of interpretation, which will be discussed in the next paragraph. In this model, the magnetic susceptibility decreases due to antiferromagnetic interactions between adjacent spins at temperatures below J/kB, and at the same time the increase in magnetic susceptibility is steeper than conventional Curie-type temperature dependence, which can be explained when the finite chain length is taken into consideration. It is noted here that
Fig. 6 First derivatives of the ESR signals of K1(AN) at 4 K, 40 K, and 280 K, and the curves simulated using Lorentzian functions. The blue empty circles are the first derivatives of the ESR signals, the red lines are the curves fitted using Lorentzian functions, and the green lines are the differences between the experimental data and the fitting results.
10044 | Dalton Trans., 2014, 43, 10040–10045
Dalton Transactions
Fig. 7
Schematic model of the finite spin chain in K1(AN).
although the influence of other phases may generally be considered in order to understand the experimental phenomena, other phases are unlikely to be involved in the present case because an ESR spectrum with only a single Lorentzian lineshape was observed, as shown in Fig. 6. It may be reasonable to imagine that the prepared intercalated phase does not have long-range ordering due to the low dimensional instability. We therefore employed a modified Bonner–Fisher model as an alternative choice by taking into account the different finite numbers of spin chains consisting of even (N) and odd (M) numbers of spins with the equivalent interaction energy ( J), whereas previously the dispersion of the chain length was neglected. The magnetic susceptibility is expressed by " # C1 J=kB T 1 ð tanhð J=2kB TÞÞN χ¼ e T 1 þ ð tanhð J=2kB TÞÞN " # C2 J=kB T 1 ð tanhð J=2kB TÞÞM þ e T 1 þ ð tanhð J=2kB TÞÞM
ð2Þ
where C1 and C2 correspond to the Curie constants of the chains with even and odd numbers of spins, respectively. A schematic model of a finite spin chain in K1(AN) is shown in Fig. 7. In this model, the antiferromagnetic interactions between two adjacent spins cause a broad hump at intermediate temperatures, while an odd number of spins increases the magnetic susceptibility at low temperatures, as shown in Fig. 8. The parameters in eqn (2) were optimized using the least squares method, and were evaluated as N = 12, M = 11, C1 = 0.168 emu mol−1 K, C2 = 0.183 emu mol−1 K, and J = 12 meV ( J/kB = 140 K). The number of spins per AN molecule (n, esti-
Fig. 8 Temperature dependence of the static magnetic susceptibility of K1(AN) and the fitting using the Bonner–Fisher model.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Dalton Transactions
mated as the summation of C1 and C2) was sufficiently large (approximately 0.93), and importantly, the estimated spin numbers were consistent, irrespective of the choice of model.
Published on 10 April 2014. Downloaded by University of Prince Edward Island on 31/10/2014 08:28:46.
4.
Conclusion
We successfully synthesized K-intercalated AN, using an improved solid-state mechanical mixing and thermal diffusion methodology, and observed its optical, magnetic and vibrational properties. All the properties of K1(AN) can be well described based on a Mott insulating state, while K2(AN) is a band insulating state. Both magnetic and optical experiments provided evidence for the Mott insulating state of K1(AN), with the transfer of one electron from K to AN. This conclusion seems reasonable when the on-site Coulomb repulsion energy, U, (1 eV) and the band width energy, W, (0.6 eV) are considered in the itinerant band limit. We did not observe thermally accessible triplet states via biradical formation in the electron localization limit. Our results demonstrate tuning of the ground state in a typical molecular organic semiconductor, and suggest that a unique magnetically ordered state with a finite number of spins can be created in K-intercalated aromatic hydrocarbons. In the future, it would be interesting to see whether a metallic system can be produced by controlling molecular alignment at various degrees of charge transfer under elevated pressure.
Acknowledgements We acknowledge Prof. K. Akiyama for his fruitful discussions. This work was supported by the Scientific Research Fund (24651128, 24684023 and 25610084) and the World Premier International Research Center Initiative (WPI) from the MEXT of Japan and the Tohoku University GCOE program. This work was also supported by the Joint Studies Program (2011–2012) of the Institute for Molecular Science.
References 1 H. Akamatsu, H. Inokuchi and Y. Matsunaga, Nature, 1954, 173, 168–169.
This journal is © The Royal Society of Chemistry 2014
Paper
2 P. W. Anderson, P. A. Lee and M. Saitoh, Solid State Commun., 1973, 13(5), 595–598. 3 H. Tanaka, Y. Okano, H. Kobayashi, W. Suzuki and A. Kobayashi, Science, 2001, 291, 285–287. 4 K. Lee, S. Cho, S. H. Park, A. J. Heeger, C. W. Lee and S. H. Lee, Nature, 2006, 441, 65–68. 5 P. A. Stunzhas, V. A. Benderskii and L. A. Blyumenfel’d, J. Struct. Chem., 1968, 8, 393–398. 6 M. Brinkmann, V. S. Videva, A. Bieber, J. J. Andre, P. Turek, L. Zuppiroli and R. Humphry-Baker, J. Phys. Chem. A, 2004, 108, 8170–8179. 7 M. F. Craciun, G. Giovannetti, S. Rogge, G. Brocks, A. F. Morpurgo and J. van den Brink, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 79(12), 125116–125123. 8 T. T. M. Palstra, R. C. Haddon, A. F. Hebard and J. Zaanen, Phys. Rev. Lett., 1992, 68(7), 1054–1057. 9 R. C. Haddon, G. P. Kochanski, A. F. Hebard, A. T. Fiory and R. C. Morris, Science, 1992, 258, 1636–1638. 10 S. Heguri and M. Kobayashi, Solid State Commun., 2010, 150, 1489–1492. 11 M. F. Craciun, S. Rogge, M. J. den Boer, S. Margadonna, K. Prassides, Y. Iwasa and A. F. Morpurgo, Adv. Mater., 2006, 18, 320–324. 12 A. Charlesby, G. Finch and H. Willman, Proc. Phys. Soc., 1939, 51, 479–528. 13 T. Kodaira, Y. Nozue, S. Ohwashi, T. Goto and O. Terasaki, Phys. Rev. B: Condens. Matter, 1993, 48, 12245–12252. 14 K. H. J. Buschow and G. J. Hoijtink, J. Chem. Phys., 1964, 40, 2501–2504. 15 M. G. Mitch, S. J. Chase and J. S. Lannin, Phys. Rev. B: Condens. Matter, 1992, 46, 3696–3699. 16 J. Räsänen, F. Stenman and E. Penttinen, Spectrochim. Acta, Part A, 1973, 29, 395–403. 17 C. Takahashi and S. Maeda, Chem. Phys. Lett., 1973, 22, 364–367. 18 D. E. Nixon and G. S. Parry, J. Phys. C: Solid State Phys., 1969, 2, 1732–1741. 19 H. J. Dauben Jr., J. D. Wilson and J. L. Laity, J. Am. Chem. Soc., 1969, 91, 1991–1998. 20 E. Wasserman, L. C. Snyder and W. A. Yager, J. Chem. Phys., 1964, 41, 1763–1772. 21 J. C. Bonner and M. E. Fisher, Phys. Rev., 1964, 135, A640– A658.
Dalton Trans., 2014, 43, 10040–10045 | 10045
