Thermal and magnetic study of exchange in the quasi-1-D molecular compound , TTF ⋅ PtS 4 C 4 ( CF 3 ) 4

Field‐dependent differential susceptibility study of TTF‐AuS4C4(CF3)4: Universal aspects of the spin‐ Peierls phase diagram J. Appl. Phys. 53, 2725 (1982); 10.1063/1.330945 An EPR study of interlayer exchange coupling in the quasi‐two‐dimensional salts, (C n H2n+1NH3)2CuCl4, with n = 1, 2, and 3 J. Chem. Phys. 74, 6018 (1981); 10.1063/1.441042 Solid solution and modulation in the molecular solid (TTF)2(Ni4C4H4)3 AIP Conf. Proc. 53, 205 (1979); 10.1063/1.31860 The molecular and crystal structure of squaric acid (3,4‐dihydroxy‐3‐cyclobutene‐1,2‐dione) at 121 °C: A neutron diffraction study J. Chem. Phys. 67, 4825 (1977); 10.1063/1.434686 [This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 131.128.120.39 On: Mon, 27 Jul 2015 21:19:22 THERMAL AND MAGNETIC STUDY OF EXCHANGE IN THE QUASI-I-D MOLECULAR COMPOUND, TTF.PtSqCq(CF3)q* J. C. Bonner University of Rhode Island, Kingston, RI 02881 T. S. Wei University of Pennsylvania, Philadelphia, PA 19174 H. R. Hart, Jr., L. V. Interrante, I. S. Jacobs, and J. S. Kasper General Electric Corp. Res. & Dev., Schenectady, NY 12301 G. D. Watkins Lehigh University, Bethlehem, PA and 18015 H. W. J. BUite Kamerlingh Onnes Laboratory, Leiden, The Netherlands


INTRODUCTION
Quasi-one-dimensional (1-0) systems have attracted much attention for their unique physical properties.We have prepared [1) a family of rr-donor-acceptor compounds of the type TTF+.MSqC4(CF3)4, where TTF+=tetrathiafulvalinium and MS4Cq(CF3)4 is a series of bisdithiolene metal complex anions with M=Cu, Au, Pt and Ni.They are magnetic insulators with much variety in magnetic properties [2].The TTF+ ions are spin-l/2 paramagnetic (g = 2) while the MSqCq(CF3)4 ions are diamagnetic for M=Au, Cu but paramagnetic for M=Ni, Pt (S=1/2, g = 2).In the M=Au, Cu compounds, a system of anti ferromagnetic (AF) chains of TTF+ spins displays a spin-Peierls transition [3,4].We examine the contrasting M=Pt compound, whose powder susceptibility [5] indicates dominant ferromagnetic (FM) subsystems with residual AF coupling at low temperatures.
We report single crystal susceptibility (X) and specific heat measurements (~).The magnetic contribution to the specific heat (em) is estimated using earlier results for the M=Au compound [6].We explore the limits of an analysis of the combined results for the M=Pt compound using a simple model which ignores differences between the two types of spin carriers.A system of FM chains is treated with "exact" numerical models while the AF interchain exchange is evaluated in a mean field approximation.It proves possible to describe the properties within reasonable limits, in terms of a dominant FM exchange intermediate between the isotropic Heisenberg and the extreme uniaxially anisotropic Ising limits, but closer to the former.

II. EXPERIMENT AND RESULTS
The synthesis and structure of TTF-PtSqCq(CF3)4 have been reported [1,7].All of the compounds of the family crystallize to form stacks in which the planar T~ and MS4Cq(CF3)4 ions alternate.The M=Pt, Cu and Au compounds are isostructural, with triclinic symmetry, resembling a distorted NaCl arrangement [4].
In our earlier work, the molecular stacking axis was suqqested as the exchanqe path both for the AF chains of the M=Cu, Au compounds and for the FM chains of the M=Pt compound.Each compound shows evidence [2] of an abrupt but magnetically undramatic change just below room temperature.Delker [8] studied its structural aspects for M=Cu.He showed that nearly equivalent diagonal directions of the room temperature facecentered "NaCl" structure became rather different below the transition.Its importance became apparent with the observation [9] that the predicted low temperature dimerization in the Cu compound occurs along the shorter of these diagonals.Thus, the AF interaction of the M=Cu, Au compounds is along this diagonal and appears to result from direct eXchange between TTF+ spins.
No similar study of the M=Pt compound has been carried out.It is, however, reasonable to assume that this structural feature persists and, with it, the AF exchange.The simplest assumption for the FM interaction is that it occurs along the molecular stacking axis.
Magnetic susceptibility measurements were carried out on a 3mg crystal using the Faraday method with techniques described elsewhere [4].Measurements were made with the field along the stacking axis b, and along two other principal directions, !* and b, of the face-centered cell.In Fig. 1, we show these data.The solid line is calculated from the model as described in Section III.Results for the three directions have been normalized arbitrarily at lOOK because of a variation (up to 19%) in magnetometer response When the sample is remounted for different orientations.The positive deviation from a Curie-law above 40K demonstrates the FM interaction, while the drop in Xb below -12K indicates some AF coupling as well as magnetic anisotropy.In ref.
[2J we raised questions as to the "completeness" of the apparent transition to a 3-D AF ordered state at the apparent inflection point in Xb at 8K.These questions remain.
The specific heat was measured utilizing a calorimeter described elsewhere [10].Data were taken on a compacted sample (464mg) using a heat-and-drift cycle time of 25 or 125 sec.In Fig. 2 we show the C data for the M=Pt compound, along with the data for Pthe isostructural M=Au member [6].A broad anomaly be- Some epr measurements (f=20GHz) were also performed, mainly at 20K.A single absorption line is seen with g-factors of 1.91~.05along the ~-axis and extremes of ->-2.05+.05 and 2.10+.05 in the plane perpendicular to c.The considerable width at half maximum of -1800 Oe, is responsible for the large error limits on the g-factors.This 10% g-factor anistropy is midway between the 20% anisotropy observed on the anion complex in a frozen glass [11] and the <0.5% anisotropy of the TTF+ ion [4] indicating coupling between the different spin carriers.It has also been suggested [12] that anisotropic ionic g-factors can introduce anisotropy in exchange of order /',g/g or (/',g/g)2.

III. SUSCEPTIBILITY ANALYSIS
The reduced susceptibility Xr for interacting chains may be written in standard mean field approximation where, X is the isolated chain susceptibility and o,r AAF is the mean field interaction parameter.The reduced susceptibility is dimensionless, i.e., X ~ X J F /Ng 2 "B 2 , where J F is the FM intrachain o,r 0 exchange, N is the number of spins per mole, and the other symbols have their usual meaning.
We com~ared two double log plots, one of the measured (X m )-vs.T and the other of (X r )-1 vs. reduced temperature kT/J F .For given Xo,r we construct curves with differing AAF and look for the best superposition of model and measurement.The "T" scaling yields JF/k and that of X-I determines g.The "best fit" A corresponds to an interchain coupling J AF where z is the number of significant neighbors.Inasmuch as we are considering anisotropic exchange (for S=1/2, there can be no crystal field anisotropy), we compare the experimental Xb with the theoretical XII .
If y is the anisotropy parameter in the Heisenberg-Ising Hamiltonian [13], y=l defines the Heisenberg and y=O the Ising exchange.We have extended earlier calculations [14,15] to produce X o .r extrapolations for various y values for infinite length systems based on rings up to 13 spins.
For each y, we can narrow the graphical comparison to determine a "best fit" J F and AAF pair, estimated to 10%.These are given in Table I.The resulting fit is, however, equally satisfactory irrespective of the y value!The method is indeterminate.In most other studies, one of the limiting cases y~O or y=l is assumed, often without apparent justification.For the Pt compound the X and g-factor anisotropy suggest some intermediate range of y.  the choice to the region close to y=0.9.The solid line of Fig. 1 presents this case with the other parameters of Table I.
The g-factors of Table I are relatively independent of y, but are low compared to the epr g-factors.Some of the discrepancy may reside in the experimental normalization which could raise values by about 0.17.Other possible sources are the crudeness of the meanfield approximation when I J AF \ is not small in comparison with J F , as well as the simplification of ignoring differences between the different spin carriers and their respective interactions.

IV. SPECIFIC HEAT ANALYSIS AND DISCUSSION
The usual problem with specific heat results is that of reliably separating the magnetic and lattice contributions, i.e., C = C + C. The lattice subtraction problem here p is m L formidable in view of the large lattice contribution due to complexity of the structure and the relatively high transition temperatures [4,6].In the case of atomic lattices that are not extremel~ anisotropic, a low temperature limiting form C = ST (following, e.g., from a simple Debye model)Lis valid for a limited temperature range.With low symmetry lattices problems arise in formulating suitable extensions of the 3-D Debye model, and much effort has been expended in methods of analysis (16).Further, in our case of molecular solids, low-temperature intramolecular excitations severely limit the range of validity of the basic T3 approximation [17].
A second, and very successful empirical approach in applicable cases, is the uSe of a diamagnetic isostructural (nearly isomorphous) analog.A corresponding state procedure [18J is used to make a small mOdification of C of the diamagnetic analog, which then represents tkat of the magnetic system under investigation.This procedure consists of writing C~= f(T/b) , where f is one unknown function for an isostructural family and b is a different scaling temperature for each compound ..
In this family we lack a diamagnetic analog because the TTF+ cation is always paramagnetic.However, the availability of C results for M=Au [6], the relatively  I.
low temperature at which the second-order transition occurs [4J, and our success in describing the magnetic properties above that transition, yield a very good approximation to the desired analog.We subtract the theoretical C for a system of AF S=1/2 Heisenberg chains [13-15¥ evaluated for \J 1/k=34K appropriate to the Au compound, and thus AF obtain CL(Au) for T > 4K which we then use to estimate C L (PE).The results for C (Pt) are very sensitive to the scaling factor b, as s~own in Fig. 3 the raw data show considerable scatter at temperatures above about 10K.Each curve shows a broad anomaly near 8K which corresponds in temperature to the apparent inflection point in X b .It.appears.t? lack the sharpness which usually appears ~n trans~t~ons to ordered magnetic states.This rounding is not due to chemical impurity or lack of data points.
Theoretical curves for C obtained numerically by extrapolation techniques, simTlar to those used for X, for a variety of anisotropy values between y=l (Heisenberg limit) and y=O (Ising limit) are shown in Fig. 4. The parameters are for the various "x-acceptable" combinations [y,J 1 listed in Table I.In contrast to previous analyses~ we concentrate on intermediate anisotropy.The form of the theoretical curves shows considerable variation with anisotropy, particularly for y>0.85.It is clear that fitting theoretical and experimental C curves can narrow down the acceptable ranges of [y,JmJ and thus remove the indeterminacy of the x-analysis: An important restriction is that J F .be adjusted so that the low temperature results perm~t an entropy matching at high temperatures.This establishes a "c -acceptability" criterion which appears as an upper li~it curve On the JF-y graph of Fig. 5, along with the x-acceptable range.The crossing of these two sets of criteria in the region 0.85 < Y < 0.9 indicates where our final best fit of theory to experiment must  (theory) for y=0.9, ~nd J F /k=66K for TTF'PtS4~4(CF3)4 lie.The "best" curve corresponds to y=0.9, JF/k = 65K and is shown in Fig. 6 in comparison with Cm(Pt) for b=53.
Error bars arising from scatter in C data are indicated for the higher temperature P region, while much smaller ones pertain at lower temperatures.The crossing of the theoretical and experimental curves down near 4K is not understood.It may be related to residual effects of the 2.1K transition in the M=Au compound.
From the relationship between AAF' J AF , J F and z, we roughly estimate I JAF'/k to be 11K (with z=2) or about one-third the value found in the isostructural M=Au, Cu compounds.This may reflect a different degree of distortion for the Pt compound and/or the average of several (possible competing) interstack exchange interactions.
We have thus obtained estimates of the parameters describing both the susceptibility and magnetic specific heat of TTF.PtS4C4(CF3)4' within the simple model proposed.While the numerical values cannot be considered to be very precise, they serve as guides until more sophisticated theoretical approaches can be developed in conjunction with future data using such microscopic probes as'epr, nmr and neutron diffraction.The C measurements were undertaken because of uncertaint~es in the interpretation of the experimental X• The C m analysis has been highly successful in determining the degree of Heisenberg-Ising uniaxial anisotropy, and may prove valuable for examining similar systems.
. A family of potential C (Pt) curves is displayed which result from trying vari~us possible C L (Pt) curves characterized by b values in the range 53-56.(Note that b(Au)=55 is the same as G in Ref. 6).The relative smoothness of these curves ~esults from a polynomial fitting:

TABLE I "
Best X Fit" Exchange Parameters for Various Anisotropies, y, in TTF•PtS4C4(CF3)4