Journal of Alloys and Compounds 338 (2002) 173–184 L

www.elsevier.com / locate / jallcom

Synthesis and characterization of Cu 2 ZrCl 6 : a thermochromic,

van Vleck paramagnet
a b b a,
Andrew M. Dattelbaum , Liang He , Frank Tsui , James D. Martin *
a
Department of Chemistry, North Carolina State University, Raleigh, NC 27695 -8204, USA
b
Department of Physics and Astronomy, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA
Received 14 November 2001; accepted 29 January 2002

Abstract

The solid state reaction of CuCl and ZrCl 4 yields the pale yellow crystalline solid of composition Cu 2 ZrCl 6 . This material crystallizes
in the trigonal space group P3¯m1 ([164) at room temperature with lattice constants a510.9092(4) and c56.1517, Z53. The structure is
shown to be related to ZrCl 3 , where every other Zr site has been replaced with Cu atoms that partially occupy the six tetrahedral
interstices surrounding the now vacant octahedral interstice. At room temperature, two copper atoms are disordered about three
crystallographically distinct sites. Two thermochromic transitions are observed for this material at 120 and 673 K, which appear to be
related to the movement of copper from the center of a tetrahedral interstice to a trigonal planar face. A second-order Jahn–Teller
distortion accounts for this structural variation. This along with the insertion of the empty Zr-4d orbitals into the already small copper (I)
based HOMO–LUMO gap afford significant mixing between ground and excited states resulting in notable van Vleck paramagnetism.
 2002 Elsevier Science B.V. All rights reserved.

Keywords: Metal halide compounds; Crystal structure; X-ray diffraction; Neutron diffraction; Magnetic measurements

1. Introduction frameworks from the phosphate-like [CuCl 4 ] 32 building

blocks with the Lewis acidic ZrCl 4 [3,5]. Unlike our prior
The chemistry of copper (I) is unique among transition work with the group 13 Lewis acids (Al and Ga), the
metals because of its distinctive electronic configuration. introduction of Zr(IV) into the cuprous halide lattice is
With respect to reactivity, the high lying and filled 3d- expected to further narrow the band gap of these wide
orbitals allow reactivity with electrophiles; at the same band gap semiconductors by inserting the t 2g -type zir-
time the low lying and empty 4s- and 4p-orbitals provide a conium 4d-orbitals into the already narrow gap between
site for nucleophillic attack. This relatively small HOMO- the Cu 3d- and 4s,p-orbitals. In prior reports we have
LUMO gap is also responsible for the wide range of described the synthesis and characterization of
coordination geometries observed for Cu(I) [1]. Though ((C 6 H 6 )Cu) 2 ZrCl 6 [6] and [H 2 NMe 2 ]CuZrCl 6 [7]. We here
this d 10 -metal may have a crystal field preference for a describe the synthesis and characterization of the parent
tetrahedral geometry, a second-order Jahn–Teller mixing compound Cu 2 ZrCl 6 . We further demonstrate that the
between the filled d- and empty s- and p-orbitals stabilizes interaction between the d 10 -Cu(I) and d 0 -Zr(IV) is respon-
two and three coordination [2]. In addition to unique sible for the observed thermochromism, and provides a
structure and reactivity, the electronic configuration of system in which to probe the nature of van Vleck paramag-
Cu(I) has profound implications with respect to optical and netism.
luminescent properties [3], as well as with respect to
mobility in ionic conductors [4]. In the course of our
continuing investigation of the structure and reactivity of 2. Experimental
Cu(I) centers in Lewis acidic metal halide lattices, we have
been interested in the possible construction of open 2.1. General procedures

*Corresponding author. All reactions were performed under an inert atmosphere

E-mail address: jdmartin@ncsu.edu (J.D. Martin). of dry N 2 in a glove box or using Schlenk line techniques.

0925-8388 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved.
PII: S0925-8388( 02 )00230-X
174 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184

ZrCl 4 (99.99%) was used as received from Aldrich. The was subsequently confirmed by the structural solution
CuCl was prepared from Cu metal and CuCl 2 (Aldrich) refined using the NRCVAX suite of programs [10]. All
according to literature methods and further purified by zirconium and chlorine atoms were located by direct
sublimation [8]. Powder X-ray diffraction measurements methods using the SIR92 program [11]. One copper site
were obtained using an Enraf-Nonius Guinier camera and (Cu(1)) was also observed in the initial direct methods
were indexed with respect to a silicon standard. Diffuse solution. Subsequent difference maps identified two addi-
reflectance spectra were taken using a Cary 3e UV–Vis tional sites that are partially occupied by copper atoms. All
spectrometer equipped with an integrating sphere and atoms were refined anisotropically. A full matrix least
measured with respect to a pressed polytetrafluoroethylene squares calculation using the 452 unique reflections (I.
powder standard. Reflectance spectra are plotted as the 2.5s (I)) gave a final refinement with R factors of R50.053
remission function F(R ` )5(12R ` )2 / 2R ` , based on the and R w 50.087.
Kubelka–Munk theory of diffuse reflectance [9]. Differen-
tial scanning calorimetry (DSC) data were collected on a
2.4. X-ray structure determination of Cu2 ZrCl6 at 157 K
Perkin-Elmer DSC 7 at a heat / cool rate of 108 / min.
Samples were placed into stainless steel, high-pressure
The crystal used for the room temperature data collec-
pans and hermetically sealed with gold foil caps. Magnetic
tion was removed from the capillary under a flow of argon,
susceptibility measurements were performed on a Quantum
covered in silicone grease, placed on the end of a glass
Design MPMS-5S SQUID magnetometer with a 5 T
fiber, and immediately transferred to the diffractometer
magnet in fused silica magnetic susceptibility containers
where it was cooled to 157 K under a stream of dry
and in DELRIN sample holders.
nitrogen. Data were then collected on an Enraf-Nonius
CAD4-Mach diffractometer with monochromated Mo(Ka)
2.2. Preparation of Cu2 ZrCl6 radiation. Lattice constants for the trigonal cell, a5
10.906(2) A,˚ c512.234(1) A, ˚ were determined from a
A sample of 0.200 g of ZrCl 4 (0.858 mmols) and 0.140 symmetry-constrained fit of 24 well-centered reflections
g of CuCl (1.72 mmols) was ground together in a mortar between 338,2u ,368 and their Friedel pairs. A unique
and pestle, then transferred into a fused silica tube (6 quadrant h, k, 6l, was collected with 2429 independent
mm3 8 cm I.D.). The reaction vessel was evacuated and reflections by v -scans for 08,2u ,608. These data were
sealed with a torch. The reaction was heated to 450 8C in a scaled to three intensity check reflections using a five-point
box furnace for 1 day, and then cooled at 1.0 8 / min to smoothing routine. An empirical absorption correction was
room temperature. A small temperature gradient in the applied using psi scan data. Systematic absences were
furnace caused the Cu 2 ZrCl 6 to crystallize on the cool end found to be consistent with the space group P3¯c1, which
of the tube. The bulk material was characterized by powder was subsequently confirmed by the structural solution
X-ray diffraction, DSC, magnetic susceptibility and diffuse refined using the NRCVAX suit of programs. In spite of
reflectance spectroscopy. the doubled c-axis and change of space group, refinement
of the low temperature data yielded essentially the same
structure solution as that found for the room temperature
2.3. X-ray structure determination of Cu2 ZrCl6 at 298 K
data, with a similar distribution of the disordered copper
atoms but a small variation in certain atom positions. All
A pale yellow single crystal (0.2430.2230.12 mm)
atoms were refined anisotropically. A full matrix least
covered in silicone grease was mounted in a Pyrex
squares calculation on 877 unique reflections [I . 1.5s (I)]
capillary and sealed under a N 2 atmosphere. Diffraction
gave a final refinement of R511.1 and R w 510.8.
data were collected on an Enraf-Nonius CAD4 diffrac-
tometer at 298 K using MoK a radiation ( l50.71073 A). ˚
Lattice constants for the trigonal cell, a510.9892(4) A, ˚ 2.5. Neutron powder diffraction
c56.1517(2) A, ˚ were determined from a symmetry con-
strained fit of 24 well centered reflections (338,2u ,358) Cu 2 ZrCl 6 (6.0 g, 0.014 mol) was sealed in a vanadium
and their Friedel pairs. A zero-layer Weissenberg photo- container of length 50 mm and diameter 6.0 mm (I.D.)
graph and axial photographs taken on the CAD-4 con- inside a dry, He-filled glove box. Data were collected at
firmed these lattice constants and gave no indication of any room temperature, and at 100 K using a closed-cycle He
superstructure. A hemisphere of data, 6h, k, 6l, with 2878 refrigerator for temperature control. Neutron powder dif-
independent reflections was then collected by the u / 2u fraction data were collected using the BT-1 32 detector
scan mode for 08,2u ,548. Data were scaled to the neutron powder diffractometer at the NIST Center for
intensity check reflections using a five-point smoothing Neutron Research reactor, NBSR. A Cu(311) mono-
curve fitting routine. An empirical absorption correction chromator with a 908 take-off angle, l 51.5402(2) A, ˚ and
was applied using psi scan data. Systematic absences were in-pile collimation of 15 min of arc were used. Data were
found to be consistent with the space group P3¯m1, which collected over the range of 3–1688 2u with a step size of
 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184 175

0.058. The instrument is described in the NCNR web site

(http: / / www.ncnr.nist.gov).
The structure was also evaluated by the Rietveld refine-
ment of the room temperature neutron diffraction data
using GSAS [12]. Lattice constants and atom positions
obtained from the CAD-4 data were utilized as starting
positions for the refinement. Doubling of the c-axis was
also considered, but it did not improve the overall fit of the
data. The full profile refinement parameters include the
zero point, 12 background parameters, three profile param-
eters plus the structural parameters. The refined structural
parameters include two lattice parameters, the 15 atomic
position parameters not dictated by special positions, the
three isotropic thermal parameters for the chlorides, and
Fig. 1. Final Rietveld refinement plot for room temperature neutron
the three fractional occupancies for the copper positions. powder data for Cu 2 ZrCl 6 . The experimental data are represented by
Pseudo-translational symmetry led to a high correlation of crosses, while the calculated pattern is shown by the solid line. The lower
the metal thermal parameters, and thus, they were all fixed trace is the difference curve between observed and calculated patterns.
to Uiso 50.03. The final agreement factors are R p 56.04%, The Bragg reflections are shown by the vertical bars.
R wp 57.07% for the full profile.
positions and selected bond distances and bond angles
from the room temperature single crystal refinement are
3. Results given in Tables 2–4, respectively. (The corresponding data
for the low temperature and neutron refinements are
3.1. Solid-state structure available from the author.) ORTEP drawings describing
the local structure around each of the three copper sites are
The solid-state structure of Cu 2 ZrCl 6 was determined by given in Fig. 2. The structure of Cu 2 ZrCl 6 consists of a
single crystal X-ray diffraction at room temperature and at complex network based on a hexagonal close-packed
157 K, and by Rietveld refinement of room temperature chloride sublattice. Two crystallographically distinct zir-
neutron powder diffraction data. A summary of these conium cations reside on 3m and 3¯m sites, respectively,
crystallographic data is presented in Table 1 and the together filling 1 / 6 of the octahedral interstices. The Zr–Cl
difference function of the Rietveld refinement of the distances around the 3m site, Zr(1)–Cl(1)52.481(3) A33 ˚
neutron powder diffraction data is given in Fig. 1. Atomic and Zr(1)–Cl(3)52.478(3) A33, ˚ and the 3¯m site, Zr(2)–

Table 1
Crystallographic data for Cu 2 ZrCl 6 at 298 K and 157 K (single crystal X-ray), and 293 K (powder neutron)
Formula Cu 2 ZrCl 6
Formula weight (g / mol) 431.22
Crystal dimensions (mm) 0.2430.2230.12 Powder
Crystal system Trigonal
Temperature (K) 298 157 293
Space group (No.) P3¯ m1 (164) P3¯ c1 (165) P3¯ m1 (164)
˚
a (A) 10.9892(4) 10.906(2) 10.9648(3)
˚
c (A) 6.1517(2) 12.234(1) 6.1411(4
V (A˚ 3) 643.37(3) 1260.2(2) 639.41(4)
Z 3 6 3
R-merge 0.020 0.053
rcalcd (mg cm 23 ) 3.338 3.386 3.358
l(Mo Ka), A ˚ 0.71073 0.71073 1.5405 (neutron)
m (cm 21 ) 78.8 80.4
No. reflns measured 2878 2429
No. reflns unique 537 1218
Ra 0.053 0.111 R p 0.061 d
Rw b 0.087 0.108 wR p 0.071 e
GOF c 2.87 2.09
a
R 5 o(F0 2 Fc ) /F0 .
b
R w 5 [o(w(F0 2 Fc )2 ) /wF 20 )] 1 / 2 .
c
GOF5[o(w(F0 2 Fc )2 ) /(No. of reflns. 2 No. of parameters)] 1 / 2 .
d
R p 5 ouIo 2 Ic u / oIo .
e
wR p 5 œ(M / oI 2o ).
176 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184

Table 2
Atomic coordinates and isotropic displacement parameters for Cu 2 ZrCl 6 (1) at 298 K
Occupancy x y z Biso
Zr(1) 1.0 1/3 2/3 0.0244(2) 1.64(5)
Zr(2) 1.0 0 0 1/2 1.52(6)
Cu(1) 0.52(5) 0.2279(3) 0.7721(3) 0.4344(6) 6.2(2)
Cu(2) 0.32(1) 0.1083(4) 0.8917(4) 0.9122(1) 6.3(3)
Cu(3) 0.17(1) 0.4242(8) 0.5758(8) 0.609(2) 5.8(5)
Cl(1) 1.0 0.4405(1) 0.5595(1) 0.2539(4) 2.31(9)
Cl(2) 1.0 0.1070(1) 0.8930(1) 0.2716(4) 2.35(8)
Cl(3) 1.0 0.2261(1) 0.7739(1) 0.7958(4) 2.56(8)

Table 3
˚ for the room temperature structure of Cu 2 ZrCl 6
Bond distances (A)
Zr(1)–Cl(1) 2.481(3)33 Cu(2)–Cl(2) 2.340(7)32
Zr(1)–Cl(3) 2.478(3)33 Cu(2)–Cl(2) 2.211(6)
Cu(2)–Cl(3) 2.354(7)
Zr(2)–Cl(2) 2.474(2)36
Cu(3)–Cl(1) 2.204(9)
Cu(1)–Cl(1) 2.308(3)32 Cu(3)–Cl(1) 2.71(2)
Cu(1)–Cl(2) 2.511(6) Cu(3)–Cl(3) 2.226(7)32
Cu(1)–Cl(3) 2.223(4)

Cl(2)52.474(2) A36 ˚ are essentially equivalent. The

copper cations reside in slightly distorted tetrahedral
interstices and are disordered over three crystallographical- Fig. 2. ORTEP drawings of the chain-like units running along the c-axis
demonstrating the local coordination with occupancy of the three respec-
ly distinct sites with occupancies of Cu(1)552(5)%, tive copper sites. (a) Cu(1), (b) Cu(2) and (c) Cu(3).
Cu(2)532(1)%, and Cu(3)517(1)% (Occupancies of
46%, 29%, and 16% were found for Cu(1), Cu(2), and
Cu(3), respectively, when the thermal parameters were cated the presence of a non-merohedral twin. Flack’s
fixed at Biso 50.08.). (Refinement of the neutron powder method of left coset decomposition was utilized to de-
data yielded occupancies of Cu(1)541(2)%, Cu(2)5 termine the possible twin laws for a merohedral twin [13].
24(2)%, and Cu(3)521(2)%.) Copper atoms 1 and 3 in Information about each possible twin law for the Laue
particular are distorted toward a trigonal face of the group 3¯m1 [14] was incorporated into the structure model
interstice with three short Cu–Cl distances of 2.2–2.3 A ˚ using the twin refinement options provided by SHELXL97
and one long distance, Cu(1)–Cl(2)52.511(6) A ˚ and [15]. None of the possible twin laws led to an improve-
Cu(3)–Cl(1)52.71 A. ˚ The bond distances around Cu(2) ment of the structure refinement.
are more regular, ranging from 2.21(1) to 2.35(1) A. ˚ All Low-temperature single crystal X-ray diffraction (157
attempts to refine the structure in lower symmetry space K) and X-ray and neutron powder diffraction (100 K) data
groups in order to accommodate a stoichiometric oc- were obtained to examine the temperature dependence of
cupancy of the copper sites resulted in the same structural the copper site occupancy. The 157 K single crystal X-ray
solution with the symmetry of the original space group data gave evidence for a doubling of the c lattice constant
P3¯m1. Neither Weissenberg nor axial photographs indi- (and thus a change in the space group), however, the
Table 4
Selected bond angles (deg.) for Cu 2 ZrCl 6 at 25 8C
Cl(1)–Zr(1)–Cl(1) 90.81(8)33 Cl(2)–Cu(2)–Cl(2) 118.6(2)32
Cl(1)–Zr(1)–Cl(3) 179.87(8)33 Cl(2)–Cu(2)–Cl(2) 97.8(3)
Cl(1)–Zr(1)–Cl(3) 89.10(7)36 Cl(2)–Cu(2)–Cl(3) 108.3(4)
Cl(3)–Zr(1)–Cl(3) 90.99(8)33 Cl(2)–Cu(2)–Cl(3) 106.1(2)32
Cl(2)–Zr(2)–Cl(2) 180.0(1)33 Cl(1)–Cu(3)–Cl(1) 100.1(4)
Cl(2)–Zr(2)–Cl(2) 90.92(8)36 Cl(1)–Cu(3)–Cl(3) 123.9(2)32
Cl(2)–Zr(2)–Cl(2) 89.08(8)36 Cl(1)–Cu(3)–Cl(3) 98.1(5)32
Cl(3)–Cu(3)–Cl(3) 105.2(3)
Cl(1)–Cu(1)–Cl(1) 99.9(1)
Cl(1)–Cu(1)–Cl(2) 101.5(2)32
Cl(1)–Cu(1)–Cl(3) 119.2(1)32
Cl(2)–Cu(1)–Cl(3) 112.6(2)
 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184 177

of about 2.9 eV. A second broad maximum in the absorp-

tion spectrum is observed at around 35 000 cm 21 . By way
of comparison, the diffuse reflectance spectrum of the
colorless [H 2 NMe 2 ]CuZrCl 6 , and CuCl, which has a band
gap of 3.2 eV [16], are also given in Fig. 4b and c,
respectively. The other starting material, ZrCl 4 , has a band
gap in the vacuum ultraviolet; beyond the limits of our
instrumentation. These spectra further provide no evidence
for the presence of any copper (II) chloride (onset of
absorption at 16 000 cm 21 ). Due to our instrumental
limitations it has not been possible to obtain spectroscopic
data for either the high-temperature or low-temperature
phases.
The thermochromic and structural phase transitions were
Fig. 3. Synchrotron powder X-ray diffraction data for Cu 2 ZrCl 6 obtained further quantified by differential scanning calorimetry
at (a) 298 K and (b) 100 K. (DSC) measurements shown in Fig. 5. Two sharp endo-
therms are observed at 673 and 735 K upon heating a
structural solution was otherwise equivalent to that de- sample of Cu 2 ZrCl 6 from room temperature to 773 K at a
scribed above for the room temperature refinement with rate of 10 8 / min. These correspond to the temperatures of
only small changes in atom positions and copper site the high-temperature thermochromic (and structural) phase
occupancies observed. The 100 K X-ray and neutron transition, and the melting point, respectively. Cooling at
powder diffraction data indicate a slightly more pro- the same rate finds that both transitions supercool by about
nounced structural change as seen by the splitting of 20 8. These same transitions were observed on multiple
several of the diffraction peaks (X-ray data are shown in heat–cool cycles confirming that no decomposition takes
Fig. 3). We have not yet solved the structure of the place under these conditions. However, the material will
low-temperature phase, and an attempt to obtain 100 K sublime at the elevated temperatures, thus requiring the use
single crystal X-ray diffraction data was prevented by of hermetically sealed, high-pressure DSC pans. Low
fracture of the crystal upon cooling. temperature DSC measurements were made from room
temperature to 93 K at a cooling rate of 10 8 / min.
3.2. Color and thermochromism However, these data are essentially featureless, given the
sensitivity of our instrument.
At room temperature, Cu 2 ZrCl 6 exhibits a pale yellow
color. Upon cooling the sample in liquid nitrogen (77 K) it 3.3. Magnetic susceptibility
becomes colorless, but returns to its original color upon
warming. Upon heating to 673 K the sample exhibits a A series of magnetic measurements were made to
sharp thermochromic transition to a burnt red color before confirm that the yellow color in Cu 2 ZrCl 6 was not due to a
it is observed to melt at 735 K. The room temperature Cu(II) impurity. In the absence of a paramagnetic impuri-
diffuse reflectance spectrum of Cu 2 ZrCl 6 , shown in Fig. 4, ty, we further wanted to investigate the van Vleck
with the onset of absorption at approximately 21 000 cm 21 paramagnetism of this material with a relatively small
is consistent with the yellow color and an optical band gap

Fig. 5. DSC data obtained upon (a) heating Cu 2 ZrCl 6 from 25 to 500 8C
Fig. 4. Reflectance data for (a) Cu 2 ZrCl 6 , (b) [H 2 NMe 2 ]CuZrCl 6 and (c) and (b) cooling to room temperature (cooling data is offset by 230 mW
CuCl, plotted as the remission function, F(R ` ) 5 (1 2 R ` )2 / 2R ` . for clarity) at 10 8 / min.
178 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184

HOMO-LUMO gap between the high-lying d 10 -orbitals of

Cu(I) and the low-lying d 0 -orbitals of Zr(IV). Measure-
ment of the magnetization was made both as a function of
temperature and of the applied field. Because any magnetic
response from this material would be expected to be quite
small, measurements were made in both fused silica and
DELRIN sample holders to verify that the observed
responses were not artifacts of the sample holder. Parallel
measurements were also made on materials from different
synthetic preparations to control for possible impurities
introduced in the synthesis.
A plot of the magnetic susceptibility, x, as a function of
Fig. 7. Plot of the magnetization of a 0.041 g sample of Cu 2 ZrCl 6 as a
temperature at applied field strengths of 0.05, 0.5, 1.0 and function of the applied field at 250 K. The inset shows the low field M vs.
5.0 T is given in Fig. 6. A diamagnetic core correction of H data obtained at 300, 180, 100, 20, and 5 K demonstrating the
1.9310 24 emu mol 21 , taken from tabulated values [17], temperature dependence of this data.
was applied to this data. A weak and nearly temperature
independent susceptibility is observed, albeit surprisingly
dependent upon the magnitude of the applied field. A
pronounced step is observed in the magnetic susceptibility at 120 K was observed in the [H 2 NMe 2 ]CuZrCl 6 measure-
data at around 120 K, which apparently corresponds to the ments, and the known hydrogen-bond induced structural
low-temperature thermochromic transition. This transition phase transition at 210 K [7] is not apparent in the
is most pronounced in the low-field measurements, and magnetic data. Although about half the magnitude, the M
tends to be masked at higher applied fields. A plot of the vs. H data for [H 2 NMe 2 ]CuZrCl 6 similarly exhibit a
magnetization as a function of the applied field is given in paramagnetism at low applied field that saturates above an
Fig. 7. These data demonstrate a paramagnetic suscep- applied field of about 0.1 T. Unlike Cu 2 ZrCl 6 the y
tibility at low applied fields. However at an applied field of intercept of the slope of the high field data is temperature
greater than about 0.1 T the paramagnetism saturates and independent for [H 2 NMe 2 ]CuZrCl 6 suggesting that both
the diamagnetic susceptibility dominates. While the slope paramagnetic and diamagnetic components of the later
of the high field response of the M vs. H curve (the material are temperature independent.
diamagnetic susceptibility) is approximately constant over
the temperature range examined, the y intercept increases
with decreasing temperature due to a temperature depen- 4. Discussion
dence of the paramagnetic contribution. A parallel set of
measurements was made on the related compound 4.1. Understanding the structure of Cu2 ZrCl6
[H 2 NMe 2 ]CuZrCl 6 , which was made from the same CuCl
and ZrCl 4 starting materials. No step in the susceptibility The complex network structure of Cu 2 ZrCl 6 can most
readily be understood by a comparison to the structure of
ZrCl 3 . The c lattice constant in these materials is virtually
identical; 6.13 A ˚ for ZrCl 3 and 6.15 A ˚ for Cu 2 ZrCl 6 ,
consistent with the hcp chloride sublattice common to both
materials. In the a–b plane, the unit cell of Cu 2 ZrCl 6
] ]
represents a Œ3 3Œ3 expansion of the ZrCl 3 unit cell as
shown in Fig. 8a and b. In ZrCl 3 , the face-sharing
octahedral chains, Fig. 8c, reside on the 6 3 -axis, which
also correspond to the edges of the unit cell. In Cu 2 ZrCl 6 ,
half of the Zr atoms are removed and are replaced by Cu
atoms that are distributed throughout tetrahedral interstices
surrounding the now vacant octahedral interstice. If the
zirconium atom at z50.5 is removed, the symmetry of this
broken chain is lowered to 3¯m and all six tetrahedral
interstices surrounding this site are equivalent; this is the
site of Cu(2) in Cu 2 ZrCl 6 . By contrast if the zirconium
atom at z50.0 is removed, the symmetry of the broken
chain is lowered to 3m. Now two crystallographically
Fig. 6. Plot of molar susceptibility, x vs. T data obtained at applied field distinct sets of tetrahedral interstices surround the vacant
strengths of 0.05–5.0 T, respectively. octahedral interstice providing the sites for Cu(1) and
 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184 179

Fig. 9. Ball and stick representation of the Cu 2 ZrCl 6 networks formed by

complete occupancy of the (a) Cu(1), (b) Cu(2) and (c) Cu(3) sites,
respectively.

Fig. 8. A structural comparison of (a) ZrCl 3 and (b) Cu 2 ZrCl 6 . In (c) the links six different chains through long, apical Cu(1)–
lowering of symmetry from 6 3 in the ZrCl 3 chains to 3¯ and 3, ˚ bonds. Full occupancy of the Cu(3) site,
Cl(2), 2.511(6) A,
respectively, depending on which Zr atoms are removed is demonstrated.
Fig. 9c, results in analogous chains that are oriented in the
2c direction and are connected to each other by the
bridging Cl(1), which is apical to the adamantane-type
Cu(3) in Cu 2 ZrCl 6 . Independently, the Cu(1) and Cu(3) building block of one chain, Cl(1)–Cu(3)52.71(2) A, ˚ and
sites form rings of three corner shared tetrahedra which in inner to a similar building block of a neighboring chain,
turn are capped by an octahedral zirconium chloride unit Cl(1)–Cu(3)52.20(1) A. ˚ While adjacent chains are con-
creating the Cu 3 ZrCl 12 adamantane-type building blocks. nected by bridging halides, in this lattice model the
These adamantane-type building blocks are then stacked Zr(2)Cl 6 octahedra are completely isolated in channels
into chains pointing in the 1c direction for occupation of constructed from the network of chains. By contrast, when
the Cu(1) site and in the 2c direction for occupation of only the Cu(2) sites are fully occupied, as shown in Fig.
the Cu(3) site, Fig. 2a and c, respectively. By contrast, the 9b, three chains of bicapped trigonal anti-prisms are linked
Cu(2) sites surrounding a 3¯m octahedral interstice form to Zr(1)Cl 6 octahedra through chloride bridges, Cu(2)–
chains of bicapped trigonal antiprisms equivalent to the Cl(3)52.35(1) A. ˚ This connectivity also leaves three
superposition of the two adamantane-type chains formed terminal Zr(1)–Cl(1) contacts in the network.
by Cu(1) and Cu(3). Unlike the structure of ZrCl 3 where The structural solution described above suggests the
the chains are isolated by van der Waals contacts, occupa- possibility of a variable copper stoichiometry, up to
tion of the tetrahedral copper sites link these chains into Cu 6 ZrCl 6 . However, off composition loadings exhibit no
three dimensional networks through inter-chain Cu–Cl discernable change in lattice constants, thus providing no
bonds. These external Cu–Cl bonds that are elongated in evidence of any significant variation in stoichiometry.
Cu 2 ZrCl 6 , likely the result of a second-order Jahn–Teller Copper rich compositions (Cu 21x ZrCl 6 ), or even occupa-
trigonal pyramidal distortion commonly observed in cupr- tion of only the Cu(2) site by greater than 0.5 would
ous halides [2]. require quite close copper–copper contacts, |2.3 A. ˚ While
Full occupancy of any one copper site would result in such edge shared tetrahedral linkages are known for
the formation of one of the three different lattices shown in cuprous chlorides, connectivity through sharing tetrahedral
Fig. 9a–c; each with the same Cu 2 ZrCl 6 stoichiometry. corners is much more common. The requirement of such
The set of zirconium and chloride positions are equivalent close copper–copper contacts in the lattice resulting from
in each figure. When the Cu(1) site is fully occupied, Fig. the full occupation of the Cu(2) site leads us to suspect
9a, the lattice consists of chains of Cu 3 ZrCl 12 building that this structure is energetically unfavorable. Neverthe-
blocks running along the 1c direction, which are con- less, we find no reason why one of the ordered lattices
nected laterally along a and b by chloride bridges to resulting from the complete occupancy of either Cu(1) or
isolated Zr(2)Cl 6 octahedra. Each Zr(2)Cl 6 octahedron Cu(3) is not energetically favored over the observed
180 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184

disordered structure. Careful examinations of the diffrac- 4.2. Color and thermochromism
tion data provide no evidence for the twinning of crystals
with different copper site occupancies. To consider the The pale yellow color observed for Cu 2 ZrCl 6 at room
possibility of the in-registry growth of crystallite domains, temperature is a result of the onset of absorption at around
each with a unique copper site occupation, the calculated 20 000 cm 21 (500 nm) as seen in Fig. 4a. This absorption
powder patterns for the three lattices obtained by the originates from the insertion of the empty zirconium d-
occupation of individual copper sites (Fig. 10b–d) were orbitals into the copper centered (Cu-d 10 to Cu-s 0 ) band
compared with the observed powder diffraction (Fig. 3). gap of cuprous chloride. We previously demonstrated that
For each of these calculated patterns the (100) reflection is the electronic localization caused by alternation of Al (III)
observed to exhibit significant intensity. But this reflection and Cu (I) in CuAlCl 4 flattens the valence and conduction
is absent in the observed room temperature powder and bands and shifts the absorption deeper into the UV than is
single crystal data. Notably, the (100) reflection is also observed for the binary CuCl [3]. The lowest energy
absent from the diffraction pattern calculated for the model absorption in CuAlCl 4 was also clearly shown to be due to
based on the fractional occupancy of all three copper the Cu-d 10 to Cu-d 9 s 1 transition. Here we assign the broad
positions (Fig. 10a). By contrast, the 100 K X-ray powder absorption maximum at 35 000 cm 21 to the copper cen-
diffraction data (Fig. 3b) exhibits a reflection at Q50.59 tered transition in Cu 2 ZrCl 6 . (The Cl-lone pair to Zr-4d
21
Å that may be related to a (100)-type reflection. This transition in ZrCl 4 is in the vacuum UV.) Unlike Al(III),
reflection is calculated to be less intense in the neutron however, the octahedral Zr(IV) introduces the low-lying
diffraction given the difference between X-ray and neutron and empty t 2g -type non-bonding orbitals to the system.
scattering factors, and as a result is not observed in the low Under the C3v symmetry of the Cu 3 ZrCl 12 adamantane
temperature neutron data. The appearance of this low Q cage-type units the zirconium nonbonding orbitals trans-
reflection along with the splitting of the other reflections is form with a 1 and e symmetry, and, the set of copper
suggestive of a lower symmetry model in which the Cu d-based Cu–Cl s * orbitals transform as two orbitals of a 1
atoms are possibly localized in certain of the tetrahedral symmetry, one a 2 orbital and three sets of orbitals with e
interstices at low temperature. Nevertheless, we conclude symmetry. Several transitions between the Cu-d 10 to the
that at room temperature any ordering of copper atoms into Zr-d 0 are allowed, and sufficient orbital overlap is possible
domains with specific site occupancies does not occur over given the Cu–Zr distance of only 3.2 A, ˚ such that this
distances of larger than at most a few unit cells. Given the metal-to-metal charge transfer is responsible for the low
apparently insignificant difference in energetic preference energy band edge, and thus the color at room temperature.
for the three copper site occupancies, we speculate that the These assignments are further supported by Extended
copper atoms have significant mobility throughout this ¨
Huckel calculations, in which the projected DOS clearly
lattice. By contrast, at low temperature, the copper atoms shows a majority of Cu and Cl character with a small
are likely localized in specific sites. amount of Zr in the valence band, whereas the much
narrower conduction band is localized on the non-bonding
Zr d-orbitals. [18]
The abrupt color changes over small ranges in tempera-
ture at around 120 and 673 K, respectively, are characteris-
tic of discontinuous thermochromic transitions and suggest
that the thermochromism of Cu 2 ZrCl 6 is due to either first-
or second-order structural phase transitions. The low
temperature DSC measurements provide no evidence for
the phase transition, consistent with a very small and
reasonably a continuous change in enthalpy across the
phase transition. Thus the 20–30 8 temperature range over
which the structural transformation appears to occur
suggests that the low temperature transition is a second
order phase transition. In addition to the proposed localiza-
tion of the copper atoms into discrete sites, based on the
low temperature diffraction data, we suspect that the
copper atoms adopt a more ideal tetrahedral geometry at
low temperature, analogous to the phase transition be-
havior we have previously reported for ((bz)Cu) 2 ZrCl 6 [6]
Fig. 10. Calculated powder X-ray diffraction patterns for the lattices of
Cu 2 ZrCl 6 (a) with a fractional occupancy of all three copper sites, (b) and [H 2 NMe 2 ]CuZrCl 6 [7]. The movement of copper
with complete occupancy of Cu(1), (c) with complete occupancy of Cu(2) away from a tetrahedral face into the center of the
and (d) with complete occupancy of Cu(3). interstice will result in a slight lowering of the Cu–Cl s *
 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184 181

the t 2 -type Cu–Cl s* orbitals of the tetrahedral interstice

are split such that an a 1 orbital is stabilized and an e set of
orbitals are destabilized as described in Scheme 1. Such a
destabilization of the HOMO by placing the copper into a
site of trigonal coordination above the high temperature
phase transition could account for the decrease in the band
gap, and thus the yellow to burnt-orange thermochromism.

4.3. van Vleck paramagnetism

The magnetic susceptibility of materials with closed

shell ground states (i.e. with no permanent magnetic
moment) yet exhibit a temperature independent or van
Vleck paramagnetism is a relatively understudied phenom-
enon. Initially described by van Vleck this temperature
independent paramagnetism observed for close shell ma-
Scheme 1. terials originates from the coupling of ground and excited
states through a magnetic field if there is spin–orbital
coupling [20]. As such, the magnitude of this effect is
greatest for systems that exhibit low-lying excited states.
HOMO (Scheme 1) thus shifting the absorption edge into The [MnO 4 ] 2 and [CrO 4 ] 22 anions are archetypal exam-
the UV to give a colorless material. ples of van Vleck paramagnetism, although this phenom-
A sharp endotherm at 673 K characteristic of a first- ena has been observed for numerous closed shell transition
order phase transition is observed by DSC upon heating metal compounds [21]. Normally resulting from in-
Cu 2 ZrCl 6 prior to the melting endotherm at 735 K, as tramolecular interactions, van Vleck paramagnetism has
shown in Fig. 5. At this phase transition an abrupt color also been measured for a series of charge transfer salts in
change from yellow to burnt-orange is observed. This which the magnitude of the paramagnetic susceptibility is
phase transition is reversible as seen by the reverse color inversely correlated to the energy of the charge transfer
change and the exotherm in the DSC observed upon bands [22]. Recently, controversy has surfaced with regard
cooling. All attempts to obtain structural data for this high to the possible van Vleck paramagnetism in In 2 ZrBr 6 ,
temperature phase have been complicated by the sublima- which was proposed to originate from an indirect elec-
tion of the material out of the X-ray beam when at elevated tronic coupling between occupied In 5s- and empty Zr
temperatures. Undoubtedly at temperatures above this 4d-orbitals through the bromide bridges [23,24]. The
phase transition there will be increased disorder and / or challenge facing all of these studies is the deconvolution of
mobility of the copper atoms. This high temperature phase the contribution of the diamagnetic core from any van
transition may be related to the thermochromism observed Vleck paramagnetic susceptibility. In most cases, a
in Ag 2 HgI 4 and Cu 2 HgI 4 [19]. In these materials the room diamagnetic susceptibility is measured that is smaller than
temperature color is assigned to iodine-to-mercury charge that expected, based on Pascal’s constants for the diamag-
transfer. The thermochromism has been assigned to an netic core [17]. Paramagnetism is thus implied after
increased width of the iodine p-bands due to metal–iodine subtraction of the diamagnetic core correction from the
interactions above the phase transition temperature when measured susceptibility. However, measurement of the
copper or silver atoms are disordered over multiple tetra- magnetization as a function of the applied field and
hedral sites. In Cu 2 ZrCl 6 , the HOMO (valence band) is not temperature allow us to separate the diamagnetic and
halide centered because the chlorine is much more elec- paramagnetic components of the susceptibility.
tronegative than iodine. However, the Cu–Cl s* orbitals Given the small magnitude of the observed magnetic
(bands) have a substantial mix of both copper and chlorine response for Cu 2 ZrCl 6 , it is difficult to unequivocally
character and thus will similarly be influenced by the distinguish an intrinsic effect of a material from a small
copper distribution and site symmetry. As the copper magnetic impurity, although, the temperature independence
atoms traverse between tetrahedral interstices, they must of van Vleck paramagnetism is notably distinct from the
pass through either a trigonal face or two coordinate edge behavior of a simple paramagnetic impurity. Thus, we
of the tetrahedron. It has been shown that such copper ion must preface this part of the discussion with the possibility
mobility is facilitated by a second-order Jahn–Teller that the observed effects are due to some kind of magnetic
distortion whereby the low-lying and empty copper 4s- and impurity. Nevertheless, based on the following evidence
4p-orbitals mix into the copper d-orbitals [2]. By moving we have attempted to rule out the likelihood that the
the copper atom into a trigonal face between interstices, observed effects are due to magnetic impurities: (1)
182 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184

Equivalent data are obtained for samples originating from perature independent. Thus we believe that the observed
different sample preparations. (2) Equivalent data are response is intrinsic to these materials and not an impurity.
obtained from samples sealed in fused silica and DELRIN The negative linear slope at high fields of the M vs. H data
sample holders, respectively. (3) Samples of Cu 2 ZrCl 6 are is equal to the diamagnetic susceptibility, xdia . The
EPR silent at room- and low-temperature. (4) Distinct, but diamagnetic susceptibility of Cu 2 ZrCl 6 , extracted from the
related behavior is observed for multiple preparations of average of the linear fits of the 2 and 10.4 to 5.0 T data
the related compound [H 2 NMe 2 ]CuZrCl 6 that was pre- obtained at temperatures between 5 and 300 K, are plotted
pared from the same CuCl and ZrCl 4 starting materials. (5) in Fig. 11, and are observed to be essentially temperature
The temperature effect on the M vs. H measurements are independent. The slight up-turn in this xdia below 100 K
distinctly different for Cu 2 ZrCl 6 and [H 2 NMe 2 ]CuZrCl 6 . may be a result of a small paramagnetic impurity. Similar
The molar magnetic susceptibility data for Cu 2 ZrCl 6 , at behavior is observed for [H 2 NMe 2 ]CuZrCl 6 . The mag-
applied field strengths of 0.05–5.0 T (Fig. 6), are essential- nitude of the measured xdia , 21.6310 24 emu mol 21 for
ly temperature independent. However, when plotted on the Cu 2 ZrCl 6 , and 21.7310 24 emu mol 21 for
expanded scale of Fig. 6 (two orders of magnitude less [H 2 NMe 2 ]CuZrCl 6 , are consistent with the values obtained
than a S51 / 2 system), subtle temperature and field effects from the sum of Pascal’s constants, 21.9310 24 emu
are observed. Immediately apparent is the field dependence mol 21 and 21.8310 24 emu mol 21 , respectively. This
of this data, and the jump in the susceptibility data at about diamagnetism can then be subtracted from the M vs. H
120 K. We note that at applied fields above about 1 T, the data to yield Fig. 12, for example, which is presumably a
‘paramagnetic susceptibility’ of Cu 2 ZrCl 6 is only a result picture of the field dependence of the van Vleck paramag-
of the measured diamagnetic susceptibility being smaller netism.
than the diamagnetic core correction. However, a para- The saturation magnetization of this van Vleck paramag-
magnetic susceptibility is measured directly under applied netism, Msat vV , can be readily determined from the y-
fields of less than about 1 T. Data obtained on the related intercept of the high field xdia curve of Fig. 7, and is
compound [H 2 NMe 2 ]CuZrCl 6 exhibit a similar field de- plotted as a function of temperature in Fig. 13. Only a
pendence, however, no transition at 120 K is observed. small decrease in this Msat vV is observed for
Thus, we believe that jump in the susceptibility of [H 2 NMe 2 ]CuZrCl 6 upon decreasing temperature from 300
Cu 2 ZrCl 6 is a manifestation of the low temperature to 20 K. A slight decrease in the y-intercept would be the
thermochromic, and structural phase transition. A similar expected effect of a small paramagnetic impurity (i.e. the
conclusion was proposed by Dronskowski [23] to explain measured slope, xdia 1 ximpurity , would be less negative
the observed magnetic response in In 2 ZrBr 6 . His conclu- resulting in a smaller y-intercept). By contrast the Msat vV
sion was later contested by Jansen et al. [24]. However, of Cu 2 ZrCl 6 is about twice the magnitude of that observed
when the data from the latter report are digitally scanned for [H 2 NMe 2 ]CuZrCl 6 and exhibits a remarkable tempera-
and expanded to the same scale as that reported in the
former (as well as subtracting the diamagnetic core correc-
tion), the two sets of data are seen to be very similar,
further supporting the idea that the structural phase transi-
tions may be manifest in the van Vleck paramagnetism.
Nevertheless, the impact of this phase transition on the
susceptibility appears to be masked at higher applied field
strengths, suggestive of a magnetic saturation.
To further probe the field dependence of this magnetic
response, we have measured the magnetization as a
function of applied field at several temperatures between 5
and 300 K for both Cu 2 ZrCl 6 (shown in Fig. 7) and
[H 2 NMe 2 ]CuZrCl 6 . These data exhibit a maximum in the
magnetization at applied field strengths of about 0.1 T. A
mixture of a ferromagnetic impurity and a diamagnetic
compound might account for such data. However, while
only 4.7310 27 g of iron (out of the total sample of 0.041
g) could account for a molar magnetization of 1310 24
emu mol 21 , it seems unlikely that multiple preparations
would yield the same very small amount of an iron Fig. 11. Lower plot: Diamagnetic susceptibility, xdia , extracted as the
slope of the M vs. H data between 0.4 and 5.0 T, as a function of
impurity. Furthermore, while the M vs. H data for
temperature. Upper plot: van Vleck paramagnetic susceptibility, xpara vV ,
Cu 2 ZrCl 6 exhibit a notable temperature dependence, inset extracted as the slope of the M vs. H data between 6200 T, as a function
of Fig. 7 (itself inconsistent with a ferromagnetic impuri- of temperature. (Note the order of magnitude difference in the scales of
ty), the equivalent data for [H 2 NMe 2 ]CuZrCl 6 are tem- these respective plots.).
 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184 183

paramagnetic susceptibility, xpara vV , from the slope of the

low field M vs. H data (60.02 T), although there is
considerably more scatter to this data, thus any conclusions
from it must be made cautiously. Nevertheless, these data,
shown in Fig. 11, suggest a xpara vV of about 1.2310 23
emu mol 21 above the thermochromic transition and a
xpara vV of about 1.6310 23 emu mol 21 below the phase
transition. A value of xpara vV ¯2310 23 emu mol 21 is
obtained for [H 2 NMe 2 ]CuZrCl 6 , although the magneti-
zation begins to saturate and thus deviate from a linear
response even by 0.01 T. Interestingly, both the colorless,
low-temperature phase of Cu 2 ZrCl 6 and the colorless
[H 2 NMe 2 ]CuZrCl 6 exhibit a larger xpara vV than is ob-
Fig. 12. Saturation of the van Vleck molar magnetization of Cu 2 ZrCl 6 at
250 K as a function of the applied field.
served for the room-temperature phase of Cu 2 ZrCl 6 that
has the smaller band gap. These data are consistent with
the above hypothesis that the low temperature phase of
ture dependence increasing approximately linearly to about
Cu 2 ZrCl 6 exhibits a more ideal local tetrahedral coordina-
120 K then reaching a plateau between about 100 and 40
tion around Cu, analogous to that observed in
K. The gradual increase in the Msat vV upon decreasing
[H 2 NMe 2 ]CuZrCl 6 . However, this trend in the xpara vV is
from room temperature to 120 K is consistent with a
inverse to that anticipated based on the band gap (i.e.
gradual lattice contraction, as seen in the room temperature
xpara vV is anticipated to be inversely proportional to the
and 157 K lattice constants (a volume contraction of about
band gap). While the optical band gap due to the Cu→Zr
2%), which induces increased orbital overlap. At 120 K
transition increases as the copper moves into the center of
the structure changes in such a way that the band gap is
a tetrahedral interstice, this motion also significantly
increased in energy out of the visible spectrum as evi-
reduces the second-order Jahn–Teller mixing of the Cu-3d
denced by the thermochromic transition. The increased
and Cu-4s,p orbitals, which lowers the energy of the empty
band gap thus results in a diminished mixing between the
Cu-4s,p orbitals. Thus, while the Cu→Zr band gap in-
ground and excited states, and thus the change in the
creases on going to the tetrahedral geometry, the Cu-
temperature response of the Msat vV . However, the plateau,
3d→Cu-4s band gap may decrease. Therefore, a combina-
instead of a decrease in the Msat vV , may indicate that the
tion of the copper centered and copper→zirconium cen-
structural change causing an increased band gap also
tered charge transfer likely account for the increase in the
results in a bonding configuration with greater coupling of
xpara vV and Msat vV of Cu 2 ZrCl 6 at low temperature.
the ground and excited states. The origin of the steeper
up-turn in the Msat vV is not clear, but may be related to the
localization of the copper sites below the second order
5. Summary
phase transition.
It is possible to extract a value for the van Vleck
The solid state reaction of CuCl and ZrCl 4 was shown to
yield Cu 2 ZrCl 6 . The structure of this compound is most
reasonably understood as a derivative of the ZrCl 3 struc-
ture-type. Here, every other metal site in a ZrCl 3 face-
sharing octahedral chain is removed, and is replaced by
copper(I) atoms, which partially occupy the six tetrahedral
interstices surrounding the now vacant octahedral inter-
stice. The multiple tetrahedral interstices provide three
crystallographically equivalent sites over which the copper
atoms are disordered at room temperature. The compound
Cu 2 ZrCl 6 is observed to undergo two thermochromic and
structural phase transitions at 120 and 473 K, respectively.
Copper-to-zirconium charge transfer, the energy of which
is determined by the geometry of the copper coordination
sphere, best accounts for the color and thermochromism.
As the copper atoms move out of a more ideal tetrahedral
Fig. 13. Plot of the y-intercept of the high field M vs. H curve (equal to interstice at low temperature toward a site in the trigonal
Msat vV ) as a function of temperature for Cu 2 ZrCl 6 and
[H 2 NMe 2 ]CuZrCl 6 . The line through the [H 2 NMe 2 ]CuZrCl 6 is the least face of the interstice at high temperature, the band gap is
squares linear fit. The line through the Cu 2 ZrCl 6 is simply to highlight the decreased and thus the absorption edge is red shifted into
change as a function of temperature. the visible range.
184 A.M. Dattelbaum et al. / Journal of Alloys and Compounds 338 (2002) 173 – 184

Coupling between the high-lying filled Cu–Cls* orbitals DOE BES contract (DE-AC02-98CH10886). This work
and the low lying and empty Zr 4d-orbitals (or to the low was supported by the National Science Foundation (DMR-
lying Cu 4s- and 4p-orbitals) provides a stage on which to 9501370, DMR-0072828, DMR-9703419 and DMR-
investigate the temperature independent paramagnetism 0108605) and instrumentation grants (CHE-9509532 and
said to result from the field induced mixing of the ground DMR-9601825). J.D. Martin is a Cottrell Scholar of the
and low-lying excited states described by van Vleck [20]. Research Corporation.
While one can never completely exclude the possibility
that a weak magnetic response is due to impurities, the
magnetic behavior of the closed shell materials Cu 2 ZrCl 6 References
and [H 2 NMe 2 ]CuZrCl 6 seem to provide new insights into
the nature of van Vleck paramagnetism. As classically [1] L. Subramanian, R. Hoffmann, Inorg. Chem. 31 (1992) 1021.
understood, the magnitude of the van Vleck paramagnetism [2] J.K. Burdett, O. Eisenstein, Inorg. Chem. 31 (1992) 1758.
[3] R.M. Sullivan, J.D. Martin, J. Am. Chem. Soc. 121 (1999) 10092.
is clearly dependent on the structure of the material as well
[4] R.A. Huggins, in: A.S. Nowick, J.J. Burton (Eds.), Diffusion in
as its band gap (or HOMO-LUMO gap). However, it is Solids, Academic Press, New York, 1975, p. 445.
reasonable to assume that like spin-only magnetization, [5] J.D. Martin, A.M. Dattelbaum, R.M. Sullivan, T.A. Thornton, J.
one must also consider saturation effects. The van Vleck Wang, M.T. Peachey, Chem. Mater. 10 (1998) 2699.
paramagnetism exhibits saturation effects similar to that [6] A.M. Dattelbaum, J.D. Martin, Inorg. Chem. 38 (1999) 6200.
[7] A.M. Dattelbaum, J.D. Martin, Inorg. Chem. 38 (1999) 2669.
described by the Brillion function for classic paramagnetic
[8] G.B. Kauffman, L.Y. Fang, Inorg. Syn. 22 (1983) 101.
systems, though here saturation occurs at very low applied [9] W.W. Wendlandt, H.G. Hecht, Reflectance Spectroscopy, Inter-
fields. As a result, most literature reported values for van science, New York, 1966, Chapter 3.
Vleck paramagnetic susceptibility reflect the saturation [10] E.J. Gabe, Y. Le Page, J.-P. Charland, F.L. Lee, P.S. White, J. Appl.
magnetization. That the colorless (i.e. larger band gap) Crystallogr. 22 (1989) 384.
[11] A. Altomare, M.C. Burla, G. Camulli, G. Cascarano, C. Giacovazzo,
[H 2 NMe 2 ]CuZrCl 6 and low temperature form of Cu 2 ZrCl 6
A. Guagliardi, G. Polidori, J. Appl. Cryst. 27 (1994) 435.
exhibit the larger values of xpara vV than the yellow room [12] A.C. Larson, R.B. Von Dreele, GSAS: General Structure Analysis
temperature phase of Cu 2 ZrCl 6 , can be accounted for if the System Manual, Los Alamos Report LAUR 86-748, Los Alamos
less distorted tetrahedral Cu site provides more effective National Laboratory, Los Alamos, NM, 1986.
overlap between the ground and excited states, relative to [13] H.D. Flack, Acta Cryst. A43 (1987) 564.
[14] Y. Le Page, J.D.H. Donnay, G. Donnay, Acta Cryst. A40 (1984)
the second-order Jahn–Teller distorted trigonal prismatic
679.
geometry. By contrast the Msat vV is apparently doubled for [15] R. Herbst-Irmer, G.M. Sheldrick, Acta Cryst. B54 (1998) 443.
Cu 2 ZrCl 6 with respect to [H 2 NMe 2 ]CuZrCl 6 because it [16] M. Ferhat, A. Zaoui, M. Certier, B. Khelifa, Phys. Lett. A 216
has twice the number of copper atoms per formula unit. It (1996) 187.
is necessary to now reexamine a variety of other materials [17] L.N. Mulay, E.A. Boudreaux (Eds.), Theory and Applications of
Molecular Diamagnetism, Wiley, New York, 1976, pp. 494–495.
that are reported to exhibit van Vleck paramagnetism to
[18] A.M. Dattelbaum, Thesis, North Carolina State University, NC,
firmly establish the saturation effects, and then develop a USA, 2000.
model consistent with the phenomenon. [19] H.-R.C. Jaw, M.A. Mooney, T. Novinson, W.C. Kaska, J.I. Zink,
Inorg. Chem. 26 (1987) 1387.
[20] J.H. van Vleck, The Theory of Electric and Magnetic Sus-
ceptibilities, Oxford, London, 1932.
Acknowledgements
[21] (a) P.W. Fowler, E. Steiner, J. Chem. Phys. 97 (1992) 4215;
(b) P.W. Fowler, E. Steiner, J. Chem. Soc. Faraday Trans 89 (1993)
The authors acknowledge Dr Paul Boyle for collecting 1915;
the single crystallographic data, Dr Barbara Reisner, Dr (c) D.P. Raychaudhuri, P.N. Sengupta, Indian J. Phys. 10 (1936)
Brian Toby, and Roger Sullivan for collecting and analyz- 245;
(d) D.P. Raychaudhuri, P.N. Sengupta, Indian J. Phys. 10 (1936)
ing the neutron powder diffraction data. Neutron powder
253.
data were obtained with support from the National Institute [22] M. Ghosh, R. Basu, Naturwissenshaften 62 (1975) 392.
of Standards general user program (award [ NIST-1965), [23] R. Dronskowski, J. Am. Chem. Soc. 117 (1995) 1991.
and synchrotron powder X-ray diffraction data were [24] M. Jansen, N. Wagner, M. Becker, U. Wedig, J. Am. Chem. Soc. 122
obtained at the National Synchrotron Light Source via the (2000) 808.