Competitive interactions and glassy state extension in lithium salt solutions

Article abstract of DOI:10.1021/jp9841440

We report measurements of high- and low-temperature limits of the liquid states of aqueous solutions for two common aqueous salt solutions. The liquid range extends from the glass transition temperature to the boiling point, and is maximized when the anion basicity is minimized. These limits and other characteristic temperatures have been determined by DTA and DSC studies of solution samples which are used to generate the most complete phase diagram yet available for the LiBr + H₂O system. The glass-formation region has been extended by developing an emulsification technique suitable for use with concentrated aqueous salt solutions. Comparison of boiling points, liquidus temperatures, and glass transition temperatures shows that the reduction of the water content below 3H₂O per Li⁺ is accompanied by important energetic effects. These are not present in the corresponding lithium acetate solutions, which exhibit relatively much lower boiling points. This is attributed to the stronger water - anion interaction in the lithium acetate system which is proportional to the anion basicity. Water activity increases and boiling points are lowered when the anions are more basic, but total cohesion, manifested by the glass transition temperature, increases. By contrast, addition of anion complexing salts such as ZnBr₂ and ZnCl₂ to LiBr-H₂O solutions to reduce the anion basicity results in increased boiling points to over 200 ?°C as well as decreased crystallization temperatures. ? 1999 American Chemical Society.

Full text of DOI:10.1021/jp9841440

J. Phys. Chem. B 1999, 103, 4159-4163
4159
Competitive Interactions and Glassy State Extension in Lithium Salt Solutions
Alwarappa Sivaraman,† Hema Senapati, and C. Austen Angell*
Department of Chemistry, Arizona State UniVersity, Tempe, Arizona 85287
ReceiVed: October 23, 1998
We report measurements of high- and low-temperature limits of the liquid states of aqueous solutions for two
common aqueous salt solutions. The liquid range extends from the glass transition temperature to the boiling
point, and is maximized when the anion basicity is minimized. These limits and other characteristic temperatures
have been determined by DTA and DSC studies of solution samples which are used to generate the most
complete phase diagram yet available for the LiBr + H2O system. The glass-formation region has been extended
by developing an emulsification technique suitable for use with concentrated aqueous salt solutions. Comparison
of boiling points, liquidus temperatures, and glass transition temperatures shows that the reduction of the
water content below 3H2O per Li+ is accompanied by important energetic effects. These are not present in
the corresponding lithium acetate solutions, which exhibit relatively much lower boiling points. This is attributed
to the stronger water-anion interaction in the lithium acetate system which is proportional to the anion basicity.
Water activity increases and boiling points are lowered when the anions are more basic, but total cohesion,
manifested by the glass transition temperature, increases. By contrast, addition of anion complexing salts
such as ZnBr2 and ZnCl2 to LiBr-H2O solutions to reduce the anion basicity results in increased boiling
points to over 200 °C as well as decreased crystallization temperatures.
Introduction
systems. At the same time the insights gained from the study
may be useful for purposes of selecting aqueous systems with
While concentrated solutions of lithium halides in water have
found a place in a variety of academic studies1-4 because of
their glass-forming ability and the wide range of dynamic
behavior they exhibit, they are also of great interest in
refrigeration and cryogenic engineering. This is a consequence
of their unusual combination of high boiling points with high
fluidities (for a salt solution) and a range of endothermic mixing
compositions which makes them especially suitable as fluids
for absorption refrigeration. This is a process in which heat is
used to drive water from the solution in one place (outside)
after which the separated water is first condensed and then
evaporated, with associated cooling, back to low activity in the
salt solution in another place (inside). Outside the United States,
such a process is used to provide some 50% of the world’s
industrial refrigeration.5 The higher the temperature at which
energy is absorbed the more efficient the overall process. A
low viscosity in the concentrated solution is a requirement for
efficient fluid circulation. The need to develop systems with
high boiling points while maintaining acceptable fluidities, and
low-temperature stability against crystallization, provides a
physical chemistry challenge to this branch of industrial
technology.
specific dynamic characteristics (for instance, the decoupling
of water molecule motion from cation motion, on one hand and/
or anion motion on the other), for further academic studies.
We have therefore carried out a simple comparison of the
extremes of the liquid state, boiling on one hand and glass
transition on the other, for two common lithium salt systems in
which concentrated solutions approaching the dihydrate can be
studied in both molten and supercooled states. (In one of these,
the key industrial system, LiBr-H2O, we have had to invoke
emulsion studies to suppress crystallization near the dihydrate
composition.) The results of this study then suggest the types
of systems in which higher boiling points might be obtained
without undue increase in the glass transition temperature and
hence viscosity. A partial study of the systems LiBr-H2O-
ZnCl2, and LiBr-H2O-ZnBr2 for which the basis was provided
long ago6 is then reported, in which the boiling range increases
and the crystallization range decreases without undue increases
in viscosity.
Experimental Section
(i) Sample Preparation. Anhydrous LiBr was prepared by
vacuum dehydration of the commonly available hydrate (SIGMA
chemicals, 99.9% pure) for 4 days at 100 °C to constant weight
without any decomposition. Aqueous LiBr solutions with
compositions ranging from 9.9 mol % (34.76 wt %) to 47.0
mol % (81.05 wt %) LiBr were prepared by dissolution of
anhydrous LiBr in pure deionized water. Since LiBr is very
hygroscopic, all solutions were prepared in a drybox. Five-gram
batches of the solution were taken and stored in stoppered Pyrex
glass vials.
The high boiling points of concentrated lithium halide
solutions are a consequence of the strong Li+-H2O ion-dipole
interaction, while the high fluidity is a consequence of the weak
H2O-Br- interaction (coupled with the geometrical simplicity
of the anion). It is of interest to examine these factors in a
systematic fashion in order to better understand the appropriate
routes to improving the efficiency of industrial refrigeration
* Author to whom correspondence should be addressed. Fax: (602) 965-
2747. E-mail: caa.asu.edu.
Lithium acetate solutions were prepared in a similar manner
using LiOAc‚2H2O (Aldrich, 99.9% pure). Solutions containing
† Current address: HYCAL Energy Research Lab Ltd., 1338A-36th
Avenue, Calgary, Alberta, Canada T2E676.
10.1021/jp9841440 CCC: $18.00 © 1999 American Chemical Society
Published on Web 04/01/1999

4160 J. Phys. Chem. B, Vol. 103, No. 20, 1999
Sivaraman et al.
ZnBr2 and ZnCl2 were prepared using the anhydrous salts from
Aldrich, 99.9% pure.
(ii) DTA and DSC Measurements. A simple in-house
differential thermal analysis (DTA) setup utilizing twin digital
voltmeters with microvolt sensitivity interfaced to a laboratory
computer, was used to investigate the phase transitions in LiBr
aqueous solutions. About 0.2 mL of the solution was pipetted
into a small DTA sample cell made from Pyrex tubing of 3
mm diameter and 7 cm in length (and containing ∼2 mg Al2O3
powder to promote boiling without superheating), a chromel
alumel thermocouple was inserted in the cell, and the cell was
then sealed at the top with Parafilm to avoid sample evaporation.
The cell was inserted in one of the two 4 mm diameter holes in
an aluminum block temperature smoother of the DTA system.
An identical Pyrex cell containing small amounts of anhydrous
Al2O3 powder as a reference with the thermocouple junction
was inserted in the second hole. The junctions of thermocouples
were maintained at fixed deep immersion in the samples.
Figure 1. Typical warm-up DTA scan (for a 40 mol % LiBr aqueous
solution) used in the construction of the phase diagram for the LiBr-
H2O system. The scan is annotated for the processes responsible for
the important thermal effects observed.
Samples were quenched separately in liquid nitrogen, for glass
formation, or in the block when rapid cooling was not needed,
and then scanned during warming at ∼10 K/min by heating
the block and simultaneously recording the emf difference and
the sample temperature.
droplet size. In the case of lower LiBr concentrations, they
proved to be stable over a period of several days.
A typical emulsion consisted of 50 wt % of LiBr aqueous
solution and 50 wt % of (matrix + surfactant) liquid. Emulsions
were created by weighing out the required amount of LiBr
aqueous solution into a capped glass bottle, then adding the
required amount of (matrix + surfactant) phase. The mixture
was then emulsified by agitation with an ultrasonic disperser
for 90 s. In cases where the liquidus of the LiBr aqueous solution
was above ambient, the mixture was first heated to above the
liquidus on a hot plate and then emulsified.
The glass transition temperature, Tg, the eutectic temperature,
peritectic temperature, the liquidus temperature, and the boiling
points (pH O ) 1 atm) were determined from the peaks in the
2
differential emf trace which was continuously monitored (real-
time) and recorded by a PC-AT (386) computer as a function
of time and temperature.
Differential thermal calorimetry (DSC) experiments were
carried out using a Perkin-Elmer Differential Scanning Calo-
rimeter, DSC 7. Twenty-five milligrams of the solution or
emulsion samples were sealed in aluminum DSC pans and
subjected to continuous cooling and heating scans at 10 K/min.
Glass transition temperatures were measured on samples that
were quenched in the DSC at 500 K/min.
(iii) Emulsification of LiBr-H2O Solutions. To extend the
glass-forming region in the LiBr-H2O system, an emulsification
technique similar to that developed for LiCl-H2O solutions by
McFarlane et al.7 was employed. In preliminary studies Span
65 was used as the surfactant and methyl cyclohexane as the
dispersion medium (matrix phase). Emulsions were prepared
by mixing together about 50 wt % LiBr-aqueous solution with
the surfactant-containing matrix phase and agitating the mixture
with a hand-held high-speed drill adaption. This resulted in a
translucent emulsion containing microscopic LiBr-H2O droplets
dispersed in the matrix liquid. The emulsions thus prepared
allowed for the measurement of their glass transition temper-
atures using the DTA setup. These preliminary emulsion
experiments served the purpose of demonstrating that hetero-
geneous nucleation could be suppressed and the supercooling
range extended by emulsification. However the emulsions were
unstable and tended to separate out on standing.
To prepare more stable emulsions, two different paraffin oils
and kerosene as matrix phases as well as different surfactants
were tested. The mechanical process of emulsification was also
modified by the use of an ultrasonic disperser rod in place of
the drill adaption. Lanolin, an emulsifying agent used widely
in the literature on supercooled aqueous solutions,8,9 though not
a pure substance, was found to be highly satisfactory as a
surfactant. Emulsions made up with a 50:50 mixture of paraffin
oil and kerosene as matrix phase with 15 wt % of lanolin were
found to be most suitable for investigation. The emulsions were
translucent rather than opaque white, implying a very small
Results
(i) Phase Diagram for the LiBr-H2O System. The com-
plete phase diagram for the LiBr-H2O system was determined.
In addition to the eutectic, peritectic, and liquidus temperatures
for the various hydrates (LiBr‚5H2O, LiBr‚3H2O, LiBr‚2H2O,
LiBr‚H2O), the glass transition temperature (Tg) and boiling
points (Tb) were also determined in order to better understand
liquid-state cohesive interactions.
A typical DTA scan, with the various thermal effects
identified thereon, is shown in Figure 1. The initial steplike
(endothermic) displacement is the glass transition which is
followed by the (exothermic) crystallization of the metastable
supercooled liquid. Then follow the eutectic, one or more
peritectic transitions depending on the composition, then a
liquidus, and finally the boiling point, which is characterized
by a sharp endothermic change.
In some regions of the phase diagram, the liquidus was poorly
defined by DTA studies due to small heat effects. In these cases
the thermal studies were supplemented with visual observations
of crystal disappearance.
The results, obtained initially by DTA and later checked by
DSC studies for the different compositions, are summarized in
a phase diagram in Figure 2, using both mol % and wt % scales.
The DSC studies were found to agree with the DTA results
within experimental error. This work extends the phase diagram
available in the literature10-12 to the monohydrate composition.
Also included are the boiling points and glass transition
temperatures which define the high-temperature and the low-
temperature limits of the liquid state, respectively.
(ii) Extension of the Glass-Forming Region by Emulsifica-
tion. The glass transition temperatures of bulk LiBr aqueous
solutions were determined up to 20 mol % LiBr. The results

Competitive Interactions in Lithium Salt Solutions
J. Phys. Chem. B, Vol. 103, No. 20, 1999 4161
Figure 4. Phase diagram for the binary LiOAc-H2O system showing
the glass transition temperatures (Tg), liquidus temperatures (TL), and
the boiling points (Tb). Note the very gradual change in boiling point
with composition.
Figure 2. Phase diagram for the LiBr-H2O system, including glass
transition temperatures, Tg (circles, bulk samples; squares, emulsions),
and boiling points, Tb. Shaded area shows domain of industrial
refrigeration interest in this binary system.
(iii) Phase Diagram for the Lithium Acetate-H2O System.
To understand the effects of anion basicity on the liquidus
temperatures and boiling points, lithium acetate aqueous solu-
tions were studied in the composition range of 0-33 mol %
LiOAc. A partial phase diagram for the LiOAc-H2O system,
which includes the glass transition temperatures and boiling
points, is presented in Figure 4.
(iv) Phase Diagrams for Pseudo-Binary LiBr-H2O Sys-
tems Containing ZnBr2 and ZnCl2. To contrast with the
behavior of LiOAc solutions vis-a`-vis lithium bromide aqueous
solutions the effects of adding Lewis acid halides ZnBr2 13 and
14
ZnCl2 to LiBr solutions in order to sequester the bromide
anions in complex anion form were investigated. A series of
pseudo-binary solutions were prepared by substituting various
amounts (up to 30 mol %) of ZnBr2 and ZnCl2 for LiBr in a
LiBr‚3H2O composition. This results in a series of solutions
where the salt:water ratio is kept constant while the LiBr is
systematically replaced by ZnBr2 or ZnCl2.
The phase diagrams for the two pseudo-binary systems
(ZnBr2-LiBr)‚3H2O and (ZnCl2-LiBr)‚3H2O are presented in
Figures 5 and 6, respectively.
Figure 3. Glass transition temperatures for bulk and emulsified LiBr-
H2O solutions.
Discussion
Referring to the data in Figure 2, it can be seen that the boiling
point, which is an indicator of the heat of vaporization, rises
steeply in the vicinity of the hydrate compositions. The trend
reflects the action of the same forces which determine the
existence of the crystalline hydrate. It might be expected that it
would become increasingly difficult to remove water from the
near neighbor coordination shells of the lower hydrates, but
instead the rate of increase of the boiling point slows down
beyond the dihydrate compositions. This could imply that the
Li+ tends to pair up with Br- at high concentrations. A similar
trend has been revealed in the extended glass range exposed in
this study.
agree with those obtained by Sare et al.1 Above 20 mol % LiBr,
the quench rate achieved by direct immersion into liquid N2 of
the Pyrex tube containing the solution sample was insufficient
to produce a homogeneous glass. Therefore, the process of
emulsification, which suppresses crystallization by sequestering
the heterogeneous nuclei in a tiny fraction of the isolated
droplets, was invoked.7 Using the emulsification technique
described above, the glass-forming region could be extended
up to 33 mol % LiBr, i.e., to the dihydrate of LiBr. The glass
transition temperatures (Tg’s) for the bulk and emulsified LiBr-
H2O solutions are shown in Figure 3 where data from DTA
scans and subsequent DSC scans are seen to be in good accord.
The kinetics of hydrate crystallization in these concentrated
LiBr-H2O solutions have been studied in detail using the
emulsion samples, and the results will be reported elsewhere.
It is of interest to see what happens to the glass transition as
the hydrate compositions are approached. We find that, after
being remarkably flat, from H2O/Li ) 11 to H2O/Li ) 6, Tg
exhibits a steplike increase in the vicinity of 20 mol % LiBr

4162 J. Phys. Chem. B, Vol. 103, No. 20, 1999
Sivaraman et al.
Figure 7. Comparison of the boiling points (Tb, open symbols) and
glass transition temperatures (Tg, solid symbols) for the binary solutions
LiOAc + H2O and LiBr + H2O, and for the pseudo-binary solutions
of tetrahalozincate composition, viz., [(LiBr)0.67 (ZnBr2)0.33]‚3H2O and
[(LiBr)0.67 (ZnCl2)0.33]‚3H2O.
Figure 5. Phase diagram for the pseudo-binary (ZnBr2-LiBr)‚3H2O
system showing the glass transition temperatures (Tg), liquidus tem-
peratures (TL), and the boiling points (Tb).
ion, increasing water activity and making its escape into the
vapor phase relatively easier. Therefore the boiling condition,
pH O ) 1 atm, is reached at a much lower temperature. The
2
change in boiling point with composition is gradual at all
compositions in the LiOAc-H2O system in contrast to the
steeply increasing boiling points in the LiBr-H2O system. By
contrast, the glass transition temperature at a given H2O/Li+
ratio, R, is much higher for the acetate solution, for instance at
R ) 9 (10 mol % LiBr) Tg is -135 °C for the LiBr system but
-100 °C for the LiOAc system. Clearly the strong interaction
between the cation-coordinated H2O and the acetate oxygen is
the source of this increased cohesion. Indeed a correlation of
solution Tg (for a given cation) with the anion basicity quantified
by the pKa value has been noted previously.1,10
Applying the same principle, an increase in boiling point
should follow from a decrease in anion basicity. To obtain an
anion of basicity lower than Br-, the Br- anions should be
complexed with a suitable Lewis acid. Here, the Lewis acids
ZnBr2 and ZnCl2 act as anion complexing agents to form
ZnBr42- (or a mixed tetrahalozincate) anions. These new anions
act as weak bases which have a markedly lower attraction for
the protons on water molecules than does the Br- anion of the
original LiBr-H2O solution. This was earlier demonstrated by
the NMR proton chemical shift measurements of Easteal et al.6
and supported by those of Sare et al.10,15 As a result the binding
of the H2O molecules to Li+ is strengthened, the activity of
H2O decreases, and an increase in boiling point follows. Of the
two additives, ZnBr2 is most effective in raising the boiling point
while ZnCl2 is most effective in lowering the liquidus temper-
ature and hence in depressing the crystallization temperature.
Figure 6. Phase diagram for the pseudo-binary (ZnCl2-LiCl)‚3H2O
system showing the glass transition temperatures (Tg), liquidus tem-
peratures (TL), and boiling points (Tb).
and then flattens out again as the dihydrate composition is
approached (Figure 3). This parallels the anomalous variation
of the boiling point which also increases rather rapidly and then
flattens out. However, the effects are more sharply registered
and also are displaced to lower salt contents in the glass
transitions, because all interaction effects are accentuated in the
low-temperature region.
It is important to note the very low boiling points exhibited
in the LiOAc-H2O system relative to those in the LiBr-H2O
system (e.g., 116 °C for LiOAc‚3H2O vs 157.5 °C for LiBr‚
3H2O), Figure 4. The reason for this difference can be attributed
to the relatively strong basicity of the acetate anion (pKa ) 4.75).
The strongly basic acetate ion pulls H2O away from the lithium
-
Monovalent anions such as AlCl4 would have even less
attraction for H2O and would therefore be even more effective
-
in raising the boiling points, but unfortunately AlCl4 is not
stable in aqueous solutions unless acidified.
The boiling points and Tg values for the binary solutions,
LiBr-H2O and LiOAc-H2O and for two pseudo-binary solu-
tions with 25 mol % salt content, Li2ZnBr4‚3H2O and Li2Zn-
(Br,Cl)4‚3H2O, are collected in Figure 7. It can be seen that the
glass transition temperature for the R ) 3 tetrabromozincate
solution (-105 °C) is indeed lower than that for LiBr‚3H2O
(-95 °C) if the extrapolation to 33 mol % ZnBr2 in Figure 5 is

Competitive Interactions in Lithium Salt Solutions
J. Phys. Chem. B, Vol. 103, No. 20, 1999 4163
valid. At lower R values the position would possibly reverse,
but in the molten salt-like conditions at the trihydrate composi-
tion and beyond, it seems even the greater mass and complexity
Acknowledgment. The authors are indebted to Gas Research
Institute (GRI), Chicago, IL, for financial support of this work
under Contract 5094-260-2893. We are thankful to Dr. Sheng-
ping Ding for his assistance in the early stages of this project.
2-
of the ZnBr4 anions is subjugated by the decreased basicity.
The basicity would be further reduced if the anion charge
were lower. AlCl4- would, if stable, be the ideal anion, but there
may be alternative monovalent anions of less common character,
e.g., SO3F-, which could serve. However, without knowledge
of their crystalline hydrate properties it is not possible to go
further in the development of these ideas here.
References and Notes
(1) Angell, C. A.; Sare, E. J. J. Chem. Phys. 1970, 52, 1058.
(2) Moynihan, C. T.; Bressel, R. D.; Angell, C. A. J. Chem. Phys.
1971, 55, 4414.
(3) Halalay, I.; Nelson, K. A. J. Non-Cryst. Solids 1991, 131, 375.
(4) Feiweier, T.; Isfort, O.; Geil, B.; Fujara, F.; Weingartner, H. J.
Chem. Phys. 1996, 105, 5737.
Concluding Remarks
(5) Johnson, W., Trane Co. Internal Report, S. Kujak, Trane Co., private
communication.
(6) Easteal, A. J.; Sare, E. J.; Moynihan, C. T.; Angell, C. A. J. Solution
Chem. 1974, 3, 807.
Since physical chemistry abounds in examples of kinetic and
thermodynamic character running in parallel, there is some
interest in the kinetic consequences of anion basicity counter-
balancing the cation entrapment of H2O molecules, as evidenced
by the low boiling point of LiOAc‚2H2O. An H2O molecule
balanced evenly between cation and anion attractions is more
likely to escape from both, and hence to diffuse with unusual
freedom. This raises the possibility of decoupled motion of water
molecules through the ionic superstructure. The free movement
of water molecules in complex structures is currently an issue
of interest to the pharmaceutical and food preservation com-
munities, so this issue is one which we feel is deserving of
follow-up research.
(7) MacFarlane, D. R.; Angell, C. A. J. Phys. Chem. 1984, 88, 779.
(8) Clausse, D.; Broto, F. J. Phys. Chem. 1976, 80, 4251.
(9) Clausse, D.; Broto, F. Colloid Polym. Sci. Lett. 1982, 260, 641.
(10) Sare, E. J. Ph.D. Thesis, Purdue University, West Lafayette, IN,
1971.
(11) Vuillard, G. Ann. Chim. (Paris) 1957, 2, 223.
(12) Vuillard, G.; Kesses, J. J. Bull. Soc. Chim. Fr. 1960, 2063.
(13) Adegoke, C. O. Int. J. Refrig. 1993, 16, 45.
(14) Iyoki, S. Int. J. Refrig. 1990, 13, 191.
(15) Sare, E. J.; Moynihan, C. T.; Angell, C. A. J. Phys. Chem. 1973,
77, 1869.