513
ELECTRONIC PROPERTIES OF TCNQ COMPLEXES barrier film is almost the same both in the anodic and cathodic direction of polarization. That would suggest that very thin films of oxide formed spontaneously probably do not rectify, at least, not before activation (flaw production?) of the surface has been achieved by some initial cathodic polarization. Here, it has been assumed that both Faradaic rectification (at filmsolution interface)3* and any possible rectification at the metal-oxide interface are negligible and do not enter into the argument. This behavior may be contrasted with that of anodically grown thicker films which are known to rectify when placed in metal-oxide-electrolyte configuration.ag Here, an implied conclusion is that thin films which are probably continuous (i.e., without flaws) before cathodic activation do not rectify whereas thicker films of the same oxide do rectify and hence39 must have flaws at which t8heh.e.r. is sustained without developing an appreciable potential drop across the film
in the cathodic direction of polarization. Most of these conclusions are, obviously, quite tentative and would need much more elaborate investigations of the matters mentioned above. Acknowledgments. The author wishes to acknowledge helpful discussions with Drs. Robert S. Alwitt and Glenn 31. Cook of these laboratories and Dr. 31, Salomon of NASA, ERL, Cambridge, Massachusetts. Thanks are also due to Nr. W. Hilchey, of Sprague Test Equipment Department, for assistance in calibration of the instruments. (38)K. S. 0.Doss and H. P. Agarwal, J . S e i . Ind. Res. India, 9B, 280 (1950); see also Proc. Indian Acad. Sci., 34A, 263 (1951); 35A, 45 (1952); also see ref 23.
(39) Symposium on "Electrolytic Rectiflcation and Conduction Mechanisms in Anodic Oxide Films," P. F. Schmidt and D. M. Smyth, Ed., The Electrochemical Society Inc., New York, N. Y., 1967.
Electronic Properties of Some TCNQ Complexes1 by A. Rembaum, A. M. Hermann, F.
E. Stewart, and F. Gutmann
Polymer Research Section. Jet Propulsion Laboratory, Californ6a Institute of Technology, Pasadena, California 91103 (Received June 1 3 , 1968)
A study of electrical properties of tetracyanoquinodimethaiie (TCNQ) complexes representing unit segments of nonconjugated as well as conjugated polymers is described. Corresponding studies with the analogous polymer complexes are presented. The model compounds choseii were a saturated donor I ,2-bis(4-pyridyl)ethane and an unsaturated donor 1,2-bis(4-pyridyl)ethylene. Analyses are presented to substantiate the chemical structure. Spectrophotometric data are in agreement with previous results. Electron spin resonance studies show the triplet nature of the complexes, and rotational anisotropy in compressed pellets denionstrates orientation of molecules or crystallites during compression. Electronic transport properties iiicludiiig the first reported measurement of the Hall effect in TCNQ complexes are described. These measurements along with conipanion studies of conductivity and therinoelectric power indicate concentration of carriers of several orders of magnitude below that of the unpaired spins. The transport data presented are interpreted in terms of band theory.
I. Introduction
1 ohn1-l cm-l, Le., the highest conductivity of organic crystals known to date. It was also shown recently2b It was recently reported that certain heterocyclic that polymeric analogs of the TCNQ salts can be salts exhibit an exceptionally high cond~ctivity.**~ prepared. One such complex, copoly (styrene),1These compounds may be represented by the general butyl-2-vinylpyridinium (TCNQ-) TCNQ had a formula D+(TCNQ-) in which D is an aromatic conductivity of ohm+ cm-1 and these polymeric molecule generally containing a nitrogen atom and (TCNQ-) symbolizes 7,7', 8,8'-tetracyanoquinodimethane in the form of a paramagnetic radical anion. (1) This paper represents one phase of research performed by the Jet Propulsion Laboratory, California Institute of Technology, Neutral TCNQ molecules may also take part in the sponsored by the National Aeronautics and Space Administration, complex in which case the electrical conductivity is Contract NAS7-100. (2) (a) R. G. Kepler, J . Chem. Phys., 39, 3528 (1963): (b) J. H. increased by several orders of magnitude. A number Lupinski and K. D. Kopple. Sctence, 146, 1038 (1964). of the TCNQ complexes exhibit room temperature (3) L. R. Melby, R. J. Harder, W. R. Hertler, W. Mahler, R. E. conductivity varying from about 10-6 up to about Benson, and W. E. Mochel, J. Amer. Chem. SOC.,84, 3374 (1962). Volume '7SPNumber 9 March 1969
A, REMBAUM, A. HERMANN, F, STEWART,AND F. GUTMANN
514
complexes could be cast from solution as homogeneous films. The mechanism of electronic transport in the above monomeric or polymeric salts is not well understood. In this report a study of the electrical properties of TCiYQ complexes representing unit segments of nonconjugated as well as conjugated polymers is described. The following two model compounds were chosen: a saturated donor 1,2-bis (4-pyridyl) ethane, and unsaturated donor 1,2-bis (4-pyridyl) ethylene. These two model compounds will be referred to as D, and D,, respectively,
1,2.bis(4-pyridyl)ethylene D"
l,Ebis(4-pyridyl)ethane D B
Both D, and D, could be incorporated int'o a polymeric chain by means of the reaction4
L
h
where n represents the number of unit segments. D, reacted with p-dibromoxylene to yield an analogous polymer. These polymers when reacted with LiTCNQ will be designated by the symbols DU2+( TCNQ-) 2 polymer and D2+(TCNQ-) 2 polymer, respectively. The polymers present the advantage of easier moldability and higher tensile strength than the corresponding monomeric analogs. Most of the electrical measurements were carried out on D, or D, mono- or diiodide salts after reaction of the latter with LiTCNQ or with a mixture of LiTCNQ and neutral TCNQ.
11. Experimental Section 1. Reagents. D, was recrystallized from benzene (mp 122'). D, was used as received (Aldrich) mp 153-155'. p-Dibromoxylene was recrystallized from benzene (mp 145-148'). All solvents were freshly distilled and TCNQ was used as received from the Dupont Go. 2. Preparation of Mono- and Bis-N-Methyl Pyridyliuiiz Derivatives of 1 ,b-Bis(4-Pyridyl)ethane. a. 1(4-Pyridyl), 2- (4'-N-rnethylpyridyliuin iodide) ethane (DJ).1,2-Bis(4-pyridyl) ethane (0.01 mol = 1.84 g), dissolved in benzene (50 ml), was added dropwise to CH,I (0.06 mol = 8.52 g) in benzene (10 ml). The mixture was heated while stirring to start the reaction, and the stirring continued for 1.5 hr a t ambient temperature. It was then kept in the dark for approximately 4 hr. The light yellow microcrystals which had precipitated out were filtered off, washed with The Journal of Physical Chemistry
benzene, and dried in vacuo. Anal. Calcd for a monoiodide salt: I, 39.2. Found: I, 39.5. b. 1-(4-Pgridyl), 2-(4'-N-inethylpyridyliunz iodzde)ethylene (our).1,2-Bis-(4-pyridyl)ethylene(0.01 mol = 1.82 g) was dissolved in benzene (50 ml) and added dropwise to CHJ (0.06 mol = 8.52 g) in benzene (10 ml). The remaining procedure was identical with the previous example. Anal. Calcd for a monoiodide salt: I, 39.2. Found: I, 38.5. c. 1 ,d-Bis(4-N-nzethylpy~idyliuin iodide)ethane(Ds12). 1,2-Bis-(4-N-pyridyl)ethane(0.01 mol = 1.84 g) was dissolved in CHJ (25 ml). The solution boiled for 8 hr while stirring and was kept in the dark a t room temperature overnight. The white-yellow crystals were filtered off under suction, washed with ether, and dried in vacuo. Anal. Calcd for a diiodide salt: C, 35.91, H, 3.88, N, 5.98, I, 54,23. Found: C, 35.25, H, 3.6, N, 6.00, I, 53.1. d . 1,R-Bis-(4-N-methylpyridyliunz iodide)ethylene ( D J 2 ) . 1,2-Bis-(4-pyridyl) ethylene (0,Ol mol = 1.82 g) was dissolved in CHJ (25 ml). The solution was boiled for 8 hr while stirring, then kept in the dark a t room temperature overnight. The orange-yellow crystals were filtered off under suction, washed with ether, and dried in vacuo. Anal. Calcd for a diiodide salt: C, 36.04; H, 3.46; N, 6.02; I, 54.45. Found: C, 35.93; H, 3.46; N, 6.20; I, 53.9. e. 1-(4-Pyridyl)R-(4'-N-methylpyridyliurn TCNQ-)ethane (D,+TCN&-), 1-(4-Pyridyl) ,2- (4'-N-methylpyridylium iodide)ethane (0.00145 mol = 0.47 g) was dissolved by heating in 96% ethanol (10 ml) and dropwise added to a boiling solution of LiTCNQ (0.002 mol = 0.42 g) in ethanol (100 ml). LiTCNQ was prepared according to a described method.8 The reaction was carried out under nitrogen. The mixture was cooled to room temperature in the dark, then kept at 0' for complete precipitation of the dark blue needles. The mixture was filtered under suction and the crystals were washed with ethanol and ether and dried in vacuo. Anal. Calcd: C, 74.41; H, 4.75; N, 20.83. Found: C, 73.55; H, 4.88; N, 20.43. fa 1-(C-Pyridyl),2-(4'-N-methylpyridylium TCNQ-)ethylene (D,+TCNQ-) . I-( $-Pyridyl), 2-(4'-methylpyridylium iodide) ethylene (0.0145 mol = 0.47 g ) was dissolved in a mixture of 96% ethanol (10 ml) and HzO (1 ml) . The solution was added dropwise to a boiling solution of LiTCNQ (0.002 mol = 0.42 g) in ethanol (100 ml) , This was followed by the procedure of the preceding sample. Anal. Calcd: C, 74.8; H, 4.27; N, 20.94. Found: C, 73.79; H, 4.52; N, 21.14. g. 1 2-Bis (4-N-methylpyridyliunz TCNQ-)ethane (D2+(TCNQ-)2. This was prepared by a procedure similar to that for the preceding sample and using )
(4) A. Remhaum. W. Baumgartner, and A. Eisenherg, J. Polymer Sci.. B , 6, 159 (198s).
515
ELECTRONIC PROPERTIES OF TCNQ COMPLEXES an excess of LiTCNQ. A n a l . Calcd: C , 73.29; H, 4.21; N, 22.50. Found: C, 74.63; H, 3.88; N, 23.32. h. 1,2-Bis (4-N-inethylpyridyliu77a T C N Q - ) ethylene DU2+(TCNQ-)2. The same procedure was used as for sample g. A n a l . Calcd: C, 73.53; H, 3.90; N, 22.59. Found: C, 73.20; H, 4.33; N, 22.18. i. Preparation of Coiizplexes Containing Neutral TCNQ. Half a mole of neutral TCKQ was combined with 1 mol of each of the following: D,+TCNQ-, D,+TCNQ-, D,2+(TCNQ-)2, and DU2+(TCT\'Q-)2 by the procedure described below. To each of the iodides, DJ, D,I, D&, dissolved in hot ethanol were added simultaneously the calculated concentrations of LiTCNQ and neutral TCNQ dissolved by a mixture (1:1) of hot ethanol and acetonitrile. The operations were carried out in a nitrogen atmosphere. Precipitation began in each case while the mixture was still hot. After cooling, the products were filtered, washed with warm acetonitrile, ethanol, and ether, and dried in vacuo. 3. Preparation of Polymers and Their T C N Q Coinplexes. The polymers containing the model compounds in the chain have not been synthesized before. The kinetics of formation and properties of similar but aliphatic polymers, termed ionenes, are the subject of a separate publication.* The aromatic ionenes used in this study were prepared by reacting D, or D, with p-dibromoxylene in stoichiometric proportions in benzene or dimethylacetamide. Bromobenzene is also a convenient solvent for the polymerization reaction carried out a t looo. The precipitated polymers were filtered off, washed with benzene, and dried in vacuo, yield 98-99%. Analysis of polymer obtained by reaction of D, with p-dibromoxylene: Calcd for a unit segment of CzoHzoN2, Br2: C, 54.0; H, 4.52; N, 6.25; Br, 35.61. Found: C, 53.10; H, 5.12; N, 6.32, Br, 32.31. The polymer was soluble in water and decomposed on heating a t 260-290". Intrinsic viscosity in aqueous 0.1 M KBr = 0.21. Analysis of Polymer Obtained by Reactions of D, with p-Dibromoxylene Calculated for a U n i t Segmenl. Calcd for C20HlsN2Br2:C, 54.8; H, 4.05, N, 6.28; Br, 35.9. Found: C, 53.59; H, 4.73; N, 7.05; Br, 30.56. Intrinsic viscosity in aqueous 0.1 A4 KBr = 0.16. The TCNQ complexes were prepared by dissolving the polymers in 50% water-methanol mixtures, to which were added solutions of LiTCNQ in methanol. A mixed solvent, methanol-acetonitrile (20: 80), was used for the preparation of polymeric salts with neutral TCNQ. Analysis of Polymer Containing DBZ+ ( T C N Q - ) 2 Units. Calcd for a unit segment of C44H28NlO; C, 75.80, H, 4.02, X, 20.1. Found: C, 75.01; H, 4.07; N, 19.45. Analysis of Polymer Containing Du2f( T C N Q - ) 2 Units. Calcd for a unit segment of C44H28T\jlO: C,
76.1; H, 3.75; N, 20.2. Found: C, 76.25; H, 3.97; N, 19.5. The elemental analyses were carried out by means of a F & M model 185 CHN analyzer. 4. Xpectrophotonzetric Data. Visible spectra of TCNQ complexes were determined by means of a Cary Model 14 Spectrophotometer in spectral grade acetonitrile as solvent. The analysis of the spectra yielded good agreement with previously published data.* As shown in Figure 1, the TCNQ radical anion absorbs 1
-- TCNQ
-Lie TCNQ-
W
0
z a
+ TCNQ
-0;'
TCNQ-
-D,f
TCNQ- 4- TCNQ
3
3
3000
4000
5000
6000
7000
BOO0
9000
WAVELENGTH, A
Figure 1. Visible absorption spectra (10-6 M ) in acetonitrile.
strongly at 4200 A ( E 24,300) and at 8420 A ( E 43,300), while neutral TCNQ absorbs only a t 3950 A ( e 63,600). The intensity ratio of the 4200- and 8420-A bands is approximately 0.57 for simple salts, and that of the 3950- and 8420-A bands is approximately 2.0 for salts containing one neutral molecule of TCNQ per complex. The spectrophotometric analysis agreed with the proposed radical anion-to-neutral acceptor ratio to within about 5%. However, the elemental analysis agreed, on the average, with the theoretical percentages of C, H, and N to about 1%. 5 . Electrical Measurenaents. Conductivity measurements were made on 0.5-in. diameter cylindrical pellets in the absence of air. Pellets prepared under pressures between 20,000 and 100,000 psi had essentiVolume 73, Number 3 March 1969
A. REMBAUM, A. HERMANN, F. STEWART, AND F. GUTMANN
516
ally identical conductivities. Electrical contact was made with vacuum-deposited gold electrodes. Using this technique, resistivity values could be reproduced within a factor of 2. Silver paste or platinum disk contact yielded much less reliable results, De+(TCNQ-) complexes were found to be unstable above 50". However, complexes containing neutral TCNQ showed identical resistivity after heating a t 90" for 24 hr. In one case only, (Ds+TCn'Q-), was it possible to carry out measurements on a single crystal. The resulting activation energy, vix. 0.14 eV, was found to be the same (within experimental error) as that obtained with the compactions, vix. 0.13 eV. The conductivity measurements were carried out in an evacuated glass cell containing a copper-constantan thermocouple. The glass cell was immersed in a dewar vessel containing the heat bath. Resistivity as a function of pressure was measured using vanadium alloy steel anvils of 0.25 in. contact diameter, The details of the apparatus were previously described.6 6 . Electron Spin Resonance (Esr). Esr measurements were made by a previously reported proceduree on samples consisting of either small chips of compacted pellets, powdered complexes, or complexes dispersed in KCI. Pressure and grinding seemed to have no significant effect on the free spin concentration as derived from the esr intensity. 7'. Seebeck Coeficient. The Seebeck coefficient was measured using a Keithley 610. BR electrometer. The sample temperature was varied by adjusting the flow of thermostatically controlled gaseous nitrogen past the sample. The thermal gradient ( A T 5") was achieved by variation of the currents through two heaters in physical contacts with the surfaces of the sample. Temperatures were measured by means of copper-constantan thermocouples connected to a Rubicon Model 2732 potentiometer. 8. Hall Effect. A schematic diagram of the Hall effect apparatus is shown in Figure 2. The bucking circuit allowed reduction of the misalignment voltage to the order of less than 0.5 mV. Resolution to some
-
I
I
extent was dependent upon sample stability. In some cases, several days of waiting was required to achieve noise and drift levels low enough to measure mobilities of the order of 0.01 to 0.1 cm2/V sec. The maximum magnetic field strength used was G kG. The VTVM employed was a Hewlett-Packard 412A or a Keithley Model 610A electrometer. An RC filter on the output of the VTVM decreased high frequency noise. The strip-chart recorder was a Varian Model G10. The accuracy of the apparatus was confirmed by the use of resistive mock-ups and an n-type germanium sample of known mobility as a reference.
111. Results 1. Resistivity (p) and Activation Energy ( E ) . The resistivities in ohm-centimeters and activation energies in electron volts of various TCNQ complexes (obtained from the equation p = poexp(E/LT) over temperature intervals ranging from 77 to 3OOOK) are listed in Table I. Table I: Resistivity ( p ) and Activation Energy (E) of TCNQ Complexes ( E is Deduced from p = po exp(E/kT)) p at 2bD,
ohm cm Ds+TCNQD.Z+(TCNQ-) e D,Z+(TCNQ-)n* gTCNQ DBa+(TCNQ--)iTCNQ DP(TCNQ-)2 polymer D,Z+(TCNQ-)2TCNQ polymer D,+TCNQ- (polycrystalline) D,+TCNQ- (single crystal) Did+(TCNQ-)2 Du2+(TCNQ-),TCNQ Df+(TCNQ-)nTCNQ Dust (TCNQ-) 2 polymer Duz+(TCNQ-)2TCITQpolymer
1.55 X 106 4 x 106 4.7
3.4 1 . 5 X lo6
92 I.I
x 104 ... 2 . 1 x 106
E . eV
0.11
...
0.04 0.035 0.15 ..,
0.13 0.14 I
,
,
13 3.8
0.04 0.035
5.2 X 106 80
0.15 I , ,
Average values are quoted with deviations being less than 30% for p and 10% for E. Such deviations are presumably due to variations in intergranular resistances.' The addition of neutral TCNQ dramatically lowers the resistivity and the activation energy presumably due to increased electron delocalization. The low activation energy of these latter complexes indicates the onset of metallic conduction. The variation of resistivity with pressure of the D,2+(TCNQ-)2TCNQ complex is shown in Figure 3. The curves were obtained on first increasing the pressure to 36 kbars then decreasing it to 1 kbar and repeating this procedure several times. Similar curves (6) A. Rembaum, J. Moacanin, and H. A. Pohl, Progr. Dielectrics, 6, 41 (1965). ( 6 ) B. E. Stewart and A. Rembaum, J. Macrom. Sci. (Chemistry)
AI, 1143 (1967).
Figure 2. Schematic diagrams of the Hall apparatus. The Journal of Physical Chemistry
(7) R. G. Kepler in "Phonons and Phonon Interactions," T. A . Bak, Ed., W. A. Benjamin Inc., New York, N. Y . , 1964, p 679.
517
ELECTRONIC PROPERTIES OF TCNQ COMPLEXES
II
0
I
I
I
2
I 4
I
3
I
I
6
5
I
7
PRESSURE”2, kbor”z
Figure 3. Resistivity vs. square root of pressure for D,+(TCNQ-)TCNQ: 0, first cycle; 0 , second cycle; X, third cycle.
a t which I is a maximum, We find for the D,+TCNQand D,+TCNQ- simple salts that T, is considerably above ambient temperature and in the range in which these salts begin to decompose. Therefore an alternate procedure for determining J was attempted. In the temperature range for which J >> k T , the product I X T is proportional to exp( -J/lcT), and a semilog plot of 1 X T lis. T-I will be a straight line the slope of which is (Jllc) log e. From such a plot (Figure 4) the measured value of J was found to be not sufficiently large in comparison with kT to make the above approximation valid. We estimate that the measured value of J is too small by a factor of approximately 2. It can only be concluded that J is probably somewhat smaller than 0.1 eV. TEMPERATURE, O C
were also obtained with all the other complexes described here. It seems worth noting that there appears to be little difference between transport properties of the polymer complexes and complexes with the corresponding monomers. Thus the polymer complexes are only slightly less conductive than those of the monomer, Also the conjugation between the pyridine rings does not increase the conductivity to any significant extent. Furthermore, TCNQ polymers with a purely aliphatic backbone exhibit similar conductivity characteristics as those containing conjugated molecules.* W. Electron Spin Resonance. The spin concentration of four representative samples and the singlettriplet separation energy J estimated from the temperature variation of signal intensity is shown in Table 11.
6
i 4
\
.E 5f
3
I
F
‘B
2
K X
c
v)
E b
IO0
e 4
3
Table 11: Temperature Dependence of Unpaired Spins Esr intensity at 2P,
spins/
mole x 10-1’
D,,+TCN Q-
0.35
D,+TCNQ-
1.5
D,+TCNQ’ -I- TCNQ D,+TCNQ- 4-TCNQ
II - 7 8.0
io3/7,
Singlet-triplet separation energy J. eV NO. 03 NO. 045
0 0
Each of the four complexes studied here contains approximately Avogadro’s number of unpaired electrons per mole a t room temperature (Table 11). Relative spin concentrations quoted are considered accurate to 30%; absolute values are known only t o within a factor of 2. Simple TCNQ salts follow a Ringlet-triplet model for paramagnetisms in which the esr intensity I is proportional to T-l[exp(J/kT 3)-l]. The singlettriplet separation energy J may be calculated from the formula J = 1.61kTm, where T, is the temperature
+
OK“
Figure 4. Temperature dependence of ear intensity for simple TCNQ salts.
In the low-temperature region of Figure 4 the triplet contribution to paramagnetism is small and the doublet impurity contribution predominates. In this region I X T is approaching n constant value independent of temperature. Complex TCNQ salts containing neutral TCNQ exhibit a Pauli spin paramagnetism characteristic of a degenerate electron gas such 11s occurs in metals.’ In this case the measured esr intensity is independent of temperature. Such a paramagnetism was found for some of our complex salts (Table 11). The shape of the esr signal of the complexed salts exhibits an interesting rotational anisotropy that to (8) D. Bijl, H. Kainer, and A. 0.Rose-Innes, J. Chem. Phys., 3 0 , 765 (1959).
Volume Y9, Number 9 March 1069
A. REMBAUM, A. HERMANN, F. STEWART, AND F. GUTMANN
518
latter case (PDC), the exciton was shown to be a Kannier exciton (as opposed to Frenkel excitons in TCNQ complexesg). Assigning anisotropic g values to our spectra on the above basis, one obtains g = 2.0033, g = 2.0024 for D,+TCNQTCKQ (Figure 5 ) ) and g = 2.0036, g = 2.0028 for D,+TCNQTCNQ. This g-factor anisotropy is thought to be due to inequivalent sites in the unit cell.ll Attempts were made to observe anisotropy in the electronic conductivity of compressed pellets, However, no conductivity anisotropy was found. 3. Carrier Mobility Measurements. The Hall E$ect and Magnetoresistance. I n order to elucidate the conduction mechanism it was considered essential to determine the mobility of the electrical carriers. This was achieved directly by Hall effect measurements.
our knowledge has not been reported previously. If a pellet is rotated in the magnetic field in such a way that the direction in which the pressure was applied to it, fi, is kept perpendicular to the magnetic field H, no line-shape change results; however, a different spectrum is observed when the pellet is oriented with fi parallel to & (Figure 5 ) . This type of anisotropy PRESSURE AXIS PERPENDICULAR TO MAGNETIC
PRESSURE AND FIELD AXIS
+
PRESSURE AND FIELD AXIS PARALLEL
COMPRESSED PELLET DENOTES DIRECTION IN WHICH PRESSURE WAS APPLIED
P
+ -1
5G
7
!-
r l O G
%INCLINED 45*T09
ello
+
Table 111: Electronic Properties of D, and D, Complexes SINGLE CRYSTAL (MONOCLINIC)
QL
%DENOTES LONG AXIS DIRECTION
Sign of
at 25O
coemcient
cmz/ V sec
Sample
Do+(TCNQ-) D "+ (TCNQ-) D,a+(TCNQ-)z D,?" (TCNQ-) z
does not occur in the siiiiple salts, The coiiiplex salts in a powdered form show a single line resonance. The /I ?I (Figure 5) is esr signal shape for pellets with similar to that of the same compound in the form of n single crystal, having axial symmetry. Compacted pellets, usually composed of many microcrystals randomly oriented, are not known t o exhibit rotational anisotropy. It would appear that under pressure either the molecules of our complex salts or small crystallites assume a preferred orientation. The spectral splittings in the case of the single crystals (or compressed pellets) are likely to be due to g-factor anisotropy. It is, however, somewhat difficult to rule out the possibility of zero-field splittings in the absence of measurements at different frequencies (which would check the dependence of the splittings on magnetic field strength), especially in view of the identification of zero-field splittings in other organic TCNQ complexes.9 However, the splittings in the present case are of the order of 1 G, compared t o splittings of -100 G for other TCNQ comple~es.~Furthermore, recent studies on the magnetic excitations in chargetransfer complexes of p-phenylenediamine-chloranillo (PDC) yielded esr spectra very similar to those shown in Figure 5 . In the case of PDC, experiments at different frequencies firmly established the g-factor anisotropy (rather than zero-field splittings) . In the The Journal of Physical Chemistry
p ~ ~ l i
Hall
D,2+(TCNQ-) 2 p oIy mer Du2+(TCNQ-)2 polymer
at 25n
+
f a I
,
a , .
(
Q, Seebeck coeff, mV/OO
Q X T.
at 25O
eV
a’)
All terms used are defined in the Appendix. The Journal of Physical Chemistry
For the diquaternary ammonium bolaform ions Rice found that the Peterlin? investigated by FUOSS,~ model of beads separated by massless rods was superior t o the rigid ellipsoid model. Using the rigid bolaform ions such as the 4,4’-biphenyldisulfonate ion, BPDSZ-, Atkinson4 found that the rigid ellipsoid model enabled one to accurately calculate the frictional coefficient of BPDS2- from the parameters of the benzenesulfonate ion, BS-. Both of these results seem valid since, in the case of the diquaternary ammonium salts a nonrigid polymethylene chain separates the charge sites, but a rigid aryl framework lies between the charge sites of the BPDS2- ion. Also, Rice, Thompson, and Nagasawa*have measured the diffusion coefficients of IGBDS, IGBPDS, and K2TPDS and found that the Perrin rigid ellipsoid gave a very accurate description of that property. I n 1959, Fuossg had proposed a method of getting a (1) Taken in part from an M.S. thesis submitted t o the Graduate School of the University of Maryland; National Bureau of Standards, Washington, D . C. 20234. (2) B. R. Staples and G. Atkinson, J . Phys. Chem. 71, 667 (1967). (3) S. A. Rice, J. Amer. Chem. Soc., 80, 3207 (1958). (4)G.Atkinson and S.Petrucci, J . Phys. Chem., 67, 1880 (1963). (5) F. Perdn, J. Phys. Rad., 7, 1 (1936). (6) Q. V. Brody and R. M. Fuoss, J. Phys. Chena., 60, 156 (1956). (7)A Peterlin, J. Chem. Phys., 47, 6 and 669 (1950). (8) G. Thomson, 9. A. Rice, and 41. Nagasawa, J . Amcr. Chem. Soc., 85, 2537 (1963). (9)R. M . Fuoss, Proc. N a t . A c a d . Sci. U.S., 45, 807 (1959).