Chem 200400104

FULL PAPER
On the Analyses of Mixture Vapor Pressure Data: The Hydrogen Peroxide/

Water System and Its Excess Thermodynamic Functions
Stanley L. Manatt* and Margaret R. R. Manatt*[a]
Abstract: Reported here are some as- (USA), 1949) have been used; there is Yamaguchi, Japan Sci. Eng. Sci. 1950,
pects of the analysis of mixture vapor significant recent interest in this 1, 41–49) who observed significant var-
pressure data using the model-free system. It was found that the original iations of temperature from place to
Redlich–Kister approach that have data is fit far better with a four-param- place within a carefully insulated appa-
heretofore not been recognized. These eter Redlich–Kister excess energy ex- ratus of the type traditionally used in
are that the pure vapor pressure of one pansion with inclusion of the pure hy- mixture vapor pressure measurements.
or more components and the average drogen peroxide vapor pressure and For hydrogen peroxide, new critical
temperature of the complex apparatus- the temperature as parameters. Com- constants and vapor pressure–tempera-
es used in such studies can be obtained parisons of the present results with the ture equations needed in the analysis
from the mixture vapor pressures. The previous analyses of this suite of data approach described above have been
findings reported here raise questions exhibit significant deviations. A prece- derived. Also temperature functions
regarding current and past approaches dent for consideration of iteration of for the four Redlich–Kister parameters
for analyses of mixture vapor pressure temperature exists from the little- were derived, that allowed calculations
data. As a test case for this analysis ap- known work of Uchida, Ogawa, and of the excess Gibbs energy, excess en-
proach the H2O2–H2O mixture vapor Yamaguchi (S. Uchida, S. Ogawa, M. tropy, and excess enthalpy whose
pressure measurements reported by values at various temperatures indicate
Scatchard, Kavanagh, and Tickner (G. Keywords: hydrogen peroxide · the complexity of H2O2–H2O mixtures
Scatchard, G. M. Kavanagh, L. B. Tick- liquid mixture vapor pressures · not evident in the original analyses of
nor, J. Am. Chem. Soc. 1952, 74, 3715– phase equilibria · thermodynamics · this suite of experimental results.
3720; G. M. Kavanagh, PhD. Thesis, water chemistry
Massachusetts Institute of Technology
Introduction temperature. Knowledge of multicomponent liquid–vapor

equilibria also plays a vital role in the development of distil-
Studies of multicomponent liquid–vapor equilibria give in- lation and other separation procedures and apparatus in in-
sights regarding the molecular forces and effects existing in dustry and chemical research.
such systems and allow determination of the excess thermo- The usual analyses of multicomponent liquid–vapor data
dynamic functions. The latter represent the differences be- start with a set of vapor pressure (VP) measurements for
tween the actual values of these functions and the values mixtures of various mole fractions of the liquids of interest,
given by the ideal solution laws at the same pressure and knowledge of the vapor pressure–temperature (VP–T) equa-
tions of the pure components and the temperatures for each
[a] Dr. S. L. Manatt,+ Prof. M. R. R. Manatt suite of mixtures mole fractions studied. Many times the
Department of Chemistry and Biochemistry
knowledge of the pure component VP–T function for one or
California State University, Los Angeles
CA 900032 (USA) more mixture components is marginal or not available over
E-mail: [email protected] the temperature range of interest. Here it is demonstrated
[+] Present address: K-2 Technical Services, 5447 La Forest Dr. that mixture vapor pressure data of good quality at a partic-
La CaÇada, CA 91011 (USA) ular temperature will allow the pure component vapor pres-
Supporting information for this article is available on the WWW sure of at least one component to be determined to high ac-
under http://www.chemeurj.org/ or from the author. Hydrogen perox- curacy. In addition, it was found that fitting of the experi-
ide and water 2nd virial coefficients calculated from the Tsonopoulos
equations[12] and the Keyes, Smith and Gerry equations[21] and hydro-
mental data can be improved significantly if the absolute
gen peroxide and water activities in mixtures at selected temperatures temperature is also made a parameter. We were led to con-
between 0 and 105 8C. sider this latter factor by the work reported some 54 years
6540 F 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/chem.200400104 Chem. Eur. J. 2004, 10, 6540 – 6557
6540 – 6557
ago by Uchida, Ogawa, and Yamaguchi[1] who observed Theoretical Considerations

temperature variations within an apparatus used for mixture
vapor pressure measurements; the significance of these au- Many approaches to the analysis of isothermal, total vapor
thorsI work has gone unrecognized by workers in the mix- pressure data of multicomponent liquid–vapor mixtures
ture vapor pressure field. have been reported.[15, 16] The Redlich–Kister expansion for
Although we have investigated several systems, here, to the excess energy,[14] which was used in the original analy-
demonstrate the use of the above mentioned observations, ses,[6h, 8] is still held to be valid and convenient.[15, 16] Thus, a
we show as an example the results for the H2O2–H2O four-parameter expansion for a binary mixture is given in
system. Knowledge of hydrogen peroxide (HP) vapor com- Equation (1), where X1 is the liquid mole fraction of water,
position above H2O2–H2O mixtures has assumed some X2 the liquid mole fraction of hydrogen peroxide, and the Bi
degree of importance recently. The vapor composition values are parameters (calories mol1 in equations used here
above such mixtures is important to know in schemes for and in original analyses[6h, 8]) to be determined from experi-
generating COH radicals[2] and in spectroscopic studies.[3, 4] mental data at various temperatures.
The recent detection of hydrogen peroxide on the surface of
Europa[5] has caused renewed interest in the vapor pressures gX E ¼X 1 X 2 ½B0 þ B1 ðX 1 X 2 Þ þ B2 ðX 1 X 2 Þ2
above liquid and solid H2O2–H2O mixtures. ð1aÞ
þB3 ðX 1 X 2 Þ3 þ . . . :
The group of George Scatchard from the 1930s to the
1960s has provided a legacy of careful experimental vapor–
or
liquid equilibria measurements on a number of systems.[6]
The measurements on the H2O2–H2O system[6h, 8] made over
gX E ¼X 1 ð1X 1 Þ½B0 þ B1 ð12 X 1 Þ þ B2 ð12 X 1 Þ2
50 years ago by Scatchard, Kavanagh, and Ticknor stand as ð1bÞ
the definitive results on this system. No other as extensive þB3 ð12 X 1 Þ3 þ . . . :
experimental measurements on this system have been re-
ported since their work. Based on the approach outlined From the relations given in Equation (2 a) and (2 b), [17, 18]
above, detailed analyses of their original data have been car- where the miE values are excess chemical potentials, the Red-
ried out which yield results that fit the measurements signifi- lich–Kister expansion Equation (1 b) leads to Equation (3)
cantly better than the original analyses. These results were and Equation (4), where m1E and m2E are the excess chemical
needed[2–5] as the basis for deriving vapor pressure–composi- potentials for water and hydrogen peroxide, respectively.
tion data for the H2O2–H2O system for compositions and
temperatures other than those exhibited in tables in the lit- m1 E ¼ gX E þ ð1X 1 Þ@gX E =@X 1 ð2aÞ
erature,[7] in the original work[6k, 8] and in currently available
m2 E ¼ gX E X 1 @gX E =@X 1 ð2bÞ
manufacturerIs literature.[9]
Hydrogen peroxide (HP) is unique because a number of
m1 E ¼ð1X 1 Þ2 ½B0 þ B1 ð14 X 1 Þ þ B2 ð12 X 1 Þð16 X 1 Þ
its physical properties such as boiling point (BP) at atmos- ð3Þ
pheric pressure and critical constants can not be measured þB3 ð12 X 1 Þ2 ð18 X 1 Þ þ . . . :
directly because of its instability. In the course of working
with the Scatchard, Kavanagh, and Ticknor data,[6h, 8] values m2 E ¼X 1 2 ½B0 þ B1 ð34 X 1 Þ þ B2 ð12 X 1 Þð56 X 1 Þ
ð4Þ
of certain hydrogen peroxide physical properties were þB3 ð12 X 1 Þ2 ð78 X 1 Þ þ . . . :
reevaluated, newer liquid density functions for both hydro-
gen peroxide[10] and water (W)[11] were used, newer vapor Equations (3) and (4) lead to the total vapor pressure
pressure data for water[11] were used, second virial coeffi- Equation (5), as detailed in standard texts,[15, 16] that is the
cients, required for nonideal gas corrections, were calculated heart of the original analyses[6h, 8] (pressures in mm Hg),
by using current, well-tested empirical equations of Tsono- where P is the total vapor pressure above a binary mixture,
poulos,[12] and data were discovered in the Kavanagh thesis[8] P1 the water vapor pressure, P2 the hydrogen peroxide
supporting a small correction for hydrogen peroxide decom- vapor pressure (PHP), T the absolute temperature, R the gas
position. Combining the pure hydrogen peroxide vapor pres- constant in appropriate units, IGC1 the imperfect gas correc-
sure data of Maass and Hiebert[13] with the pure hydrogen tion for water, and IGC2 the imperfect gas correction for hy-
peroxide vapor pressure data from the present work, new drogen peroxide.[19] The Scatchard groupIs initial analyses of
hydrogen peroxide VP–T equations have been derived, one the H2O2–H2O experimental mixturesI vapor pressures only
of which appears to be valid even close to the hydrogen per- considered three parameters B0, B1, and B2 with a fourth
oxide critical region. From the fitting of the Scatchard linear temperature dependence folded into B0, that is, B0 =
groupIs data to four parameter Redlich–Kister analyses (a B0’ + kt where k was a constant and t the temperature in
model-free approach),[14] equations for the temperature Celsius. The gas law deviations terms used in the present
functions of these latter parameters were derived. From analysis have the forms given in Equations (6) and (7),
these functions values for the excess Gibbs energy, excess where b1 and b2 are second virial coefficients for water and
entropy and excess enthalpy of mixing were calculated; the hydrogen peroxide, respectively, V1 and V2 are the corre-
values of the latter two parameters show the complexity of sponding liquid molar volumes at temperature T, d12 =
H2O2–H2O mixtures that was not revealed by the excess 2 b12b1b2 where b12 is the second virial cross coefficient
functions reported in the original work.[6h, 8] and Y1 is the vapor mole fraction of water.[15, 16]
Chem. Eur. J. 2004, 10, 6540 – 6557 www.chemeurj.org F 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 6541
FULL PAPER S. L. Manatt and M. R. R. Manatt
P ¼P1 X 1 exp½ðð1X 1 Þ2 =RTÞðm1 E þ IGC1 Þ units used here are atmospheres and mm Hg (where 1 at-
2 E
ð5Þ mosphere = 1.01325 bar and 1 mm Hg = 1.33322 millibars =
þP2 ð1X 1 Þexp½ðX 1 =RTÞðm2 þ IGC2 Þ
1.3332237 L 102 Pascals).
IGC1 ¼ ðb1 V 1 ÞðPP1 Þ=RTPð1Y1 Þ2 d12 =RT ð6Þ In our initial consideration of the three sets of mixture
vapor pressure data from Scatchard et al.[6h, 8] we obtained
IGC2 ¼ ðb2 V 2 ÞðPP2 Þ=RTPY1 2 d12 Þ=RT ð7Þ fits in reasonable agreement with their results using their
dated water VP–T equation from Keyes,[20] their tempera-
The calculation of the virial coefficients will be discussed tures of 333.16, 348.16, and 363.16 K, their pure hydrogen
below. The initial analyses by ScatchardIs group[6h, 8] only in- peroxide vapor pressure estimates, their dated virial coeffi-
cluded the first terms on the right of Equations (6) and (7) cient equations[21] for water and hydrogen peroxide, their
and, as pointed out above, really only three Redlich–Kister use of only one gas imperfection term (two terms seems to
parameters in their fits. Also, subsequent calculations of be the norm now[15, 16]), their use of three Redlich-Kister pa-
tables of H2O2–H2O mixturesI vapor pressures by them,[6h, 8] rameters, their use of a linear temperature term folded into
reviewers of their work,[7] and commercial tables[9] were all the Redlich–Kister B0 term and no consideration of a de-
formulated without consideration of any gas law deviation composition correction to the mixturesI vapor pressures.[22]
term although the fit parameters used were derived with Because of the number of experimental mixture vapor pres-
one. It was found here that if the data is fit initially without sure measurements at a particular temperature, it also
these corrections, significantly different fit parameters seemed valid to carry out a second set of fits including the
result. Later work from the Scatchard group always included pure vapor pressure for hydrogen peroxide as a parameter.
the full Equations (6) and (7) correction terms[6j, k] in fitting This led to Redlich–Kister parameters substantially different
mixture vapor pressure data.
Here Equations (6) and (7)
were included both in the fits Table 1. Total vapor pressure for H2O2–H2O mixtures; all pressures in mm Hg and temperatures in 8C.
and all subsequent calculations Thesis measurements Present work
[a]
on the mixture vapor pressures. temp. mole measured calcd[c] decomposition calcd
As detailed below, with four (real [b] fraction vapor mole vapor minus iterated corrected minus
temp.) water fraction water pressures measured temp. pressures[d] measured[g]
Redlich–Kister parameters, iter-
44.51 0.4860 0.8727 27.47 0.06 – 27.397 –
ation of PHP and T and all the
(44.486)
gas imperfection terms, signifi- 60.01 0.0381 0.1028 19.43 0.13 59.996 19.357 0.032
cantly better fits to the data re- (59.991) 0.1577 0.4301 26.21 0.11 26.137 0.111
sulted along with three values 0.3169 0.7049 39.79 0.33 39.717 0.282
0.4221 0.8243 52.01 0.37 51.937 0.288
for the pure hydrogen peroxide
0.5925 0.9151 76.21 1.05 76.137 0.065
vapor pressure that are signifi- 0.7190 0.9603 99.25 0.72 99.177 0.187
cantly different from those ob- 0.7964 0.9896 114.82 0.32 114.747 0.233
tained in the initial analyses. 0.9095 0.9946 135.35 0.36 135.277 0.067
The imperfect gas corrections
75.01 0.0404 –[e] 42.28 0.44 75.072 42.207 0.008
in the pressure range consid-
(74.999) 0.1428 0.3619 53.35 0.51 53.277 0.005
ered are in general small but 0.2540 0.6093 69.86 0.15 69.787 0.092
not insignificant. The second 0.4249 0.8166 105.30 0.10 105.227 0.465
terms on the right of Equa- 0.5037 0.8886 127.09 1.03 127.017 0.308
0.5101 0.8924 128.88 1.00 128.807 0.279
tions (6) and (7) always appear
0.6759 0.9532 180.44 0.09 180.307 0.266
to be substantially less than the 0.7223 0.9733 196.43 0.24 196.357 0.197
first terms. 0.8028 0.9833 225.29 0.19 225.217 0.354
0.9255 0.9964 267.24 0.47 267.167 0.090
90.01 0.0403 0.1142 84.92 0.30 90.019 84.847 0.029

The Fitting to the (90.011) 0.1582 0.4018 109.63 0.75 109.557 0.108
Experimental 0.3454 0.7284 165.37 0.32 165.297 0.274
Measurements 0.4882 0.8484 227.24 0.31 227.167 0.358
0.5020 0.8658 234.54 0.08 234.467 0.016
0.6743 0.9541 331.99 0.86 331.917 0.394
Table 1 exhibits the original
0.8046 0.9765 411.73 0.64 411.657 0.350
data and the data corrected for 0.9006 0.9911 471.07 0.04 470.997 0.134
decomposition as discussed 105.01 0.5015 0.8506 413.33 2.35 – 413.257 –
below. To allow more ready (105.026)
comparison with the original sum of squares of pressure errors 7.0051[f] 1.4277
absolute mean of pressure errors 0.4757[f] 0.1919
work,[6h, 8] extensive tables in
the literature[7] and currently [a] Thesis temperatures are 0.018 lower, because t = T273.16 was used instead of t = T273.15. [b] Corrected
thesis temperatures. [c] From “temperature-smoothed” parameters. [d] 0.073 mm Hg decomposition correction
distributed manufactureIs litera- subtracted from each thesis pressure measurement. [e] Insufficient sample for analysis in original work; value
ture on H2O2–H2O mixture used in the present work, 0.1085, has negligible effect on fit. [f] These are 12.532 and 0.524, respectively, when
vapor pressures,[9] the pressure measurements at 44.5 and 105 8C included. [g] Fits made with four Redlich–Kister parameters.
6542 F 2004 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org Chem. Eur. J. 2004, 10, 6540 – 6557
Analyses of Mixture Vapor Pressure Data 6540 – 6557
from the former fits but essentially the same pure hydrogen tion was first reported some 54 years ago by Uchida,
peroxide vapor pressures as previously reported.[6h, 8] Ogawa, and Yamaguchi[1] in their studies of benzene–tolu-
It is clearly evident from Tables 1 and 2 that the “temper- ene mixtures and several other mixtures. They found that in-
ature-smoothed” parameters (to be explained below) do not dividual thermocouples gave rather different temperatures
reproduce the experimental mixture vapor pressures well at depending on their location in the apparatus and on the rate
the three temperatures in view of the vapor pressure experi- of heating, rate of condenser cooling liquid flow and the sur-
mental error estimates in the thesis[8] of 0.01 to face area of the condensing vapor. These authors could find
0.05 mm Hg (and better than 0.01 8C). The fit errors in the no experimental settings that completely eliminated the
original work (shown in Table 1 and Table 2) with the un- small temperature differences within the volume of their
smoothed parameters derived from the individual sets of well-insulated apparatus. They did not consider iteration of
data are much smaller (sum of squares of vapor pressure the temperature in the analysis of their data.
errors 3.806, absolute mean error 0.327 and standard devia- It should be noted that usually mixture vapor pressure ap-
tion 0.390) than what is calculated from the “temperature- paratuses are calibrated using water and multijunction ther-
smoothed” parameters (7.006, 0.420, and 0.524, respective- mocouple systems. The heating system is usually set to ach-
ly). Also, the linear temperature term of the original work ieve a desired temperature based on the relation of EMF to
for B0 does not compensate for the temperature dependen- a VP–T equation for water.
cies clearly evident in the parameters shown in Table 2. In unpublished mixture vapor pressure work by one of
On comparison of the fits of the present work and those the present authors an array of eight precision-matched
of the original work,[6h, 8] the one shortcoming that stands thermocouples was employed in the usual, complicated
out in the latter then is the fact that results were “smoothed boiler-condensation system used in mixture vapor pressure
with respect to temperature.” What this phrase means is measurements.[23] Water was used for the temperature cali-
that the averages of each of the three parameters at the brations. The system was used for determination of mixture
three temperatures were taken as constants with a linear vapor mole fractions of binary systems where one compo-
temperature correction for B0. The results in the previous nent was very flammable. However, the electronics in this
work for B0’ and k in the expression B0 = B0’ + kt for each apparatus allowed individual thermocouples to be sampled.
data set were not given, but the “temperature-smoothed” This latter provision was made in light of the observations
result B0 = 752 + 0.97t was given. Thus, the extent of the by the former Japanese workers.[1] Differences from 0.01
temperature variation of B0’ and k is not clear in the original to as large as 1.00 K from location to location in the
work.[6h, 8] volume of the well insulated apparatus (also fitted with a
An aspect of the fitting of mixture vapor pressure data circulation fan) were found to exist in studies of various sol-
that appears not to have been recognized heretofore is that vent mixtures and even in the water calibrations. These dif-
the temperatures in the complex glass systems used in such ferences again depended strongly on the rate of heat input
studies, even though carefully insulated from ambient tem- and condenser coolant flow rate and could not be eliminat-
peratures, are not constant at every location. This observa- ed completely. In the original Scatchard, Kavanagh, and
Table 2. Redlich–Kister parameters, PHP values, T values, and goodness of fit parameters.
Source
and T[a] B0[b] B1[b] B2[b] B3[b] PHP[c] AME[d] SDE[e] SSQPE[f]
thesis (three parameters with B0 = B0’ + kT)
333.16 690.5 109.4 12.6 – 17.56 0.4625 0.5385 2.0374
348.16 686.3 56.4 16.5 – 39.14 0.2960 0.3505 1.1234
363.16 661.4 92.5 42.0 – 77.93 0.2288 0.2981 0.6451
group errors totals 0.3265 0.3901 3.8059
thesis (temperature-smoothed parameters)
333.16 694.0 85.0 13.0 – 17.6 0.4238 0.5510 2.1277
348.16 679.0 85.0 13.0 – 39.1 0.4220 0.5462 2.8698
363.16 665.0 85.0 13.0 – 77.9 0.4125 0.5279 2.0082
group error totals 0.4196 0.5241 7.0057
group error totals with 317.66 and 378.16 K points 0.4757 0.6811 12.5318
present work (three parameters)
333.290 689.5965 128.6024 14.3819 – 17.457 0.2586 0.3370 0.7949
348.208 689.5556 60.8988 29.0154 – 39.056 0.2116 0.2799 0.7049
363.095 659.4681 88.7518 31.8563 – 77.900 0.3368 0.4243 1.2601
present work (four parameters)
333.146 732.1492 86.7641 86.6240 139.8383 18.304 0.1579 0.1976 0.2733
348.222 680.0356 70.3986 10.7520 27.9217 38.652 0.2064 0.2667 0.6403
363.169 641.6787 108.6281 63.0625 59.4221 76.340 0.2079 0.2703 0.5115
[a] In K. [b] Units of Bi values are cal mol1 here as in the original work (references[6h] and [8]). [c] Pressure in mm Hg. [d] Absolute mean error of ex-
perimental vapor pressure measurements minus calculated in mm Hg. [e] Standard deviation of latter differences. [f] Sum of squares of experimental
vapor pressure measurements minus the calculated.
Ticknor work[6h, 8] a 20-junction thermocouple system was very insignificant contributions to the goodness of fit.
employed, but no details regarding the placement of the in- Table 1 and 2 exhibit the parameter results plus fits from the
dividual thermocouples was given or how matched in preci- original work and fits using three and four Redlich–Kister
sion the individual thermocouples were. It has been found parameters with iteration of T and PHP ; the fits made with-
here that the goodness of the fits of their data for H2O2– out any corrections for gas imperfection also gave different
H2O mixtures are extremely sensitive to very small varia- parameters and somewhat different PHP values (not exhibit-
tions of temperature (i.e. 0.002 K). ed here). As the temperature was iterated, the water vapor
Thus, it is argued here that various mole fraction mixtures pressures going into Equation (5) were calculated from An-
give rise to mixture vapor pressures reflecting an average toine equations[26] with a small added correction which was
temperature in the apparatus with the temperature slightly a part of the fit equations in the fitting program. For each T
different at different points in the apparatus volume for the the sum of squares of the pressure residuals were monitored
same EMF setting, no matter how carefully the system is in- and the T varied to give a minimum. At the same time the
sulated from ambient temperature and how carefully the water and hydrogen peroxide vapor mole fractions calculat-
input energy and condensation parameters are held constant ed were put back in the fit equations in the imperfect gas
or regulated. For mixtures these small temperature differen- correction terms.[25] As can be seen the final values of T are
ces possibly might be due to each mole fraction mixture ex- not much different from the corrected values of T of the
hibiting slightly different heat transfer properties stemming original work. The pure hydrogen peroxide vapor pressures
from the heat of vaporization differences of the compo- from the three-parameter Redlich–Kister fits made here are
nents.[24] Thus, it is felt that it is totally reasonable to expect essentially the same as those in the original work and the fit
the goodness of fit for mixtures vapor pressure data at a to the experimental data somewhat better than the latter
nominal temperature to exhibit a dependence on very small analyses.
changes of temperature. Other sets of mixture vapor pres- The classic, two-parameter VP–T Equation (8) (pressure
sure data from the Scatchard group[6g, j, k] have been tested in mm Hg), which is valid over small temperature ranges,
and in every case the fits were very sensitive to very small where A and B are constants, was used as a test of the pure
temperature variations of the order of 0.002 K. Always vapor pressures obtained in the various fits.
significant improvements in the fits to the observed mixture
vapor pressures were found upon iteration of the tempera-
ture but always the best fit temperature was close (i.e.
log10 PHP ¼ A þ B=T ð8Þ
within 0.3 K) to the temperature stated in the original
works. It is suggested here that the effect first observed by
Uchida, Ogawa, and Yamaguchi[1] represents an unrecog-
nized source of error in the analysis of vapor pressure data In Table 3 some of these results, which include those from
for multicomponent systems. The implication of this sugges- the original work and the fits (always using two gas imper-
tion is that the iterated temperature arrived at represents an fection terms) using three parameters and four parameters,
average for a suite of vapor pressure measurements of vari- are shown. The pure hydrogen peroxide vapor pressure fits
ous mixture mole fractions for a particular setting of heater- to Equation (8) from the former two fits all exhibited devia-
condensation parameters. The temperature differences be- tions far exceeding the estimated experimental mixture
tween mixture mole fractions will be small and also repre- vapor pressure errors of 0.01 mm Hg for pressures above
sent an error contribution to any pure component vapor 100 mm Hg and between 0.02 and 0.05 mm Hg below
pressure determined and to the resulting fit parameters. 100 mm Hg.[8] However, as shown in Table 3, the three pure
The final fits carried out in the present work used the new hydrogen peroxide vapor pressures from the latter four-pa-
constants and equations derived below, the pure hydrogen rameter analyses exhibit a very precise fit to the simple two-
peroxide vapor pressure as a parameter, the iteration of the parameter Equation (8) in the 60–90 8C range. The latter fit
absolute temperature, the inclusion of two gas imperfection errors to Equation (8) were somewhat less than the Equa-
terms, the use of four Redlich–Kister parameters, and the tion (8) fit errors of the new water vapor pressure data[11]
iteration of the water and hydrogen peroxide vapor phase over the same temperature range or over the same vapor
mole fractions.[25] These latter two parameters makes only pressure range.
Table 3. Fits of PHP values (mm Hg) from experimental data with the two-parameter equation log10PHP = A + B/T, where A and B are least-square con-
stants determined from three values of PHP at the values of T indicated.
T[a] PHP[b] PHPEq (8)
DPEq (8)
PHPEq (9)
DPEq (9)
T[b,c] PHP[c] PHPEq (8)
DPEq (8)
333.15 17.56 18.014 0.454 17.530 0.030 333.146 18.304 18.296 0.008
348.15 39.14 38.698 0.442 38.446 0.694 348.222 38.652 38.558 0.008
363.15 77.93 78.044 0.114 78.236 0.306 363.169 76.340 76.330 0.002
absolute mean error 0.337 0.343 0.006
standard deviation 0.452 0.509 0.008
[a] In K. [b] From three-parameter Redlich–Kister fits of references [6h, 8]. [c] From fits with four Redlich–Kister parameters, with hydrogen peroxide
vapor pressure as a fifth parameter and with temperature and hydrogen peroxide vapor mole fraction iterations.
Hydrogen Peroxide Vapor Pressure–Temperature The values of mixturesI vapor pressures and hydrogen
Equations peroxide vapor mole fraction are strongly influenced by the
VP–T equation(s) used in mixture vapor pressure Equa-
The original workers[6h, 8] used three pure hydrogen peroxide tion (5). The differences of mixture vapor pressures between
vapor pressure values[27] and a modified Ramsey–Young the present results and those of the original analysis are
treatment[28] to derive the four-parameter hydrogen peroxide larger (ranging from 10–23 %) at the lower temperatures (0–
VP–T equation given in Equation (9) (pressure in mm Hg), 40 8C) and at large hydrogen peroxide mole fractions as ex-
and, as discussed below, it was initially used here to estimate hibited in Table 4. Deviations of the hydrogen peroxide
the hydrogen peroxide boiling point. vapor mole fractions as high as 29 % were found in the
lower temperature range (0–40 8C) (see Table 5) which is
log10 PHP ¼44:57604025:3=T2:9960 log10 T that most important temperature in COH radical generation
ð9Þ
þ0:0046055T work.[2] These deviations indicate the shortcomings of the
previously reported four-parameter hydrogen peroxide VP–
Here a different and perhaps more valid approach has T Equation (9).[6h, 8] Deviations at higher temperatures were
been taken to derive hydrogen peroxide VP–T equations. in general smaller. The satisfactory agreement with the re-
The three pure hydrogen peroxide vapor pressures from the sults of GiguRre and Maass[30] at 30 8C for H2O2–H2O mix-
four-parameter Redlich–Kister fits discussed above and the tures (average error of 0.15 mm Hg or 1.9 %) supports
21 pure hydrogen peroxide vapor pressure data points of the validity of the above hydrogen peroxide VP–T equations
Maass and Hiebert[13] were next combined to derive VP–T and also the Redlich–Kister parameter temperature func-
equations. We realized that Maass and Hiebert estimated tions for the lower temperature range[31] that are derived
errors[13] of 0.1 mm Hg and 0.1 K are significantly great- below. The limitations of Equations (10) and (11) have been
er than those estimated for the Scatchard groupIs mixture tested several ways. As discussed below, the Equation (10) is
vapor pressure measurements, but it was felt that successive not valid either in the critical region or at a Tr (Tr = T/TcHP
least squares fitting of these data combined with three PHP where TcHP is the hydrogen peroxide critical temperature) of
values from the fits of the present work would flag one or 0.7, giving an unrealistic PcHP (hydrogen peroxide critical
more data points in substantial error which could be thrown pressure), and an wHP (hydrogen peroxide acentric factor)
out in the next iteration; always the three PHP values deter- which is way too low. Table 6 compares the vapor pressures
mined here were kept. Four- and seven-parameter VP–T for Equations (9), (10), and (11) over the range 0 to 160 8C.
equations were sought. Variant data points of these former The deviations of Equation (9) from the results of Equa-
workers exhibiting fit errors greater than 0.3 mm Hg were tions (10) and (11) are very significant at most temperatures
eliminated in successive iterations. The criteria of a mini- except in the 60–90 8C range. The differences between Equa-
mum of 10 points and inclusion always of the three PHP tions (10) and (11) between 0 and 90 8C are not significant,
values from the present work were set for the determination but above 90 8C the PHP from Equation (10) significantly ex-
of the four- and seven-parameter VP–T equations. The fit- ceeds that from Equation (11). As discussed below, the PcHP
ting and data point elimination procedure cut off at 11 re- calculated from Equation (11) is very close to the corre-
maining points when the absolute mean deviation was sponding state theory[32–35] estimate of 249.4 atm, so Equa-
0.080 mm Hg and the standard deviation was 0.12 mm Hg; tion (11) seems valid for vapor pressure estimates in the
these values are close to the corresponding errors estimated higher temperature range even up to the critical region.
by Maass and Hiebert.[13] Attempts to derive similar VP–T A more telling test of the range of validity of Equa-
equations for hydrogen peroxide using the vapor pressures tions (10) and (11) comes from calculations of the hydrogen
from the three-parameter Redlich–Kister fits or the Wilson peroxide heat of vaporization, Hv , in regions where the
treatment[29] fits (not exhibited here), combined with the higher order virial coefficients (i.e. third and higher ones)
Maass and Hiebert[13] data, yielded no satisfactory equations. are not important, from Equation 12[36, 37] where VL and Vsat
Thus, the four-parameter Redlich–Kister excess energy ex- are the molar volumes of the liquid and saturated vapor, re-
pansion with two gas imperfection terms leads to much im- spectively, Psat is the corresponding vapor pressure and bHP
proved fits, PHP values at three different temperatures that is the hydrogen peroxide virial coefficient.
exhibit extremely good fits to Equation (8) over the range
60–90 8C and, combined with selected data of Maass and DH v ¼ RT 2 ð1V L =V sat Þð@ln Psat =@TÞð1 þ bHP =V sat Þ ð12Þ
Hiebert,[13] give the seven- and four-parameter pure hydro-
gen peroxide VP–T equations given in Equation (10) and It was found that the DHv values from the vapor pressure
(11) (pressures in mm Hg; temperature in K). Equation (10) above 90 8C did not diminish at the rate one
would expect and by 115 8C actually were increasing signifi-
log10 PHP ¼24:84363511:54=T4:61453 log10 T cantly, as shown in Table 7. This behavior is counter to the
3:60245 103 T7:73423 106 T 2 ð10Þ expected behavior for a DHv . Careful scrutiny of the first
8 3 13 4
and second derivatives of Equation (10) indicated significant
þ1:78355 10 T 2:27008 10 T deviations from those of Equation (11) (and Equation (9)
too) above 90 8C. Thus, although Equation (10) is valid from
log10 PHP ¼38:85723627:72=T1:2133 log10 T
ð11Þ 0 to 90 8C, above the latter temperature Equation (11) was
4:74132 102 T used in the present work.
Table 4. Total vapor pressure (mm Hg) of hydrogen peroxide-water mixtures; % error with original work indicated in parenthesis.
Temp. Mole fraction hydrogen peroxide in liquid phase
[8C] 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
0.0 0.352 0.486 0.714 0.996 1.326 1.727 2.225 2.822 3.470 4.082 4.585
(22.7) (16.8) (16.9) (14.0) (8.9) (4.0) (1.2) (0.5) (0.6) (0.5) (0.1)
10.0 0.777 1.045 1.510 2.077 2.738 3.540 4.531 5.713 6.995 8.207 9.212
(19.2) (14.0) (14.6) (12.3) (7.7) (3.4) (0.9) (0.3) (0.5) (0.5) (0.1)
20.0 1.612 2.137 3.014 4.094 5.351 6.868 8.736 10.958 13.364 15.642 17.542
(15.8) (11.3) (12.3) (10.6) (6.6) (2.7) (0.5) (0.1) (0.4) (0.4) (0.1)
25.0 2.274 2.986 4.175 5.636 7.338 9.388 11.907 14.898 18.136 21.204 23.770
(14.1) (9.9) (11.1) (9.7) (5.9) (2.3) (0.3) (0.1) (0.3) (0.4) (0.1)
30.0 3.166 4.119 5.709 7.664 9.940 12.679 16.040 20.022 24.331 28.417 31.844
(12.5) (8.4) (9.9) (8.7) (5.3) (1.9) (0.0) (0.2) (0.3) (0.4) (0.1)
40.0 5.919 7.542 10.268 13.643 17.586 22.328 28.132 34.991 42.401 49.438 55.365
(9.4) (5.3) (6.9) (6.3) (3.6) (0.8) (0.6) (0.6) (0.1) (0.3) (0.1)
50.0 10.590 15.382 21.860 28.652 35.615 43.266 52.118 62.245 73.041 83.399 92.588
(6.5) (15.9) (22.2) (21.8) (17.6) (12.3) (7.8) (4.7) (2.6) (1.0) (0.0)
60.0 18.218 25.743 36.141 47.209 58.602 71.049 85.301 101.439 118.518 134.876 149.502
(3.8) (12.5) (19.4) (19.8) (16.4) (11.8) (7.7) (4.7) (2.7) (1.0) (0.0)
70.0 30.262 42.098 58.303 75.796 93.986 113.747 135.997 160.735 186.601 211.373 233.839
(1.4) (10.3) (17.3) (18.5) (15.9) (11.9) (8.2) (5.2) (2.9) (1.1) (0.0)
75.0 38.540 54.919 72.945 93.363 116.345 141.730 169.228 198.510 229.112 260.069 289.246
(0.2) (12.0) (16.0) (16.4) (14.7) (11.7) (8.1) (4.7) (2.0) (0.5) (0.0)
80.0 48.719 61.703 76.878 96.843 122.895 155.072 192.337 233.065 275.451 317.359 355.327
(0.8) (0.6) (0.5) (1.0) (0.7) (0.1) (0.0) (0.2) (0.3) (0.2) (0.0)
90.0 78.236 96.906 119.670 148.859 186.631 233.284 287.480 346.897 408.847 470.163 525.921
(2.6) (0.4) (0.4) (1.1) (1.1) (0.7) (0.4) (0.4) (0.4) (0.3) (0.0)
100.0 116.181 147.075 180.413 222.488 276.409 342.676 419.550 503.934 592.176 679.865 760.000
(4.24) (0.6) (1.0) (1.5) (1.4) (1.0) (0.7) (0.6) (0.6) (0.3) (0.0)
110.0 172.781 213.359 256.643 310.878 380.173 465.315 564.238 673.126 787.415 901.504 1006.40
(5.7) (3.5) (5.0) (6.4) (7.0) (7.2) (7.2) (7.3) (7.3) (7.1) (6.8)
120.0 251.34 313.00 378.46 459.85 563.08 689.25 835.38 996.03 1164.67 1333.25 1488.72
(6.9) (3.0) (2.9) (3.1) (2.7) (2.1) (1.6) (1.2) (0.9) (0.5) (0.0)
130.0 358.29 441.84 530.33 639.93 778.63 947.95 1144.08 1359.94 1586.97 1814.54 2025.32
(8.0) (4.2) (4.1) (4.1) (3.6) (2.8) (2.1) (1.6) (1.1) (0.5) (0.0)
140.0 501.34 612.44 729.90 874.93 1058.05 1281.42 1540.23 1825.42 2125.96 2428.09 2709.15
(8.9) (5.3) (5.1) (5.1) (4.5) (3.5) (2.6) (1.9) (1.3) (0.6) (0.0)
150.0 689.57 834.81 988.12 1176.90 1414.83 1704.88 2041.12 2412.11 2803.93 3199.00 3568.19
(9.6) (6.3) (6.1) (6.0) (5.3) (4.2) (3.1) (2.3) (1.5) (0.7) (0.0)
160.0 933.62 1120.49 1317.47 1559.47 1864.01 2235.14 2665.65 3141.42 3645.06 4154.50 4632.83
(10.2) (7.1) (7.0) (6.9) (6.0) (4.9) (3.7) (2.7) (1.8) (0.8) (0.0)
Concerning the Temperature Dependence of the (378.176 K), respectively, were assumed to have errors of
Redlich–Kister Parameters the same magnitude as the other mixture vapor pressure
measurements. Third, the following assumptions were made
As exhibited in Table 2 for the various fits to the experi- regarding the temperature dependence of the B1, B2, and
mental data with either a three- or four-parameter Redlich– B3 : B1 can be fitted to a Lorentzian curve from the three
Kister approach, all the parameters are significantly temper- values obtained in the fits by Equation (13), where the Ci
ature dependent in the range of 60–90 8C. The linear T func- values are constants and the three values the B2 and B3 pa-
tion for B0 and the “temperature smoothing” offered by the rameters can be fitted to sigmoid curves of the general form
Scatchard group insufficiently represent what is really hap- given in Equation (14), where the Cij values are constants.[39]
pening. The percentage contributions of each term in the
Redlich–Kister chemical potential expansion and the imper- B1 ¼ C0 þ C1 C2 =pðC2 2 þ ðTC3 Þ2 Þ ð13Þ
fect gas correction terms as a function of T were evaluated
for various liquid mole fractions at the three nominal tem- Bi ¼ C0j þ C1j =½1 þ expfC2j ðTC3j Þg ð14Þ
peratures 60, 75, and 75 8C. Results from these calculations
revealed that the predominant term by far is B0. Thus, it The parameters for Equations (13) and (14) are shown in
would seem that its temperature dependence would be the Table 8. For the two temperatures of 317.636 and 378.176 K
most important to characterize in some detail. To do this, the values for Bi were then calculated from the above equa-
first it was assumed that none of the Redlich–Kister param- tions. These values, along with the two corresponding liquid
eters are discontinuous; there is no evidence otherwise. mole fractions, the temperature-dependent liquid molar vol-
Second, the two mixture vapor pressures at the corrected t umes, the appropriate virial coefficients, and the water and
values[38] of 44.486 8C (317.636 K) and 105.026 8C hydrogen peroxide vapor pressures (calculated from vapor
Table 5. Mole fraction of hydrogen peroxide above H2O2–H2O mixtures; % error with original work indicated in parenthesis.
Temp. Mole fraction hydrogen peroxide in vapor
[8C] 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0.0 0.0028 0.0087 0.0208 0.0430 0.0793 0.1350 0.2217 0.3698 0.6371
(27.5) (27.3) (28.6) (27.9) (23.9) (16.8) (9.0) (4.7) (5.9)
10.0 0.0032 0.0099 0.0232 0.0472 0.0861 0.1451 0.2356 0.3870 0.6518
(22.9) (23.0) (24.5) (24.1) (20.4) (13.8) ( 6.5) (3.0) (4.7)
20.0 0.0036 0.0110 0.0256 0.0514 0.0928 0.1548 0.2489 0.4031 0.6651
(18.3) (18.7) (20.6) (20.5) (17.1) (10.9) (4.3) (1.6) (3.8)
25.0 0.0038 0.0116 0.0267 0.0535 0.0960 0.1595 0.2553 0.4109 0.6715
(16.0) (16.6) (18.6) (18.7) (15.5) (9.5) (3.6) (1.0) (3.5)
30.0 0.0040 0.0121 0.0279 0.0555 0.0992 0.1642 0.2616 0.4186 0.6779
(13.7) (14.4) (16.6) (18.8) (13.9) (8.3) (2.5) (0.5) (3.2)
40.0 0.0044 0.0134 0.0300 0.0593 0.1054 0.1736 0.2749 0.4354 0.6924
(8.2) (9.4) (12.2) (13.1) (10.8) (6.0) (1.2) (0.1) (3.1)
50.0 0.0046 0.0138 0.0315 0.0623 0.1111 0.1835 0.2909 0.4578 0.7137
(0.28) (2.1) (6.2) (8.5) (7.7) (4.4) (1.3) (1.3) (4.1)
60.0 0.0052 0.0151 0.0337 0.0657 0.1157 0.1899 0.2997 0.4693 0.7240
(2.2) (0.9) (2.5) (4.6) (4.0) (1.4) (0.8) (0.4) (3.8)
70.0 0.0060 0.0170 0.0366 0.0693 0.1200 0.1949 0.3057 0.4750 0.7255
(0.8) (0.5) (0.8) (1.4) (0.3) (2.0) (3.4) (1.5) (2.6)
75.0 0.0064 0.0185 0.0389 0.0719 0.1239 0.1040 0.3233 0.4924 0.7172
(1.2) (2.7) (2.2) (1.0) (0.0) (0.2) (0.1) (0.7) (0.8)
80.0 0.0069 0.0193 0.0401 0.0741 0.1284 0.2125 0.3352 0.5006 0.7108
(2.0) (2.0) (0.9) (0.2) (0.3) (1.2) (1.7) (1.0) (0.8)
90.0 0.0076 0.0211 0.0435 0.0797 0.1367 0.2229 0.3455 0.5067 0.7098
(1.1) (1.4) (0.6) (0.1) (0.3) (0.9) (1.0) (0.3) (2.2)
100.0 0.0083 0.0229 0.0469 0.0849 0.1439 0.2316 0.3542 0.5133 0.7131
(0.1) (0.6) (0.0) (0.5) (0.3) (0.1) (0.0) (1.2) (2.8)
110.0 0.0090 0.0246 0.5000 0.0899 0.1508 0.2401 0.3633 0.5215 0.7185
(1.1) (0.4) (0.9) (1.2) (1.0) (0.5) (0.6) (1.7) (3.1)
120.0 0.0096 0.0263 0.0531 0.0947 0.1576 0.2488 0.3730 0.5308 0.7255
(2.7) (1.7) (1.9) (2.0) (1.6) (0.9) (0.8) (1.8) (3.0)
130.0 0.0103 0.0280 0.0562 0.0995 0.1644 0.2575 0.3829 0.5405 0.7328
(4.2) (3.0) (2.9) (2.7) (2.0) (1.2) (0.9) (1.7) (2.8)
140.0 0.0110 0.0297 0.0593 0.1044 0.1712 0.2661 0.3926 0.5500 0.7400
(5.4) (3.9) (3.5) (3.1) (2.2) (1.2) (0.8) (1.5) (2.5)
150.0 0.0118 0.0315 0.0625 0.1093 0.1781 0.2748 0.4023 0.5593 0.7470
(6.34) (4.6) (4.0) (3.4) (2.3) (1.1) (0.6) (1.2) (2.2)
160.0 0.0125 0.0334 0.0658 0.1144 0.1851 0.2834 0.4119 0.5685 0.7538
(7.0) (5.1) (4.2) (3.4) (2.2) (0.9) (0.31) (0.9) (1.9)
Table 6. PHP in mm Hg calculated from various equations.

Temp. [8C] 2-parm.[a] 4-parm.[b] 7-parm.[c] Thesis 4-parm.[d] Y-parm.X-parm.[e] Thesis 4-parm.X-parm.[e]
0.0 0.411 0.356 0.352 0.272 0.004 (1.1) 0.080 (22.7)
10.0 0.865 0.781 0.777 0.627 0.004 (0.5) 0.150 (19.3)
20.0 1.731 1.614 1.612 1.357 0.002 (0.1) 0.255 (15.8)
25.0 2.407 2.274 2.274 1.952 0.000 (0.0) 0.322 (14.2)
30.0 3.309 3.163 3.166 2.769 0.003 (0.1) 0.397 (12.5)
40.0 6.069 5.910 5.918 5.360 0.008 (0.1) 0.558 (9.4)
45.0 8.102 7.948 7.958 7.325 0.010 (0.1) 0.633 (8.0)
50.0 10.720 10.579 10.590 9.899 0.011 (0.1) 0.691 (6.5)
60.0 18.299 18.214 18.218 17.521 0.004 (0.0) 0.697 (3.8)
70.0 30.279 30.277 30.262 29.839 0.015 (0.0) 0.423 (1.4)
75.0 38.529 38.566 38.540 38.427 0.026 (0.1) 0.112 (0.3)
80.0 48.693 48.753 48.719 49.077 0.033 (0.1) 0.358 (0.7)
90.0 76.283 76.261 76.269 78.200 0.008 (0.0) 1.931 (2.5)
100.0 116.666 116.181 116.452 121.058 0.271 (0.2) 4.877 (4.20)
105.0 143.068 142.091 142.675 149.118 0.584 (0.4) 7.027 (4.9)
110.0 174.512 172.781 173.886 182.528 1.105 (0.6) 9.747 (5.6)
120.0 255.748 251.342 254.546 268.650 3.205 (1.3) 17.308 (6.9)
130.0 367.759 358.294 366.117 386.760 7.824 (2.2) 28.466 (7.9)
140.0 519.610 501.338 518.448 545.601 17.110 (3.4) 44.263 (8.8)
150.0 722.267 689.575 724.145 755.430 34.569 (5.0) 65.855 (9.6)
160.0 988.812 933.624 999.372 1028.100 65.748 (7.0) 94.475 (10.1)
[a] Calculated from Equation (8). [b] Calculated from Equation (11). [c] Calculated from Equation (10). [d] Calculated from Equation (9). [e] Percent de-
viation indicated in parenthesis; up to 90 8C X is 7 and Y is 4; above 90 8C X is 4 and Y is 7.
Table 7. Calculations of Hv (in cal mol1) from Equation (24) for hydrogen peroxide which was used as a test 348.222 K to 391.463 K, a 2o
for Equations (9), (10), and (11). polynomial (parameters P2i) was
Temp. [8C] Equation (9) Equation (11) Equation (10) %diff. Diff. II and III Experimental fit to the B0 values at 348.222,
col.I col.II col.III I and II (%diff. II and III) 363.169, and 378.176 K; above
0.0 12 932 12 126 12 246 6.6 120 (1.0) 12 620[a] 391.463 K B0 was taken as a
25.0 12 587 11 878 11 913 6.0 35 (0.3) 12 340[b]
constant of 612.961. The com-
45.0 12 331 11 695 11 688 6.0 7 (0.1)
60.0 12 149 11 567 11 550 5.0 17 (0.1) posite curve is shown in
75.0 11 977 11 446 11 442 4.6 4 (0.0) Figure 1 and the parameters for
90.0 11 815 11 330 11 368 4.3 38 (0.3) the various temperature regions
100.0 11 712 11 255 11 340[c] 4.1 85 (0.8) are given in Table 8. The B0
105.0 11 662 11 218 11 332[c] 4.0 114 (1.0)
110.0 11 613 11 182 11 330[c] 3.9 148 (1.3)
minimum at about 324 K and
115.0 11 565 11 146 11 333[c] 3.8 187 (1.7) increase at lower temperatures
150.0 11 260 10 903 11 498[c] 3.3 595 (5.6) are consistent, as mentioned
[a] W. T. Foley, P. A. GiguRre, Can. J. Chem. 1951, 29, 895–903. [b] P. A. GiguTre, B. G. MorTssette, A. W. above, with the good four-pa-
Olmos, O. Knop, Can. J. Chem. 1955, 33, 804–820. [c] Note here that Hv from Equation (10) passes through a rameter Redlich–Kister fit
minimum and then begins to increase with temperature, which is wrong. found in the present work to
Table 8. Parameters for the Redlich–Kister temperature functions and

their first derivatives for B1, B2, B3 and the four regions of B0 ; see Equa-
tions (13), (14), and (16)–(21).
Para- Temperature Curve type Constants
meter range
Bo 273.150–317.636 K Lorentzian C0 = 666.8830
C1 = 2499.584
C2 = 8.261924
C3 = 327.4487
317.636–348.222 K average of above P10 = 17 418.34
Lorentzian and 2nd P11 = 109.9125
order polynomial P12 = 0.1663847
348.222–391.463 K 2nd order polynomial P20 = 6110.401
P21 = 28.08669
P22 = 0.03587408
391.463–433.150 K B0 = 612.9613
B1 273.150–433.150 K Lorentzian C01 = 126.7385
C11 = 2558.776
C21 = 12.33364
C31 = 343.1050
B2 273.150–433.150 K sigmoid C02 = 63.18354
C12 = 149.9278
C22 = 0.4745954 Figure 1. Composite B0 function. Red curve is Lorentzian; blue curve is
C32 = 348.1642 the average of a Lorentzian and a second-order polynominal; green
B3 273.150–433.150 K sigmoid C03 = 59.42228 curve is second-order polynominal; purple curve is linear region; the ex-
C13 = 199.2644 perimental points are crosses. The units of B0 here are calories mol1
C23 = 0.8321514 (1 calorie = 4.184 Joules) to correspond to the other Bi results in Table 2.
C33 = 346.2121
the limited vapor pressure results of GiguRre and Maass[30, 31]

for HP–W mixtures at 30 8C.
For calculations of mixture total vapor pressures and
pressure Equation (10) for the lower temperatures and vapor mole fractions, a master program was developed. The
Equation (11) for the higher temperatures) allowed B0 to be program includes sections that calculate the water vapor
varied until the observed mixture vapor pressures were ob- pressures from a seven-parameter equation and, depending
tained. Thus, two more B0 values based on experimental on the temperature range, hydrogen peroxide vapor pres-
mixture vapor pressures were obtained. From much experi- sures from four- or seven-parameter equations, temperature-
mentation it was determined that no single function could dependent liquid molar volumes, temperature-dependent
be found that would fit all five of the B0 values. Thus, sever- virial coefficients, and the four temperature-dependent Red-
al functions were used to describe the behavior of B0. From lich–Kister parameters. The outputs are mixture vapor pres-
273.150 to 317.636 K, a Lorentzian was used; it was fit to sures and vapor phase mole fractions calculated for the
the B0 values at 317.636, 333.146 and 348.222 K (parameters liquid mole fraction water or hydrogen peroxide at any in-
Ci). The average of a 2o polynomial (parameters P1i) and a crement desired. For comparisons the program also calcu-
Lorentzian (the same from the fit over the latter tempera- lates mixtures vapor pressure data and vapor phase mole
ture range) was used from 317.636 to 348.222 K. From fractions using the parameters and equations from the origi-
nal analyses.[6h, 8] Optional outputs are any or all of the other @B1 =@T ¼ 2ða1 =ð1 þ a1 2 Þ2 ÞC11 =pC12 2 ð19Þ
quantities mentioned above and the percentage contribu-
tions for each term in the chemical potential expression. where a1 = (TC13)/C12
The only inputs are the absolute temperature and a small
correction term to the seven-parameter water vapor pres- @B2 =@T ¼ C21 C22 a2 =ð1 þ a2 Þ2 ð20Þ
sure equation given below. These results are exhibited in
Tables 4 and 5 with indications of the deviations from the where a2 = exp(C22(TC23))
original work. As discussed above deviations as high as
@B3 =@T ¼ C31 C32 a3 =ð1 þ a3 Þ2 ð21Þ
22 % for mixturesI vapor pressures were found. Hydrogen
peroxide vapor phase mole fraction deviations as high as
where a3 = exp(C32(TC33))
29 % were found at the lower temperatures and around am-
bient temperatures which are those where much COH radical Using the constants in Table 8 to evaluate Equation (1 b’)
generation work[2, 3] and spectroscopy[3, 4] are carried out. and, using the derivative expressions above in Equa-
tion (15), sXE values were calculated. The excess heat of
mixing, hXE, is calculated from the relation hXE = gXE +
Energy, Entropy, and Enthalpy of Mixing TsXE. For comparison all three quantities were also calculat-
ed using the “temperature-smoothed” constants from the
The first of these quantities, gXE, can be found by substitut- original work.[6h, 8] Figures 2 and 3 exhibit these results for
ing the temperature-dependent Redlich–Kister parameters
into Equation (1 b), where the terms Bi(T) are functions of
temperature.
gE ¼X1 ð1X 1 Þ½B0 ðTÞ þ B1 ðTÞð12X 1 Þ þ B2 ðTÞð12X 1 Þ2

þB3 ðTÞð12X 1 Þ3
ð1b0 Þ
In the original work B0 = B0’ + kt, where B0’ was “tempera-

ture-smoothed,” and the other Bi values were taken as the
“temperature-smoothed” constants. The excess entropy, sXE,
is obtainable from Equation (15), where the constants come
from Table 8.
sX E ¼@gX E =@T ¼ X 1 ð1X 1 Þ½@B0 ðTÞ=@Tþ

@B1 ðTÞ=@Tð12 X 1 Þ þ @B2 ðTÞ=@Tð12 X 1 Þ2 þ ð15Þ
3
@B3 ðTÞ=@Tð12 X 1 Þ
When the full temperature functions obtained here for

the Bi values are inserted in Equation (15), a rather cumber-
Figure 2. H2O2–H2O excess functions at 75 8C. The red curves are from
some expression results. Also, remember that B0(T) has four the present work; the blue curves are from equations in references
parts. The derivatives of the B0(T), B1(T), B2(T), and B3(T) [6h, 7, 8]; the gE curves from the latter workers and the present work are
functions are given in Equations (16)–(21). nearly identical.
273.15 to 317.636 K
the nominal temperatures of 60, 75, and 90 8C. It can be
@B0 =@T ¼ 2ða0 =ð1 þ a0 2 Þ2 ÞC1 =pC2 2 seen that there are very significant differences between the
ð16Þ
where a0 ¼ ðTC3 Þ=C2 original work and the present treatment, especially for TsXE
and hXE in the region of 75 8C. These differences point up
317.636 to 348.222 K the inadequacy of the previous results.
However, the gXE values from both treatments are not
that different at all temperatures. The equations in the origi-
@B0 =@T ¼ 0:5½2ða0 =ð1 þ a0 2 Þ2 ÞC1 =pC2 2 þ P11 þ 2P12 T
nal treatment yield a temperature-independent hXE with
ð17Þ only very slight changes in gXE and TsXE with temperature.
The major contributions to the 75 8C sXE in the present anal-
348.222 to 391.463 K (P2iIs from fit to 2o polynomial) ysis are the slopes of the parameters near 75 8C. If one were
to consider approximate linear functions of temperature for
@B0 =@T ¼ P21 þ 2P22 T ð18Þ B0, B2, and B3 and the slope of a Lorentzian for B1 only
over the 75 8C region, TsXE values nearly the same as those
T > 391.463 K @B0/@T = 0.0 obtained from the more complex temperature functions de-
Figure 3. H2O2–H2O excess functions at 60 and 90 8C. The red curves are Figure 4. H2O2–H2O excess functions at 25 and 50 8C. The red curves are
from the present work; blue curves are from equations in references from the present work; blue curves are from equations in references
[6h, 7, 8]. [6h, 7, 8].
rived above result. Thus, from the very negative TsXE values solution interactions. Although the ammonia–hydrazine
in this temperature region it appears that some rather signif- system (A–HA) might be seen as a similar pair, and al-
icant changes in the H2O2–H2O solution interactions are though there are more hydrogen bonding possibilities in
taking place at certain mole fractions. These changes are not that system, the types of hydrogen bonds involved are much
reflected in the activity functions (given in the Supporting weaker[41] so the solution interactions can be expected to be
Information). Figure 4 and Figure 5 exhibit the excess quan- less for ammonia–hydrazine mixtures than for the case of
tities at four other temperatures. Noteworthy are the TsXE H2O2–H2O mixtures.
maximum around 50 8C and the complexities seen at 55 8C. The vapor pressures of H2O2–H2O mixtures have received
Further investigation of H2O2–H2O mixtures in all these little attention since the work of Scatchard, Kavanagh, and
temperature regions by other techniques should reveal the Ticknor[6h, 8] of over 50 years ago. From the detailed evalua-
nature of the solution interaction changes hinted at by the tion of these authorsI work, we conclude that the experi-
TsXE behaviors seen here. mental work was carefully done and the experimental data,
outside of several minor problems, is worthy of a contempo-
rary reanalysis using the approach outlined here and current
Discussion and Conclusions data fitting methodology. The only significant aspect of the
data with which we take issue concerns their neglect of a
Besides the current interest in the H2O2–H2O mixtures,[2–5] small correction for hydrogen peroxide decomposition; Ka-
the present results are of interest because the H2O2–H2O vanagh exhibited results[8] which he felt indicated that hy-
system is probably the simplest binary system with extensive drogen peroxide decomposition was unimportant, but we
hydrogen bonding where the hydrogen bonding of each feel his results warrant that a small constant decomposition
pure component should be very similar and because the de- correction be made to the original published data over all
rived thermodynamic properties might shed light on existing the temperatures.
ing.[27] It has been demonstrated here that these pressures

can be parameters in the analyses of the data. Further it was
shown above that fits with only three Redlich–Kister param-
eters, although leading to pure hydrogen peroxide vapor
pressure values essentially the same as found in the original
analyses, give pure hydrogen peroxide vapor pressure values
that fit poorly to two-, four-, or seven-parameter hydrogen
peroxide VP–T equations (see Table 3 and Table 4).
As shown here the goodness of the fits of mixture vapor
pressure are very sensitive to small changes (of the order of
0.002 K) of temperature. The temperature iteration to a
least-squares minimum of the vapor pressure errors suggest-
ed here leads to much improved fits. The temperature ar-
rived at represents an average temperature seen by a suite
of mixtures where the average temperature of a particular
mole fraction mixture is slightly different at various loca-
tions within a well insulated apparatus, even though the
input energy and condensation parameters may have been
always set to give the same output of a multi junction ther-
mocouple system. This situation was first recognized by
Uchida, Ogawa, and Yamaguchi[1] some 54 years ago, but it
seems not to have been noted by other workers measuring
mixture vapor pressures. As mentioned above, the former
workers did not consider iteration of the temperature in the
analysis of their data. However, these workersI observations
were incorporated by one of the present authors in the
design of an apparatus for mixture vapor pressure measure-
ments[23] and led us in the present work to consider iteration
of temperature in our reanalysis of the Scatchard, Kava-
nagh, and Ticknor data.[6h, 8] As suggested above, possibly a
major contribution to this effect arises from small differen-
ces in heat transfer for mixtures of different mole fractions
Figure 5. H2O2–H2O excess functions at 40 and 55 8C. The red curves are
due to differences in the heats of vaporization of the compo-
from the present work; blue curves are from equations in references nents.[24] This effect then represents a source of error in any
[6h, 7, 8]. parameters determined from such data and needs to be
taken into account.
The findings discussed above and the demonstration that
the pure hydrogen peroxide vapor pressure can be extracted
In the present work various hydrogen peroxide physical from the mixture vapor pressure data should have signifi-
constant estimates, several other constants and several equa- cant practical importance in mixture vapor pressure studies.
tions needed in the fitting have been updated as described In the past for vapor pressure studies of pure hydrogen per-
below. These include the following: use of new value for the oxide and other pure substances it was deemed necessary to
gas constant; new estimates of the hydrogen peroxide criti- carry out laborious purifications to obtain materials as close
cal constants based on the theory of corresponding as possible to 100 % purity. Previous studies of such hydro-
states;[32–35] use of water vapor pressure and liquid density gen peroxide materials at temperatures much above 50 8C
data taken from a recent steam table update;[11] use of appear to have been plagued by decomposition, whereas in
newer hydrogen peroxide liquid density results;[21] use of the the case of mixtures that are 90 % or so in hydrogen perox-
empirical Tsonopoulos treatment for estimation of the hy- ide, the decomposition can be kept to a very low level as
drogen peroxide second virial coefficients.[12] was done by Scatchard, Kavanagh, and Ticknor.[6h, 8] Thus,
A weakness in the original analyses most definitely was for studies of mixture vapor pressures of two components at
the “temperature smoothing” of the Redlich–Kister parame- a particular nominal temperature, if a good VP–T equation
ters obtained by averaging the parameters from the separate for one component is known, the pure vapor pressure of the
fits to the data at the three nominal temperatures 60, 75, other component should be determinable from the mixture
and 90 8C. That all the parameters are substantially tempera- vapor pressure data. If enough different mole fraction mix-
ture-dependent is evident (see Table 2) even in the original tures are studied at a nominal temperature, it also should be
work. The linear temperature dependence of B0 previously possible in some cases to determine the pure vapor pres-
suggested[6h, 8] is definitely insufficient. The original descrip- sures of both components.[42] Also, The approach demon-
tion of the estimations of the pure hydrogen peroxide vapor strated here can be extended to mixture vapor pressure data
pressures from the three sets of data is somewhat confus- involving more than two components.
The use of four Redlich–Kister parameters in the fitting Estimation of Constants, Derivation of Density
of the three sets of data considered herein led to much im- Equations and Review of Data
proved fits and to three pure hydrogen peroxide vapor pres-
sure values that fit the two-parameter VP–T Equation (8) Concerning the experimental data: The experimental work
very precisely over the range 60–90 8C. Combined with the displayed in detail in the Kavanagh thesis[8] has been care-
pure hydrogen peroxide vapor pressure data reported by fully scrutinized.[43] Detailed study of the rates of decompo-
Maass and Hiebert,[13] reasonable four- and seven-parameter sition of hydrogen peroxide have been reported by
VP–T equations could be derived. It was found that the Schumb.[44] It was not expected[8] that the decomposition
seven parameter equation is not valid above 90 8C, but the rates for hydrogen peroxide would be as low as was found
four parameter equation appears to give reasonable pres- in this latter study due to the large and complicated surface
sure estimates even up to the critical region. of the vapor pressure apparatus. For various mole fraction
Attempts at fittings with five Redlich–Kister parameters mixtures at temperatures of 60, 75, and 90 8C measurements
and with PHP and T iterations, for reasons that are not clear, were made by Kavanagh[8] of the pressure rise in the experi-
did not lead to stable solutions, reasonable pure PHP values, mental system during the time of a usual run. The small
or much improved agreements with the experimental data. pressure rise observed at all temperatures did indeed signifi-
Perhaps, as suggested elsewhere[15, 16] these sets of data may cantly exceed what would be expected from the work of
not be accurate and extensive enough to warrant use of Schumb.[44] However, it was concluded that the amount of
more parameters than six. decomposition “would not affect the results appreciably.”[8]
To derive temperature functions for the four Redlich– We disagree with this conclusion as it appears that the aver-
Kister parameters, B0, B1, B2, and B3, some assumptions, de- age extent of decomposition is nearly the same at all the
scribed above, were adopted regarding the nature of the three temperatures and all the mole fractions studied. This
temperature dependencies. These led to relative simple tem- perhaps indicates some catalytic region of constant area at
perature equations for these functions. The Gibbs excess en- some site within the apparatus. Thus, a small correction of
ergies of mixing, gXE, were calculated from these tempera- 0.073 mm Hg was applied to the original vapor pressure
ture-dependent Redlich–Kister parameters; the values measurements in the present analysis. The original experi-
found indicated negative deviations from ideal mixtures at mental data are exhibited in Table I along with the data cor-
all temperatures. The excess entropy, sXE, was calculated uti- rected for this decomposition.[38] As mentioned above the
lizing the temperature derivatives of the four Redlich– mixture vapor pressures are given here in mm Hg
Kister parameters equations; the sXE behavior is driven by (1.0 mm Hg = 1.3332237 L 102 Pa) to facilitate comparisons
the nature of the derivatives of the temperature dependen- with the original data and analyses,[6k, 8] the results in an ex-
cies of the four Redlich–Kister parameters. It varies from a tensive review of the original work[7] and currently available
very small positive value near 0 8C to a maximum positive commercial data sheets describing the vapor pressures of
value near 50 8C to a very negative minimum near 75 8C to HP–W mixtures.[9] In other places pressure units used are at-
small negative values at 120 8C for large XHP and small posi- mospheres or bars (1.0 atm = 1.013250 bar).
tive values at 120 8C for small XHP. The sXE behaviors in the
40, 55, and 75 degree regions (see Figure 2 to Figure 5) Calculation of molecular volumes and water vapor pressure
appear very complex and warrant further investigation by equation: Temperature-dependent values of water and hy-
other techniques. The naive linear temperature correction to drogen peroxide liquid molecular volumes are required for
B0 put forth in the original work[6h, 8] could give no hint of the gas law deviation corrections and for the empirical cal-
the complex behavior our reanalysis of the experimental culations of the virial coefficients described below. The orig-
data suggest. Also, the results in the original work lead to a inal work[6h, 8] used the hydrogen peroxide liquid density re-
temperature-independent hXE which is certainly wrong. sults of Huckaba and Keyes[45] extrapolated from results at 0
It is hoped that the results presented here and the recent and 20 8C to the region 60–90 8C; this extrapolation seems a
interest in H2O2–H2O mixtures[2–5] will stimulate further little unrealistic. In the present work the newer liquid densi-
studies of this system. A better characterization of H2O2– ty equation [Eq. (22)] of Easton, Mitchell, and Wynne-
H2O mixtures at temperatures below 60 8C and down to am- Jones,[10] based on measurements at 0, 10, 25, 50, and 96 8C,
bient temperatures should be feasible; such results would was used, where a is the density of water at t [oC], w the wt.
test further the assumptions and analyses made in the pres- fraction of hydrogen peroxide (here w = 1.00) and b, c, and
ent work regarding the parameter temperature dependen- d are functions given elsewhere.[7, 10]
cies and should provide further pure hydrogen peroxide
vapor pressure values. It would appear that a vast amount 1HP ¼ a þ bw þ cw2 þ dw3 ð22Þ
of mixture vapor pressure data in the literature could be
fitted substantially better in light of the results presented
The water liquid density function developed here is based
here.
on the density data contained in the new steam table.[11]
Equation (23) represents the simplest equation that gave a
precise fit to the experimental data.
1W ¼ 1:0000 þ c1 T þ c2 T 2 þ c3 =T þ c4 =T 2 ð23Þ
However, two sets of ci values were required. The first set mates are very close. The other hydrogen peroxide boiling
fit over the range 0 to 52.5 8C and the second set over the point estimate was derived by the Scatchard group[6h, 8] by a
range 52.5 to 160 8C. The first set of parameters were de- somewhat modified Ramsey–Young[28] treatment that yield-
rived from 14 values which included data at 18 intervals ed the four-parameter hydrogen peroxide VP–T Equa-
from 0 to 8 8C and at 108 intervals to 50 8C. These parame- tion (9) (pressure in mm Hg) exhibited above.
ters reproduced the density maximum around 4 8C. The Equation (9) gives a boiling point of 150.2 8C. This value
second set of parameters were derived from 12 values start- gives a TcHP value of 730.2 K, which was used initially in the
ing at 55 8C with values at 10 8 intervals from 60 8C to 160 8C. present work to obtain the first estimated values of the
Possibly these two sets of parameters might be of use to second virial coefficients from the empirical equations de-
others in certain cases so they are shown in Table 9. The scribed below. The results of the new fits with four Redlich–
Kister parameters discussed above yielded significantly dif-
ferent values for the three pure hydrogen peroxide vapor
Table 9. Parameters for liquid water density functions based on new steam
table density data[11] fitted by equation 1w = 1.000 + c1T + c2T2 + c3/T + c4/T2. pressures at the three nominal temperatures of “60”, “75”
T range [K] c1 c2 c3 c4
and “90” oC. These values and selected data from Maass and
3 6
Hiebert[13] were combined, as described above, to give the
273.15–328.15 3.171838 L 10 4.279879 L 10 367.0906 59.47127
328.15–433.15 4.718129 L 104 1.709110 L 106 27.28696 7352.166
four- and seven-parameter hydrogen peroxide VP–T Equa-
tions (11) and (10) yielding boiling points of 153.155 8C
(426.305 K) and 151.478 8C (424.628 K), respectively. These
boiling points led to very slightly different critical constants,
water and hydrogen peroxide liquid density equations de- slightly different virial coefficients and different Redlich–
scribed above then were used to calculate the liquid molecu- Kister parameters, but no significant changes in the pure hy-
lar volumes at various temperatures using the molecular drogen peroxide vapor pressures arising from the second
weights of 18.016 for water and 34.016 for hydrogen perox- rounds of fitting to the experimental measurements by four
ide. Redlich–Kister parameters and iteration of T and PHP.
The seven-parameter VP–T equation (24); pressure in With TcW = 647.3 K[46] and with the assumption that the
mm Hg) for water derived from the new steam table data[11] ratio TcHP/TcW is approximately equal to the ratio TbHP/TbW ,
was used in the simulation program described above. TcHP values of 736.6 or 739.5 K are yielded, depending on
whether the boiling point is 151.478 or 153.155 8C, respec-
log10 PW ¼19:3891272861:9133=T3:2418662 log10 T tively. For comparison applying this concept to the ammo-
1:0799994 104 T7:9189289 106 T 2 ð24Þ nia–hydrazine pair, which are certainly also molecules
having significant hydrogen bonding interactions in the
þ1:5411774 108 T 3 8:1926991 1012 T 4
liquid, the hydrazine critical temperature, TcHZ, is calculated
as 653.9 K compared with a literature value of 653 K;[46] this
From comparisons with the steam table data a very small
result supports the corresponding state theory estimate[32–35]
correction to this equation needed to be made above 100 8C.
of TcHP made above.
Hydrogen peroxide boiling point, critical constants, and Several suggestions have been made for estimation of the
Pitzer acentric factor: The conditions for the experimental critical pressure. One is that insertion of the critical temper-
determination of hydrogen peroxide critical constants prob- ature into a VP–T equation will give the critical pressure.
ably will never be realized. For empirical calculations of Kavanagh[8] calculated a PcHP value of 214 atm from Equa-
virial coefficients needed for the gas law corrections terms tion (9), which seems too low and is less than the 218.3 atm
estimates of these parameters are required. A good estimate PcW of water.[46] For the ammonia–hydrazine pair (A–HZ)
of the hydrogen peroxide critical temperature, TcHP, can be the former has a PcA value of 112.0 atm[46] and the latter a
made based on corresponding state theory.[32–35] This ap- PcHZ value of 145 atm.[46] Another pair of molecules that sug-
proach says that the ratio of the atmospheric boiling points gests that the value of PcHP should be significantly greater
of water, TbW, and its critical temperature, TcW, is the same than that of PcW is the methylamine–methyl hydrazine pair
for a closely related substance, in this case the ratio TbHP/ (MA–MH), in which the former has a PcMA value of
TcHP. The required boiling point of pure hydrogen peroxide 73.3 atm and the latter a PcMH value of 81.3 atm.[46] Using
at atmospheric pressure (TcHP) has not been measured. the seven-parameter hydrogen peroxide VP–T equation
However, three estimates have been reported. Maass and [Eq. (10)], PcHP is estimated as 1837 atm which is unreasona-
Hiebert[13] estimated this boiling point in two ways. From bly high and higher than any known value for a critical pres-
the classic, two-parameter VP–T equation (pressure in sure. Thus this seven-parameter VP–T equation definitely is
mm Hg), given in Equation (8) they calculated a value of invalid close to the critical region. However, the four-param-
151.1 8C. Their other estimate was derived from the eter VP–T equation [Eq. (11)] gives a value of 255.0 atm. In
Ramsey–Young extrapolation approach,[28] which assumes the original work PcHP was also required for calculation of
that the ratio between temperatures at which a substance the virial coefficients. Although the above Kavanagh value
whose vapor pressure is to be extrapolated and a reference was exhibited as PcHP, the value used for calculation of the
substance with similar solution properties or interactions has virial coefficients appears to have been based on the as-
the same vapor pressure is taken as a linear function of tem- sumption that the ratio PcW/TcW is approximately equal to
perature. A value of 151.4 8C was calculated; thus, both esti- the ratio PcHP/TcHP. The TcHP of 730 K calculated in the origi-
nal work[8] gives a PcHP value of 247 atm. For compounds Table 10. Constants used in calculations of virial coefficients.
that have very similar interactions in the liquid there is sup- TBP[a] Tc[a] Pc Vc w
port for this approximation from the law of corresponding water [b]
373.150 647.3 218.3 atm 3
57.1 cm mol 1
0.344
states.[32–35] This assumption then gives PcHP as 248.4 atm for hydrogen peroxide 426.305 739.5 252.2 atm 83.0 cm3 mol1 0.417
the boiling point at 151.5 8C and 249.4 atm for the boiling [a] Temperature in K. [b] Taken from reference [46].
point at 153.2 8C. This assumption applied to the ammonia–
hydrazine pair gives a PcHZ value of 180.4 atm, which is 24 %
higher than the literature value of 145 atm.[46] On considera- this section and which are required for calculations of the
tion of the pair methylamine–methyl hydrazine, a PcMH virial coefficients.
value of 96.7 atm is calculated, which is 19 % higher than It should be kept in mind that all the constants estimated
the literature value of 81.3 atm.[46] Its effect in calculations in this section are used to calculate virial coefficients that go
below being small, PcHP was taken as 252.2 atm, the average into the imperfect gas law correction terms of Equations (6)
derived from the four-parameter VP–T equation [Eq. (11)] and (7), which are expected to make only small contribu-
and the corresponding state theory estimate based on the tions to the complete chemical potential expressions Equa-
153.2 8C boiling point. tions (3) and (4) at the pressures involved here. To verify
The final two constants required in the Tsonopoulos virial this in the present work model calculations were carried out
coefficient calculations[12] are the hydrogen peroxide critical which showed that the imperfect gas correction terms con-
volume, VcHP, and the Pitzer acentric factor,[47] wHP. Here tribute at most 1–2 % to the mixture vapor pressures in the
two estimates of VcHP have been made. The first was made temperature and pressure ranges considered. Depending on
from the Equation (25) (volume in cm3) of Tyn and the mole fraction and temperature, these terms, however,
Calus,[48, 49] where Vb is the liquid molecular volume at the can contribute as much as 20 % to a particular, small total
normal boiling point and Vc the critical volume. chemical potential term.
V b ¼ 0:285ðV c Þ1:048 ð25Þ Calculations of second virial coefficients: For use in the cal-
culation of the second virial cross coefficient, bHP–W, values
For water Equation (24) gives VcW as 53.7 cm3 mol1 which of the so-called binary critical constants TcHP–W, PcHP–W and
is 5.9 % lower than the experimental value of wHP–W are required. The following mixing rules [Eq. (27)–
57.1 cm3 mol1.[46] When applied to hydrogen peroxide using (29)], recommended by Tsonopoulos and Heidman,[12b] have
the liquid density functions discussed above, which gives a been used.
molecular volume of 27.6 cm3 mol1 for the boiling point
153.2 8C, a VcHP value of 77.4 cm3 mol1 is obtained. We wHPW ¼ 0:5ðwW þ wHP Þ ð27Þ
chose to scale this value up by the deviation noted for water
1
to give a VcHP value of 82.3 cm3 mol1. Another estimate can T c HPW ¼ ðT c W T c HP Þ =2 ð1kHP-W Þ ð28Þ
be made from corresponding state theory arguments[32–35]
that the ratio of the liquid molecular volumes of similar liq- 4 T c HPW ðPc W V c W =T c W þ Pc HP V c HP =T c HP Þ
Pc HPW ¼ ð29Þ
uids at their atmospheric boiling points are in the ratio of ðV c W 1=3 þ V c HP 1=3 Þ3
their critical volumes; this gives VcHP as 83.7 cm3 mol1. The
average of these two estimates, 83.0 cm3 mol1, has been If two polar substances are very similar in chemical
used in the present work. nature and not too different in size, kHP–W can be taken as
The Pitzer acentric factor, wHP, reflects the noncentral zero which was done here. The virial coefficients, bW, bHP,
nature of intermolecular forces. It has been defined by and bHP–W, were calculated from the Tsonopoulos Equa-
Equation (26)[47] where Ps is the vapor pressure calculated tion (30),[12] for which [Eq. (31)–(33)] are applicable.
from a vapor pressure equation at T with the reduced tem-
perature, Tr, of 0.7 and Tr = T/Tc. bi Pi c =RT i c ¼ f ð0Þ ðT r Þ þ wi f ð1Þ ðT r Þ þ fð2Þ ðT r Þ ð30Þ
w ¼ log10 ðPs =Pc ÞT r ¼0:7 1:000 ð26Þ fð0Þ ðT r Þ ¼0:14450:330=T r 0:1385=T 2r0:0121=T 3r
ð31Þ
0:000607=T 8r
For water Equation (26) gives a w value of 0.344, which is
the accepted value.[46] Using the seven- and four-parameter fð1Þ ðT r Þ ¼ 0:0637 þ 0:331=T 2r0:423=T 3r0:008=T 8r ð32Þ
VP–T equations for hydrogen peroxide derived below and
Equation (26), we calculated values of 0.244 and 0.417, re- fð2Þ ðT r Þ ¼ 0:0109=T 6r ð33Þ
spectively. The former value seems way too low, which indi-
cates that even at a Tr of 0.7 (575.7 K, 242.5 8C) the seven- In the above equations bi is the virial coefficient, Pic and
c
parameter VP–T equation [Eq. (10)] is not valid. One can T the appropriate critical pressure and temperature, respec-
i
again draw a comparison with the ammonia–hydrazine pair, tively, R the gas constant (units cm3 mol1 K1), wi the Pitzer
for which the w values are 0.244 and 0.316, respectively.[46] acentric factor, and Tr the reduced temperature, T/Tic. Here
The 0.417 value for wHP given by the four-parameter VP–T the pressure units of Pic are bars (1.0 atm = 1.01325 bar =
equation [Eq. (11)] then seems reasonable and has been 760.0 mm Hg). A program was written to calculate the bi
used here. Table 10 summarizes the constants estimated in values as a function of temperature for both water and hy-
drogen peroxide. These results along with the results for [17] See: H. C. Van Ness, M. M. Abbott, Classical Thermodynamics of
these parameters calculated from equations from the origi- Nonelectrolyte Solutions with Applications to Phase Equilibria,
McGraw-Hill, New York, 1982, pp. 200—202.
nal work[8, 21] are in the Supporting Information. [18] G. M. Wilson, PhD thesis, MIT (USA), 1953, p 56.
[19] The exponent terms in Equation (5) need to be divided by RT. For
the chemical potentials, m1E and m2E, the RT must be T L 82.0578/
41.2925, while RT for the imperfect gas correction terms is 760 L T L
82.0518.
Acknowledgment [20] F. G. Keyes, J. Chem. Phys. 1947, 15, 602 – 612.
[21] Calculated from an equation given by: F. G. Keyes, L. B. Smith,
We thank Dr. Art Lane and Dr. Stan Sander of the Jet Propulsion Labo- H. T. Gerry, Proc. Am. Acad. Arts Sci. 1936, 70, 319 – 364.
ratory, California Institute of Technology, for bringing this problem to [22] The data processing package used was “PSI-Plot” available from
our attention and for providing copies of relevant reprints and theses. We “Poly Software International,” P. O. Box 60, Pearl River, NY 10965
thank Prof. Harry Gray, Prof. Jack Roberts, and Prof. Fred Shair of The (USA). The version used to fit equations like Equation (5) above
California Institute of Technology, Prof. Alan Cowley, The University of must be “Version 3.” This version has a choice of three nonlinear
Texas, Austin, Prof. John Prausnitz, University of California, Berkeley, curve fitting techniques: Powell, Marquardt—Levelburg, and Sim-
and Dr. Ash Cooper, ICI, Alderley Edge for their careful reading of and plex methods. Newer versions of this software package only have
helpful comments on this work and for encouraging us to publish it. We the Marquardt-Levelburg method; we found that the latter does not
also wish to thank a reviewer for detailed comments and suggestions. give solutions to most iterative equations that have been tried in our
work. With “Version 3” sometimes both the Powell and the Mar-
quardt–Levelburg methods will not converge or fail many times in
[1] S. Uchida, S. Ogawa, M. Yamaguchi, Japan Sci. Rev., Eng. Sci. 1950, initial fittings where reasonable starting parameters are not known.
1, No. 2, 41 – 49 (in English). However, the Simplex method with somewhat arbitrary guesses was
[2] See for example: a) L. F. Keyser, J. Phys. Chem. 1982, 86, 3439 – found to yield reasonable initial parameters. When the fit was get-
3446; b) J. J. Lamb, L. T. Molina, C. A. Smith, M. J. Molina, J. Phys. ting close, it was observed that the Simplex method in “PSI-Plot”
Chem. 1983, 87, 4467 – 4470; c) S. P. Sander, J. Phy. Chem. 1984, 88, appears to wander for equations such as Equation (5). Thus, the
6018 – 6021; d) G. L. Vaghjiani, A. R. Ravishankara, N. Cohen, J. final fits were always made with the Powell method which always
Phys. Chem. 1989, 93, 7833 – 7837. led to stable solutions. It was noted that the final parameters from
[3] A. L. Lane, S. P. Sander, Jet Propulsion Laboratory; personal com- the Powell method vary slightly with the input parameter estimates.
munications. In our work the practice adopted was to use parameter estimates of
[4] J. M. Nicovich, P. H. Wine, J. Geophys. Res.[Atmos.] 1988, 93, 2417 – five significant digits which were iterated until no changes were ob-
2421. served in the fifth decimal place of final parameters. To reduce the
[5] a) R. W. Carlson, M. S. Anderson, R. E. Johnson, W. D. Smythe, possibility of roundoff error, only four decimals points are shown in
A. R. Hendrix, C. A. Barth, L. A. Soderblom, T. B. McCord, J. B. Table 2 and these are the numbers used in all the simulation data
Dalton, R. N. Clark, J. H. Shirley, A. C. Ocampo, D. L. Matson, Sci- presented. Equation (5) was programmed into the “PSI-Plot” non-
ence 1999, 283, 2062 – 2064; b) J. F. Cooper, R. E. Johnson, B. H. linear least-squared curve fitting script language with the parameters
Mauk, H. B. Garrett, N. Gehsels, Icarus 2001, 149, 133 – 159. B0, B1, B2, B3, and PHP. To start, the nominal T value reported in the
[6] a) G. Scatchard, C. L. Raymond, H. H. Gilman, J. Am. Chem. Soc. previous analyses[6h,8] was used. Next T was changed in small incre-
1938, 60, 1275 – 1278; b) G. Scatchard, C. L. Raymond, J. Am. Chem. ments and improvements in the observed minus calculated mixture
Soc. 1938, 60, 1278 – 1287; c) G. Scatchard, S. E. Wood, J. M. Mochel, vapor pressures were sought. Each change of T causes a change in
J. Phys. Chem. 1939, 43, 119 – 130; d) G. Scatchard, S. E. Wood, J. M. PHP and PW. Attempts to incorporate the two required VP–T equa-
Mochel, J. Am. Chem. Soc. 1939, 61, 3206 – 3210; e) G. Scatchard, tions along with Equation (5) proved too cumbersome and appeared
S. E. Wood, J. M. Mochel, J. Am. Chem. Soc. 1940, 62, 712 – 716; to exceed the capability of the nonlinear curve fitting script lan-
f) G. Scatchard, S. E. Wood, J. M. Mochel, J. Am. Chem. Soc. 1946, guage in “Version 3” of “PSI-Plot.”
68, 1957 – 1960; g) G. Scatchard, S. E. Wood, J. M. Mochel, J. Am. [23] S. P. Vango, S. L. Manatt, unpublished work carried out at the Jet
Chem. Soc. 1946, 68, 1960 – 1963; h) G. Scatchard, G. M. Kavanagh, Propulsion Laboratory, California Institute of Technology.
L. B. Ticknor, J. Am. Chem. Soc. 1952, 74, 3715 – 3720; i) G. Scatch- [24] In the work of reference [23] the following observations were made
ard, L. B. Ticknor, J. Am. Chem. Soc. 1952, 74, 3724 – 3729; j) G. that are suggestive of this explanation for mixtures. For solvent mix-
Scatchard, F. G. Satkiewiez, J. Am. Chem. Soc. 1964, 86, 130 – 133; tures of various mole fractions, the temperature differentials were
k) G. Scatchard, G. M. Wilson, J. Am. Chem. Soc. 1964, 86, 133 – 137. mole fraction dependent when the energy input, rate of reflux and
[7] W. C. Schumb, C. N. Satterfield, R. L. Wentworth, Hydrogen Perox- coolant flow rate were carefully held constant. However, tempera-
ide, ACS Monograph Series, Reinhold Publishing Corp., New York, ture differences were even observed for pure water and pure sol-
1956, pp. 221 – 231. vents but they were significantly less than for mixtures. In the analy-
[8] G. M. Kavanagh, PhD thesis, MIT (USA), 1949. sis of the data in the Jet Propulsion Laboratory work temperature
[9] FMC, Technical Bulletin 67. iteration was not done, but the temperature was always taken from
[10] M. F. Easton, A. G. Mitchell, W. F. K. Wynne-Jones, Trans. Faraday a thermocouple at one particular location in the apparatus. Also for
Soc. 1952, 48, 796 – 801. the most part pure component vapor pressures were taken from
[11] L. Haar, J. S. Gallagher, G. S. Kell, NBS/NRC Steam Tables, Hemi- manufactureIs tables but in several cases vapor pressures were deter-
phere Publishing Corp., New York, 1984. mined so all these results certainly had errors, based on the observa-
[12] a) C. Tsonopoulos, AIChE J. 1974, 20, 263 – 272; b) C. Tsonopoulos, tions in the present work, probably of the order of those existing in
J. L. Heidman, Fluid Phase Equilib. 1990, 57(3), 261- 276. the original Scatchard, Kavanagh, and Ticknor work[6h, 8]
[13] O. Maass, P. G. Hiebert, J. Am. Chem. Soc. 1924, 46, 2693 – 2700. [25] The Scatchard group (references [6h, 8]) stated that the measured
[14] O. Redlich, A. T. Kister, Ind. Eng. Chem. 1948, 40, 345 – 348. vapor mole fractions, Yi’s, could not be measured to nearly the same
[15] For example, see J. M. Prausnitz, R. N. Lichtenthaler, E. G. de Aze- accuracy as the liquid mole fractions and mixture vapor pressures.
vedo, Molecular Thermodynamics of Fluid-Phase Equilibria, 3rd ed. However, the Yi’s are calculated accurately from the equation Yi = Pi
Prentice Hall, New Jersey, 1999, pp. 225 – 264. Xiexp[(Xj2/RT)(miE + IGC i)]/P.
[16] For other texts see: a) E. Hala, J. Pick, V. Fried, O. Vilim, Vapor- [26] a) C. Antoine, Compt. Rend. 1888, 107, 1143 – 1145; b) See also:
Liquid Equilibrium, Pergamon Press, Oxford, 1967; b) H. C. Van G. W. Thomson, Chem. Rev. 1946, 46, 1 – 39.
Ness, M. M. Abbott, Classical Thermodynamics of Nonelectrolyte [27] In the original analyses of the three sets of mixture vapor pressures
Solutions with Applications to Phase Equilibria, McGraw-Hill, New at the nominal temperatures of “60,” “75” and “90”8C it was ac-
York, 1982; c) J. S. Rowlinson, F. L. Swinton, Liquid and Liquid knowledged by Scatchard, Kavanagh and Ticknor that the vapor
Mixtures, 3rd ed., Butterworth Scientific, London, 1982. pressures of pure hydrogen peroxide at these latter temperatures
are required.[6h,8] They first obtained these vapor pressures by graph- small correction term obtained graphically from plots of the differ-
ical extrapolations of the mixture vapor pressures to 1.00 mole frac- ences between the pressures from the above equation and the new
tion hydrogen peroxide. There was no mention of what the starting steam table data.[11] The fit to a quadratic relation in the Celsius
extrapolated pure hydrogen peroxide vapor pressures were. The temperature with inclusion of all 17 thesis EMF values exhibited
Redlich–Kister constants were then obtained “using a successive ap- average errors significantly greater than 0.02 8C. Sifting the data
proximation form of the method of least squares.” “Between steps easily eliminated six of the thesis EMF points and the resulting 11
in the successive approximation for the constants, new values for the points yielded a fit with an average error 0.005 8C. The stated tem-
hydrogen peroxide vapor pressure were calculated by use of the im- peratures for the various runs (44.50, 60.00, 75.00, 90.00 and
proving constants and the experimental pressures.”[6h, 8] The pure hy- 105.00 8C) (with errors estimates of about 0.02 oC) were achieved
drogen peroxide vapor pressures resulting from the iterations to a by observing the thermocouple system EMF, while setting the appa-
stationary sum of the squares of the vapor pressure residuals were ratus heater system to give an EMF corresponding to that of the de-
not given. The values for pure hydrogen peroxide vapor pressures at sired temperature as given by the quadratic relation determined
the three nominal temperatures were calculated from the final fit from the water VP–T equation. Although the temperatures estimat-
parameters using only “the two measurements most concentrated in ed in the present work (44.495, 60.004, 75.016, 90.031, and
hydrogen peroxide. The two values obtained were averaged with
105.050 8C with an error estimate of 0.005 8C) differ only slightly
double weight being given to the determination more concentrated
from the thesis ones, the important thing to keep in mind here is
in hydrogen peroxide ….”[6h, 8] No mention was made of how many
that the water vapor pressure differences between the Keyes equa-
iterations were required or of any estimates of the error. The state-
tion[20] values and an equation derived here based on the new steam
ment was made that, “The final extrapolation for the vapor pres-
table data[11] (old steam equation, 70.034, 149.400, 289.100, 525.724,
sures was made with least squared constants before they had been
and 905.700 mm Hg versus new, 70.039, 149.447, 289.307, 526.380,
temperature smoothed.”[6h, 8] The meaning of “final extrapolation” is
and 907.443 mm Hg) affect the goodness of fit significantly as the
not clear to us. The values for the pure hydrogen peroxide vapor
water vapor pressures multiply the exponentials of the excess
pressures found by Scatchard, Kavanagh, and Ticknor are shown in
Table 3 along with the errors of their fits to Equations (8) and (9). energy terms. The differences in temperature discussed here, howev-
Fits were also carried out in the present work using the Wilson[29] er, become unimportant because, as discussed above, the average
approach with a third polar parameter and two gas imperfection temperature of a suite of mixtures vapor pressure measurements of
terms (with PHP as a parameter and T iteration). Results similar to different mole fractions made at a particular heater setting dictated
those from a three parameter Redlich–Kister treatment were ob- by a particular EMF setting can be determined in the fit to the mix-
tained for the goodness of fit for the “60” and “75” degree data, but ture vapor pressure data.
the fit for the “90” degree data was poor and the values for the pure [39] For the Lorentzian and sigmoid curves four parameters are required
hydrogen peroxide vapor pressures for the “75” and “90” degree ex- to be determined from only three data points. The following itera-
perimental data were substantially different, 38.98 and 79.24 mm Hg, tive procedure was used with “PSI-Plot”: two parameters were
respectively, from those (see Table 3) from the three-parameter guessed and two parameters were fit to the three data points; in the
Redlich–Kister treatment. next stage the fit parameters were input and two parameters guessed
[28] W. Ramsay, S. Young, Philos. Mag. 1886, 21(5), 33 – 51. originally were fit; repetition of this procedure, while monitoring
[29] G. M. Wilson, J. Am. Chem. Soc. 1964, 86, 127 – 130. the least square residuals and the parameters, was stopped when the
[30] P. A. GiguRre, O. Maass, Can. J. Res. 1940, 18B, 181 – 193. parameters were unchanged in the sixth decimal place.
[31] These same authors presented data at 45 and 60 8C but they had se- [40] The activity coefficients, gi values, were calculated at a number of
rious reservations about its quality as they were plagued by decom- temperatures including 60, 75 and 90 8C. Although the differences
position. Calculations using the parameters derived in the present between the gi values calculated from the previous work and the
work gave results in very poor agreement with these latter data sets present work are small at 75 8C, the latter gi values exhibited no
which seems to confirm the serious experimental problems these au- anomalies that might reflect on the large TsXE values calculated in
thors had at these two temperatures. Our parameters at 60 8C this temperature region. This is consistent with the gi values coming
should certainly be correct because of the good fit to the data of ref- essentially from gXE. The gXE values of the previous and present
erences [6h, 8]. works are close at all temperatures. The equations used to calculate
[32] F. G. Keys, J. Am. Chem. Soc. 1938, 60, 1761 – 1764. the gi values are given in reference [15], pp. 236–237.
[33] See Reference [15] pp. 104–113. [41] See for example: L. Pauling, The Nature of the Chemical Bond,
[34] E. A. Guggenheim, J. Chem. Phys. 1945, 13, 253 – 261. 2nd ed., Cornell University Press, Ithaca, New York, 1948, pp. 301 –
[35] G. W. Castellan, Physical Chemistry, Addison-Wesley, Palo Alto 304 and p. 333.
1964, pp. 41 – 43.
[42] We have verified this possibility in one case using the Scatchard and
[36] C. F. Curtiss, J. O. Hirschfelder, J. Chem. Phys. 1942, 10, 491 – 496.
Raymond data on the ethanol-cholorform system.[6b] Possibly the
[37] G. S. Kell, G. E. McLaurin, E. Whalley, J. Chem. Phys. 1968, 48,
reason it works for the latter data is because there were several
3805 – 3813.
measurements for mole fraction greater than 0.9 for both compo-
[38] Several questions regarding the original temperature calibrations
nents.
arose. The centigrade temperature (International Practical Celsius
[43] Several typo errors in the experimental data in the thesis,[8] the
Temperature) was taken as t = T273.16 in the original work, where-
review,[7] and the original publication[6h] were noted. In Table 1 of
as now the accepted equation is t = T273.15 where in both cases T
the thesis the EMF in run #3 should be 0.035814 v and the tempera-
is the International Practical Kelvin Temperature (See: The Interna-
tional Practical Temperature Scale of 1968, Metrologia 1969, 5, 35 – ture should be 44.599 8C; the pressures in runs 7 and 8 should be
44). This is a very small difference, but as discussed above the quali- 394.711 and 354.182 mm Hg, respectively; in Table 2 the mole frac-
ty of the fits to the experimental data is very sensitive to T. The tion at 60 8C of 0.1577 should have a pressure 26.21 mm Hg and the
original temperature calibrations of the apparatus were based on fourth mole fraction should be 0.4221; in reference [7] in Table 16
EMF measurements with a 20 junction thermocouple system and a the mole fraction at 60 8C of 0.8423 should have a pressure of
water VP–T equation reported by Keyes[20] some 57 years ago. In 26.21 mm Hg and the eighth mole fraction at 75 8C should be 0.7460;
the present work the recent revised water vapor pressure values in reference [6h] the 44.50 and 105.00 8C smoothed pressures should
(and water liquid densities used below) recommended by Haar, Gal- be 27.41 and 415.68 mm Hg, respectively.
lagher and Kell were used.[11] In the present analysis for the temper- [44] W. C. Schumb, Ind. Eng. Chem. 1949, 41, 992 – 1003.
ature determinations from experimental water vapor pressure values [45] C. E. Huckaba, F. G. Keyes, J. Am. Chem. Soc. 1948, 70, 2578 – 2500.
(Pw in mm Hg) the new water vapor pressure data for 11 points [46] For an extensive compilation of boiling points, critical constants,
58C about each estimated temperature was fit by a separate Antoine acentric factors, and other molecular constants, see: R. C. Reid,
equation[26] of the form ln Pw = A + B/(T + C) + Corr. where A, B J. M. Prausnitz, B. E. Poling, The Properties of Gases & Liquids,
and C are least-square determined constants and Corr. is a very 4th ed., McGraw-Hill, New York, 1987, pp. 656 – 732.
[47] a) K. S. Pitzer, J. Am. Chem. Soc. 1955, 77, 3427 – 3433; b) K. S. [49] See: R. C. Reid, J. M. Prausnitz, B. E. Poling, The Properties of
Pitzer, D. Z. Lippman, R. F. Curl, C. M. Huggins, D. E. Peterson, J. Gases & Liquids, 4th ed., McGraw-Hill, New York, 1987, pp. 53 – 55
Am. Chem. Soc. 1955, 77, 3433 – 3440. .
[48] M. T. Tyn, W. F. Calus, Processing 1975, 21, 16 – 17. Received: February 3, 2004
Revised: July 3, 2004
Published online: November 18, 2004

Chem 200400104

Uploaded by

Copyright:

Available Formats

Chem 200400104

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chem 200400104

Uploaded by

Copyright:

Available Formats

FULL PAPER

On the Analyses of Mixture Vapor Pressure Data: The Hydrogen Peroxide/

Stanley L. Manatt* and Margaret R. R. Manatt*[a]

Introduction temperature. Knowledge of multicomponent liquid–vapor

ago by Uchida, Ogawa, and Yamaguchi[1] who observed Theoretical Considerations

90.01 0.0403 0.1142 84.92 0.30 90.019 84.847 0.029

Table 6. PHP in mm Hg calculated from various equations.

Table 8. Parameters for the Redlich–Kister temperature functions and

the limited vapor pressure results of GiguRre and Maass[30, 31]

gE ¼X1 ð1X 1 Þ½B0 ðTÞ þ B1 ðTÞð12X 1 Þ þ B2 ðTÞð12X 1 Þ2

In the original work B0 = B0’ + kt, where B0’ was “tempera-

sX E ¼@gX E =@T ¼ X 1 ð1X 1 Þ½@B0 ðTÞ=@Tþ

When the full temperature functions obtained here for

ing.[27] It has been demonstrated here that these pressures

You might also like