Studies of molecular complexes.


Material Information

Studies of molecular complexes.
Physical Description:
xiv, 168 leaves. : ill. ; 28 cm.
Jao, Tze Chi, 1940-
Publication Date:


Subjects / Keywords:
Complex compounds -- Spectra   ( lcsh )
Molecules   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis--University of Florida.
Bibliography: leaves 163-167.
Statement of Responsibility:
By Tze Chi Jao.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000580807
notis - ADA8912
oclc - 14087758
System ID:

This item is only available as the following downloads:

Full Text




'" .



To my parents


my wifce- Carmen
* 9


Special thanks are for Dr. Willis B. Person for his wise counsel,

constant enthusiasm, and encouragement throughout the entire project.

I should also like to express thanks for their help to Dr. Keith E.

Gubbins, Dr. Yngue Ohrn, Dr. Thomas M. Reed and Dr. John R. Sabin.

I am grateful to the University of Puerto Rico (MayagUiez Campus)

for financial support of my graduate studies at the University of

Florida and also for support from the National Science Foundation

(Research Grant No. GP 17818).

Support of computing expenses by the College of Arts and Sciences

of the University of Florida is gratefully acknowledged.

I should like to express my sincere thanks to Dr. Shigeo Kondo,

Mr. Robert Levine, Mr. James H. Newton, Mr. Gary Peyton and Miss

Barbara Zilles for their friendship and assistance.

Finally, I express my appreciation to Mrs. James H. Newton for

her patience in typing this manuscript.



ACKNUOWLEDGEMENTS ............................................

LIST OF TABLES ...........................................

LIST OF FIGURES .............................. .............

ABSTRACT ..................................... ...............


I. INTRODUCTION ................... ............ ..........


Introduction ........................... ..............

Source of Reagents ....................................

Preparation of Chlorine Solutions ....................

Liquid Cells .........................................

BENZENE SOLUTIONS .....................................

Experimental Procedures ...............................

Analysis of the Ultraviolet Spectroscopic Data from
Chlorine-Benzene Solutions ............................

SOLUTIONS ...............................................

Introduction ...........................................

Experimental Procedure ................................

Results of the Raman Measurements on Chlorine Solutions

Absolute Raman Intensity of Chlorine in Carbon
Tetrachloride ....................... ........... .....



















Chapter Page

BENZENE SOLUTIONS ..................................... 69

Experimental Procedure ................................ 69

Analysis of the Experimental Results .................. 88


Introduction .......................................... 107

General Expression for the Integrated .Collision-
Induced Absorption Intensity ........................... 108

N12 The Number of Collision Pairs .................... 108
+ 12 .
Evaluation of U12 d) ............ ....... 113

Explicit Expression for the Integrated Collision-
Induced Absorption Intensity ......................... 116

Evaluation of [(a /a~ )0 F2 + a2 (3F /31)0 2 ) for
the "Axial" Chlorine-Benzene Pair ................... 116

Actual Calculation of the Infrared Intensity of Chlorine
in Benzene Solution for Collision Pairs in Different
Orientations .................... ................. 120

Evaluation of [a' FP2 + a2 F] 2) for Chlorine-
Carbon Tetrachloride Collision Pairs ................. 125

SOLUTIONS ............................................ 130

Introduction ......................................... 130

Theory of the Raman Intensity Enhancement Caused by
Electrostatic Interaction (Bernstein's Collision-
Complex Theory) ....................................... 131

Calculation on the Raman Intensity Enhancement for
Chlorine-Benzene and Chlorine-Carbon Tetrachloride
Pairs in Different Solute-Solvent Orientations ........ 138

Discussion ............................. ....... 146



Theory of the Pre-Resonance Raman Effect .............. 148

Application of the Pre-Resonance Raman Effect Theory
to the Interpretation of the Raman Intensity Data
of Chlorine in Benzene Solutions ..................... 152

APPENDIX ............................................... ........ 157'

REFERENCES ................................. .................. 163

BIOGRAPHICAL SKETCH ...................... .................... 168


Table Page

I. Source and Purity of the Reagents ....................... 12'

II. Extrapolated Values at Time Zero of the Normalized
Absorbance (Ac/CA) for Different Chlorine Solutions ....... 28

III.The Values of Ke280, K, and c280 for Three Different Sets
of Ultraviolet Data from Chlorine-Benzene Solutions
Obtained from the Scott Plot .............................. 35

IV. Values of KE K, and e for Chlorine Solutions from
Scatchard Plots .................................. 39

V. Observed Raman Shifts, Half-Band Widths and Relative
Intensities of Chlorine Solutions in C6HR-CC1 Solvent
6 6 4
Mixtures as a Function of Benzene Concentration ........... 52

VI. Depolarization Ratio of Chlorine in Different Chlorine
Solutions ................................................. 65

VII.Integrated Infrared Molar Absorption Coefficients (A) and
the Parameters of the Lorentzian Functions for the Cl-Cl
Vibration of Chlorine in Benzene Solutions ................ 100

VIII.Parameters Used for the Calculation of the Collision-
Induced Infrared Intensity of Chlorine in Benzene or
Carbon Tetrachloride Solutions ............................ 121

IX. Calculated Collision-Induced Infrared Intensity for
Chlorine-Benzene Pairs in Different Orientations .......... 122

X. Calculated Collision-Induced Infrared Intensity of Chlorine-
Carbon Tetrachloride Pairs in Two Different Orientations .. 127

XI. Coefficients of the 1/R Terms in the Expressions for
S2 2
A(a ) and A(y A) (Eqs. 7-21a and 7-21b) for Several
Different Solute-Solvent (AB) Orientations ................ 143

XII.The Calculated Values of A(RAB 2 (YAB )R'
N /N and the Enhancement of Intensity (AP) ............. 145
12 A


A-i. Parameters Used for the Calculations of the Pair-
Correlation Functions of Chlorine-Benzene and
Chlorine-Carbon Tetrachloride Pairs ...................... 160

A-2. The Potential Function ua(R12, wl' '2)/kT for Chlorine-
Benzene Pairs as a Function of Relative Orientation
(at T = 2980K) .......................................... 161

A-3. The Potential Function Ua(R12, W', 2)/kT for Chlorine-
Carbon Tetrachloride Pairs (at T = 298K) ................ 162





Figure Page

1. The NMR spectrum of the residue (redissolved in CC1 )
after benzene solutions were evaporated .................. 16

2. Schematic diagram of the infrared liquid cell ............ 20

3. Ultraviolet spectra of chlorine in benzene taken at
different concentrations and at different times .......... 25

4. Plots of the normalized absorbance of chlorine at 280 nm
vs. time for different chlorine solutions (around 0.001 M). 27

5. Ultraviolet spectra of chlorine solutions in a 30% (v/v)
C6H6 and 70% (v/v) CC14 solvent .......................... 31

6. Replotted spectrum of 0.027 M chlorine solution in 30%
(v/v) CH6 and 70% (v/v) CC14 mixture (0.1 mm path-
length) ... .... ................ ......... .............. 32

7. Scott plots of the complex of chlorine and benzene ....... 34

8. Scatchard plots for the complex of chlorine with benzene;. 38

9. Raman spectrum of the C6H6-CC -CHC13 mixture at a ratio of
6:3:1 .................................................... 44

10. Raman spectrum of 0.12 M chlorine in a C6H -CC1 -CHC1
mixture at a ratio of 8:1:1 .......... .. .............. 47

11. Raman spectrum of 0.5 M chlorine in 6:4 C H -CC1 ........ 48

12. Raman spectrum of the chlorine-free 6:4 C H -CC1
66 4
solution ................................................. 50

13. Plot of the Raman spectral half-band width of chlorine vs.
the concentration of benzene (CD) ........................ 54

14. The relative Raman intensity of chlorine I as a
function of the benzene concentration (M) ................ 57
1 o
15. Plot of l/[I I ]R( vs.-pB /p for chlorine in
benzene solutions ....................................... 60

1 o
16. Plot of p /[(I I R ] vs. (p + C C) for
B R(v) R(v) B A
Eq. 4-4.......... ... ..... ...... ........................ 64

17. Infrared spectra of chlorine (about 0.6 M) in benzene
solutions as a function of the exposure to fluorescent
lights............................ .................. ..... 71

18. Infrared spectra of 0.33 M chlorine in benzene. The path-
length of the cells for all studies is 3 mm............... 76

19. Infrared spectrum of chlorine in 60% (v/v) benzene and
50% (v/v) carbon tetrachloride. The pathlength is 3 mm
for all spectra....................... ............. 78

20. Infrared spectrum of chlorine solution in 20% (v/v)
benzene and 80% (v/v) carbon tetrachloride. Pathlength
is 3 mm .................... .............. .......... 80

21. Spectrum of chlorine in carbon tetrachloride. Path-
length is 3 mm for all spectra............................. 82

22. Infrared spectra of chlorine in carbon tetrachloride. The
pathlength is 3 mm for all measurements................... 85

23. Infrared spectrum of chlorine in carbon tetrachloride solu-
tion with the 6 mm pathlength liquid cell................. 87

24. Replotted infrared spectra of chlorine in benzene solutions;
the concentration of chlorine in each solution is 0.5 M
and the pathlength is 3 mm................................ 90

25. Lorentzian curve (F) fitted to the observed absorption
spectrum of chlorine in benzene (I)....................... 93

26. Lorentzian curve (F) fitted to the observed absorption
band (I) of chlorine in 60% (v/v) benzene and 40% (v/v)
carbon tetrachloride.................................... 95

27. Lorentzian curve (F) fitted to the observed absorption
band (I) of chlorine in carbon tetrachloride.............. 97

28. Scott plot of the infrared data for the complex of
chlorine with benzene (neglecting the intensity of "free"
chlorine in the presence of carbon tetrachloride)......... 103

29. Scott plot of chlorine complex with benzene, using a
recalculated intensity of chlorine, as given by Eq. 5-9,
and described in the text................................. 105



Figure Page

30. Coordinate system used in defining the orientation of
linear molecules............................................ 110

31. Coordinate system used in defining the orientation of
chlorine and carbon tetrachloride......................... 126

32. Coordinate system and symbols used for deriving the
electrostatic potential due to a dipole.................. 132"

33. The relative orientation between cartesian coordinates
(x, y, z) and the polar coordinate unit vectors
(eR e ) .......... ...... ............ 133
iRm 0e lo o o e e~ O e O OO~ o o O e I

Abstract of Dissertation Presented to the
Graduate Council of the University of Florida in Partial
Fulfillment of the Requirements for the Degree of Doctor of Philosophy



Tze Chi Jao

March, 1974

Chairman: Willis B. Person
Major Department: Chemistry

Molecular complexes of chlorine in benzene solutions were studied

by ultraviolet, Raman and infrared spectroscopic techniques. Ultra-

violet spectroscopic studies verify the order of magnitude of the

previously reported equilibrium constant for the assumed benzene-
chlorine 1:1 complex. Its value, K = 0.025 + 0.015 liter mole1,

agrees quite well with values obtained from Raman and infrared spectro-

scopic data.

Careful experimental measurements were made of the absolute Raman

and infrared intensities of the CI-C1 stretching vibration of chlorine

in solutions of benzene in carbon tetrachloride. Specially designed

infrared long path liquid cells, inert to chlorine, were constructed

and used to obtain the absorption spectra of chlorine in the mixtures

of benzene and carbon tetrachloride. The resulting infrared studies


were more accurate than previous work, and the collision-induced infrared

absorption spectrum of C12 in CC14 could also be observed.

The v 'venumber of the Raman band of C12 shifts uniformly from
-1 -1
530 cm benzene solution to 543 cm in carbon tetrachloride solu-

tion, alti.>ugh the half band width clearly broadens at a 1:1 ratio of

benzene to carbon tetrachloride. The infrared absorption by chlorine

shifts slightly from 527 cm-1 in pure benzene to 532 cm1 in 2.26 M

benzene in CC14; with a big jump to 545 cm- for chlorine in pure

carbon tetrachloride. This difference between the Raman and infrared

spectra for chlorine in benzene solutions suggests that the infrared

absorption is from only the completed chlorine, while the Raman band

is a composite of two unresolved bands, one for the completed chlorine

and the other from free chlorine. The absolute infrared intensity and

the relative Raman intensity for the C1-C1 vibration of chlorine both

increase approximately by a factor of five from those for the Cl2

solution in carbon tetrachloride to the solution in benzene.

In order to interpret these observed Raman and infrared spectra,

theoretical calculations were made of the effect on the Raman and

infrared intensities from direct electrostatic interactions. The basic

theory of collision-induced infrared absorption intensity by Van

Kranendonk and Fahrenfort and of the Raman intensity enhancement by

Bernstein was applied, using statistical mechanics of liquid structures

with angularly dependent pair-correlation functions of chlorine in

benzene solutions. The isotropic and anisotropic effects were taken

into consideration for the calculations of both Raman and infrared

intensities. The calculated intensities were then compared with the


experimentally measured values.

The electrostatic effect predicts a maximum Raman intensity

enhancement of 100% for chlorine from gas phase to solution in CC14,

and 134% for solution in benzene, compared with an observed enhancement

of more than 400, from solution in carbon tetrachloride to solution in

benzene. The contribution of the electrostatic 'effect to the infrared

intensity of chlorine in benzene was predicted to be, at most, 50% of

the total measured intensity of 333 cm mmole-I. The vibronic charge-

transfer effect may be responsible for the intensification of infra-

red absorption while a pre-resonance Raman effect involving the charge-

transfer absorption band may explain the enhancement of the Raman band.




Complexes of halogens with benzene have been studied quite

intensively in the last two decades by both experimental and theoretical

methods (1-34). The general subject of molecular complexes has been

reviewed by several authors (35-40). The experimental studies include

those by ultraviolet, visible, Raman and infrared spectroscopic techni-

ques, while the theoretical studies are concerned with the mechanism

of the interaction and the theory of the associated experimental

phenomena, and particularly the relative importance of the electro-

static and charge-transfer effects. The complexes of iodine with

aromatic donors have been studied quite thoroughly by these methods.

Less attention has been paid to the complexes of bromine and chlorine

with benzene or other aromatic donors, because the latter are more


In the first careful study by Andrews and Keefer (6) the complex

of chlorine with benzene was found to exhibit an additional strong ul-

traviolet adsorption band near 278 nm, which is absent in the spectrum

of each individual constituent solution. From their ultraviolet data,

they found by the Benesi-Hildebrand method (2) that the equilibrium

constant of the complex is about 0.033 liter mole- with a maximum

-1 -1
absorptivity e for the complex at 280nm of 9090 liter mole cm .
The equilibrium constant for the complex of iodine with benzene


was found (2) to be about 0.17 liter mole- with E of about 15000
-1 -1
liter mole -cm so that the complex of chlorine with benzene is

weaker than that for iodine with benzene, in agreement with the expect-

ed Lewis acid strengths of chlorine and iodine.

In an attempt to explain the results of these ultraviolet spectro-

scopic studies (1, 2) of iodine with benzene, Mulliken (11) introduced

the "charge-transfer" resonance structure theory. This theory described

the ground state electronic wavefunction N of a donor-acceptor

complex approximately by a combination of two resonance structure

functions V and ~Y:

N (D'A) = aTV(D,A) + bt~(D+ A) .(1-1)
N 1
(no-bond) (dative)

Here a and-b are the coefficients of the no-bond and dative structures,

respectively. In the ground state of a weak complex, a is expected

to be approximately 1.0 and b expected to be less than about 0.1.. The

stability of the complex depends on the extent of the mixing between

the wavefunctions of the no-bond and dative structures.

If the ground state structure of a complex is given by VN' then

according to the "charge-transfer" theory, there is an excited state

which is called a charge-transfer state, given by

V = b 0Y(D,A) + a T1(D+ A) (1-2)

The coefficients b and a are determined by the quantum theory require-

ment that the excited state wavefunction be orthogonal to the ground

state function:

J T~ d = 0.

The electronic absorption frequency of the new band formed in the

complex corresponds to the energy difference between the ground state

(N) and this charge-transfer excited state (V) of the complex. The

charge-transfer theory also explains the characteristically high in-

tensity of the electronic absorption band of the complex (for

further discussion see Ref. 37).

The first infrared study of the complex of chlorine with benzene

was made by Collin and D'Or (13). A new weak and relatively broad

absorption band was observed near 526 cm-1 for solutions of chlorine

dissolved in benzene. The Raman shift for chlorine (35 C12) in carbon

tetrachloride was known (22) to be at 548 cm-1. More quantitative

studies of the infrared spectrum of chlorine in benzene were carried

out by Person and associates (18, 21).

An attempt was made by Friedrich and Person (26) to interpret the

changes in vibrational frequency and intensity of the halogen-halogen

stretching vibration when the halogen molecule, an ao acceptor (37),

complexes with benzene, a br donor (37) (or with other electron donor

molecules) in terms of charge transfer theory (11). They postulated

that a relationship existed between the vibrational frequency shift

(Av) and b the coefficient of the dative wavefunction:

FIN = (b2 + abS1) = Ak/k0 = 2Av/v0 (1-3)

Here FN is the weight of the dative structure in TN (FN = b + 2abS01)
IN N lN 01"
S is the overlap integral between T and P k is the force constant
01 0 1
and v0 is the vibrational wavenumber of the isolated molecule, while

Ak and Av are the changes (k0 k or v0 v, respectively) in the


In the following, we shall summarize some of the material from

Ref. 41 relating to the theories of charge-transfer and of electro-

static effects for the interpretation of the changes in infrared

intensity of halogens forming complexes with benzene. The experi-

mental absolute infrared intensity (A.) of the ith normal vibrational
mode of any molecule can be related to the dipole moment derivative

(Dp/E1) by (41)

2 2 2
A. = Nig./3c (Op/2.) = K(ap/3a.) (1-4)
i i i 1

Here N is the Avogadro's number, g. is the degeneracy of the ith

normal vibration at wavenumber v., (Dp/*.) is the magnitude of the
1 1
dipole moment derivative for the ith normal vibration with respect

to normal coordinate (5 ). The integrated molar absorption coeffi-
cient Ai is defined experimentally by:

'A = (l/nt)'fn(I /I)dv (1-5)
1 0
Here n is the concentration of the absorbing molecules in liter mole

a is the pathlength in cm, I and I are the transmitted and incident

intensities of monochromatic light at wavenumber v .

Based on charge-transfer theory (11), Friedrich and Person (26)

argued that the dipole moment derivative for the C1-Cl stretching

vibration in the complex is given by

ap/a = '91 /3 + (2b).(3b/8 ) Ef | (1-6)
1 N i i 1 0

Here p/8a represents the magnitude of change in the C1-Cl dipole
N i
moment of the uncomplexed Cl molecule when the Cl-Cl coordinate (E.)
2 i
-n -l +
changes; Iu 11 0 is the difference between the dipole moment in the
dioemmn 0nh

dative state p~ and that in the no-bond state i0, and is approximately

equal to .VN' the electronic transition moment for the charge-transfer

absorption band. The second term (ab/~C.) is the change in the coeffi-

cient b as the C1-C1 bond length changes (C ) and gives the vibronic

charge-transfer effect.

The derivative (3p/9E.) is related to the derivatives with
respect to the internal coordinates (R.) by:
+ +
p/aci = Z L.ji(p/3R) (1-7)

Or conversely,
-1 +
p/R. = E L Op/S ) (1-8)
3 i ij i

Here L.. is the jith coefficient from the normal coordinate transforma-
1 -
tion, while L is the corresponding element from the inverse transfor-
nation. In analogy with Eq. 1-6 we have

8p/aR.= ap/3R + (2b)(9b/lR) ~i 0 (1-9)

Assuming a8 /3RR is not different from the dipole moment derivative

of the free molecule, it has been shown (41) from Eq. 1-9 that the

vibronic contribution (M ) to the infrared intensity change, for the
special case Rj = R the stretching coordinate of the X-Y bond of a

completed halogen molecule, can be obtained from the following equation:

Md = (ap/R) ( N/R ) (2FNFIN)(aEv/3R1)/A -1


Here FN is the weight of the no-bond structure in TN (F N a + abS )
ON N ON 01
and is related to FIN by FON + FIN = 1; A is the difference between the

energies of the dative and no-bond structures and Ev is the vertical

electron affinity of the acceptor, and appears because b depends on E ,
so that Db/DR is related to 3E /PR The comparison between the
1 A i
calculated M and the observed values is shown in Table 1.8 of Ref.
41. The agreement is qualitatively good, but the calculated values

in some cases are larger than experimental by a factor of 2 to 5.

The defect of the model of Friedrich and Person arises from two

sources: one is the oversimplified assumption that 3y /aR can be
N 1
approximated by 3p O/R the dipole moment derivative of the free
molecule, the other is because aE V/R (and, to a lesser extent, the
A 1
other parameters such as FN ) cannot be obtained a priori.
IN -
In an alternative treatment, Hanna and associates (30, 31) attempt-

ed to interpret the change in the infrared intensity of halogens in

benzene as a purely electrostatic effect. They estimated an induced

dipole for chlorine in a complex arising from the interaction of the

field along the six-fold z-axis from the benzene molecule with the

polarizable halogen molecule:

i = (1/2) a (E + E). (1-11)

Here a is the polarizability of the halogen parallel to its axis (in

the z direction); E is the field from the benzene at the nearest

halogen atom (X), and E is the field at the halogen atom (Y) further

away from the benzene. Taking the derivative of Eq. 1-11 with respect

to the internal coordinate R of the halogen,

(O /R ) = (1/2)(a '/R )(E + E )
i 1 1 2

+ (1/2) a'[(DE 1/R ) + (3E /aR )] (1-12)
1 1 2 1

The calculated infrared intensity for the halogen completed with benzene

from Eq. 1-12 is compared with the observed values (31). Again the

agreement is good (within a factor of 2).

The values from the model of Hanna and associates (30, 31) appear

to be the right order of magnitude, but the parameters needed for this

calculation are not easy to determine. For example, the polarizability

derivative (aa /aR ) of chlorine was estimated from the semi-empirical
Lippincott model (42); the experimental studies reported here (Chap. IV)

found that this estimate is too large. There is also some considerable

uncertainty in the parameters chosen for E and E since estimation of
1 2
the correct values requires quite good molecular wavefunctions of the

benzene molecule.

The subject of electrostatic "collision-induced" infrared

absorption is an old one, having been studied by Van Kranendonk (43),

Fahrenfort (44) and others (45). In the original formulation of

collision-induced infrared absorption given by Van Kranendonk (43)

and Fahrenfort (44), homonuclear diatomic molecules or nonpolar linear

molecules under high pressure are predicted to have induced infrared

absorption due to (a) an atomic distortion effect and (b) a quadrupole

distortion effect. The former arises from mutual repulsion of electron

clouds at small intermolecular separation, while the latter comes from

the interaction between one polarizable molecule and the electric

field generated by the quadrupole moment of the other molecule. Hanna

and associates (30, 31) have considered the second effect in the

benzene-halogen case. In line with the correct collision-induced in-

frared absorption theory (43, 44), the estimated induced infrared

absorption intensity for chlorine in benzene solution should be obtained

as an appropriate statistical average over all intermolecular orienta-

tions, and not just for one orientation, as assumed by Hanna et al.

Using the same argument as Hanna and associates (30,.311, Kettle

and Price (33) applied the theory of collision-induced far infrared

absorption (46-50) to interpret quantitatively the observed results

of their studies on solutions of iodine and bromine in benzene. They

reported that the intensities of the far-infrared absorption of iodine

and bromine in benzene solutions could be adequately explained by con--

sidering only quadrupole-induced dipole moments, but they had to

assume different values for the parameters from those given by Hanna

and Williams (31).

Meanwhile, a Raman spectroscopic study of iodine in benzene solu-

tion had been carried out by Klaeboe (27). He observed only one Ramian

band in the iodine solution, and did not see a Raman shift for the

uncomplexed iodine. Later, Rosen, Shen andStenman (29, 32) made a

more systematic study of the Raman spectrum of iodine in benzene solu-

tion. They observed a uniform change in frequency of the Raman shift

for iodine as the benzene was increasingly diluted by the addition of

the inert solvent; e.g., n-hexane. No change in band shape was ob-

served on dilution. They concluded that the reason for not resolving

two Raman peaks, one for the completed iodine and the other for an

uncomplexed one, was the weakness of the charge-transfer interaction

between iodine and benzene, so that iodine could interact with more

than one donor. If the observed single Raman band for iodine in

benzene is due to the weak charge-transfer complex, then the weaker

charge-transfer complex of chlorine with benzene may exhibit the

same uniform change in vibrational frequency with dilution for both

Raman and infrared spectra.

SAfter reviewing these previous studies of complexes of halogens

with benzene, we see there is considerable conflict in the interpreta-

tion of the observations of infrared absorption of halogens in benzene.

In order to understand better the difference in the interpretations

of the spectroscopic phenomena associated with the complexes of halo-

gens with benzene, we decided to re-investigate the benzene-chlorine

system. We remeasured the infrared intensity of the complex of

chlorine with benzene under carefully controlled experimental condi-

tions, and also studied the infrared absorption spectrum of chlorine

in carbon tetrachloride, which is not expected to form a charge-trans-

fer complex with chlorine, in order to compare it with the one in

benzene. On the other hand, we re-examined the theory of induced in-

frared absorption more carefully. With the improved liquid structure

theory (51, 52) recently available, we applied the collision-induced

infrared absorption theory of Van Kranendonk (43) and Fahrenfort (44)

in order to determine whether intensities observed for chlorine in both

the benzene and carbon tetrachloride solutions could be explained

quantitatively by this theory alone, without any charge-transfer effect.

Secondly, we extended the work of Rosen, Shen and Stenman (29, 32)

by studying the Raman spectrum of chlorine in benzene-carbon tetra-

chloride solutions in order to understand better the nature of weak

charge-transfer complexes of halogens with benzene. There are three

reasons for our choice of chlorine instead of bromine or iodine for

the Raman work: (1) the complex of chlorine with benzene is much weaker

than a complex of iodine with benzene (6), so that the negative results

for benzene-iodine could be even worse for benzene-chlorine, (2) the

chlorine solution absorbs less of the Raman exciting line because the

solution is more transparent, so it could be observed easier in the

Perkin-Elmer LR-1 spectrometer, and (3) the infrared absorption of

chlorine is in a region which can be studied with less difficulty (13,

17) than for the benzene-iodine solutions.

We repeated the study of the ultraviolet spectrum reported by

Andrews and Keefer (6) for two reasons: first, we wanted to see if the

equilibrium constant of the complex of chlorine with benzene changes

with concentration of chlorine, since the Raman and infrared experi-

ments required a high.concentration of chlorine, while the equilibrium

constant obtained by Andrews and Keefer was presumably determined in a

dilute solution (they did not report their concentrations); secondly,

error analysis (53, 54) of the determination of equilibrium constants

for weak molecular complexes has shown that equilibrium constants as

small as that reported for chlorine with benzene cannot be determined

with any meaningful accuracy. For a complex with saturation function

(s, the fraction of completed Cl2 in the solution) between 0.01 and

0.1, the relative error in both the equilibrium constant and absorptiv-

ity (E) will vary between + 10 and + 100%, respectively (52a). It is

thus of interest to compare equilibrium constants obtained from these

three different methods (ultraviolet, Raman, and infrared spectra) in

order to see the order of agreement that can be achieved.




In this chapter we shall discuss the experimental procedures that

were common to all of the different spectroscopic studies. These in-

clude the discussion of the source and purity of-reagents, the prepara-

tion and handling of chlorine solutions, and the cells used, including

a description of the specially constructed infrared cells. In the

following chapters we shall then describe in detail the special experi-

mental procedures for each different (ultraviolet, Raman arid infrared)

spectral study.

Source of Reagents

The reagents used for all experiments in this work are listed in

Table I. All the solvents were used without further purification. How-

ever, these solvents did not have impurities detectable at the conditions

of our spectral studies. The two different grades of chlorine gas did

not show any differences in our spectra.

Preparation of Chlorine Solutions

The chlorine solutions for all the different spectral studies were

prepared in the same way. Chlorine gas was introduced into the solution

through a gas dispersion tube connected by Tygon tubing to a trap filled

with glass wool to filter any solid impurity and then to the flask of

chlorine. The flow rate of the chlorine gas was not regulated by a





chlorine gas

benzene and
carbon tetrachloride


sodium thiosulf-ate
(Na2S203 5H20)

potassium iodate

Potassium iodide

research or ultra-high
purity grade

(1) spectrophometric grade (1)

(2) spectro-quality reagent (2)

analytical reagent

analytical reagent

analytical reagent

reagent grade


Matheson Gas

Chemical Works

Matheson Cole-
man and Bell

Allied Chemical

Fisher Scientific

Fisher Scientific

Fisher Scientific


University of


regulator, but was controlled in such a way that the concentration of

the chlorine was around 0.1 M after bubbling Cl for five minutes,
around 0.2 M after 10 minutes, and so on. It may be safe to say that

this flow rate is about 5 bubbles per second. The bubbling process

usually took 5 to 30 minutes depending on the concentration desired.

As will be discussed in more detail later, the most serious difficulty

with the chlorine solution was preventing the formation of photochemical

product. Since this product does not absorb ultraviolet light nor

have Raman scattering in the same spectral region as does the completed

chlorine, a small amount does not interfere with those studies. How-

ever, it does absorb near the Cl-Cl infrared-absorption and the im-

purity is especially bothersome there.

The chlorine solutions were prepared under ordinary room lights,

and the flow rate adjusted as described above during the ultraviolet

and Raman experiments. With the experience gained from these two.

experiments, it was easy to work in the dark in order to prepare

chlorine solutions for the infrared experiments, where the photochemical

product interfered more seriously. However, when the solutions were

exposed to light, we could easily detect formation of large amounts of

photochemical product, since the solution would first become a little

cloudy, clearing again with time, because the product is very soluble

in benzene and in carbon tetrachloride. If this cloudiness was detected,

we would then discard the solution and use a freshly prepared one. In

the dark, we could monitor the solution by feeling the flask to detect

the solution heating up, since we believe that heating was always an

indication of extensive photochemical reaction.

Because of this well-known (55) photochemical reaction between

chlorine and benzene, occurring in daylight or under fluorescent lights,

it was necessary to keep the room as dark as possible. Actually, we had

found that the chlorine concentration in the benzene solution under or-

dinary fluorescent lights decreased by about 8% in 2 hours when the

stock solution was about 0.2 M. The reason for the decreasing concen-

tration of chlorine was partly due to the photochemical reaction, but

possibly was also due to chlorine gas escaping from the flask even

though the flask was stoppered. After evaporating benzene solutions of

chlorine that had been exposed to light, the solid residue was dissolved

in carbon tetrachloride. The NMR spectrum of this solution showed the

residue was mainly hexachlorocyclohexane (CH C1 6) because of the peak

at 4.7 ppm (see Fig. 1).

To make sure the Raman spectrum and the infrared spectrum that were

observed for the chlorine solution were actually due to the chlorine

molecule and not to any compound formed between chlorine and the sol-

vent, we removed the chlorine gas from the solvent after running the

spectrum either by bubbling nitrogen through the solution in the case

of Raman experiments, or by pumping out the chlorine gas (from the

solution in the cell) through a vacuum line, in the case of infrared

experiments. Following this treatment the spectrum of the clear

solution was then taken, so that absorption by the photochemical pro-

duct could be detected.

We have mentioned briefly earlier that the photochemical product

(C6H6C16) gives more serious problems for the infrared experiment. The
absorption of the hexachlorocyclohexane near 510 cm could distort

the spectrum of chlorine by overlapping the two bands. Therefore, ex-

treme care is necessary to avoid exposing the sample to light. However,





S0 u






0) .
(-1 *


N C0
, ooo

0 a

4.J* *n

S a <

*rc 0-a 0

0 r4
(-) P.4
4 *( 0
r-- ur
4-4 ri ofl














one method which can inhibit the formation of C6 H6C6 was to add some

oxygen gas in the solvent before chlorine gas was introduced. Oxygen

was reported to be a radical quencher (56). Nevertheless, we could not

completely inhibit the photochemical reaction by this procedure.

Every chlorine solution was freshly prepared for each experiment.

The concentration of chlorine was determined by withdrawing a 5 ml

portion of chlorine solution from the stock solution and transferring

it into a prepared solution containing excess potassium iodide. The

iodine released by the reaction with chlorine was titrated with standard

aqueous thiosulfate. The analysis of this stock solution was done

before and after the Raman and infrared spectra were taken. As a check,

occasionally, a 1 ml portion of chlorine solution from the sample cell

was withdrawn after its spectrumhad been taken, and its concentration

was determined. The concentrations of the chlorine solutions in the

two cases (from the cell or from the stock) were not significantly

different, but the concentrations after the experiment were about 5-6%

lower than at the beginning.

Liquid Cells

We used a set of matched silica cells with a pathlength of 1 cm

for the ultraviolet spectroscopic study of chlorine solutions at low

concentration (about 1.0 x 10-3 M), and a silica cell with light path-

length of 0.1 mm for studying chlorine solutions of higher concentration

(about 1.0 x 10-1 M). The 1 cm ultraviolet absorption cells were rec-

tangular ones with ground-glass stoppers, while the 0.1 mm one was con-

structed with platinum and tantalum parts, with silica windows, and

assembled with teflon spacers. All of them were supplied by Beckman.

A standard 2.5 ml Raman liquid cell from Perkin-Elmer was used

for the Raman experiments.

.For the infrared experiments, specially constructed liquid cells

were made. In order to eliminate the solvent spectrum, we built two

fairly well-matched sets of liquid cells, one with a pathlength of 3 mm

and the other with 6 mm pathlength; potassium bromide windows (25 mm

in diameter and 3 mm thick) were used. Chlorine did not react very

rapidly with KBr in contrast with the solvent we used. This was tested

by placing the windows in the chlorine solution for one hour. They

did not show a significant change in their infrared spectrum from the

original KBr background. However, when we tried intentionally to -ex-

tend the window contact with the chlorine solution to 24 hours, the

baseline did change and we saw actual corrosion of the KBr salt plate.

We had tried to use AgC1 windows of 1.5 mm thickness, and found the

thin windows were too soft to resist the pressure difference on evacua-

tion of the cell.

The infrared liquid cell is shown in Fig. 2. The spacer was made

of teflon. Between the KBr window and the thick spacer, we

inserted a thin Teflon spacer in order to avoid damage by the rigid

contact with the KBr window and the Teflon spacer and in order to mini-

mize leaking of the chlorine solution from the cell. To prevent the

KBr windows from cracking when they were fastened by two brass tubes

to form the liquid cell, an 0-ring was placed between the brass tubing

and the KBr window. Finally, we compressed all these components by

the two outer brass plates, each with four drilled holes, and tightened

them with bolts. A hole of the exact dimension of the glass tubing

adaptor to the vacuum line was cut into the side of the Teflon spacer.

The filled 3 mm pathlength liquid cell contained 2.5 or 3.0 ml of






41 41- 0

O -0

000 I 0C

04 wa00m r-44
o o 00

1 Cd CO 4 rid -
E r-1 0 w r0)

*r-i cl cn o
o o-^

6 d



Experimental Procedures

The ultraviolet spectrum of the complex of chlorine with benzene

dissolved in carbon tetrachloride was studied as a function of benzene

concentration at two different chlorine concentrations, one on the order

of 0.001 M, and the other in a short path cell (0.1 mm) for solutions

around 0.1 M. Five different benzene-carbon tetrachloride mixtures

were prepared, ranging from pure benzene to pure carbon tetrachloride,

and the chlorine was dissolved in each.

For the chlorine solutions with concentration around 0.001 M, a

fairly well-matched set of silica cells was used in the spectral

studies. Before any solution was prepared, the Cary Model 15 ultra-

violet spectrophotometer was turned on and allowed to warm up. At

this point, we started preparing 50 ml of aqueous potassium iodide

solution (containing 2 grams of KI) necessary for the titration of the

chlorine solution, and the different preparations of solvent needed

to dilute the stock chlorine solution. The chlorine solution around

0.1 M was prepared as described in Chap. II. The time was recorded

when a 5 ml stock solution was pipetted into the flask containing excess

potassium iodide solution; immediately following, we pipetted a 10 ml

stock solution into a 125 ml flask containing 90 ml of the same solvent.

This solution was diluted to 1/10 the original concentration by adding

10 ml of this solution to a 125 ml flask containing 90 ml of the sol-

vent.* Three more dilutions were made to form three different final

solutions ranging in concentrations from 0.0005 M to 0,0001 M in

chlorine, each with final volume of solution around 60 ml. Each flask

containing a chlorine solution was stoppered properly with a glass

stopper. At this point, we recorded the baseline of the solvent vs.

solvent with the double beam spectrometer. The spectrum (from 320 to

250 nm)' of each chlorine solution was then recorded, proceeding from

higher to lower concentration, recording the time at the beginning of

each spectrum. We repeated each measurement at least three times,

proceeding from higher to lower concentration by refilling the sample

cell solution from the flask. When all spectra were obtained, the

chlorine stock solution already added to the potassium iodide solution

was then titrated.

Spectra of chlorine dissolved in pure benzene for several different

concentrations of chlorine (each studied as a function.of time) are

shown in Fig. 3. The time interval between recording any two successive

spectra of the same solution was about 20 minutes. The concentration

of chlorine indicated for solutions A, B, and C are the values deter-

mined by titration of the stock solution, combined with the known

volume ratios on dilution, but those values are probably not correct

concentrations for the solutions at the time the spectra were taken.

From measurements of the baseline (D in Fig. 3) before and after all

spectra of the chlorine solutions were taken, we notice that the

spectrum is not reliable below 280 nm, where absorption by pure benzene

in the sample and reference beams reduces the signal to zero. Each

sample spectrum shown in Fig. 3 was measured in a fresh solution formed

by refilling the sample cell from the flask as described above. The

decrease in absorbance of each chlorine solution with time (for

example, from A-l to A-3) was most probably due to the changing

chlorine concentration, not because of the photochemical reaction

between chlorine and benzene (in such dilute solutions), but rather

because of the escape of the chlorine gas from the solution into the

vapor phase in the flask.

The normalized absorbance at 280 nm (defined as the observed
absorbance A divided by the concentration of chlorine CA) of each

chlorine solution was plotted as a functicnof time. The functions

were quite linear as can be seen in Fig. 4. There is a considerable

uncertainty in the extrapolated values of the normalized absorbance

at time zero because of the long extrapolation. The non-uniform

slope for different chlorine solution plots could be due to the diffi-

culty in defining uniquely the procedure for handling the chlorine

solutions. When we repeated some measurements for one chlorine solu-

tion from freshly prepared solutions, a different slope of the plot

was obtained. The best least-squares line through the data in Fig. 4

was used to obtain the extrapolated values of A /CA at time zero. The

results are shown in Table II. These values are then analyzed by the

Benesi-Hildebrand or Scott method (as described below) to obtain the

formation constant K, and the molar absorptivity e for the complex.
For the more concentrated chlorine solutions (around 0.1 M), we

used a single silica cell (described in Chap. II) of 0.1 mm pathlength,

measuring against air as the reference. Since these solutions were

prepared directly without successive dilution as described before, we

modified the procedure slightly from the one previously described.

Fig. 3. --

Ultraviolet spectra of chlorine in benzene taken at
different concentrations and at different times.
(A) 0.00053 M, (B) 0.00027 M, (C) 0.00013 M (all in
C12), (D) baseline; (1), (2), and (3) are the order
of successive measurements on fresh solutions (see

4... \

I- ,, 1




-' \


Nt.\ (1)

Ss-- (2)
.. (3)

before any spectrum was taken

(D) e
./ after all spectra were taken

I[ --- "





\ "9'

0.2 L










44 O

0g 0

$4-H >f > >

U Q 0 0
0 0

0 0 0 0

0 u0

0- 4 '0 '0
0 U0

*w : i
Tl ~ r)~





Concentration of
Benzene (M)




Extrapolated Valuesa
Absorbance at 280 nm

of Normalized
(Ac/CA) x 10-








a. Values obtained from the intercept (in Fig. 4) of the best least-
squares line. The uncertainty of each value is + 4% (twice the
standard deviation).

This time the concentration of chlorine in the cell could be determined

at a time much closer to that of the spectral measurement since the

concentration was high enough to be accurately determined. When the

sample cell was filled up each time with the syringe and placed in the

sample compartment of the spectrometer for the measurement, a 5 ml por-

tion of the chlorine solution was withdrawn within one minute and

pipetted into the flask containing excess potassium iodide solution.

The measurement was repeated three to four times with fresh solutions

from the flask. As before, we measured the baseline before and after

the sample spectrum was obtained. Three spectra of chlorine solutions

of different concentrations in a 30% (v/v) C6H6, 70% (v/v) CC4 solvent

are shown in Fig. 5. The spectrum A was for a 0.105 M chlorine solu-

tion, spectrum B for 0.053 M chlorine, and spectrum C for a 0.026 M

chlorine solution; spectra D were taken of the solvent in the cell

before and repeated after the spectrum of one of those chlorine solu-

tions was measured. A replotted.spectrum for 0.027 M chlorine solution

in a 30% (v/v) C6H6, 70% (v/v) CC4 mixture is shown in Fig. 6. The

spectrum is uncertain below 275 nm due to solvent absorption, so that

a clear determination of the wavelength of maximum absorbance cannot

easily be made, although it appears from Fig. 6 to be near 275 nm.

Analysis of the Ultraviolet Spectroscopic Data from Chlorine-Benzene

The data at 280 nm from both dilute and concentrated chlorine

solutions were analyzed using the Scott equation (14),

CAC /A = C/E280 + 1/KE280 (3-1)

Here Z is the pathlength in cm, CD is the initial donor concentration

Fig. 5. --

Ultraviolet spectra of chlorine solutions in a 30% (v/v)
CH6 and 70% (v/v) CC1I solvent. (A) 0.105 M, (B) 0.053 M,
( ) 0.027 M in C12, (D) baseline (solvent vs. air) path-
length = 0.1 mm.












Fig. 6. -- Replotted spectrum of 0.027 M chlorine solution in 30%
(v/v) C6H6 and 70% (v/v) CC14 mixture (0.1 mm path-

-1 I' ~-----I

(M), CA is the initial acceptor concentration (M), Ac is the absorbance

of the complex at 280 nm, e280 is the molar absorptivity of the complex

at 280 nm and K is the equilibrium constant. We assumed the absorption

at 280 nm is due to the charge-transfer absorption of the one-to-one

complex of chlorine with benzene. The reason absorbance at 280 nm

is studied is because that wavelength is the closest to the maximum

absorbance that can be studied before the solvent absorption becomes

too great.

For the data obtained from the dilute solution in the 1 cm path-
length cell, we used the extrapolated values of Ac/CA at time zero

(Table II) for the Scott plot, while for those obtained from the con-

centrated solutions in the short (0.1 mm) pathlength cell, we used the

direct absorbance readings and concentrations for the plot. At the

same time, we re-analyzed Andrews and Keefer's ultraviolet data (6)

by the same Scott plot. The three different sets of the ultraviolet

data were plotted on the same graph, and shown in Fig. 7. The error

bars for points obtained from the long pathlength cell were estimated

from the standard deviation of the least-squares fit to the extrapola-

tion plot (Fig. 4), and from the uncertainty involved in the concentra-

tion determination. The error bars for the points obtained from the

short pathlength cell are the standard deviations of three to four

repeated measurements. There was no way to estimate the uncertainties

from Andrews and Keefer's data since they did not report their experi-

mental conditions.

From Fig. 7 we can say that for chlorine solutions of low concen-

tration in chlorine ( 0.001 M) the agreement between Andrews and

Keefer's result and ours is quite good. In particular, we have the

6.0 -




0.0 I
0.0 5.0 10.0

CD (M)

Fig. 7. -- Scott plots of the complex of chlorine and benzene.
O and for 1 cm pathlength, D and --for 0.1 mm
pathlength cell, A and -- for Andrews and Keefer's
data (Ref. 6).



Boo o
4 uoo 0
0 t 0

8 +1 o a
0 o U )
o0 0 0 d
Ss -~- i. 0
S. w 0 *+-4

oQ 'I
O E4 41-C o

H C- 44
r-4 C
0 0
M a m

Q b .0 0
O 0 -t 00 0
[ Cto oou
0 60 aa ,t t
44 C s r-O Cn r-i
E- O a

M oo i4
W4 0 0
1 W40 0a ,o + 0 *-

M O 0- 1 *4 44

pu a *- oc +4
0 0 H N P4 (1 )

O E- E -'4-
.aO e N 3
cMoo ( >C < o >


4. a 414
0 U 00 4J T* 0
OE- 00 0 H- 4

O 00 00 ar C
d N ** C
i-3> 00 C 1-4
> 3 u cun + co U m
M ( r-H +1 0 N N 0 0)
C0 6 N c
04 00 rl 0 r-1
0 N r1 O H)

E-O0l -4
rM 4-4 00
Cn *-4 a 1


I 0

0 N) ao U ) C

Ii C-)N N 0 00
0 0 0 E EH 01

same intercepts of the straight lines which determine the product K280

The difference between the plots of the low and the high concentrations

of chlorine solutions may not be significant even though the factor of

the activity coefficients for solutions of different concentrations

could be different (57). However, the experimental uncertainties

were so large, we are not in a position to give any definite conclu-

sions about this point.

The values of Ke280, K, and E280 from the Scott plots for the

three different sets of ultraviolet data from chlorine solutions are

shown in Table III. For each set, the constants were calculated from

the best least-squares line. The upper and lower limits of uncer-

tainties listed for each constant were twice the calculated standard


Despite the fact that the experimental uncertainties were large,

the equilibrium constant K of the chlorine-benzene complex is believed

to be 0.025 + 0.015 liter mole-1. The large uncertainty in the value

of K is also expected theoretically (53, 54). Nevertheless, the

order of the magnitude of K indicates this complex is indeed a very

weak one. It is worthwhile to note that the saturation fractions

(defined as s = Cc/CA, where CA is the concentration of completed C12)

are between 0.1 to 0.25 in benzene for the above K values.

It has been suggested by Deranleau (54a) that the Scatchard plot

is a better method for the analysis of spectral data from weak molecu-

lar complexes. In order to check the reliability of the values ob-

tained from the Scott plot, we also used Scatchard's method to analyze

the short pathlength ultraviolet data and Andrews and Keefer's data.

The reason for not analyzing the long pathlength cell data was

because the values of the parameters Ks280,' 280' and K of this system

were within the range of those obtained from the short pathlength

cell data, and those from Andrews and Keefer's data.

The Scatchard equation is given (54a) by

0 0
A /CCD = K(280 Ac/C) (3-2)
c AD 280 c A

Here CD is the equilibrium concentration of the donor, (C = CD, the

total concentration of the donor for solutions with excess donor), k,
CA, Ac K and E280 are the same as defined for Eq. 3-1. We calculated
o o o o
Ac/ACCD and Ac/RCA for each CA at a particular CD and plotted Ac/kCACD
vs. A /ZC for the five different values of C The results are shown
- c A D

in Fig. 8. For the 0.1 mm (short pathlength) cell data, we obtained
0 0 0
the average values of A /CACD and A /kCA by averaging all A /CACC
C A D A c A D
and A /AC values, respectively, at each C The error bars in Fig.8
c A D
were obtained from the scatter of the measurements about the average
o o
value. There were five sets of (A /C ACD, A /CA ) values at each of
c AD' c A
the five different C values. We estimated K and K280 from the slope
D 280
and the intercept of the best least-squares line through these points.

We applied the same technique to analyze Andrews and Keefer's data

to estimate K, KE278 and E278. The calculated parameters of the

Scatchard plots are shown in Table IV. The upper and lower limits

were twice the calculated standard deviations.

When we compare Table III and IV, we see that the values of KE ,

K and e are not significantly different. Again, the value of K from

the short pathlength cell may be lower than the value from the more

dilute solutions. However, a line can be drawn through the error bars

for these data (Fig. 8) that includes the value of K from the long




0o 1 I I I
0 5.0 10.0 15.0 20.0 25.0

A /ICA (x 102)
c A

Fig. 8. -- Scatchard plots for the complex of chlorine
with benzene. (1) 0.1 mm short pathlength
data; (2) 0 from Andrews and Keefer's data


v v


Short Pathlengtha
(0.1 mm)

228 + 8

Andrews and Keefer

269 + 12c

0.0064 + 0.0048c

35,000 + 100,000c

0.034 + 0.01c

7,300 + 4,400

Evaluated at v = 280 mm.

Evaluated at v = 278 mm.

The upper and lower limits were twice the calculated standard

_ _


pathlength studies. We conclude that the value of Ke is 280 + 40,
with K = 0.025 + 0.015 liter mole-1 and E280 = 13,000, possibly from

values as low as 5,000 to values as high as 35,000 liter cm-lmole-.

In concentrated solutions, K may possibly be somewhat smaller. It

is not possible to reach more definite conclusions about these values

from this very weak complex (54).




We are particularly interested in studying the Raman frequency

shifts and the Raman intensity change of chlorine in solution as the

composition of the solvent is gradually changed from benzene by the

addition of carbon tetrachloride. The spectral profiles of the CI-Cl

stretching vibration as a function of the composition of the mixture

C6H6-CCI4 were carefully examined in order to understand more about the

nature of the complex of chlorine with benzene. As a check of the

reliability of the equilibrium constant of the complex determined from

the ultraviolet spectroscopic measurements, we analyzed the Raman in-

tensity data both by the method of Rosen, Shen and Stenman (32) to

estimate the equilibrium constant and also by the method of Bahnick and

Person (58). The absolute Raman intensity of chlorine in carbon tetra-

chloride was carefully determined.

Experimental Procedure

For this study, a Perkin-Elmer LR-1 Raman spectrometer was used with

a Ne-He laser with a minimum out-put of 2.7 mw. The Raman shift (A cm-)

is proportional to the grating position read in mechanical units (called

drum numbers) from the linear spectrometer scale. The actual Raman

shift was obtained from a calibration curve of wavenumbers vs. drum

number using the known wavenumbers of the emission lines of a Ne lamp.


Three lines at 650.669 nm, 653.308 nm and 659.918 nm (or 433.1 cm ,

495.1 cm and 648.5 cm from the Raman exciting line at 15,802.7 cm

or 632.8 nm) were chosen for this purpose because they had been well

studied (59). The Ne lamp source was a Pen-Ray quartz lamp operated

with a 115 volt 60 cycle/second power supply (Model No. SCT2, with

maximum current of 4 amperes from Ultra Violet Products, Inc., San

Gabriel, California). In practice, we placed the lamp in the sample

cell position in the sample compartment of the Raman spectrometer,

opened the mechanical slit to 5 microns and recorded the spectrum just

as though we were making a Raman measurement except the laser was not

turned on. As mentioned in Chap. II, the standard Perkin-Elmer 2.5 ml

multiple-path cell was used.

Because of the chemical instability of the chlorine solution, it

was desirable to work with low chlorine concentrations (around 0.1 M)

and to scan the spectrum quite rapidly. In order to obtain a spectral

resolution of 7.5 cm-1 at a 5% peak-to-peak noise level, the spectral

scan rate was about 6 cm- per minute, or 15 to 20 minutes to measure
-1 -I
a complete chlorine Raman spectrum from 480 cm to 600 cm.

Before studying the Raman spectra of chlorine solutions, we investi-

gated the solvent background in the region where the Raman band of

chlorine would appear. The Raman spectrum for the solvent mixture

of C6H6-CC14-CHC13 in a 6:3:1 ratio is shown in Fig. 9. The Raman

shift for chlorine is expected to appear between the two bands (vl of
carbon tetrachloride at 461 cm on the low frequency side and v18 of
benzene at 606 cm- on the high frequency side) with only slight over-

lap with these two solvent bands. Nevertheless, that overlap results

in the loss of most of the spectral information from the wings of the










0 o o 0 0
co 0 -It C4

SlINfI av1mIaxv

chlorine band. The extent of the overlap of the chlorine band with the

two solvent bands is shown in Fig. 10, where the spectrum was obtained

from a 0.12 M chlorine solution in the mixture of C6H -CCI -CHCI3 at a

ratio of 8:1:1.

Secondly, since we knew that it was very difficult to prevent the

photochemical reaction from occurring, especially when the chlorine

solution was irradiated by the laser in order to obtain the Raman spec-

trum, we studied the Raman spectrum of the photochemical product

(C6H6C16) in order to determine its interference with the chlorine

band. We found that the hexachlorocyclohexane (and also any other

unidentified photochemical products) did not have any observed Raman

shift near the chlorine band. This was done by an experiment in which

we let the chlorine solution (around 0.5 M) in a 6:4 C6 H-CC14 mixture

stand under the fluorescent room lights for several hours before tak-

ing the Raman spectrum of the solution. Afterwards we eliminated the

chlorine by bubbling N gas through the same solution (but not from

the sample cell solution) and measured the Raman spectrum of the result-

ing chlorine-free stock solution. We knew that the photochemical pro-

duct did form in the experiment, since the solid residue after evaporat-

ing the solvent was dissolved in carbon tetrachloride and the NMR spec-

trum showed its existence. The Raman spectrum of this particular

chlorine solution (approximately 0.5 M) in the C6H6-CC14 mixture at a

ratio of 6:4 is shown -in Fig. 11. The Raman spectrum of the chlorine-

free stock solution at the same experimental condition is shown in

Fig. 12. The apparent reduction in the Raman intensity, both for v18

of benzene and v of carbon tetrachloride bands in the chlorine solu-
tion (Fig. 11) compared to the intensities in the colorless solution
tion (Fig. 11) compared to the intensities in the colorless solution







"- 4)-
0 *


0 "

* 0

0 l
I- -

SIINI xiw ) ImIv





Si8 of C6H6

20% C_ 2

579 490


Fig. 11. -- Raman spectrum of 0.5 M chlorine in 6:4
C6H6-CC4 .

(Fig. 12), is most probably due to the absorption of the existing or

scattered light in the dark-colored solution. Within the experimental

error we can say that there is no observable Raman band due to photo-

chemical products in this spectral region (between v of CC4 and v18

of C6H6).

At this point we were ready to measure the Raman spectra of chlorine

solutions. We warmed up the spectrometer for one hour. During this

period, we prepared 100 ml of solvent, which always contained 10 ml of

chloroform. When the spectrometer was ready, we bubbled chlorine into

the solvent for 5 minutes (to prepare a solution about 0.1 M in

chlorine). A 5 ml sample of the chlorine solution was withdrawn for

titration and the Raman liquid cell was immediately filled with the

chlorine solution and placed in the spectrometer. After the Raman

spectrum of that chlorine solution was recorded, a 5 ml sample of solu-

tion was again withdrawn from the stock solution for concentration

determination. We also determined (once only) the chlorine concen-

tration for some solution taken directly from the Raman cell after its

Raman spectrum had been recorded. The result was the same as the

concentration from the stock solutions within the experimental error.

The depolarization ratio (p) of the Raman band of chlorine was

measured for three different chlorine solutions (one of chlorine in

pure C H6, one in a 1:1 C6H6- CC14 mixture, and one in pure CC14). The

concentrations of these chlorine solutions were not determined but they

were believed to be around 0.2 M. For these measurements, we used an

Ahrens prism placed between the sample housing and the monochromator

as described in the Perkin-Elmer manual (60). The spectral resolution

was the same as before. The intensity of the parallel component was



60% L




Fig. 12. --








Raman spectrum of the chlorine-free 6:4


6 6 4

- ---- ------- ---- ---

measured first by adjusting the experimental parameters so that the

maximum Raman scattering of this component was around 70% on the

chart paper scale. (We did not turn the gain higher because we

wanted to keep the peak-to-peak noise level less than 10%.) Then we

measured the intensity of the perpendicular component. In order to

compensate for the fluctuation of the laser power and the change in

chlorine concentration during the measurement, we measured again the

intensity of the parallel component immediately after we had measured

the perpendicular one. To obtain the depolarization ratio, we divided

the band area of the perpendicular component by the average band area

of the two measurements of the parallel component. Because the Raman

band of chlorine was weaker as more carbon tetrachloride was added to

the solvent mixture, the peak-to-peak noise level was higher for Cl
in CC1 As we tried to increase the amplifier gain in order to obtain

a comparable signal for different chlorine solutions, the depolariza-

tion ratio of the Raman band of chlorine had larger uncertainty (for

solutions containing more CC1 ). In particular, the noise level was

so high that the intensity of the perpendicular component could not

be measured with certainty for the chlorine solution in pure carbon

tetrachloride. Hence, only an upper limit can be given for p for Cl
dissolved in pure CC1 .

Results of the Raman Measurements on Chlorine Solutions
The Raman shifts (in cm ) observed for chlorine solutions are

shown in Table V as a function of the benzene concentration. The
values listed here are believed to be accurate within + 0.5 cm and

were obtained from the positions of the band maxima. The concentration

of benzene was estimated from a knowledge of the volume of benzene

J C H -



OH 000 0 0 0 0
Sq r-i
OH 44 C oc o C)
1 HC) +1 +1 +1 +I +1 +1 +1

U W ( *

u r- rr-rr r r4

H z


IO c

C1s 4' cfC .0 P r 0 l

Co 00
Ho co o o o vi o :
m + + 1 +1 +i

* m O 6

H H <

H 4 h
M En 4J
1 o 0
I 0 0
EI "- 0 -

cU3 H c 00. C' 1o o rC1 o L N m
z M cS 0 C 0'4 0 C4 Cf Q0
HH (0 *0

E- C,)
S( c o 0 Z p C



0 0
4- a 0 e
0 E

u0 -c N cm '
t 4 rl J0 r* *5 01 CM i

added, and the total volume of the solvent. The concentration of pure

liquid benzene was taken to be 11.3 M at room temperature. The concen-

tration of benzene in the solvent mixture was calculated by multiplying

this number by the volume fraction of benzene in the mixture (assuming

no volume change as benzene was dissolved in CC1 ).

At our rather poor spectral resolution of 7.5 cm we could not

observe (for any solvent mixture) two clearly separated Raman peaks,

one for completed chlorine and the other for the uncomplexed molecule,

so we examined the spectral profile of the chlorine band as a whole.

The half-band widths of the chlorine solutions were measured and are

also shown in Table V. The corresponding values are plotted vs. the

concentration of benzene in Fig. 13, where we see clearly the broaden-

ing of the chlorine Raman band that occurs as the ratio of C6H6 to CC14

approximates 1:1. This behavior may be an indication of two overlapping

bands, one for completed chlorine and the other for the uncoriplexed


The relative integrated intensity IR(v) of the Raman band of

chlorine was defined as given by Bahnick and Person (58),

I =I/I M
R(v) V RlefM (4-1)

Here I is the band area of the chlorine band, I is the band area
v Ref
of the 366 cm chloroform reference band (an internal standard, always at

10% by volume), and M is the total molar concentration of chlorine.

The band area was measured by a planimeter (Keuffel and Esser Co.).

The most difficult thing in defining the band area was the decision

on how to draw a baseline. We estimated by different assumed baselines

S20.0 ---T--f --- p
iI 20.0 -

15.0 -

S 0.0


5.0 10.0

CD (M)

Fig. 13. -- Plot of the Raman spectral half-band width of
chlorine vs. the concentration of benzene (C ).

that the choice of the baseline might lead to an uncertainty of + 5-10%

in band area. In general, we drew a baseline through the average back-

ground noise level in the two wings of the band. The relative inte-

grated intensities obtained according to Eq. 4-1 for chlorine solutions

are shown in column 5 of Table IV. The corresponding values of I
are plotted in Fig. 14 as a function of benzene concentration.

The uncertainties in the values of the relative intensities of

chlorine were estimated by propagation of errors. From Eq. 4-1 the

relative error is given by

2 2 -2 1/2
61R() /I = [(I /I ) + (I /I +(6M/M)
R(v) R(v) v v ref ref
From the scatter in measurements, we estimated that the individual

error is + 4.3% for (61 /I ), + 6.8% for (Iref/Iref) and 4.6% for

(6M/M), so that the uncertainty in the relative intensity (I6R /IR()
R(v) R(v)
was then found from Eq. 4-2 to be + 9.3%.

From Fig. 14 we see that there is a drastic intensification of the

relative Raman intensity of the C1-C1 stretching vibration of chlorine

as the solvent is changed from pure CC4 to pure benzene. The en-

hancement in intensity is found to be approximately by a factor of 5

based on a total chlorine concentration or by a factor of 20 based on

the concentration of completed chlorine. (Note: The fraction of

chlorine in pure benzene solution that is completed is about 0.2 of

the total concentration, based on a value for the equilibrium constant

of 0.03 liter mole-1.) There appear to be only two possible explana-

tions for this dramatic intensity increase -- one due to the non-

specific solvent effects and the other due to the effects from the

formation of the charge-transfer complex. A more detailed discussion

Fig. 14. --

The relative Raman intensity of chlorine IR(v)
as function of the benzene concentration

E3 Measured values

Calculated values from K and I determined
fTom Rosen plot and measured value of






C (M)



will be given in Chap. VII.

Rosen, Shen and Stenman (32) have derived an equation which can

be used to analyze the Raman spectroscopic data from the complex, which

may not necessarily be a charge-transfer complex. Their equation was

derived from a statistical mechanical treatment assuming that the

properties of the acceptor are a statistical average over all possible

configurations between acceptors and donors weighed by an appropriate

distribution function. For a one-to-one complex, the Benesi-Hildebrand

type equation for Raman intensity data of the acceptor was found by

Rosen, Shen, and Stenman to be

I 0 a
/[(IR IR()] = BIRK B + I/IR (4-3)

Here I is the total relative intensity of the Raman band of the
acceptor at some particular donor concentration, I is the relative
intensity of the acceptor Raman band at zero donor concentration (i.e.,

the pure CC4 solution), PB is the molar concentration of the pure donor,
PB is the molar concentration of the donor in the solvent mixture, IR

is the Raman intensity statistically averaged over all possible orienta-

tions of the one-to-one interactions between donor and acceptor, and
I 0
K /pB ( = K) is the statistically averaged "equilibrium constant" over

all the one-to-one interactions.

In applying Eq.4-3, we assumed that only one-to-one interactions

between chlorine and benzene have a significant effect on the Raman

spectrum. Using the relative Raman intensity data for chlorine in

this system, reported in the last column of Table IV, a plot of Eq. 4-3

is shown in Fig. 15, where the error bar for each point indicated the

uncertainty of the measured relative Raman intensity of chlorine. The





) 0


o ,

L, r

4 1










I I I. I

o o 0 0 0


constant IR is found from the intercept of Fig. 15, by a least-squares

+ C
fit to be 54.7 3 7 (the upper limit is undefined because the standard

deviation is larger than the magnitude of the intercept), and the con-
I o 0
stant K IR/PB is calculated from the slope to be 1.2 + 0.15. Taking
I o
the ratio, we estimated that the equilibrium constant K (= K /pB) is

0.022 + 0.055 liter mole-1 (the lower limit is undefined because the
upper limit of I is undefined). The undertainties here are the
standard deviations. Even though the uncertainty in K is large, its

value (0.022 liter mole-1) is almost identical with the equilibrium

constant determined from the ultraviolet spectroscopic measurements

given earlier in Chap. III.

In order to gain some confidence in our measured relative Raman

intensities of chlorine in different benzene solutions, we substituted
0 I 0
the values found here for IR and K /pB back into Eq. 4-3 together with
the measured value for I (2.75 liter mole ) to obtain values for

IR(v) vs. PB to be compared in Fig. 14 with the measured values of

IR(v). Comparing these calculated values of IR() 's with the observed

ones, we see that the measured I ('s are reasonably good.
Earlier, Bahnick and Person (58) had derived a different expression

for the analysis of the Raman spectroscopic data of a complex to obtain

the formation constant. By analogy to the derivation of the method

for analysis of ultraviolet spectroscopic data by Tamres (61), they

obtained an equation for a one-to-one complex (assumed to be in a

particular configuration):
1 I 0 1
pB/R(v) -R(v =(P + CA -C)/[I IR + /KR R(v)
R (4-4)

e I ), a the same as hose defined fr E. 4-3
Here I -R I -R(W I are the same as those defined for Eq. 4-3; p B is
K^V/ K.W/ RB

still the molar concentration of the donor and C is the total molar
concentration of the acceptor, while C is the molar concentration of

the complex (or the molar concentration of the completed acceptor for

the one-to-one complex). The difference between Eq. 4-3 and Eq. 4-4

is that the former is equivalent to the Benesi-Hildebrand equation (2),

while the latter is similar to the Scott equation (14).

In order to apply Eq. 4-4, a trial value of K has to be assumed,

and C is then computed for each solution. The left-hand side of Eq. 4-4

is then plotted vs. (pg + CA C), and the best straight line is

fitted to the points by least squares. The value of K calculated from

this line was used to recompute C and so on until K converges to a

constant value (for more detailed procedure, see Ref. 58). Since C for

the complex of chlorine with benzene is very small (since it was found

from the ultraviolet spectroscopic data that the equilibrium constant
0 0
was very small), (pB + CA C) is almost equal to (pB + CA), so we

used the value of K (0.022 liter mole-1) obtained from the plot of

Eq. 4-3 to calculate C and then made a least-squares fit to the points
calculated. The formation constant K and the values of I were then

estimated from this best straight line (as shown in Fig. 16) to be

0.034+ 0.045 liter mole-1 and 38.9 180.0 respectively.
0.03 15.0
Thus both methods for analyzing the Raman intensity data gave

almost the same equilibrium constant K for the complex of chlorine with

benzene and nearly equal values for the relative molar Raman intensity
0 0
IR for the completed chlorine. The slight differences in K and IR

obtained from the two methods could be due just to the differences

between the two procedures (Benesi-Hildebrand and Scott), since the

two methods weigh the experimental points differently.

The results of the depolarization ratio measurements (p) for the

C1-C1 stretching vibration measured in these different chlorine-

benzene solutions are shown in Table VI. p was estimated from the ratio

of the band area of the perpendicular component to that of the parallel

one. As mentioned in the experimental procedures section, the perpen-

dicular band of chlorine in carbon tetrachloride was weak and the

noise level was so high that it was impossible to obtain a reliable

value of p in this solvent. We believe these results show a tendency

for the depolarization ratio to increase as the benzene concentration

increases, possibly because more chlorine molecules are completed as

more benzene is added. However, we cannot make a definite conclusion

about whether any real increase in the depolarization ratio occurs as

benzene is added, since the value of p for chlorine in carbon tetra-

chloride could not be determined accurately. In order to give some

indication as to what the value of p for chlorine in carbon tetrachlo-

ride might be, we listed the gas phase value in Table VI.

Absolute Raman Intensity of Chlorine in Carbon Tetrachloride

It is possible to obtain the absolute Raman intensity and hence

the values of the average polarizability derivative a' and of the

anisotropic polarizability derivative y' from the measured relative

Raman intensity. This had been demonstrated first by Bernstein and

Allen (62) and confirmed later by Long, Gravenor and Milner (63). In

order to calculate a' and y' we have to know both P (which will be

defined in the following) and the depolarization ratio p of the band

at v. Bernstein and Allen (61) showed that a standard intensity P

(for a Raman band at v) of a compound could be defined by comparing

- 2.0







(pB + c C)

Fig. 16. --

I 0
Plot of p /[(I R( I ()] vs. (p + C C) for
Eq. 4-4.) B A
Eq. 4-4.

_~I ~I III(_~~



Concentration of
Benzene (M)


of chlorine

0.27 + 0.03

0.25 + 0.04

< 0.22c+ 0.04


gas phase


a. p = band area of perpendicular component divided by band area of
parallel component.

b. In carbon tetrachloride.

c. It was difficult to measure the perpendicular band of chlorine
in this solvent with certainty, so the upper limit is listed
(see text).

d. See Ref. 63, 64, 65 and others.


this band with the measured intensity of the.459 cm- band of carbon

tetrachloride in the pure liquid:

s [ 2 ,2 compound '2 ,2 CC14
P = [45a + 7y 2 /45a + 7y ] (4-5)
Vv 459
,2 (2 CC1
By taking arbitrarily the value of [45a + 7y ] 4 to be 1,
s -1
PS of the v = 366 cm- chloroform band was estimated to be 0.28 by Long
Gravenor and Milner (63). From our measurement, we obtained the rela-

tive molar intensity of chlorine in carbon tetrachloride with respect
to the 366 cm band of 1.25 M chloroform to be I = 2.75 + 0.26 (as
given in Table V Analytically, this means that

[ 2 ,2 C12 ,2 ,2 CHC13
[45a + 7y ] /[45a + 7y ]3 =
543 366

(2.75 + 0.26) x 1.25 = 3.46 + 0.30 (4-6)

From the value of PS for chloroform (63) and the results in Eq. 4-6,
we then obtained the standard Raman intensity of the C1-C1 stretching
vibration at 543 cm to be:

s [452 2 C12 ,'2 ,2 CC14
P = [45a + 7y ] /[45a + 7y ] = 0.969 + 0.084.
543 543 459 (4-7)
,2 --
Often in the literature, y2 of the 459 cm band of carbon tetra-

chloride has been assumed to be zero [for example, Long, Gravenor and
5 2
Milner (63)] defined Ps explicitly with the assumption that (y') is 0

for this band. We verified the reasonableness of this assumption by
calculating (y ) 2 from Bernstein's value 0.015 for p, finding that
7y contributed only 2% to the total intensity of this particular

band. Therefore, we can assume in practice that:

[45a2 + 72 CC14 = [452 1 CC14 (4-8)
[45a'2 + 7y245 = [45a 45
459 459
,2 CC14
The value of [45a ] had been measured in liquid CC1 and found (64)
459 4
to be (33.71 + 9.63)x 10-8 cm /g. From this value and Eq. 4-7 we obtain

[45a2 + 7y2] 12 = (32.3 + 10.0) x 10 cm /g (4-9)

(Here the major uncertainty is in the value for the absolute intensity

of the CC4 band.)

As has been mentioned, the depolarization ratio of chlorine in

carbon tetrachloride was difficult to determine with our spectrometer.

However, the gas phase value (p = 0.14) (65) for chlorine is most probably

the lower limit for the values (see Table VI). Using this value, we

then obtained

,2 ,2 v2
p = 3y /(45a + 4y )= 0.14 (4-10)

Solving Eqs. 4-9 and 4-10, we find that the absolute intensity of chlo-

rine is given by

a2 = (0.512 + 0.159) x 10-8 cm4/g = [(3a/3 ) ] (4-11a)

w2 -8 4 2
y = (1.32 + 0.41) x 108 cm /g = [(y/B ) ] (4-1lb)
4 4 ,2 ,2
We may convert from unit of cm /g to cm by multiplying a and y
each by the reduced mass of chlorine (29.426 x 10 g). When this was

done, we found

,2 -30 4 2
a = (0.15 + 0.047) x 10 cm = [(/a/Dr ) ] (4-12a)
1 0
,2 -30 4 2
Y = (0.39 + 0.12) x 10- cm = [(y/r ) ]2 (4-12b)
1 0


a = (3.9 + 0.6) x 1016 cm2 (4-13a)

-16 2
y = (6.2 + 0.9) x 1016 cm (4-13b)

However, the depolarization ratio of chlorine in benzene was measured

to be 0.27 (see Table VI) which may be an upper limit for p for chlorine

in carbon tetrachloride. With this value we found

-16 2
a= (3.3 + 0.46) x 1016 cm (4-14a)

y = (8.2 + 1.2) x 10 cm (4-14b)

We believe that the depolarization ratio of chlorine in carbon

tetrachloride may be much closer to the value found for the gas phase

than it is to the value obtained in benzene solution, since carbon

tetrachloride is expected to be a relatively inert solvent. Therefore

the values of a and y for chlorine in carbon tetrachloride given by

Eqs. 4-13a and 4-13b are expected to be more reliable, although the

range of possible values for these parameters allowed by the uncertain-

ties in p (compare Eq. 4-13 to Eq. 4-14) is not very large.



Experimental Procedure

The concentrations of chlorine solutions required for infrared

studies (0.3 M to 0.9 M) were higher than those for ultraviolet and

Raman studies. At these concentrations, we found that the photochemical

reaction occurs very easily. When we exposed the solution of chlorine

in benzene to fluorescent light, and determined the infrared spectrum

as a function of the time after it was prepared, an increase in the

concentration of photochemical product was observed as illustrated in

Fig. 17. Since it was impossible to prepare the solution and to fill

the sample cell in complete darkness, there was no way to prevent the

photochemical reaction from occurring. At last we tried adding oxygen

gas to the solution to act as a radical quencher (56). As a result,

the photochemical reaction was inhibited considerably, if not completely.

A small amount of iodine in addition to oxygen was reported to be even

more effective in quenching the radicals (56). Since iodine itself

forms a stronger complex with benzene than does the chlorine, we did

not try to use iodine as radical quencher in our studies for fear of

further complicating the system.

The infrared spectrometer we used was a Perkin-Elmer Model 621.

In order to minimize the change in chlorine concentration during the





00 7-
M *r- -A

) 0 0o

0 ,C
t 41 r)

0 0
.0 0

(3 4-4 4)

04 0

0 41
C 0 4)

$O P

0 m
4 i

u r0
(o <
a) i

0 0

o o

0 0
4. 4!1

m0 00
4-1 r4
x u

4l 1.1

w o
S 0 0)
S 44

- oi o
0 U 0)

3 4-1 4-1

S0 0

01 0 0
S0 0

o V :

a) 0 0

0 3
0) 0 0
04 *}
0 0

3 p

m -, u





So 00

(%) 33NV~~IZiISN~M~

course of recording the spectrum and the growth of the photochemical

product sufficient to distort the absorption band from the C1-C1

stretching vibration, we chose to sacrifice the signal-to-noise ratio

(S/N), reducing it to around 125, with a spectral resolution of about

2 cm-1. The total scan time for each measurement from 650 cm1 to

350 cm-1 was about 15 to 20 minutes.

Before the preparation of the chlorine solution, we warmed up the

spectrometer and covered the sample compartment with a black.polyethy-

lene sheet. The baseline was recorded beforehand to give the apparent

transmittance of solvent vs. solvent in the matched cells described

earlier in Chap. II. At this point, we prepared the chlorine solution,

withdrew a 5 ml portion for determination of the chlorine concentration,

and immediately filled the liquid cell with the chlorine solution using

a micropipette (5 3/4 inches long, P5205-1 Scientific Products, Evans-

ton, Illinois). The sample cell was carried to the spectrometer .in a

box covered with a black polyethylene sheet. All lights in the room

were turned off (note the sample preparation room was separated from

the laboratory containing the spectrometer), since the light emitted

from the infrared glower source was sufficient to permit the alignment

of the cells in the sample compartment. A special holder was made to

fit the sample compartment so that we could reproducibly align the

sample each time with little light. After the spectrum of the chlorine

solution was taken, the sample cell was placed in the box mentioned

before, and connected to the vacuum line in order to pump out the

chlorine. The chlorine solution always filled the cell up to the top

of the glass tubing, so that the pumping process did not reduce the

liquid level below the upper edge of the light path. It usually took

about half an hour to eliminate the chlorine gas from the solution.

The baseline spectrum of clear solution vs. solvent was then recorded.

The baseline always changed a little from that recorded at the begin-

ning but not enough to affect the total band area of the chlorine ab-

sorption by as much as 2%. Most importantly, if enough of the photo-

chemical product was present in the original chlorine solution, we

could detect it by its infrared absorption spectrum in the second clear

solution after the chlorine was removed. If that product was detected,

then we would have to repeat the measurement. Occasionally, instead

of pumping out the chlorine gas in the sample cell, we withdrew a 1 ml

portion of the chlorine solution and determined the concentration of

chlorine by titration to find out how much had been lost by both the

photochemical reaction and the escape of chlorine gas from the cell.

We had actually tried three liquid cells with different path-

lengths. The 1 mm liquid cell was made specially of tantalum metal

in our machine shop. Its construction was not much different from an

ordinary standard round liquid cell with Teflon spacer (Barnes cell

# 0004-035) such as that used to take the spectrum shown in Fig. 17.

Since the intensity of the chlorine absorption (spectrum B in Fig. 17)

with this 1 mm pathlength cell was so small, it was difficult to

observe especially the very weak absorption by chlorine in benzene

solutions diluted very much with carbon tetrachloride. Furthermore,

it was difficult to distinguish between the baseline shift due to the

photochemical reaction and the absorption by the chlorine. The 6 mm

pathlength liquid cell described in Chap. II also had been used to

measure the absorption by chlorine, but the benzene absorption in this

long path cell was so high in this region, and the actual signal-to-

noise ratio was so low, that we could not measure a reliable chlorine

absorption band with this 6 mm cell. Hence, the 3 mm pathlength liquid

cell (also described in Chap. II) was the most suitable for this kind

of measurement. The pathlength (3 mm) was obtained by measuring the

thickness of the Teflon spacer with a micrometer. Since the absorp-

tion by carbon tetrachloride was less than that by benzene in the

infrared region near the chlorine absorption band, we were able to

measure the absorption by chlorine dissolved in carbon tetrachloride

in the 6 mm pathlength liquid cell by making a special effort (more

discussion will be given later in this chapter).

The spectrum of 0.33 M chlorine in benzene in the 3 mm cell is

shown in Fig. 18 (recorded vs. pure benzene in the reference beam in

a matched cell), where it is compared with the spectrum of benzene in

that same cell, recorded with air in the reference beam, and with a

baseline of benzene vs. benzene. Benzene absorbs almost totally above

575 cm-1 and below 410 cm-1, so that we lost all the spectral informa-

tion for the chlorine-benzene solution outside the region from 410 to
575 cm However, within that spectral region, we were able to work

at quite a good signal-to-noise ratio (a ratio of 125), and so we

could obtain a reliable spectrum. (The spectra were recorded in the

transmittance mode rather than in the absorbance mode because opera-

tion of our Perkin-Elmer 621 in the absorbance mode was not reliable

at the time of the experiment.)

Spectra of chlorine solutions in benzene diluted by carbon

tetrachloride are shown in Figs. 18-21. As benzene was gradually

diluted by the addition of carbon tetrachloride, less solvent absorp-

tion was found above 575 cm-1, so that more spectral information was











10 0



. .


O 0

01 4-4

4 -4
M r-I
0 0

H 0






(%) U3twiNNlsItrvdi




0 c




0 'zt
r4 I



N e
cg C

4 In

44 <4

*H 0
-1 c


o o
o ~w

0 0

N *rI

0 4
U 0
a ,d

O O 0 0 0
0C 4T 10 00 0




0 a

N 0






.r U )


4jH H

O( 0 :

4 l4 C3
r. 0 0

1-40 U 0

B t C
10 6

0 >
> l
H o

0 -
* C

O *H
*H 0

o 0 0
a >

0 0 0
-1 01
o o H

o 0 ) w
. Ur >c
0o 0 0

CD0)( 0
P0 U
di CI
F9 0

/j /' o
(, (


/ \ :
i -



(%) aoTVIIITSV.lh

0 0













4 ,

0-4 0.

0 W

4- (






W 0
, *O

03 C

0 0H


4-1 4-1 c

0 0
C H4
) 0a)

o-l 0

o HO
- en3
CO f 4.

0 0 0 0 0 0
,3- o0 00 0


unveiled concerning the high frequency wing of the chlorine absorption

band. However, the'strong absorption by carbon tetrachloride in the

3 mm cell below 500 cm-1 made it difficult to observe the low frequency

wing of the chlorine absorption band. This situation is illustrated in

Fig. 20 and Fig. 21. The weak absorption band near 510 cm-1 in Fig. 20B

was due to the hexachlorocyclohexane.

One of the most exciting things in this work was the observance of

the infrared absorption band of chlorine in carbon tetrachloride shown

in Fig. 21. To the author's knowledge, this was the first time that

the infrared absorption by chlorine has been observed for a chlorine

solution in a solvent whose molecules have tetrahedral symmetry. This

spectrum is quite different from that of chlorine in benzene (Fig. 18)

or those of chlorine in different benzene-carbon tetrachloride mixtures

(Figs. 19-20). It is very broad and weak. If one did not use a long

pathlength liquid cell, this broad and weak absorption could easily be

confused with some baseline shift of unknown origin. To make clear that

the absorption in Fig. 21B is due to chlorine in carbon tetrachloride,

we measured this absorption band at several different concentrations

of chlorine, as shown in Fig. 22. The dependence of the absorption on

the concentration of chlorine was clearly demonstrated. To be more

convincing, the spectrum of 0.33 M chlorine in carbon tetrachloride was

measured with the longer pathlength (6 mm) cell. The spectrum obtained

is shown in Fig. 23 where the absorption is seen to be twice as large

(within experimental error) as that shown for the same solution in

Fig. 21 in a 3 mm cell.

*rW 4
80 '-



0o 4

c a


0u 0




H 0


-a 0

4 H

0 (0

4 -

14 Ci
o 0

0 a4
4d 0

0) 41)
41 r.

0 1 3

r- d V

co 0

0 a >
4 0

S -4 01
a) ) 0d

9 O O
o p0 0

C0 1

0a 0 0
41 U4 1 .

0 0 0

0 4
r a) 0

XO 41

*c 0 0

o ) U i 03

m ud Q

0 0 0 0 0 0
"10 00 0



O 0 4
o 0

41 0 9
O 4 4
0 0 U
0H 0

4 U >

* J o 4

So ma 0
U0 0 >


.ol a) o P r- -
0 C r-0 0
C c
S*d d (
0- 0 *- 0

S-1 C

Q0 ,O O ,0

j4a 4- 4- 4-

S+ 0 0 0

0 3 3


Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E212PTYD0_5O8WIZ INGEST_TIME 2012-03-09T01:18:55Z PACKAGE AA00003944_00001