Solubility of Hydrogen in a White Oil

B. A. SCHOFIELD, Z. E. RING* and R. W. MISSEN$

Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, Ontario, Canada M5S 1A4

The solubility of hydrogen in a white mineral oil has been measured at pressures of 1.48 to 10.44 MPa at each of
four temperatures (573,603,623,643 K), in conjunction with a study of hydrodesulphurization in a trickle-bed reactor.
The solubility varies from a hydrogen mole fraction of 0.056 at the lowest { T, P ) to 0.262 at the highest. The solubility
increases with increasing pressure and temperature. It can be represented by an equation suggested by using Krichevsky-
Ilinskaya and van’t Hoff expressions.
~- ~~

La solubilitC de I’hydrogkne dans une huile minCrale blanche a CtC mesurCe a des pressions allant de 1.48 10.44
MPa a chacune des quatre tempkratures (573, 603, 623 et 643 K), en parallkle avec une Ctude de I’hydrodCsulfurisation
dans un rCacteur B lit ruisselant. La solubilitC varie d’une fraction molaire d’hydrogkne de 0,056 pour le plus bas [ T, P ]
a 0,262 pour le plus ClevC. La solubilitC augmente avec I’augmentation de la pression et de la tempkrature. Elle peut
Ctre reprisentee par une Cquation combinant les expressions de Krichevsky-llinskaya et de van’t Hoff.

Keywords/phrases: hydrogen solubility; white mineral oil; dependence on pressure and temperature; n-decane

A s part of a study of trickle-bed reactors, including

hydrodesulphurization (Ring and Missen, 1986, 1989,
1991), we present here solubility data for hydrogen in a white
mineral oil at conditions of temperature (573 to 643 K) and
pressure (1.5 to 10.5 MPa) which include those for com-
mercial hydrotreating operations. A white oil, a readily
obtainable material, consisting of saturated paraffins and
naphthenes was chosen as the liquid carrier for diben-
zothiophene (DBT) in the hydrodesulphurization reaction to
simulate a hydrocarbon stream with appropriate boiling
characteristics (588-793 K) and relatively low coking
tendency.
The primary purpose of the phase equilibrium (solubility)
measurements was to establish, for the conditions used in
the reactor, the role of DBT as the limiting reactant, and to
indicate that the reaction might then be expected to be pseudo-
first-order in DBT, as subsequently confirmed kinetically
(Ring and Missen, 1989). The solubility data reported here,
however, are for a range of temperature and pressure con-
ditions greater than that used in the operation of the trickle-
bed reactor. Solubility data for hydrogen in n-decane are also
given for comparison with literature data.
Experimental methods
APPARATUS
AND PROCEDURE
The solubility-measuring apparatus was essentially the
same as that of Wisniak et al. (1974). It consisted of a high-
pressure equilibrium cell, and an ambient-pressure burette
system for volumetric measurement of gas and liquid.
A four-litre stainless-steel autoclave (Autoclave Engineers,
Inc., Erie, PA; rated at 21 MPa and 615 K) was used as the
equilibrium cell (the same vessel as was used for the kinetics
experiments; Ring and Missen, 1989). It was equipped with
sampling and purge ports, a Magnedrive high-speed stirrer
*Present address: Shell Canada Limited, Box 2 100, Oakville, Ontario
L6J 5C7.
‘Author to whom correspondence should be addressed.
for internal agitation, a pressure gauge, and a jacket-type
heater and cooling coil for temperature adjustment and con-
trol. Pressure was measured by a Bourdon-tube gauge (CPW
Hydro-Poise, Toronto, ON) to k0.03 MPa; the gauge was
calibrated by a dead-weight tester. Temperature was con-
trolled to k0.5 K by a Thermo Electric (Brampton, ON)
AN-12 Controller operating in conjunction with the flow of
liquid to the cooling coil. Temperature was measured by a
Type J thermocouple and was shown on a digital indicator
to 0.1 K (Doric Trendicator 400A, San Diego, CA).
The autoclave was connected to the burette system. To
determine the amount of gas dissolved in a measured volume
of liquid after equilibration, a liquid sample was transferred
from the autoclave to a 50-cm3 glass burette at ambient
pressure, and gas flashing from the liquid passed to a con-
necting 250-cm3 glass burette. Both burettes were immersed
in a thermostated bath containing water at 313.2 + 0.1 K.
The volume of gas was measured by displacement of a
column of mercury in the receiving burette maintained at
ambient pressure; the volume was converted to conditions
of 293 K and 0.1013 MPa, and was corrected to allow for
the solubility of gas at 0.1013 MPa. The volume of liquid
remaining was measured.
Preliminary experiments were used to establish the time
required for equilibration in the autoclave, and the precision
of the solubility determinations. These involved two serics
of measurements for hydrogen in the white oil at 643 K. in
which equilibrium at 9.75 MPa was approached from both
a higher pressure (1 1.13 MPa) and a lower pressure (8.37
MPa), and solubility was measured at times from IS to 120
min. For all eleven determinations in the two series made
at times of 30 min and greater, the mean deviation was 2.5%.
The means for the two series differed by 0.6%. The time
of 30 min. was then chosen as a standard time for equilibra-
tion in the autoclave.
Solubility measurements for hydrogen in the white oil were
made over the (total) pressure range 1.5 to 10.5 MPa at each of
four temperatures: 573, 603, 623, and 643 K. Measure-
ments were also made for hydrogen solubility in n-decane
at 503 K for comparison with the data of Sebastian et al.
(1980).

822 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING. VOLUME 70, AUGUST. 1992
 TABLE1
Comparison of Measured Solubility of Hydrogen in n-Decane at
503 K
This Work Sebastian et al. (1980)
1.48 0.0178
1.96 0.0303 1.94 0.0289
3.27 0.0557 3.29 0.05 11
5.20 0.0828 5.18 0.0818
6.99 0.1111
10.10 0.1507 10.12 0.1550
MATERIALS
The materials used were hydrogen, n-decane and the white
oil. A cylinder of UHP hydrogen gas (99.999 + %) was
obtained from Linde Air Products. The n-decane (olefin-free,
99 + %) was obtained from Fluka Chemical Corp. (St.
Louis, MO). The transparent white oil was supplied by Shell
Canada Limited as White Oil D . It is made up of saturated
aliphatic and alicyclic hydrocarbons; the sample used had
a density of 878 kg/m3 at 15"C, a viscosity of 31 mPa * s
at 40"C, an average molar mass of 383 kg/kmol, and a
boiling-point range of 315-520°C, this last making it suitable
for the hydrodesulphurization reaction studies. All materials
were used without further treatment.
Results and Discussion
The measured solubilities of hydrogen in n-decane at 503 K
are given in Table 1, together with the results of Sebastian
et al. (1980) for comparison, for (total) pressure ( P ) up to
10 MPa. The solubility is given as mole fraction of hydrogen,
XH2,which was calculated from the volumes of gas and
liquid measured, as described by Herskowitz et al. (1983).
The mole fraction of hydrogen remaining in the liquid in the
solubility train (at 313 K, 0.1013 MPa) was considered to
be negligibly small. The mean deviation of the six values
of this work from graphically smoothed values of Sebastian
et al. (1980) is 5 % .
The measured solubilities of hydrogen in the white oil at
1.48 to 10.44 MPa and 573 to 643 K are given in Table 2.
This solubility is also given as mole fraction of hydrogen,
which was calculated from the volumetric measurements
using the density and molar mass for the oil given above.
The solubility increases from a hydrogen mole fraction of
0.056 at the lowest { T, P 1 to 0.262 at the highest. At a given
pressure, the solubility increases with increase in tempera-
ture, as in the case of hydrogen solubility in n-decane (Sebas-
tian et al., 1980). It also increases with increase in pressure
at a given temperature.
The behaviour of the solubility data in Table 2 as a func-
tion of T and P was explored in three ways for the purpose
of fitting. The best fit was given by a cubic power series
in P with coefficients linearly dependent on T; this required
6 constants and multiple regression resulted in a standard
deviation dxH,) = 0.0035. More thermodynamically
rational approaches were also used with the dependence on
P given by forms suggested by the Krichevsky-Karasrnovsky
and Krichevsky-Ilinskaya equations (Prausnitz et al., 1986),
and the dependence on T by a van't Hoff expression, as used
by van Vuuren et al. (1988) for a similar situation. If we
assume that the vapour is pure HZ, that the fugacity
THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING, VOLUME 70, AUGUST, 1992
coefficient of H2 is 1 (it ranges from 1 .O to 1.05 for the con-
ditions used here; see e.g., Sandler, 1989), and that P > >
the vapour pressure of the oil, then the Krichevsky-Ilinskaya
equation, combined with the van't Hoff expression, can be
written as
xH, = P exp [ A + BxH?(xH, - 2) + C P ]exp ( - D / T )
....................................... (1)
This requires the use of 4 constants, A , B , C , and D. The
Krichevsky-Kasarnovsky expression for the effect of P is
simpler with B = 0; the equation corresponding to Equa-
tion (1) involves 3 constants, and is usually more appropriate
for relatively small solubilities. The use of Equation ( l ) , with
( A , B , C , D)T = (0.5222, 2,253. 0.009484 MPa-'. 2103
K)T, resulted in u(xH,) = 0.0044; the simpler form ( B =
0) gave u(xH,) = 0.006. We conclude that Equation ( I ) is
the best representation, considering the goodness of fit, the
number of constants, and the rationality of the basis.
Nevertheless, the physical interpretation of the constants is
blurred by their interdependence in the representation of thc
effects of T and P.
In spite of differences in conditions (Tand P ) and in com-
position, as reflected, for example, by average molar mass.
comparisons may be made with other measurements of Hz
solubility in similar relatively ill-defined liquids. The com-
parisons given below use the ratio R = xH, (this WOrk)/.YH,
(literature) or R' , a similar ratio using solubility expressed
in specific mass units (where the molar mass is not known).
van Vuuren et al. (1988) have measured the solubility of
H2 in a Fischer-Tropsch reactor wax (molar mass 464
kglkmol) at 438 to 516 K and 0.3 to 1 MPa. From their equa-
tion for the effect of T, extrapolation to our lowest value,
573 K, gives R = 1.3 at 1.48 MPa; conversely, the same
value is given by extrapolation of Equation (1) to 516 K at
1.48 MPa. Ding et al. (1985) have used a coal liquid (SRC
11, molar mass 221 kg/kmol) at 420 to 680 K and 4 to
11 MPa. For corresponding values of T and P, the ratio R
varies from about 2.6 at 4 MPa to 2.0 at 10.5 MPa. Simi-
larly, Lin et al. (1981) have used particular fractions from
SRC-I1 liquid. For an oil fraction with molar mass
252 kgikmol, R is 2.5 at 542 K and 10 MPa. As a final com-
parison, we use the results of Prather et al. (1977) for H2
solubility in creosote oil (obtained from distillation of a coal
pyrolysis liquid), with a boiling range of 175 to 350°C. At
573 K, the ratio R' is 1.4 at 7 MPa and 1.3 at 10 MPa. In
all these comparisons, the solubility of hydrogen is greater
in the white oil.
The solubility data were used to indicate that the concen-
tration of hydrogen in the oil in the operation of the trickle-
bed reactor would be sufficiently great to make DBT the
limiting reactant in the liquid-phase hydrodesulphurization
reaction
(C,jH&S + H2 = (C6H5)2 + H2S ... . . (2)
The concentration of DBT in the liquid feed corresponded
to 470 mg/kg as sulphur. From the solubility at 603 K (the
lowest temperature used in the reactor) and 10 MPa (the pres-
sure used), the molar ratio of hydrogen to sulphur in the
liquid of density 708 kg/m3 (at 603 K; Ring and Missen.
1989) is about 65, compared with the stoichionietric value
of 1. Even if the equilibrium value represented by the solu-
bility were not reached in the operation of the reactor.
involving a thin film of liquid, the indication is that the actual
value was > > 1. This meant that a particular form of the
822
 TABLE2 Abbreviation
Measured Solubility2 of Hydrogen in White Oil D , xH?
DBT = dibenzothiophene
T, K
References
P, MPa 573 603 623 643
Ding, F. X., S . H. Chiang and G. E. Klinzing. “Hydrogen Solu-
1.48 0.056 0.061 0.070 0.073
bility in Coal Liquid (SRC-11)”. Fuel 64. 1301-1305 (1985).
2.17 0.078 0.085 0.094
Herskowitz, M., J. Wisniak and L. Skadman, “Hydrogen Solu-
2.86 0.084 0.094 0. I04 0.116
bility in Organic Liquids,’’ J. Chem. Eng. Data 28, 164-166
4.24 0.113 0.129 0.140 0.154
(1983).
5.61 0.139 0.157 0.170 0.183
Lin, H. M., H. M. Sebastian, J. J . Simnick and K. C. Chao. “Solu-
6.99 0. I67 0.183 0.197 0.207
bilities of Hydrogen and Methane in Coal Liquids”, Ind. Eng.
7.68 0.194 0.211 0.221
Chem. Process Des. Dev. 20, 253-256 (1981).
8.37 0.205 0.221 0.227
Prather, J. W., A. M. Ahanger, W. S. Pitts, J . P. Henley, A. R.
9.06 0. I92 0.214 0.229 0.242
Tarrer and J . A. Gruin, “Solubility of Hydrogen in Creosote
9.75 0.224 0.239 0.245 Oil at High Temperatures and Pressures”, Ind. Eng. Chem.
10.44 0.216 0.235 0.249 0.262
Process Des. Dev. 16, 267-270 (1977).
“Each entry is the mean of two determinations (average mean devi- Prausnitz, J. M., R. N . Lichtenthaler and E. G. de Azevedo.
ation was 0.0036). “Molecular Thermodynamics of Fluid-Phase Equilibria”, 2nd
ed., Prentice-Hall, Englewood Cliffs, NJ (1986), pp. 376-380.
model for the overall effectiveness factor relating to a non- Ring, Z. E. and R. W. Missen, “Trickle-Bed Reactors: Effect of
volatile limiting reactant (Ring and Missen, 1986) could be Wetting Geometry on Overall Effectiveness Factor”, Can. J .
used. Further support for this conclusion was obtained kinet- Chem. Eng. 64, 117-124 (1986).
ically; the reaction was found to be pseudo-first-order in DBT Ring, Z . E. and R. W. Missen, “Trickle-Bed Reactors: An
(Ring and Missen, 1989). Experimental Study of Partial Wetting Effect”. AIChE J . 35,
1821- 1828 (1989).
Ring, Z. E. and R. W. Missen, “Trickle-Bed Reactors: Tracer
Acknowledgements Study of Liquid Holdup and Wetting Efficiency at High Tem-
perature and Pressure”, Can. I. Chem. Eng. 69, 1016-1020
Financial assistance has been received from the Natural Sciences ( 1 99 1).
and Engineering Research Council of Canada and from the Univer- Sandler, S. I., “Chemical and Engineering Thermodynamics”, 2nd
sity of Toronto. We are grateful to Shell Canada Limited for ed., Wiley, New York (1989) p. 224.
supplying the white oil used and values of its properties. Sebastian, H. M., J. J. Simnick, H-M. Lin and K-C. Chao. “Gas-
Liquid Equilibrium in the Hydrogen + n-Decane System at
Nomenclature Elevated Temperatures and Pressures”, J. Cheni. Eng. Data 25.
68-70 (1980).
A = constant in Equation ( I ) , dimensionless van Vuuren, D. S . , J. R. Hunter and M . D. Heydenrych. “The
B = constant in Equation (l), dimensionless Solubility of Various Gases in Fischer-Tropsch Reactor Wax,”
c = constant in Equation (I), MPa-’ Chem. Eng. Sci. 43, 1291-1296 (1988).
D = constant in Equation ( I ) , K Wisniak, J., M. Herskowitz, R. Leibowitz and S . Stein, “Hydrogen
P = total pressure, MPa Solubility in Aqueous Solutions of Sugars and Sugar Alcohols”.
R = x H L(this work)/xH, (literature) J. Chem. Eng. Data 19, 247-249 (1974).
R’ = ratio as for R , buf with solubility in g H2/g oil
T = temperature, K
-
‘H2 - equilibrium (liquid) mole fraction (solubility) of H2

Greek letter
Manuscript received May 8, 1991; revised manuscript received
u(xHJ = standard deviation in fitted values of xH, December 9, 1991; accepted for publication December 23, 1991.

824 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING. VOLUME 70. AUGUST. I992