Appl. Magn. Reson. 8, 173-180 (1995)

Applied Magnetic Resonance ~'. Sprmger-Verlag 1995 Printed in Austria

13C MAS NMR Investigations of Alkali Doped C60 M. K a n o w s k i l, H. Werner 2, R. Schl/igl 2, H.-M. Vieth 1, and K. Lª

1

1nstitut fª Experimentalphysik, Freie Universit~it Berlin, Berlin, Germany 2 Institut fª Anorganische Chemie, Universit~tt Frankfurt a. M., Frankfurt, Germany Received April 14, 1994

Abstraet. Results of ~3C MAS NMR measurements of the Rb~Cs0system (x = 2.75, 3, 4, 6) and the A6C60 compounds (A = K, Rb, Cs) are presented. Special attention was paid to sample preparation in order to suppress effects of impurities and lattice defects due to imperfect C£ starting material. The ~3C MAS NMR measurements of the RbxC60system demonstrate the usefulness of this method to reveal valuable information about its phase diagram. The existence of underdoped Rb3C60is proved. Well resolved lines in all investigated A6C60 compounds confirm the orientational order of the C60 ions. An assignment of the signals to the three magnetically inequivalent carbon atom positions in the crystal structure is proposed.

1. Introduction The characterization o f different superconducting Rb3C60 samples by X-ray diffraction [1] and ac-susceptibility [2] revealed that the sample properties depend strongly on the quality o f the C60 starting material used in the preparation procedure. Moreover, intensive work on sample preparation o f alkali doped C60 compounds demonstrated the necessity to use carefully sublimed C60 in order to ensure satisfying reproducibility and to reduce effects o f lattice defects and impurities like remaining solvent molecules [3]. Nearly all research on the phase diagram o f the RbxC60 system was done by X-ray diffraction and homogeneous phases for x = 1, 3, 4, 6 are documented [4-6]. The X-ray diffraction method needs coherent scattering domains o f a considerable size to detect and identify a phase. In contrast 13C N M R is sensitive to the local environment o f the 13C nuclei and therefore can give valuable supplementary information. ~3C MAS N M R is an important technique to reveal structural and dynamical information in organic solids and it has been successfully used in the characterization o f alkali doped C60 [7-10] so far. We applied this method mainly to aim a more reliable knowledge o f the phase diagram o f the RbxC60 system especially o f the nominally underdoped Rb3C60 phase and about orientational order o f the C60 ions in

174

M. Kanowski

e t al. :

the alkali metal doped fullerenes. All these properties are of great interest because of the intense discussion of the phase diagram and effects of order and disorder in doped C60 compounds also with respect to superconductivity (see [11] for a detailed review).

2. Experimental C60 was produced in an arc generator, separated from the higher fullerenes by liquid chromatography, and vacuum dried during a temperature ramp up to about 680 K to remove the main part of the solvent. All subsequent sample handling was done in a glove box under argon atmosphere because of the extreme air and moisture sensitivity of these materials. The vacuum dried C60 was sublimed in two further steps. Under a dynamic vacuum at about 780 K in a first step the residual solvent molecules and other impurities are removed. The second sublimation stage was performed in a vacuum sealed quartz ampoule at about 880 K. This was essential for a low defect crystalline growth. The alkali saturated A6C60 phases were prepared in a two-zone quartz ampoule sealed under vacuum. About 1 g of the doubly sublimed C60 was placed at one end and the respective alkali metal in a slight excess of the stoichiometric amount at the other end. Then the ampoule was heated in a tube fumace for 5 days while a temperature difference between the C60 powder (510 K) and the alkali metal (500 K) prevented the condensation of alkali vapour on the powder. The RbxC60 compounds were prepared by mixing Rb6C60 with the appropriate amount of sublimed C60 and subsequent armealing at 750 K for 3 days. After every preparation step the sample material was checked by X-ray diffraction.

For the 13C MAS NMR measurements about 100 mg of the sample material were transferred into a pyrex rotor insert (Wilmad Glass Co.), evacuated, a n d a t first sealed about 4 cm away from the final sealing constriction of the insert due to the heat sensitivity of the samples. After that a special apparatus rotated the sample slowly around its vertical axis and while simultaneously cooled with liquid nitrogen the sample was finally sealed off. All :3C NMR measurements were performed at 7.05 T field (75.47 MHz resonance frequency) with a BRUKER CXP 300 spectrometer equipped with a 7 mm double bearing MAS probe. The rotor insert was put into the rotor together with some KBr powder for fine adjustment of the magic angle. Spinning frequencies up to 4 kHz were applied. For data acquisition a rotor synchronous lr/2-TR-~r echo sequence (TR = rotor period) was used in order to minimize phase and baseline distortions. Typical 1r/2-pulse lengths were 2.5 gs. The pulse sequence was fully phase cycled for quadrature detection, also to remove the free induction decay from the echo. Due to the relative long delay (TR = 250 gs at VROT = 4 kHz) in the echo sequence for control purposes the free induction decays were recorded in order to ensure that no signal contributions with short T z times are suppressed in the spectra. Spin lattice relaxation times T1 were measured with the saturation recovery method.

t3C MAS NMR Investigations of Alkali Doped C60

175

3. Results

3.1. RbxC6o Compounds Table 1 summarizes the values for the isotropic chemical shifts, the centerband widths, and the spin lattice relaxation times T1 of the room temperature 13C MAS NMR spectra of the RbxC60 compounds for x = 0, 3, 4, 6 (RbC60 data are reported from [12]). In order to research the phase diagram Table 1 provides the necessary information for the interpretation of spectra of samples with other nominal stoichiometries. Fig. 1 shows the 13C MAS NMR spectrum of Rb3C60 together with that of the underdoped Rb3C60 phase Rb2.75C60 as an example. Both spectra show a broad inhomogeneous centerband line in the range from 175 to 205 ppm. Tycko [10] reported a similar result for Rb3C60. The small line at 181.6 ppm in the Rb3C60 spectrum belongs to Rb4C60 (see Table 1). This very small amount (< 1%) ]ies within the systematic error of the stoichiometry. A test measurement with a longer repetition time verified that both samples contain neither C60 nor Rb6C60. 13C MAS NMR is especially sensitive to C60 due to its very narrow linewidth compared to the other phases. The absence of other phases proves that the nominally underdoped sample is a homogeneous phase and consequently Rb3C60 is n o t a line phase. X-ray investigations [1] and ac-susceptibility studies support this result. The latter show similar superconducting transition temperatures (30.7 K for Rb3C60 and 30.1 K for Rb2.75C60) and superconducting volume fractions for both samples [13].

3.2. A6C6oCompounds Figure 2 shows the spectrum of Rb6C6o and Fig. 3 the centerbands of the A6C6o spectra with A = K, Rb, Cs. The overall spectra of K6C60 and Cs6C60 (not shown) have sideband intensity ratios very similar to those of Rb6C6o. In the case of K6C6o a 13C MAS NMR spectrum was already published by Rachdi et al. [8] with the same

Table 1. Characteristic data of the room temperature t3C MAS NMR spectra of the Rb~C60 phases. Isotropic chemical shifts (ppm) relative to TMS C60 143.0 RblC60 175 Rb3C6o from 175 to 205 Rb4C60 181.6 Rb6C60 155.4, 156.8

Width of the centerband (ppm)

T~ (s)

0.05 -30 1.2 2.0

95 1 0.25 0.09 >120

Remarks

data from [12], measured without MAS see Fig. l see Figs. 2 and 3, T1 is estimated

176

M. Kanowski et al.:

,I 300

,

I 200

,

I 100

I

ppm

Fig. 1. ~3C MAS N M R spectra of Rb3C6o (120000 scans, repetition time 2 s, VROT = 4050 Hz) and Rb 27sC6o (4800 scans, repetition time 2 s, VRoT = 4030 Hz). Stars denote spinning sidebands. The small line at 181.6 ppm in the Rb3C6o spectrum belongs to Rb4C60. Chemical shifts are given relative to TMS.

RbsCso

.

:,1,:

Lj. ,

I 300

,

,

,

,

I 250

,

,

~

,

I 200

,

,

,

,

I

,

,

150

,

,

I 100

,

,

,

,

I 50

,

,

,

,

I

,

0

ppm

Fig. 2. t3C MAS NMR spectrum of Rb6C60 (7654 scans, repetition time 30 s, VgoT= 2265 Hz). Stars denote spinning sidebands. The small lineat 181.6 ppm belongs to Rb4C60. Chemical shifls are given relative to TMS.

13C MAS NMR Investigations of Alkali Doped C60

177

CssCeo i

t

I

160

I

i

ppm

i

I

~

9

15O

Fig. 3. Centerbands of the ~3C MAS NMR spectra of A6C60 compounds (A = K, Rb, Cs). Chemical shifts ate given relative to TMS.

line positions and intensity ratios as observed in our measurements. The well resolved lines in all the three A6C60 compounds (Fig. 3) have their origin in magnetically inequivalent C atom positions. Their occurrence confirms the orientational order of the C60 ions in the bcc lattice of these compounds first revealed by Riet-veld refinement of X-ray diffraction data [4, 14]. There are three inequivalent carbon atom positions in the crystal structure. In each case 48 carbon atoms are in C2 and C3 position and the remaining 24 of the 120 carbon atoms of the unit cell in C1 position [14]. Hence, the ~3C MAS NMR spectrum should exhibir three lines with an intensity ratio of C1 : C2 : C3 = 1 : 2 : 2 as observed for K6C60 (Fig. 3). In the spectra of Rb6C6o and Cs6C60, however, only two resolved lines are present. The intensity ratio of the two lines of about 3 : 2 manifests the assumption that due to its smaller intensity the signal from the C1 position coincides either with the signat of the C2 or the C3 position. Small uncertainty remains because of the very long spin lattice relaxation times Tl in the A6C60 compounds, however a spectrum of Rb6C60 measured with 90 s repetition time instead of 30 s as in Fig. 2 shows the same intensity ratio of the two resolved lines in the centerband. From spectra with different acquisition delays we estimate that Tt exceeds 120 s.

178

M. Kanowski et al.: 4. Discussion

4.1. RbxC6o Compounds Despite the fact of the usefulness of 13C MAS NMR in revealing valuable information regarding the phase diagram an interesting question is why the 13C MAS NMR spectrum of Rb3C60 shows such a broad distribution of isotropic chemical shifts (including Knight shift contributions). According to the Rietveld refinement of the Rb3C60 structure [5] one would expect three resolved lines a s a consequence of orientational order of the C60 ions in the lattice similar to the situation in the A6C60 compounds. Lattice defects o f the C60 material or impurities are a rather unlikely reason for such a distribution because of the high quality of the starting materials. Therefore, the broad centerband line should be discussed as an intrinsic property. The occurrence of only one line can be interpreted a s a result of the absence of orientational ordering. In this case the more or less random orientations of the C60 ions would result in an inhomogeneously broadened line. But the width of the observed signal is about one order of magnitude larger (about 30 ppm, Fig. 1) than the difference of the signals from inequivalent C atom positions in Rb6C60 (about 2 ppm, Fig. 3). Although Rb3C60 (fcc) and Rb6C60 (bcc) can not be compared directly, the absence of orientational order is unlikely to explain such a large width. More probably, it can be attributed to off-center positions of the rubidium ions in the octahedral interstitial sites of the C60 fcc lattice. Their existence was suggested by 87Rb NMR [15] and later shown by X-ray absorption investigations [16]. Further work, however, is necessary to elucidate the nature of these Rb ion displacements and their influence on the electronic structure. Since the orientational order of the C60 ions could be masked by the structural disorder due to the Rb ions in the octahedral interstitial sites the unresolved centerband should not be taken as evidence for the absence of orientational order in Rb3C60.

4.2.

A6C6o

Compounds

The situation in the A6C60 compounds is different. Here the resolved lines in the centerband confirm the orientational order and it remains to assign the signals to the different carbon atom positions in the crystal structure. We propose a line assignment in Fig. 4 where the chemical shifts of the A6C60 compounds versus the nearest C60-C60 distances are plotted. The ionic salt character of the A6C60 compounds [17] motivates the choice of the nearest C60-C60 distance as parameter for the plot, implying that the electrostatic interaction between C60 anions and alkali cations is the main factor for the stabilization o f the lattice and that the lattice constants scale nearly linearly with the ionic radii. Therefore, it is expected that the electronic structure probed through the chemical shift at the 13C nuclei undergoes a monotonous variation from K6C60 to Cs6C60. AS-

1~C MAS NMR Investigations of Alkali Doped C60

179

158 ppm 157 156 155 154

9

153 ~" 9.8

C2

[] C 3 I

9.9

lo',o

lo',1

1012

Ceo- eso (~,)

Fig. 4. Plot of the chemical shit~ vs. the nearest C60-C60distances (K: 9.85 A, Rb: 9.98/~, Cs: 10.20 A). The legend proposes an assignment of the signals in Fig. 3 to the three inequivalent carbon atom position of the orientationally ordered C60 ions in the A6C60compounds. The notation of the carbon atom positions is according to [14]. The lines are guides for the eye.

suming coinciding lines for Rb6C60 and Cs6C60 as explained above, the assignment of the signal with the hall intensity of the other two signals to the carbon C1 position is u n a m b i g u o u s (crosses in Fig. 4) and has to be taken as spectroscopical evidence. This signal shows the largest change in the chemical shift (about 4 ppm). For the other signals this shift varies less. It is reasonable to suppose that the change in the electronic structure of the C60 ion surface is mainly mediated over the bonds. Since C2 and C1 are at neighbouring positions in the carbon network whereas C3 and C1 are only indirectly connected over two bonds [14] one can assign the line with the smallest chemical shift variation (squares in Fig. 4) to the C3 position and the other one (circles in Fig. 4) to the C2 position. The large chemical shift change for the C1 atoms leads to the conclusion that the change in the electronic structure due to the lattice dilatation by introducing larger alkali metal ions is most pronounced at the C1 carbon atom posifions. For K6C60 a local density approximation calculation [18] gives a nonuniform distribution of the electrons at the different carbon atom positions. The calculated slight preference for the excess charge on the C60 ion to be associated with C1 atoms is explained a s a consequence of those C atoms being closest to the K ions. I r the upfield shifted C1 signal in K6C60 (153.4 ppm) is assumed to be a result of this extra charge, one would expect similar upfield shifted lines in the spectra of Rb6C60 and Cs6C60since in these compounds the C1 carbons are also closest to the alkali ions. This is not the case and indicates that the change of the electronic structure of these insulators is more complex and does not scale in a simple manner. Nevertheless, our results clearly show a redistribution of the electrons on the C60 ions regardless to all line assignment considerations.

180

M. Kanowski et al.: References

[1] Werner H., Schl6gl R.: in preparation. [2] Beanitz M., Heinze M., Straube E., Werner H., Schl6gl R., Thommen V., Gª H.-J., Lª K.: Physica C 228, 181 189 (1994) [3] Werner H.: Doctoral Thesis, Universit~t Frankfurt a. M. [4] Zhu Q., Zhou O., Coustel N., Vaughan G.B.M., McCauley Jr. J.E, Romanow W.J., Fischer J.E., Smith III A.B.: Science 254, 545-548 (1991) [5] Stephens EW., Mihaly L., Wiley J.B., Huang S.-M., Kaner R.B., Diederich E, Whetten R.L., Holczer K.: Phys. Rey. B 45, 543 546 (1992) [6] Zhou O., Cox D.E.: J. Phys. Chem. Solids 53, 1373-1390 (1992) [7] Yannoni C.S., Wendt H.R., de Vries M.S., Siemens R.L., Salem J.R., Lyerla J., Johnson R.D., Hoinkis S., Crowder M.S., Brown C.A., Bethune D.S.: Synth. Metals (in press) [8] Rachdi F., Reichenbach J., Firlej L., Bemier R, Ribet M., Aznar Z., Zimmer G., Helmle M., Mehring M.: Solid State Commun. 87, 547-550 (1993) [9] Bemier R, Rachdi F., Ribet M., Reichenbach J., Firlej L., Lambert J.M., Zahab A., Belahmer Z., Azner R. in: Springer Series in Solid-State-Science, vol. 117: Electronic Properties of Fullerenes (Kuzmany H., Fink J., Mehring M., Roth S., eds.), pp. 348-353. Berlin Heidelberg: SpringerVerlag 1993. [10] Tycko R.: J. Phys. Chem. Solids 54, 1713-1723 (1993) [11] Fischer J.E., Heiney RA.: J. Phys. Chem. Solids 54, 1725-1757 (1993) [12] Tycko R., Dabbagh G., Murphy D.W.: Phys. Rey. B 48, 9097-9105 (1993) [13] Beanitz M., Heinze M., Werner H., Kanowski M., Lª K.: in preparation. [14] Zhou O., F i s c h e r J.E., Coustel N., K y c i a S., Z h u Q., McGhie A.R., R o m a n o w W.J., McCauley Jr. J.R, Smith III A.B., Cox D.E.: Nature 351, 462M64 (1991) [15] Walstedt R.E., Murphy D.W., Rosseinsky M.: Nature 362, 611-613 (1993) [16] Nowitzke G., Dumschat J., Wortmann G., Werner H., Schl6gl R.: Mol. Cryst. Liq. Cryst. 245, 321-326 (1994) [17] Tanigaki K., Hirosawa I., Mizuki J., Ebbesen T.W.: Chem. Phys. Letters 213, 395-400 (1993) [18] Erwin C.S., Pederson M.R.: Phys. Rey. Lett. 67, 1610-1613 (1991)

Author's address: Dr. Martin Kanowski, Freie Universitfit Berlin, Institut fª WE l, Arnimallee 14, 14195 Berlin, Germany

Experimentalphysik

Verleger Sprmger-VerlagKG. Sachsenpla~z 4 6. A-1201 Wten -- Herausgeber Prof K M Sahkho'... Za~o~sky PhyslcalTechmcal InsHute. Siblrsky tracl 10 7. 420029 Kazan RussLa Redaktlon Za,.o~sk) PhyslcaI-TechmcalIn,,lllule. S~blr~.k) tracI 10.'7, 420029 Kazan. Russla Hersle[lerSatz und Umbruch Zavolsky Physlcal-TechmcalInslltule. SIblr,,ky tracl 10 7. 420029 Kazan RussJa Offsetdruck Druckeret Ferdmand Berger&Sohne Gesellschaft m b H, A-3580 Horn Verlagsort Wlen -- Herstellungsort Horn -- Prmted m Ausma Offenlegunggem 825 Abs 1 bas3 MedLengesetz Unternehmen~gegenstand Verlag ',on wlssenschafthchen Buchern und Zeltschnften An der Sprmger-Verlag KG lsl betelhgt Dr Konrad F Spnnger. Sachsenplatz 4 6, A-1201 Wlen. als Kommand~tlst.,'u 52.38~ Geschaftsfuhrer Dr Konrad F Sprmger. Prof Dr Dlemch Gotze. lng Wolfram F Joos. Dtpl-Kfm Clau,, Mtchaletz und Rudolf Slegle. alle Sachsenplatz 4 6. A-1201 Wten