Professional Documents
Culture Documents
The far infrared vibration spectra of Fe&h, FeC13, Fe&h, and FeClr in
several matrices have been studied by effusion technique using a modified
Knudsen cell with two stages. IR vibrational spectra have been obtained in
argon matrices whereas only parts of the spectra were obtained in nitrogen
and cyclohexane matrices. Assignments for the vibrational spectra for all
molecules are given, based in part on normal coordinate analysis and observed
isotopic splittings. Thermodynamic functions of the natural isotopic mix-
tures of Fe&&,, FeCl$, and FeClz in the ideal gaseous state are presented.
I. INTRODUCTION
1Abstracted in part from PhD thesis of R. A. Frey, Swiss Federal Institute of Tech-
nology, Zurich, Switzerland, 1969.
2 Present address: Texas Technological University, Dept. of Chemistry, Lubbock, Texas.
260
MBTRIX SPECTRA OF IRON CHLORIDES 261
The first spectroscopic investigation of the gas phase at 5ZO”K was reported
by Wilmshurst (11). This revealed four strong absorptions which were assigned
in terms of l&, structure. The band contours reported by Wilmshurst appear to
be quite complex and leave some doubt about the assignment in view of the many
hot bands to be expected for the Fe&%, molecule at 520%.
Many metal dihalides have been investigated so far including FeCh (12-18).
In most cases, except for TiFz , for which a Csu geometry based on observed in-
frared vibrational spectrum and vibrational isotope effect has been suggested
(17), molecular symmetry D,h has been found to be in good agreement with
observed spectra. Specifically for FeCl? a linear structure has been derived from
analysis of infrared gas (12) and matrix spectra (18), as well as from electronic
gas and matrix spectra (19). Also the existence of the dimer molecule Fe&h
has been inferred from infrared spectra (12, 18) and from mass spectroscopic
investigations (20). However, the available information seems not to afford a
relevant conclusion about the structure of Fe2C14.
In this paper we report the results of an investigation of the matrix spectra of
Fe&& , FeCL , FeCh , and Fe&h obtained from molecular effusion from a
double stage Knudsen cell, which allows separate study of gaseous species in
equilibrium with the solid and under superheated conditions. From dilution and
annealing experiments, fairly complete spectra were obtained. The observed
spectra were assigned on the assumption of molecular symmetry Dsh,Dab,D,h ,
and Dfh,respectively, which in turn are supported by normal coordinate analysis
and partially resolved isotopic splittings observed on some bands. Based on vibra-
tional assignments and molecular models thermodynamic functions for the na-
tural isotopic mixtures in the standard gas phase were calculated. Both statistical
entropies and specific heat were found to deviate substantially from values
derived from thermodynamic experiments by earlier workers.
All FIR spectra have been taken on a Perkin-Elmer Model 301 spectrophotom-
eter in double beam operation from 680 to 30 cm-’ and single beam operation
from 30 to 20 cm-‘. Spectral slit width under conditions of highest resolution was
approx 0.7 cm-‘. Besides conventional recording the spectra were digitally re-
corded using a sampling system shown in Fig. 1. Wavelength was digitized by
taking a proportional voltage from a 25-turn high resolution potentiometer con-
nected to the wavelength drum. For digitizing transmittance, the voltage of the
instruments strip chart recorder was measured by a digital voltmeter. Sampling
rate was determined by the display of the counter (hp 524C) which was also used
for triggering of the voltmeter and generation of the printing command. Both
digital wavelength and transmittance information were aut,omatically punched
on IBM cards. Sampling conditions were chosen according to rules given by
aALGOL 60 Program “Smooth” developed by Mr. J. Keller, Swiss Federal Institute of
Technology, Zurich, Switzerland.
262 FREY, WERDER, AND GiiNTHARD
Dig.Voltmeter
HP
3440 A
Recorder
Voltage/Frequency Counter
Converter
FIQ. 1. Digital sampling system
Porchet et al. (21) ; in most cases sampling periods between 2 and 25 set were
applied. Punched data were processed by a computer program for smoothing,
interpolation, peak finding, and plotting.3 For investigation of matrix spectra,
stainless stee1 cryostats of the type described by Werder (2%‘) were used. Tem-
perature measurements were made with copper constantan thermocouples whose
hot junctions were kept in a water triple point apparatus as recommended by
the National Bureau of Standards (23). By this procedure an accuracy of 0.5”K
at 5°K was achieved.
Matrix and sample gas were condensed on either CsI or crystal quartz windows.
For production of the sample gas, a Knudsen cell with two chambers was con-
structed from stainless steel cf. Fig. 2.
Two chamber Knudsen cells have earlier been described by Akishin et al. (%J)
and by Berkowitz et al. (26). The cell used in this work differs from these designs
in several respects. In chamber 1 solid FeCb was evaporated to produce an equi-
librium two phase system at temperature 2’1. The vapor leaks through tube T
into chamber 2 held at temperature Tz . By choosing appropriate temperatures,
TI and TZ , a, wide range of conditions for chemical equilibrium may be achieved.
In particular superheated gas mixtures in chamber 2 allow investigation of
chemical gas equilibria by study of the spectra of the particles effusing from
chamber 2 into the matrix. To achieve uniform temperature distribution and
high impedance in the down stream, chamber 2 was equipped with a demount-
able system of three shields S forcing the vapor to interact strongly thermally
with the walls.
High purity commercial argon and nitrogen (total impurity ~5 ppm) were
used for preparation of matrices at LHe temperature after passing through a LN2
trap before entering the high vacuum system. In order to obtain specific M/A
ratios the matrix gas flow was determined with a needle valve (Edwards LBA 1)
and the sample vapor flow was determined by the temperature, Tl and Tz ,
of the Knudsen cell system.
MATRIX SPECTRA OF IRON CHLORIDES 263
Preparation of matrices with sufficiently high M/A ratio and low light scatter-
ing for investigation of weak bands proved rather critical in this work. Low con-
densation rates seemed to be the most important condition for obtaining low
scattering matrices (24-hr total condensation time). However, in addition, each
of the four molecules under study required some specific conditions to obtain well
behaved matrix spectra.
III. RESULTS
FIG. 3. Isolation effects of Fe&l6 in Ar, 5”K, CsI window (M/A ratios estimated)
MATRIX SPECTRA OF IRON CHLORIDES 265
creasing M/A ratio band broadening becomes dominant and additional bands
appear (Fig. 3A and B) which may be assigned to polymeric species. The same
new bands also appear if high M/A matrices are annealed at temperatures
-40°K. This interpretation is borne out also by comparison with the solid state
spectrum of FeCh , which has been studied in this work to some detail, but is
not reproduced here. Though the bands ascribed to polymers of FezC16exhibit a
tendency to approach positions of the crystal bands, low M/A FezCls spectra
are still clearly different from the crystal spectrum.
The band at 440 cm-’ is the most sensitive to perfect isolation. Since this band
just overlaps the absorption gap of crystal quartz, its dependence on isolation
provided a convenient check for sufficient isolation in matrices deposit.ed on
crystal quartz.
In Fig. 4 the low frequency part of the Fe&Ye spectrum in the argon matrix on a
crystal quartz window is shown. Below 100 cm-’ the crystal window produced
strong interference fringes which could, however, not be eliminated. The absorp-
tion bands at 210 and 150 cm-l did not appear regularly in the spectra and hence
very probably originate from residual impurities. The remaining bands at 118.5,
T d
Ii -
/
140 an-’ loo !50 3
FIG. 4. Fe&l6 in Ar, 5”K, l&0.1-mm quartz wedge, Q and Ar refer to absorptions due
to quartz and argon, respectively.
266 FREY, WERDER, AND GiiNTHARD
115.7, and 98.7 cm-’ may be assigned to Fe&%, , since they behave similarly to
the band at 280 cm-‘, though 10 times less intense.
The lower part of Fig. 4 shows the FIR absorption bands of Fe&l6 in an argon
matrix more clearly, although the doublet at 117 cm-’ is not fully resolved. The
Fe&g in Nitrogen
TABLE I
ASSIGNMENT
OF FUNDAMENTALS
OF Fe&&
vi obs (cm-r)
Assignment (normal mode) Species pi talc (cm-r)
Ar matrix Gas”
8 cf. Ref. (II), however, the assignment is revised to agreement with the one given here.
MATRIX SPECTRA OF IRON CHLORIDES 267
peak at 27 cm-’ and the rising background absorption below 25 cm-’ are due to
the spectrographs 100% line. The weak absorption at 24.3 cm-’ was proven to
belong to Fe&&, since it disappeared on annealing and therefore cannot originate
from the interference pattern of the quartz window or the water vapor absorption
band present at this wavelength.
As shown in Fig. 5 the 24.3 cm-’ band appears with higher intensity in a
nitrogen matrix in which also the 117 cm-’ doublet can clearly be recognized.
Again the bands between 30-25 cm-’ belong to the background absorption and
do not disappear after annealing the matrix.
All the bands assigned to Fe&la were found to be only slightly affected by
different matrix gases used in this work. This point will be discussed later to-
gether with the other halides. For a collection of absorption frequencies as-
signed to FezUs cf. Table I.
---+I
I I
FIG. 6. FeClz in Ar, 5”K, CsI window, M/A M 150, effusion temperature 580°C. For
explanation of dl , dl’ and dz see Sects. 3.2 and 3.3.
268 FREY, WERDER, AND GiiNTHARD
already observed by Carlson et al. (18), which leads to the conclusion that at
least two different molecular species are present in the matrix. It is well known
from mass spectroscopic investigations of FeCL vapor (.$?O), that it contains
dimeric molecules. Therefore, one has to expect absorption bands of dimer and
possibly polymer FeClz to show up in matrix spectra. Indeed some of the broad
bands in Fig. 6 point to polymer absorption bands. The spectrum shown in Fig.
6B was obtained by annealing the matrix at about 30°K for 10 min and subse-
quent cooling to 5°K. It features, as compared with spectrum A at least two
different band systems which are related in intensity, namely the bands around
490 and 475 cm-’ (d ecreasing intensity on annealing) on the one hand and the
bands near 440, 325, and 260 cm-’ (increasing intensity) on the other hand.
The annealing effect very probably consists in the removal of monomeric FeCls
species by polymerization, so that the bands decreasing in intensity are to be
assigned to FeClz , whereas the rising bands will be due to dimer and polymer
molecules. The same conclusion, including two bands at 110 and 88 cm-’ has been
drawn by Carlson et al (18), but he did not report any absorption between 400
and 118 cm-‘. Further, in contrast to Carlson’s observation, who did not report
generation of new bands on annealing the matrix, the following observation may
be stated comparing the spectra in Fig. 6: The doublet denoted by- dl centered at
436 cm-’ resists the annealing process with no remarkable intensity change, the
same feature is shown by the doublet dz at 250 cm-’ if we disregard the disappear-
ing of the splitting in spectrum 6B. On the other hand a very pronounced intensity
increase occurs in the 450, 320, and 265 cm-’ region including a very prominent
band at 444 cm-’ and several sharp bands around 460 cm-‘. We thus have to
conclude that beside the dimeric molecule Fe2C14 s:her pF>lymeric species are
present in the matrix. Finally it should be mentioned that the spectra of Fig. 6,
including the very complex behavior on annealing, was accurately reproducible in
detail in different experiments. The spectrum in Fig. 7 was obtained from FeClz
vapor at high M/A ratio and shows two band systems at 490 and 475 cm-‘,
supporting the assumption that these bands are due to monomer FeCh only. Both
band groups have very similar shapes though different intensities and are found
to fit well each by itself to the isotopic splitting pattern to be expected for FeCln .
Closer inspection of the ratio of the overall absorption intensity of the two band
systems at 490 and 475 cm-’ (cf. Fig. 6 and 7) reveals considerable variations.
This fact is used below as an important argument in the analysis of FeC& spectra.
In Fig. 8, the absorption spectrum of FeClz in a NZ matrix at M/A = 1.50 is
shown. By comparison with Fig. 6 and 7 the band systems at 445,430, and 415
cm-’ in Kz may be considered to correspond to the systems at 490, 475, and 440
cm-’ in Ar, respectively. Obviously the FeCl? spectra are subjected to large
matrix shifts, in contrast to Fe&&, , where negligibly small matrix shifts were
observed.
The frequencies of the bands observed in the matrix spectra are collected in
MATRIX SPECTRA OF IRON CHLORIDES 269
0 L I I I I I I I I I I I
500 cm-l 450 400
FIG. 7. FeClz in Ar, 5”K, CsI window, M/A cz 400
Table II. They agree within experimental error limits whith the measurements
published by Carlson et al. (18).
TABLE II
ASSIGNMENT OF FeCls INFRARED MATRIX SPECTRUM
8P 88.9
88.4
88.1
87.5
87.0
- 66Fe3%12 330.0
498.8 (482.2) S4Fe35Clz 499.0
496.1 (?) 64Fe35Cla7C1 496.1
494.0 (477.6) 56Fe36C12 494.1
491.4 (474.9) 56Fe35C137Cl 491.2
488.2 (471.9) SBFe37C12 488.1
L, I I J
FIG. 9. Fe&la, FeC13, FeClz in Ar, 5”K, CsI window, M/A M 200 for Fe&l+ The num-
bers in parentheses refer to the temperatures in the first and second chamber, respectively,
of the Knudsen cell. For explanation of dl and dl’see Sect. 3.3.
band at 450 cm-’ can be attributed either to FeL% or to FeClz and its polymers.
Particularly in the lower spectrum of Fig. 9 the two doublets do and dr’ of spec-
trum A in Fig. 6 are present in the 440 cm-’ region. Hence in the region above
160 cm-’ only one band at 467 cm-’ may reliably be assigned to FeCL since the
origin of bands at 367 and 450 cm-’ is not quite clear.
The FIR part of the spectrum of the Fe&le-FeCls-FeClz mixture is reproduced
in Fig. 10. Again the bands at 328 and 280 cm-’ belong to Fe&h,, whose low fre-
quency absorption bands however cannot be expected to be discernible under
the conditions of this spectrum (cf. Fig. 4). Since FeClz and Fe&l4 were found to
absorb near 110 and 88 cm-’ (18), the bands at 116 and 102 cm-’ may be as-
272 FREY, WERDER, AND GUNTHARD
signed to FeCb . The sporadic weak band at 91 cm-’ cannot definitely be as-
signed to one of the simple species and probably is associated with polymeric
FeClz or FeClS4. Absorption band frequencies attributed to FeC13 are listed in
Table III.
Discussion of the results reported above are based on the following arguments:
(i) qualitative comparison between related molecules of similar type;
(ii) use of normal coordinate analysis for determination of force constants
and derivation of correlation between spectra of molecules related by a simple
association reaction. For normal coordinate calculations, a method and program
developed by Hunziker (27) was used.
FIG. 10. Fe&&, FeC13, FeClz in Ar, 5”K, 1.0&l-mm quartz wedge, M/A z 159
for Fe&L,. The numbers in parentheses refer to the temperatures in the first and second
chamber, respectively, of the Knudsen cell. Q and Ar refer to absorptions due to quartz
and argon, respectively.
TABLE III
ASSIGNMENT OF FeCla INFRARED MATRIX SPECTRUM
vi obs (cm-r)
Assignment (normal mode) Species Yi talc (cm-*)
Ar matrix
4A possible explanation for the bands of 456, 367 (cf. Fig. 9) and 91 cm-r (cf. Fig. 10)
can be given in terms of a bridged molecule ClFeChFeClz, formed by a recombination
between monomeric species FeCh and FeCla.
MATRIX SPECTRA OF IRON CHLORIDES 273
(iii) analysis of isotopic splittings and band shapes resulting from incompletely
resolved splittings in cases where isotopic effects in the spectra have actually been
observed.
a = 2.11 8
b = 2.28 8
w = 128’
I3 = 92O
TABLE IV
INTERNAL COORDINATESUSED IN NORMAL COORDINATEANALYSIS OF Fe&Is
VIBRATIONALSPECTRUM
Force constant
Coordinate Symmetry coordinate
Value
Notation
(mdyn/&
turned out, that the following force constant matrices allow sufficiently accurate
description of the observed spectra
and may uniquely be derived from the observed spectra. As may be seen from
Eq. (2) only two interaction constants are included in this force field, namely f9
(interaction between Fe-Cl stretching coordinates of the ring and of the FeClz
groups) and fro (between Fe-Cl stretches of Fee12 groups). Table I gives a com-
parison of observed and calculated frequencies and furthermore the frequencies
calculated for the infrared inactive species AI,, A, , BI, , Bz, , Big . It should be
mentioned that the frequencies of these modes lie within a range of 10 cm-’ for
several slightly modified force fields which can reasonably be derived from our
data.
A short comment concerning the assignment presented in Table I seems to be
in order. In terms of group vibration two of the Fe-Cl ring stretching modes are
expected to be infrared active as well as two Fe-Cl stretches of the FeClz groups.
These modes are most likely to be assigned to the four strong absorption bands at
46X,406,328, and 280 cm-’ (cf. Fig. 3 and Sect. 3.1). It has been found for a series
MATRIX SPECTRA OF IRON CHLORIDES 275
of bridged molecules, i.e., BZHB (68), AlzFo (69), Al$& (SO), NbzCllo ($I), that
bond stretching force constants of MX in the outer MX, groups are nearly twice
as large as in the MzX4 ring. Therefore the assignment of the four bands mentioned
above seems straightforward as given in Table I. Table V gives a compilation of
the force constants fl , fi, j’s, and fd for the sake of illustration of this rule. No
analogous rule for the bending modes is known, and one then has to rely on nor-
mal coordinate conclusions and on analysis of vibrational spectra of the molecules
listed in Table V. The low lying ring puckering mode ~10 is strongly intermixed with
the rocking mode of the FeCh group of &, symmetry. These two modes may be
associated with the very low frequency band at 24.3 cm-’ and the band at
118.5 cm-’ (it may be noted here that the assignment of the bands at 118.5 and
115.7 cm-’ to the modes yg and ~18, respectively, given in Table I, is not straight-
forward, interchanging the assignment for these two modes yields only a slight
deterioration of the overall fit of calculated frequencies with the observed spec-
trum). Since the 115.7 cm-l band probably does not belong to the same symmetry
species as the 118.5 cm-’ band, the former may be assigned as the MXz bending
mode of Bau symmetry. The infrared spectra do not provide information about
the A, and B,, mode and therefore yield no determination about the force con-
stant fe . The value listed in Table IV is based on the analogies mentioned, and
has no further support, as do the frequencies of the normal modes v6 and ~1~which
mainly concern the twisting modes of the FeCh groups.
The matrix spectra exhibit several very weak absorptions (445, 427, and
315 cm-‘) which may be assumed to originate from breakdown of selection rules
by matrix fields or from the many isotopic species with noncentrosymmetric
symmetry which must occur in the natural mixture of FezCle molecules. Among
these bands only the 315 cm-’ absorption is M/A independent and therefore is
the only one included in the assignment proposed in Table I. Nevertheless the
presence of several isotopic species is revealed by the complicated band envelope
of some of the Fe&l6 absorption bands, notably the 406 cm-’ band if takenwith
TABLE V
MX-STRETCHINQ AND BENDING FORCE CON!TANTS FOR BRIDGED Lh MOLECULK+
(mdyr~/A)
of the intensity of these bands on annealing the sample reveals that other poly-
meric species than Fe&l4 preferably are formed during annealing. This assump-
tion seems to be supported by the striking absorption rising in Fig. 6B at 445 cm-’
as well as by the pronounced absorption increase at 265 and around 320 cm-’ in
the same spectrum. A further detail caused by annealing may be noted: The
disappearing of the doublet dl’ (which together with the doublet cl1is also present
in the lower spectrum of Fig. 9 and is assigned to Fe&&, too) going from spec-
trum A to spectrum B in Fig. 6 seems to go parallel with the disappearing of the
splitting of the absorption band at 250 cm-‘. This may be understood in terms of
matrix site effects, suggesting the doublet dl’ and the lower component of the
doublet clz to correspond to a E’e2C14molecule in a relatively instable matrix site.
However, the absorption bands at 438 and 250 cm-l are further regarded as ab-
sorptions of Fe&l4 in these regions.
To obtain further support for assignment, normal coordinate analysis based
partly on force constants transferred from Fe&le was used. No structural data
for Fe&J, have apparently been published. For the following we assume D2hsym,-
metry, the same geometry for the FezClz ring as for FezCls and r (Fe-Cl) = 2.11 A
for the FeCl groups. Then the fundamentals are classified according to 3A, + 2&,,
+ 2&, + &, + 2& + 2&u. Using the same internal coordinates for the ring
as for Fe&k,, one in-plane, one out-of-plane and one stretching coordinate for
each of the FeCl groups, one may associate with these the force constants ji , f3 , f5 ,
f7 , fs ,and& , respectively, defined in an analogous way as given in Table IV. Com-
paring the force constants obtained above for Fe&Is, FeC& , and FeClz , one is
tempted to transfer the values of fi , j’6 , f7 , and 18found for Ve2C16as a first approx-
imation. Furthermore the three absorption bands at 438, 325, and 250 cm-’ as-
signed to Fe&l4 probably are to be considered as Fe-Cl bond stretching modes
(ring and outer FeCl groups) predominantly associated with principal force con-
stants ji , fi , and j3 . Starting from this assumption one is then lead to the assign-
ment collected in Table VI and the following values for the force constants are ob-
tained:fi = 1.172, jz = 1.692, j3 = 0.575,jb = 0.208,j~ = 0.512,js = 0.159 mdyn/
A. The fit between observed and calculated frequencies is only slightly affected by
the bending constant f3, but is considerably improved by introducing an interac-
tion constant jiz between the inner and outer FeCl stretching coordinates; for
fi2 a value of -0.163 mdyn/W was found. As shown in Table VI, some of the
calculated normal frequencies depend only on transferred force constants. Indeed
most of the calculated frequencies below 200 cm-’ are only slightly affected by
fl , f2 , and fiz and therefore should be considered as estimates.
give a summary of the matrix shifts observed. These are collected in Table VII
and reveal the following remarkable behavior:
(i) Fe&l6 exhibits considerable gas-matrix shifts of the same order as observed
for LiF, though Fe&X, has no permanent dipole moment. Surprisingly the lowest
FeCl stretching mode w shows a negative gas-matrix shift, whereas the higher
modes have positive shifts. On the other hand, Fe&l6 vibrational modes experi-
ence only small shifts between Ar, Nz , and cyclohexane matrices.
(ii) FeClz and very probably also Fe&h show very small gas-matrix shifts,
TABLE VI
ASSIGNMENT OF Fe&l4 INFRARED MATKIX SPECTRUM
vi obs (cm-l)
Assignment (normal mode) Species Y, talc (cm-.‘)
Ar matrix
* Values depending only on force constants transfered from Fe&Id force field.
TABLE VII
MATRIX SHIFTS FOR SOME INFRARED ABSORPTION BANDS OF Fe&la , FeClt ,
AND FezC14
FerCle
FeClz Fe&
%3 VI1
va r13 VI6 Yl?
but drastic shifts going from Ar to PIT2matrices, although they are also nonpolar
molecules.
V. THERMODYNAMIC FUNCTIONS
5.1 Fed%
Kubaschewski (32) gives a compilation of available data of Fe&&, comprising
the standard entropy and heat capacity of the solid and the sublimation enthalpy
at melting point (580°K). Kangro and Petersen (8) published data on the heat
capacity of the gas. Using standard thermodynamical procedures, the entropy of
the ideal gas state may be obtained. For calculation of thermodynamic function
in the rigid rotor-harmonic oscillator approximation the molecular data and force
constants discussed in Sect. 4.1 first were used to calculate the thermodynamic
functions of the eight most abundant isotopic species (D2h-56Fe?C1F,, 0.075;
C,-56FeFC1FC1, 0.22X; Czv-56Fe~C1~7C1, 0.114; two C1-56Fe,35C1t7C12 , 0.165
each ; two C2,-56Fe235C1437C12, 0.083 each; C2,56Fe2”5C1,37Cl, , O.OS3). There-
from, the mechanical equilibrium concentrations in a natural isotopic mixture of
these eight isotopes were calculated, observing the condition that the ratio of all
chlorine isotopes 35C1:“7C1 = 3. From the equilibrium composition of the gas ap-
proximate mixing effects for molar entropy and free enthalpy were determined.
The equilibrium concentrations are dominated by the symmetry numbers, as
may be expected and are practically temperature independent. The results are
listed in the practically relevant temperature range in Table VIII and XI. In
TABLE VIII
THERMODYNAMIC FUNCTIONS OF Fe&l6 (natural isotopic mixture, ideal gas,
1 atm, cal/‘K.mole)
Table XI also included is the second law entropy, which obviously deviates sig-
nificantly from the statistical value. It seems appropriate to discuss briefly error
limits. First, from Kubaschewski’s data for the entropy at 58O”K, SE,, (.580) = 1.55
f 7 eu is obtained, which agrees with the statistical entropy St, = 154.S eu. The
latter value is based on the assignment given in Table I, in which the normal
modes of species A,, A,, &, , Bt, , and B,, originate from normal coordinate
calculations solely. Among these the low lying twisting modes I+, and v16depend
strongly on the twisting force constant j8. If je is varied within a range of 0.2 5 js
I 0.6, Siti varies less than 2 eu. Among the other thermodynamic functions,
- (C” - Ho’)/T varies by 1 -. e 2 cal/“K~mole at 29X°K, whereas the remaining
ones remain practically unaltered. Furthermore, if the matrix frequencies are
converted to gas frequencies by assuming a linear relation vFas = VAr + C (VAr -
300) between gas and argon matrix frequencies (cf. discussion in Sect. 4.1) Si, in-
creases by approx 1 eu. On the other hand, little information is available about the
accuracy of the gas specific heat, on which the large discrepancy between S,“, and
ASi,at 298°K depends critically, owing to
We therefore suggest for the normal entropy the value given in the last row of
Table XI.
5.2 FeCh
No second law data for this molecule seem to be available. For computation of
statistical thermodynamic functions molecular data listed in Sect. 4.3 and all four
isotopic species with 56Fe are used in the same procedure as outlined for FezCls .
The results are collected in Table IX. The internal consistency of the data given
TABLE IX
THERMODYNAMIC FUNCTIONS OF FeC13 (natural isotopic mixture, ideal gas,
1 atm, cal/“K.mole)
in Tables VIII and IX may be checked by comparison with data published on the
dissociation process
-Fe&&(g) + 2 FeC&(g) = 0
From AHis = 32.1 kcal/mole (9, SZ) one may infer by standard thermodynamic
relations and the data given in Tables VIII and IX, that for FeCb (g) SL (932) =
103.8 eu, whereas statistically X:,(932) = 103.4 eu is obtained. Again the errors
in S$ (298) originating from the use of matrix frequencies and calculated normal
frequencies should not exceed 1 eu.
Using the molecular data discussed in Sect. 4.3 and considering all three isotopic
species with 56Fe, the statistical thermodynamic functions collected in Tables X
TABLE X
THERMODYNAMIC FUI\‘CTIONSOF FeC12 (natural isot,opic mixture, ideal gas,
1 atm, cal/“K.mole)
TABLE XI
ST.\ND.%RDENTROPY OF IRON HALIDES (natural isotopic mixture, ideal gas,
1 atm, cal/“K.mole)
S*t
On 127.4 82.2 69.2
Zb 8 4 3
s !,P 133.2 78.7 f 4
A!P 128 z!z 2 82.5 f 1 69.5 f 1
and XI are derived. The statistical value deviates significantly from the value ob-
tained from the standard entropy of the solid and the sublimation enthalpy at
298%, which yields sh(298) = 78.7 f 4 eu. Carlson et al. (18) estimated the
statistical entropy of NiClz (g) to X$ (850) = 85.6 eq5 whereas the second law
entropy is L&(850) = 8.5.4 eu. For FeCl:! (g), L$ (850) = 84.2 eu (cf. Table XI),
a value which by comparison with Carlson and co-workers’ NiClz value, supports
the approximate correctness of the statistical value. We therefore presume that
the sublimation and vapor pressure data for FeClz are doubtful and should be re-
examined. However, low lying crystal field levels of the 3d5 system (19) might
contribute noticeably to thermodynamic functions and the errors indicated in
Table XI may be larger.
5.4 Fe&‘14
The data derived from matrix spectroscopy cannot be considered as reliable
enough to warrant a compilation of thermodynamic data.
ACRNOWLEDGMENT
We express our gratitude to t.he Swiss National Foundation (project Nr. 5161) and to
Messrs. Sandoe, Basle, for financial support of this work. Furthermore, we gratefully
acknowledge the assistance of Messrs. H. Abplanalp and J. Keller in various phases of
this work. In particular we thank Mr. P. Groner for va.lua.hle help with the manuscript.
REFERENCES
1. K. J. PALMERAND N. ELLIOT, J. Bmer. Chem. Sot. 60. 1862 (1938).
2. N. WOOSTER, F. j. Kristallogr. 83, 35 (1932).
3. K. GEIERBERGER,2. Anorg. AZZg. Chem. 268, 361 (1949).
4. I. LEFICOWITZet al., J. Phys. Chem. Solids 29, 1859 (1968).
5. E. Z. ZASORIN AND N. G. RAMBIDI, J. Struct. Chem. 4, 836 (1963).
6. K. JELLINEKet al., 2. Phys. Chem., Abt. A 146, 305 (1929).
7. H. SCHAFER,2. Anorg. AZZg. Chem. 269, 53, 65 (1949).
8. W. KANGRO BND E. PETERSEN, 2. Anorg. AZZg. Chem. 261, 157 (1950).
9. W. KANGRO AND H. BERNSTORFF,2. Anorg. AZZg. Chem. 263, 316 (1950).
IO. E. Z. ZASORIN et al., Akad. Nauk. SSSR 2, (5), 705 (1964).
11. J. K. WILMSHURST,J. Mol. Spectrosc. 6, 343 (1960).
I,%‘. W. KLEMPERERet al., J. Chem. Phys. 36, 2879 (1962).
1~. A. BiicHLEn AND W. KLEMPERER,J. Chem. Phys. 40,347l (1964).
14. D. E. MILLIGAN et al., J. Chem. Phys. 42, 902 (1965).
16. A. SNELSON,J. Phys. Chem. 70,320s (1966); 72, 250 (1968).
16. A. LOEWENSCHUSS,A. RON, .~ND 0. SCHNEPP,J. Chem. Phys. 49, 272 (1968).
17. J. L. MARGRAVE et al., J. Chem. Phys. 61, 2648 (1969).
18. K. D. CARLSON et al., J. Chena. Phys. 49, 4379 (1968).
19. C. W. DEKO~K AND D. M. GRUEN, J. Chem. Phys. 44, 4387 (1966); 46, 1096 (1967),
49, 4521 (1968).
5In Carlson and co-workers’ paper (f8) the inclusiou of the isotopic effects is not men-
tioned.
“54 FREY, WERDER, AND GiiNTHARD