Tribute to Founders: John M. Prausnitz.

AIChE Journal
Thermodynamics and Molecular-Scale Phenomena DOI 10.1002/aic.14818

Dew-point Measurements for Water in

Compressed Carbon Dioxide
Christopher W. Meyera and Allan H. Harveyb
a
Sensor Science Division, National Institute of Standards and Technology, Gaithersburg, MD,
20899, USA, [email protected]
b
Applied Chemicals and Materials Division, National Institute of Standards and Technology,
Boulder, CO, 80305, USA, [email protected]

ABSTRACT
When transporting CO2 for sequestration, it is important to know the water dew point in order to avoid
condensation that can lead to corrosion. We have constructed a flow apparatus to measure the water
content at saturation in a compressed gas. A saturator humidifies the flowing gas by equilibrating it
with liquid water. Then, a gravimetric hygrometer measures the water mole fraction of the humid gas.
We report dew-point data for H2O in CO2 on six isotherms between 10 °C and 80 °C at pressures from
0.5 MPa to 5 MPa. Our uncertainties in water content at the dew point (expanded uncertainty with
coverage factor k=2) are on average 0.3%, significantly smaller than in any previous work. The data have
been analyzed to extract the interaction second virial coefficient; our values are consistent with the
theoretical estimates of Wheatley and Harvey but have a much smaller uncertainty.

Introduction
Knowledge of the thermophysical properties of mixtures containing CO2 and H2O is important
for many applications. One such property of current interest is the water dew-point temperature in
CO2, which is dependent on the moisture content and pressure of the gas. Knowledge of this property is
needed for CO2 transportation in pipelines for carbon capture and sequestration (CCS),1 because CO2
captured from power plants contains moisture. Before this CO2 is transported, its moisture content
must be reduced to the point where it will not condense and cause corrosion. The relation between the
dew-point temperature, water content, and pressure is important for determining the degree to which
the CO2 must be dried before transportation. Better understanding of thermodynamics of mixtures
containing CO2 and H2O is also important for the design and optimization of advanced power cycles that
facilitate carbon capture.2,3
This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process which may lead to
differences between this version and the Version of Record. Please cite this article as
doi: 10.1002/aic.14818
© 2015 American Institute of Chemical Engineers (AIChE)
Received: Jan 29, 2015; Revised: Mar 25, 2015; Accepted: Mar 30, 2015
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 2 of 41

25 March 2015, revision for AIChE Journal

The dew-point temperature is defined as the temperature at which liquid water and water in
the vapor phase are in equilibrium. TDP is related to the water-vapor amount (mole) fraction yw and gas
pressure p by iteratively solving the equation

e (T )
yw = f (T, p ) , (1)
p

where T is the absolute temperature and it is understood that T = TDP. Here, e(T) is the saturated water
vapor pressure at T.4 The water vapor enhancement factor f(T, p) reflects departures from ideal solution
behavior and non-ideal gas effects5 and is a gas-dependent property. Unfortunately, for CO2-H2O
mixtures there is limited theoretical information and only scattered experimental data on f(T, p) at the
temperatures relevant for pipeline transport.

Existing experimental data for the water vapor enhancement factor f(T, p) in compressed CO2 at
moderate pressures (below about 10 MPa) were reviewed in 2011 by Wheatley and Harvey.6 Especially
at relatively low temperatures relevant to pipeline transport, the existing experimental data7-11 were
found to be mutually inconsistent. A recent experimental study12 also covered the region of interest,
but the dew-point measurements at low temperatures had large uncertainties.

The value of f(T, p) can be calculated5 using relevant virial coefficients found in a series
expansion for the deviation of a fluid from ideal-gas behavior:

p
= 1 + Bρ + Cρ 2 + ... , (2)
ρRT
where p is the pressure, ρ is the molar density, and R is the molar gas constant. B and C are the second
and third virial coefficients, respectively. For pressures up to a few MPa, it is sufficient for many
purposes to truncate Eq. 2 after the C term, or even after the B term at sufficiently low pressure.

For a binary mixture, B and C are rigorously given as a function of amount fractions y1 and y2 by

Bmix = y12 B11 + 2 y1 y2 B12 + y22 B22 , (3)

Cmix = y13C111 + 3 y12 y2C112 + 3 y1 y22C122 + y23C222 . (4)

For CO2-H2O mixtures, the virial coefficients B11 and B22 correspond to the second virial coefficients of
pure CO2 and pure H2O, respectively, and C111 and C222 correspond to the third virial coefficients of pure
CO2 and pure H2O, respectively. The values of these coefficients are well known.13-15 Recently, Wheatley
and Harvey6 calculated the interaction second virial coefficient B12, which describes the interaction

2
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 3 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

between one H2O molecule and one CO2 molecule. B12 was derived as a function of temperature using a
potential-energy-surface obtained from high-level quantum calculations. With appropriate
thermodynamic manipulations, Eqs. 2, 3, and 4 lead to expressions for the fugacity coefficients needed
to describe equilibrium between coexisting phases, enabling calculation of f(T, p). The value of f(T, p)
depends strongly on B12 at all pressures. At high pressures, f(T, p) also depends on C112 and (to a much
lesser extent) C122.

Because of limitations on the accuracy of quantum calculations in reasonable computer time,

the potential-energy surface of Wheatley and Harvey6 had a significant amount of uncertainty, which
propagates into uncertainty in B12. This uncertainty becomes larger at lower temperatures, and at
pipeline temperatures is large enough to lead to uncertainty in the dew-point temperature on the order
of 3-5 K for pressures on the order of 5 MPa (or uncertainty on the order of 15% for the water mole
fraction at a fixed temperature).

In order to reduce this uncertainty, we have measured dew-point compositions along several
isotherms with an accuracy that greatly exceeds that of previous work. The measurements were taken
over the range 10 °C ≤ t ≤ 80 °C (t is the Celsius temperature) and 0.5 MPa ≤ p ≤ 5 MPa. The resulting
data provide a test of the accuracy of the B12(T) derived by Wheatley and Harvey, and allow more
accurate modeling of this important system.

Experimental
Our dew-point measurements were performed in a facility that has two main components. The
first component is a saturation system in which liquid water is equilibrated with the working gas at a
precisely controlled temperature and pressure. The saturator temperature is the dew-point
temperature corresponding to the pressure in the vessel and the water amount fraction generated. The
second component is the NIST gravimetric hygrometer,16 an apparatus that measures the water amount
fraction in a gas by separating the water from the gas using desiccants and subsequently determining
the amount of each independently. By using the value of yw measured by the gravimetric hygrometer
and the values of T and p measured in the saturator, f can be determined by manipulating Eq. 1 to yield

yw p
f (T, p ) = . (5)
e (T )

3
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 4 of 41

25 March 2015, revision for AIChE Journal

CO2 and H2O sources

The CO2 used came from a single gas cylinder. The purity of the CO2, as given by the gas
supplier, was 99.993%. Principal impurities were N2 (<50 ppm), H2 (<5 ppm), Ar (<5 ppm), CO (<5 ppm),
and water (<5 ppm). We did not perform a separate analysis of the gas purity. A high-purity regulator
was used with the CO2 cylinder.

The water in the saturator came from a commercial distillation system. The water source for
the distillation system was tap water that passed through a paper filter, a carbon filter, and a scale-
removal filter. The distilled water was stored in a polypropylene container before being placed in the
saturator. The water in the saturator was changed weekly.

Manifold and pressure control

The manifold between the gas cylinder and the inlet of the saturation system was composed of
electropolished stainless-steel tubing of outer diameter 0.635 cm and wall thickness 0.09 cm. The gas
flowed from the cylinder through a needle valve and then into the saturation system at the desired
pressure and flow rate (typically 1 to 5 standard liters per minute). The gas also flowed in parallel
through a commercial pressure controller, which kept the downstream pressure constant to within
0.01%. The upper limit of the controller was 7 MPa. The needle valve was employed because the
pressure controller could not otherwise control the pressure for the flow rates used.

The water amount fraction generated from the saturation system usually corresponded to a
dew-point temperature that was below the ambient temperature (at ambient pressure). For this case,
the manifold between the outlet of the saturation system and the inlet of the gravimetric hygrometer
was electropolished stainless-steel tubing of outer diameter 0.635 cm and wall thickness 0.09 cm. For
the case when the dew-point temperature was above ambient temperature, commercially-made heated
flexible tubing was used and controlled at approximately 120 °C.

Saturation system
The saturation system, shown in Figure 1, consisted of three elements: a heat exchanger, a pre-
saturator and a final saturator. The gas flowed through each of these elements in the order mentioned.
All elements were immersed in a commercial temperature-controlled bath of volume 95 L. They were
placed in the section of the bath that was unobstructed by stirrers; this section had horizontal
dimensions 69.9 cm×27.9 cm and depth 33 cm (a volume of 64 L). Based on measurements performed
in a similar bath,17 we estimate the bath temperature was stable to within 0.001 °C over the range

4
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 5 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

10 °C ≤ t ≤ 80 °C. We estimate it was uniform to within 0.001 °C over the range 10 °C ≤ t ≤ 40 °C and
within 0.008 °C over the range 40 °C ≤ t ≤ 80 °C. We also estimate that the temperature stability and
uniformity of the saturators were those of the bath.

The heat exchanger was designed to bring the temperature of the flowing gas to the
temperature of the saturator. It was composed of a single electropolished stainless-steel tube of outer
diameter 0.635 cm and wall thickness 0.09 cm. The length of the heat exchanger was 10 m. It was bent
into a spiral coil of diameter 23 cm. The exchanger was attached to the pre-saturator using VCR†
fittings.

The pre-saturator and final saturator had identical designs but different functions. The pre-
saturator performed almost all of the saturation. The final saturator made small final adjustments to
the saturation and was the location where the pressure and temperature measurements took place.
This separation of functions is common to metrology-grade saturators. It is done because the large
amount of evaporation in the pre-saturator causes unacceptable temperature non-uniformities. Very
little water evaporates in the final saturator, so the temperature inside is very uniform.

The design of the pre-saturator and final saturator, shown schematically in Figure 2, was based
on the “dish” design used by Hyland and Wexler.18 Each saturator was made of two machined disks of
316L stainless steel of diameter 19.4 cm that were clamped together by 12 stainless-steel bolts of
diameter 1.25 cm. The bottom disk was 3.5 cm thick and contained a cylindrical cavity of diameter 13.1
cm and depth 1.78 cm (volume 239 mL). This cavity was typically filled with 120 mL of distilled water,
resulting in a fill level of about 0.9 cm. The water was filled through a tube of diameter 0.635 cm
attached to the bottom of the cavity; this tube was normally sealed with a valve except when filling or
emptying the saturator water. The center of the cavity bottom was 0.13 cm deeper than on the sides to
ensure that all the water could drain through the fill tubes.

The gas entered each saturator from the top of the cavity through a 0.635 cm tube located at a
distance of 6.0 cm from the axial center. The gas exited the saturator through a 0.635 cm tube located
in the top of the cavity at a distance of 1.5 cm from the axial center. All fittings used to attach tubing to
the saturator were of the VCR type. The gas flowed through a set of circularly-concentric channels of
1.25 cm width as it passed through the cavity (see Figure 2). The channels were created by circular

†
Certain commercial products are identified in this paper but only in order to adequately specify the procedure.
Such identification neither constitutes nor implies recommendation or endorsement by either the U.S. government
or the National Institute of Standards and Technology.

5
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 6 of 41

25 March 2015, revision for AIChE Journal

channel separators of thickness 0.13 cm and height 0.63 cm that were part of the top piece of the
saturator. The gas passed from the outer channels to the inner channels through openings in the
channel separators of width 1.25 cm. The cavity was bounded by the disk material on the top, sides,
and bottom with thickness 2.54 cm, 3.14 cm, and 1.71 cm, respectively. The two disks were sealed
together by a Viton O-ring with thickness 0.16 cm and diameter 14.3 cm. Finite-element analysis was
used to determine that the saturator could withstand 11 MPa of pressure inside the cavity.

A photograph of the final saturator is shown in Figure 3. An expansion (needle) valve separated
the final saturator from the manifold leading to the gravimetric hygrometer, allowing pressurization of
the saturation system. The axis of the knob to the valve was vertical so that the knob could be adjusted
using an attached rod protruding outside the bath. An industrial platinum resistance thermometer (PRT)
was immersed in a well of diameter 0.318 cm and depth 2.22 cm located in the center of the top disk.
The PRT was calibrated by comparing against a reference PRT calibrated by the NIST Thermodynamic
Metrology Group. A tube leading to a commercial strain-gauge pressure transducer was attached to the
top disk 1.5 cm from the axial center (this tube was sealed for the pre-saturator). The transducer had a
range of 6.9 MPa and was calibrated by the NIST Thermodynamic Metrology Group.

Gravimetric hygrometer
This NIST gravimetric hygrometer has previously been described in detail.16 Briefly, it separates
the water from the gas in water collection tubes, using desiccants to perform the separation.
Subsequently, the water mass mw and the gas mass mg are determined independently. The water
amount fraction is then determined by
nw mw M w
yw ≡ = , (6)
nw + ng mw M w + mg M g

where nw and ng are the amounts of water and gas in moles, respectively, and Mw and Mg are the molar
masses of water and gas, respectively.

For each measurement of yw, a nominal amount of moist gas is passed through the gravimetric
hygrometer. This amount is determined so that approximately 1 g of water will be collected during the
passage. As the moist gas passes through the gravimetric hygrometer, the water is trapped in three
collection tubes containing the desiccant magnesium perchlorate (Mg(ClO4)2). The remaining dry gas is
directed to a gas collection system.

6
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 7 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

The design of the water collection tubes is shown in Figure 4. The tubes have caps that allow
them to seal the tubes while mounted inside the hygrometer’s manifold. The caps are sealed to the
tubes by compressing elastomer O-rings using special tools mounted in the manifold. When sealed, the
caps ensure that the gas inside the collection tubes will not mix with the ambient air. A second set of
elastomer O-rings seals the tubes to the manifold. All O-rings mentioned above are Teflon-encapsulated
to prevent absorption/desorption of moisture and gas in the elastomers.

After the tubes are mounted in the hygrometer manifold, the manifold is purged and flushed
with fresh dry gas of the same type used during the measurement. This process is repeated three times
to ensure all ambient air and moisture are removed from the manifold. The purging is performed with a
molecular drag pump to keep the manifold clean.

The masses of all three collection tubes are measured both before and after the gas is passed
through them. The masses are measured by comparison against a standard mass using a commercial
electronic balance with a range of 200 g and a resolution of 10–5 g. The mass measurements are
corrected for buoyancy. The mass change of the gas in the tubes is determined and subtracted from the
total mass change, yielding the mass change due to water collection (mw). Only the first two tubes are
intended to collect water; the third tube is intended to verify that all water was collected in the first two
tubes. In the collection tubes, one batch of desiccant is used for several measurements; once the first
tube collects over 10 g of water, the desiccants in the first two tubes are replaced.

The design of the gas collection system is shown in Figure 5. The system consists of two prover
pistons. As gas is collected in the prover tubes, the pistons rise. With gas access controlled by
computer-controlled pneumatic valves, the two prover pistons alternate in collecting gas, allowing
continuous automated collection. A laser interferometer system is used to accurately determine the
vertical position of the pistons. Since the inner diameters of the prover tubes are known from earlier
dimensional measurements, the volume of gas under each piston can be calculated by its vertical
position. Once a piston reaches a certain height (58 cm), a valve automatically opens the gas to the
other piston. As the second piston rises, pressure and temperature measurements are made
underneath the first piston. Afterwards, a valve opens to allow the gas collected under the first piston
to escape, allowing the piston to fall to its original position. A second measurement of pressure and
temperature under the piston is made. Afterwards, the first piston is kept in place until its gas-access
valve is reopened. The temperature and pressure measurements and the gas’s equation of state14 are
used to calculate the density of the gas underneath the piston. The volume and density measurements

7
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 8 of 41

25 March 2015, revision for AIChE Journal

are then used to calculate the total gas mass increment. At the end of the measurement of yw, all gas
mass increments are summed to provide the total gas mass.

For the measurements presented in this paper, the gravimetric hygrometer was used as
described in Ref. 16 with a few exceptions. First, instead of using Mg(ClO4)2 in the first two water
collection tubes and phosphorous pentoxide (P2O5) in the third, Mg(ClO4)2 was used in all three tubes.
This change was made after determining that P2O5 did not capture any measureable amount of water
that was not already captured by the Mg(ClO4)2. Secondly, the two sets of O-rings used to seal the water
collection tubes were made of Teflon-encapsulated elastomers rather than ordinary elastomers. This
change prevented absorption/desorption of moisture and gas in the elastomers, and made a significant
improvement on the repeatability of the measurements. Finally, for measurement of water amount
fractions corresponding to dew points higher than ambient temperature at ambient pressure, we
heated the gravimetric hygrometer manifold section leading to the first water collection tube. Heating
tape was wrapped around this section and it was heated to 95 °C to prevent any moisture condensation
inside.

When the gravimetric hygrometer manifold was partially heated, special steps were taken to
ensure that the pressure in the water collection tubes was at ambient both before and after the moist
gas was passed through. After the manifold was heated but before the gas passage, the tube caps were
opened and the pressure in the manifold was increased to about 120 kPa for a few seconds using dry
gas; the manifold was then briefly vented to the room to reduce the pressure to ambient. After the gas
passage and with the tube caps still open, the manifold was cooled to ambient temperature. The
pressure in the manifold was then increased to about 120 kPa for a few seconds using dry gas;
afterwards the manifold was briefly vented to the room to reduce the pressure to ambient.

Measurement uncertainty
The total uncertainty of a quantity v is related to the n individual uncertainty components wi
through the general law of error propagation:19

 ∂v  n −1 n
∂v ∂v
u(w i ) u(w j ).
n
u(v)2 = ∑   u(w i )2 + 2 ∑ ∑ ri , j (7)
i =1  ∂w i  i =1 j = i +1 ∂w i ∂w j

The relevant quantities and the derivatives ∂v/∂wi may be found by expanding the differential dv:
n
∂v
dv = ∑ dwi . (8)
i =1 ∂wi

8
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 9 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

The total uncertainty for the water-vapor enhancement factor f(TDP, p) is obtained by applying
Eq. 8 to Eq. 5, which yields
yw p py
df = dp + dy w − 2w de . (9)
e e e
In de, we can separate out the differential relating to the uncertainty of its calculating equation decalc
from that relating to the uncertainty of the temperature from which it is calculated:
de
de = de calc + dTDP . (10)
dTDP

Combining Eqs. 9−10 and dividing by f gives

df dp dy w decalc 1 de
= + − − dT . (11)
f p yw e e dT

The relative uncertainty ur(f) in the water vapor enhancement factor may then be expressed as

u( f )2 u ( p ) u ( yw )
2
u (e
2
)
calc 2
 1 de 
2

 u (T ) .
2
ur ( f ) ≡
2
2
= 2
+ 2
+ + (12)
f p yw e2  e dT 
The second term on the right side of Eq. 11 can be related to the uncertainties in mw and mg using Eq. 6
to yield:

dy w  mg M w  dm  mg M w  dmg
=  w −  . (13)
y w  ( mw M g + mg M w )  mw  mw M g + mg M w  mg
 
Substituting Eq. 6 into Eq. 13,
dy w dm dm
= (1 − y w ) w − (1 − y w ) g . (14)
yw mw mg

The second term on the right side of Eq. 12 may then be expressed as
2
u(y w )2 2 u(m w )
2
2 u(m g )
= (1 − y w ) + (1 − y w ) . (15)
yw2 mw 2 mg 2

The relative uncertainty in the water vapor enhancement factor may then be expressed as

u (e )
2
u ( p)
2 2 2 calc 2
2 u ( mw ) 2 u ( mg )  1 de 
( ) ( )  u (T ) .
2
ur ( f ) =
2
2
+ 1 − y w 2
+ 1 − y w 2
+ + (16)
p mw mg e2  e dT 

The relative uncertainty for the water mass, determined by comparison of the water collection
tubes against a standard mass using an electronic balance with a sensitivity mass (see “Gravimetric
hygrometer” section), is given in Eq. 11 of Ref. 16:

9
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 10 of 41

25 March 2015, revision for AIChE Journal

10 ρ a 2 [Vt − Vs ]  u ( pa ) 
2
u ( mw ) u ( Ta )  1 dρ a 
2 2 2 2

( )
2
= ⋅  + +   u RH 
mw 2 mw 2  pa
2
Ta 2  ρ a d ( RH )  
 . (17)
17u ( m ′ ) u ( msens ) u ( mesc )
2 2 2

+ 2
+ 2
+
mw mw mw 2

Here, Vt is the volume of the collection tube, Vs is the volume of the mass standard, pa is the ambient
pressure during the mass measurement, Ta is the ambient temperature during the mass measurement,
ρa is the density of the ambient air, RH is the relative humidity of the ambient air, m' is the measured
mass of a collection tube, and msens is the mass of the sensitivity mass. Finally, mesc is the mass of the
water that may have escaped the water collection tubes (assumed to be 0 g).

The uncertainty of the gas mass is given in Eq. 13 of Ref. 16:

u ( mg ) u ( pg ) u ( Tg )
2 2 2
u ( A) u ( ∆z )
2 2

= + + + . (18)
mg2 A2
∆z 2
pg2
Tg2

Here, A is the cross-sectional area of the prover tubes, pg and Tg are the pressure and temperature of
the gas in the prover tubes, respectively. Also, ∆z is the piston displacement at which the piston prover
stops rising and the gas volume is measured, and its uncertainty is given by

u(∆z )2 = u(z1 )2 + u(z2 )2 . (19)

Here, z1 is the piston position before gas enters the prover tube and z2 is the piston position after the
gas stops entering the tube.

The uncertainty budget for the elements listed above is shown in Table 1. The absolute
uncertainty values are shown in the third column of the table, and the contribution of the element to
ur(f)×100 is shown in the fourth column. Since the highest value of yw measured is ≈0.1, most
contributions are nearly constant. However, the contribution from T varies from 2.0×10−2 to 3.5 ×10−2.
The contribution for mesc is negligible for most values of yw measured for this paper. The largest
contribution (for yw = 5.8×10−4) was 1.3×10−2 and the smallest (for yw = 9.7×10−2) was 6.9×10−5. The
relative combined expanded uncertainty with a coverage factor of k=2, Ur(f) = 2ur(f), ranged from
Ur(f)×100=0.10 to Ur(f)×100=0.12. For simplicity, we estimate Ur(f) to be Ur(f)×100=0.12 for
measurements made with this facility. Additional contributions to the uncertainty, specific to the
experiments conducted in this work, will be discussed in a subsequent section.

10
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 11 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

Validation of the facility

Since the saturation system of the dew-point measurement facility had never been used before,
we first performed measurements of f(T, p) in CO2-free air to validate the system. These values of f are
well-known and can be calculated using Eq. 18 of Ref. 5. Therefore, we considered the agreement of
our measurements of f with these values to be an appropriate validation of the saturation system.

The measurements were performed with the saturator temperature controlled at 21.67 °C. The
gas used was CO2-free air. The gas supplier specified the air to have CO < 1 μL/L, CO2 < 0.1 μL/L, H2O <
0.5 μL/L, NOx < 50 nL/L, O2 20–21%, SOx < 50 nL/L, and total hydrocarbons < 0.05 μL/L. We did not
perform a separate analysis of the gas composition. A high-purity regulator was used with the air
cylinder.

The pressures used were over the range 0.25 MPa to 6.5 MPa. The results of the comparison
are shown in Figure 6. Here, the relative difference, (Δf)r, given by
(Δf)r = [f(measured) − f(calculated)] / f(calculated) , (20)

is shown as a function of pressure. The agreement of all points is within 0.2, and the standard deviation
of all the Δf/f values is 0.10 %, which is consistent with the uncertainty of the facility.

The measurements shown in Figure 6 were performed with flow rates ranging from 2 to 6
standard liters per minute (SLM). For example, the three measurements shown at 1 MPa were
performed with flow rates of 2.0 SLM, 3.4 SLM, and 3.8 SLM, and these measurements showed no
systematic variation of f with flow rate. The good agreement of all measurements with the calculated
values demonstrates that the system saturates fully for flow rates over this range.

Results
We performed measurements of f(T, p) using CO2 on six isotherms: 10 °C, 21.7 °C, 30 °C, 40 °C,
60 °C and 80 °C, and 0.5 MPa ≤ p ≤ 5.0 MPa. For most isotherms, the data were taken at ten pressure
values from 0.5 MPa to 5.0 MPa in increments of 0.5 MPa. However, for the 10 °C isotherm, data were
not taken above 4 MPa because CO2-H2O hydrates are known to form in this range.20

For most measurements, the manifold connecting the saturation system to the first desiccant
tube of the gravimetric hygrometer was not heated, because the generated dew-point temperature at
ambient pressure was below the ambient temperature. However, for pressures of 0.5 MPa to 1.5 MPa
on the 60 °C isotherm and pressures of 0.5 MPa to 4.0 MPa on the 80 °C isotherm, the dew-point
temperature was above ambient temperature. For this case, we heated the tube connecting the final

11
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 12 of 41

25 March 2015, revision for AIChE Journal

saturator to the inlet of the gravimetric hygrometer to 120 °C and we heated the gravimetric
hygrometer tube between the inlet and the first desiccant tube to 95 °C to prevent water condensation
in these areas.

Shown in Figure 7 is a comparison of the data for CO2 and air. Here, f(T, p) is shown as a
function of p on the isotherm t=21.7 °C. The air data are those from our validation experiments shown
in Figure 6. The values of f for CO2 are considerably higher than those for air, with values of (f−1)
ranging from 3 to 6 times greater. These higher values show that the attractive interaction between
water molecules and CO2 molecules is considerably stronger than the average interaction between
water molecules and the ensemble of molecules present in air.

Values for all the data are listed in Table 2, and these values are plotted in Figure 8. The data
are presented for six isotherms: 10 °C, 21.7 °C, 30 °C, 40 °C, 60 °C and 80 °C. As shown in the figure, the
value of f increases nearly proportionally with pressure but with some upward curvature. The curvature
is very small at the higher temperatures but becomes more apparent as the temperature is decreased.
For a given pressure, the value of f decreases as the temperature increases.

Thermodynamic Analysis
Description of phase equilibrium
At vapor-liquid equilibrium in the mixture of CO2 (component 1) and H2O (component 2), the
fugacity of each component in the liquid phase must equal that in the vapor phase. For water, we can
write this rigorously as:21

γ 2 x2 f 2pure (T , p) = φ2 y2 p , (21)

where x and y denote mole fractions in the liquid and vapor phases, respectively, γ2 is an activity

coefficient representing the deviation from ideal-solution behavior, f 2pure (T , p ) is the fugacity of pure

water at the temperature and pressure of the equilibrium, and φ2 is the fugacity coefficient that
describes the deviation from ideal-gas behavior in the vapor phase. Since the liquid phase at the
moderate pressures of interest here will be nearly pure water, it is reasonable to assume γ2 = 1. Also,
for convenience we write x2 as (1−x1) for the binary mixture, resulting in our main working equation:

(1 − x1 ) f 2pure (T , p ) = φ2 y2 p . (22)

12
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 13 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

It is common to compute f 2pure (T , p ) by a multistep process in which the saturation pressure of

pure component 2 at temperature T is corrected by a vapor-phase φ2 at saturation of the pure

component and a Poynting correction for the effect of pressure on the fugacity. In this case, however,

we can compute f 2pure (T , p ) directly from the international standard equation of state for water as

implemented in a NIST Standard Reference Database.15,22

To a first approximation, the liquid-phase mole fraction of CO2, x1, can be computed from
Henry’s law, where the Henry’s constant kH is defined as the limiting ratio of the fugacity to the liquid-
phase mole fraction:
f1
k H = lim , (23)
x1 → 0 x1

where f1 = φ1y1p is the fugacity of component 1. For sparingly soluble gases in water at pressures of a
few MPa, it is adequate to remove the limit in Eq. 23 and calculate x1 directly from the fugacity.
However, CO2 is more soluble in water than most gases, so we add a correction term (known as the
Krichevsky-Kasarnovsky equation) for the effect of gas pressure exceeding the vapor pressure of water:21

ln
f1 v
= ln k H + 1
∞
( p − e(T ) ) , (24)
x1 RT

where v1∞ is the partial molar volume of the solute (CO2) at infinite dilution in the solvent (H2O). For

v1∞ , we use the value 34.2 cm3/mol that Harvey et al.23 adopted based on their analysis of the limited
data available near ambient conditions. We use values of kH as a function of temperature from the
critical evaluation of Carroll et al.24 At the highest pressures considered in this work, the Krichevsky-
Kasarnovsky correction reduces the solubility x1 from the Henry’s-law value by relative amounts on the
order of 5 %. Because x1 is small compared to x2, this correction corresponds to a tiny but not entirely
negligible increase in x2 and therefore in the equilibrium fugacity of water.

The most important term in Eq. 22 for our thermodynamic analysis is the fugacity coefficient φ2.
At the moderate pressures of our experiments, it is sufficient to describe the vapor phase with the virial
expansion truncated after the third virial coefficient, as described in the Introduction. The fugacity
coefficient for component 2 is then given by25

13
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 14 of 41

25 March 2015, revision for AIChE Journal

 p 
ln φ2 = [2( B12 y1 + B22 y2 ) − Bmix ]  +
 RT 
2
, (25)
3  3 2   p 
 2 (C112 y1 + 2C122 y1 y2 + C222 y2 ) − 2 Bmix ( B12 y1 + B22 y2 ) −  Cmix − 2 Bmix    RT 
2 2

    

where Bmix and Cmix are the mixture second and third virial coefficients given by Eqs. 3 and 4,
respectively. With appropriate permutation of subscripts, Eq. 25 is also used to compute φ1 for the
fugacity of CO2 in the solubility calculation of Eq. 24.

Virial coefficients
Equation 25 contains seven virial coefficients: the pure-component values of B and C for each
component and the cross coefficients B12, C112, and C122. If these virial coefficients are known, Eqs. 22-25
may be used to calculate values of f(T,p). For pure CO2, accurate values of B and C may be obtained
from the reference-quality equation of state of Span and Wagner.14 For pure water, values of B from
the correlation of Harvey and Lemmon13 and values of C from the reference equation of state of Wagner
and Pruß15 are used. The values of B12, C112, and C122, are not well known. However, using the data from
Table 2 along with Eq. 22, it should be possible to determine these values.

When using Eq. 22 to determine virial coefficients from the values of T, p, and f over the range
0 MPa to 5 MPa, we originally hoped we could assume C112 = C122 = 0 and use the data to determine
values of B12. However, when analyzing the data it became clear that this assumption was not valid. We
demonstrate this in Figure 9, which shows the relative deviation of the f(T, p) data to the calculated
values, (∆f)r, where the calculations use values of B12 predicted by Wheatley and Harvey6 and assume
C112 = C122 = 0. Although the assumption appears to be valid on isotherms at 30 °C and 40 °C, the
quadratic deviations at other isotherms clearly show that higher-order virial coefficients are needed for
accurately calculating f(T,p).

In principle, B12, C112, and C122 could all be simultaneously fitted to our experimental data.
However, they differ widely in their influence on the phase equilibrium. Because our pressures are not
too high, B12 is more important than either of the higher-order C coefficients. Also, because y2 is
generally quite small, the contribution of C122 is small compared to C112, to the point where the
sensitivity of φ2 to C122 is almost negligible. We therefore chose to adjust only B12 and C112 to our data
and use estimated values of C122.

We estimated values of C122 at our experimental temperatures by the procedure of Hyland and
Mason,26 which takes advantage of the strong water-water interaction to consider the vapor phase as a

14
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 15 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

mixture of (nonpolar) gas molecules, water monomers, and water dimers. The dimerization of water is
related to its second virial coefficient B22, and a Lennard-Jones potential is used to describe the
interaction of the nonpolar gas with the water dimer. Hyland and Mason applied their method to the
water-air interaction; we adapted it to our system by using effective Lennard-Jones parameters for CO2
from Tee et al.27 and by replacing their older values of B22 with those from a current correlation.13
Hyland and Mason estimate that their technique is only accurate to within a factor of two, but this is
adequate for our purposes given the aforementioned insensitivity of the equilibrium to C122. Our
estimates for C122 are given in Table 3.

With these estimated C122, we first derived initial estimates of B12 and C112 (and their
uncertainties, as described below) based on our experimental values of f on each isotherm, using a
simplified version of Eq. 22 that neglected the correction given by Eq. (24) and approximated

f 2pure (T , p ) by applying a simple Poynting correction to the vapor pressure. We first varied C112 until
the deviation (∆f)r was linearly proportional to p. The optimized value of C112 was that for which the
standard deviation of (∆f)r from a best-fit line, constrained to have a value of (∆f)r = 0 at p = 0, was
minimal. For this determination of C112, we nominally used B12 from Wheatley and Harvey,6 along with
our estimated value of C122 and the values of B11, B22, C111, and C222 from the references mentioned at the
beginning of this section. When performing this optimization, the values of (∆f)r above p = 3.5 MPa on
the 10 °C and 21.7 °C isotherms were not used, as a good fit was not possible; for the 10 °C isotherm,
this may be due to the proximity to the region of CO2-H2O hydrate formation; we do not have a clear
explanation for the deviation of the high-pressure points on the 21.7 °C isotherm.

Once the initial estimate of C112 was determined, we estimated B12 by minimizing the standard
deviation of the ensemble of (∆f)r values. When performing the optimization of B12, the values (∆f)r
values above p = 3.5 MPa on the 10 °C and 21.7 °C isotherms again were not used. For the
determination of B12, we used the estimate of C112 described in the previous paragraph and the
estimated value of C122 along with the values of B11, B22, C111, and C222 from the references mentioned
above.

With these estimates of B12 and C112 as an initial guess, we then performed a final two-
dimensional optimization of their values, using the full model of Eq. (22) described earlier in this section.
The optimized values of B12 are given in Table 4, and the optimized values of C112 are given in Table 3.
The values of (∆f)r determined using the optimized values of B12 and C112 and our estimated values of
C122 are plotted in Figure 10. All values are within the bounds of the plots except for three of the points

15
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 16 of 41

25 March 2015, revision for AIChE Journal

not used in the fits: p = 4.0 MPa on the 10 °C isotherm ((∆f)r = 1.97%) and p = 4.5 MPa and 5.0 MPa on
the 21.7 °C isotherm ((∆f)r = 2.31% and 3.10%, respectively).

Uncertainty Analysis for CO2-H2O Measurements

The standard deviation of the points shown in Figure 10 should in principle be within the
estimated relative standard uncertainty for the facility described here. This relative uncertainty,
calculated above, is ur(f) × 100 = 0.06. However, the standard deviations of the (∆f)r values plotted in
Figure 10, shown in Table 5, are in general larger than this value. We describe here other elements that
may contribute to the uncertainty of our measurements with CO2-H2O mixtures and provide a final
uncertainty estimate.

One uncertainty element particular to CO2-H2O mixtures is the effect from the process of
carbonation. The unusually high solubility of CO2 in water results in a process of CO2 absorption in the
water once pressurization begins. Before each measurement was taken, the saturator was at ambient
pressure. Once the saturator was pressurized, our measurements typically began within 20 minutes.
Therefore CO2 absorption in the water may well have been occurring during our measurements, with
the possible effect of the formation of tiny water droplets that may have been carried away by the
flowing CO2, resulting in an excessively large measured value of f.

Another uncertainty element not discussed above is the exchange of air with CO2 in the water
collection tubes of the gravimetric hygrometer when they are briefly cracked open to bring them to
ambient pressure. This procedure was done to ensure proper determination of the gas mass inside the
tubes. In general, the tube pressure before the cracking was always higher than ambient, so CO2 leaked
out during the cracking. However, if the cracking was not performed carefully, there was always a
possibility of a small amount of air leaking in, resulting in an excessively small measured value of f.

Because of these uncertainty elements, whose values are impossible to estimate, we have opted
to increase our uncertainty estimates for f by adding them in quadrature with the standard deviation
values shown in Table 5. Since these values differ for different isotherms, we used individual
uncertainties for each isotherm. These uncertainties are given in Table 5 as well. We note that the
relative uncertainty in the saturated gas compositions yw is approximately the same as that for f.

The uncertainty for the virial coefficients may be estimated by examining the sensitivity of the
(∆f)r values to changes in B12 and C112. The sensitivity of (∆f)r to these virial coefficient values is shown in

16
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 17 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

Figure 11. Here, values of (∆f)r are shown for B12 = −113 cm3/mol and C112 = 5000 cm6/mol2 (the
optimized values). Values are also shown for B12 = −109.7 cm3/mol and C112 = 2000 cm6/mol2. These
latter values were obtained by first adjusting the value of C112 by 3000 cm6/mol2 and afterwards
adjusting B12 so that the (∆f)r values at 5 MPa would agree. We see that the maximum variation in the
two sets of (∆f)r × 100 values is approximately 0.3. This is approximately the value that we have
estimated for the relative combined expanded uncertainty of our determinations of f at this isotherm.
The difference between the two values of B12 is 3.3 cm3/mol and that for the two values of C112 is 3000
cm6/mol2. We therefore estimate the expanded uncertainty of B12 to be U(B12) = 3 cm3/mol and that for
C112 to be U(C112) = 3000 cm6/mol2 on this isotherm. We performed a similar procedure for estimating
U(B12) and U(C112) on the other five isotherms, using the respective values of U(∆f)r for determining the
amounts to perturb C112 and B12. Our values for U(C112) and U(B12) are listed in Table 3 and Table 4,
respectively. It should be noted that C112 and B12 are correlated (and therefore their uncertainties are
not independent); a higher C112 implies a lower B12 and vice versa.

In Figure 12, we compare the values of B12 determined here to those predicted by Wheatley and
Harvey6 and those from other work.7-10,28,29 Our results are slightly smaller in magnitude than the
predictions of Ref. 6, with the ratio of our values to those of Ref. 6 varying from 0.96 to 0.99.
Nevertheless, our values are well within the uncertainties of those of Ref. 6. We also note that the B12
values and error bars for other experiments shown in Figure 12 were mostly estimated by Wheatley and
Harvey6 with a method that neglected third virial coefficients; as we have seen in this work (see Figure
9), this assumption would introduce additional error for data sets with data extending above
approximately 2 MPa.

Discussion
While not all of the previous experimental studies of the dew point of H2O in compressed CO2
reported uncertainties,7-12 those that did had uncertainties of several percent in the water content of
the vapor phase. There is no reason to expect that the uncertainties of the other studies were any
better. In addition, the existing data are not mutually consistent, as is evident in values of B12 derived
from the data and shown in Figure 12. In contrast, our experimental uncertainties, as discussed above,
are on average 0.3% (relative combined expanded uncertainty with coverage factor k=2, approximately
corresponding to a 95% confidence interval). The use of a high-precision gravimetric hygrometer,

17
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 18 of 41

25 March 2015, revision for AIChE Journal

originally designed for humidity standards, has allowed us to reduce the uncertainty in knowledge of
this important quantity by nearly an order of magnitude.

Another result of this work is the validation of the theoretical potential-energy surface and
B12(T) of Wheatley and Harvey,6 as shown in Figure 12. That work had relatively large uncertainties in its
derived values of B12 at the temperatures studied here; our results (with much smaller uncertainties) lie
almost in the center of their uncertainty band. This implies that Wheatley and Harvey might have been
overly pessimistic in their uncertainty estimates, or perhaps that they were fortuitous in their
theoretical approximations. In any event, the validation of their theoretical results over the
temperature range studied in this work means that their B12 should also be valid at higher and lower
temperatures, enabling those B12 to be used as a boundary condition for modeling mixtures (especially
gas-phase mixtures) containing H2O and CO2 over a wide range of temperatures. There does appear to
be a small systematic offset between our results and those of Wheatley and Harvey,6 so it might be
reasonable to reduce the magnitude of their B12 values by 2% for use in such modeling.

In contrast to B12, the values of C112 derived in this work were not determined with high
accuracy. We found that small perturbations in the input data, within the scatter of our measurements,
could affect our C112 in the second, and sometimes even the first, significant digit. In addition, the
optimized values of C112 in Table 3 do not vary smoothly with temperature as should be the case. Our
results also do not quantitatively line up with an extrapolation of the results of Patel et al.,29 who
reported some C112 at higher temperatures. We do not recommend that our C112 from Table 3 be used
in modeling; they should be viewed as qualitative results that may also be accounting for other
experimental effects (such as nonideality in the liquid phase at large CO2 solubilities) that appear at
higher pressures. It might be possible to use modern high-accuracy potential-energy surfaces6,30 and
integration techniques31 to derive theoretical values of C112 that are more reliable than the values we
give here, although this would require a physically reasonable model for three-body forces. Recent
work by Schultz et al.32 attempts to perform such calculations for C112 and C122.

While this work has greatly improved knowledge of the dew points of these mixtures up to 5
MPa, that upper limit is still below pressures that might be encountered in sequestration operations.
Such operations might be as high as 20 MPa in the pipeline and 50 MPa in compression for geologic
sequestration.33 These results will still be useful in modeling the high-pressure systems; a physically
based model (such as a modern equation of state) that is constrained by good data at moderate
pressures will likely be fairly reliable at higher pressures. However, it would be desirable to have some

18
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 19 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

high-pressure data of similar accuracy. Our gravimetric hygrometer could be used for such experiments,
but it would require a new high-pressure saturation system, perhaps similar to the high-pressure air
saturator designed by Hyland and Wexler.18

Acknowledgments
This paper is a contribution of the National Institute of Standards and Technology and is not
subject to copyright in the United States. This work was supported by the U.S. Department of Energy,
National Energy Technology Laboratory, under DOE Interagency Agreement DE-FE0003931. A.H. thanks
John Prausnitz for teaching him that even those of us who specialize in theory and modeling should
appreciate the value of good experimental work. C.M. thanks John Prausnitz for producing excellent
chemical engineers to help understand the results of good experimental work.

19
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 20 of 41

25 March 2015, revision for AIChE Journal

Literature Cited
1. Haszeldine RS. Carbon Capture and Storage: How green can black be? Science. 2009;325:1647-
1652.

2. Dennis RA, Harp R. Overview of the U.S. Department of Energy’s Office of Fossil Energy Advanced
Turbine Program for Coal Based Power Systems with Carbon Capture. Proceedings of ASME Turbo
Expo 2007, ASME Press, New York, 2007;2:1093-1104.

3. Figueroa JD, Fout T, Plasynski S, McIlvried H, Srivastava, RD. Advances in CO2 capture technology –
The U.S. Department of Energy’s Carbon Sequestration Program. Int J Greenhouse Gas Control.
2008;2:9-20.

4. Wagner W, Pruss A. International equations for the saturation properties of ordinary water
substance. Revised According to the International Temperature Scale of 1990. J Phys Chem Ref
Data. 1993;22:783−787.

5. Hyland RW, Wexler A. Formulations for the thermodynamic properties of dry air from 173.15 K,
and of saturated moist air from 173.15 K to 372.15 K, at pressures to 5 MPa. ASHRAE Trans.
1983;89-IIa:520−535.

6. Wheatley RJ, Harvey AH. Intermolecular potential energy surface and second virial coefficients for
the water–CO2 dimer. J Chem Phys. 2011;134:134309.

7. Coan CR, King Jr. AD. Solubility of water in compressed carbon dioxide, nitrous oxide, and ethane.
Evidence for hydration of carbon dioxide and nitrous oxide in the gas phase. J Am Chem Soc.
1971;93:1857-1862.

8. Gillespie PC, Wilson GM. Vapor-Liquid and Liquid-Liquid Equilibria: Water-Methane, Water-
Carbon Dioxide, Water-Hydrogen Sulfide, Water-nPentane, Water-Methane-nPentane. GPA
Technical Report RR-48. Tulsa, OK: Gas Processors Association, 1982.

9. Bamberger A, Sieder G, Maurer G. High-pressure (vapor + liquid) equilibrium in binary mixtures of

(carbon dioxide + water or acetic acid) at temperatures from 313 to 353 K. J Supercrit Fluids.
2000;17:97-110.

10. Valtz A, Chapoy A, Coquelet C, Paricaud P, Richon D. Vapour-liquid equilibria in the carbon dioxide-
water system, measurement and modelling from 278.2 to 318.2 K. Fluid Phase Equilib.
2004;226:333-344.

20
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 21 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

11. Koglbauer G, Wendland M. Water Vapor Concentration Enhancement in Compressed Humid Air
Measured by Fourier Transform Infrared Spectroscopy. J Chem Eng Data. 2008;53:77-82.

12. Hou S-X, Maitland GC, Trusler JPM. Measurement and modeling of the phase behavior of the
(carbon dioxide + water) mixture at temperatures from 298.15 K to 448.15 K. J Supercrit Fluids.
2013;73:87-96.

13. Harvey AH, Lemmon EW. Correlation for the second virial coefficient of water. J Phys Chem Ref
Data. 2004;33:369-376.

14. Span R, Wagner W. A new equation of state for carbon dioxide covering the fluid region from the
triple point temperature to 1100 K at pressures up to 800 MPa. J Phys Chem Ref Data.
1996;25:1509-1596.

15. Wagner W, Pruß A. The IAPWS formulation 1995 for the thermodynamic properties of ordinary
water substance for general and scientific use. J Phys Chem Ref Data. 2002;31:387-534.

16. Meyer CW, Hodges JT, Hyland RW, Scace GE, Valencia-Rodriguez J, Whetstone JR. The second-
generation NIST standard hygrometer. Metrologia. 2010;47:192-207.

17. Meyer CW, Hodges JT, Huang PH, Miller WW, Ripple DC, Scace GE. Calibration of Hygrometers
with the Hybrid Humidity Generator. NIST Special Publication 250-83, 2008.

18. Hyland RW, Wexler A. The enhancement of water vapor in carbon dioxide-free air at 30, 40, and
50 °C. J Res Nat Bur Stand. 1973;77A:115-131.

19. ISO, Guide to the Expression of Uncertainty in Measurement, Geneva: International Organization
for Standardization, 1993.

20. Yang SO, Yang IM, Kim YS, Lee CS. Measurement and prediction of phase equilibria for water + CO2
in hydrate forming conditions. Fluid Phase Equilib. 2000;175:75-89.

21. Prausnitz JM, Lichtenthaler RN, Gomes de Azevedo E. Molecular Thermodynamics of Fluid-Phase
Equilibria (3rd edition). Upper Saddle River, NJ: Prentice Hall, 1999.

22. Lemmon EW, Huber ML, McLinden MO. REFPROP: Reference Fluid Thermodynamic and Transport
Properties, NIST Standard Reference Database 23 (Version 9.1). Gaithersburg, MD: National
Institute of Standards and Technology, 2013.

21
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 22 of 41

25 March 2015, revision for AIChE Journal

23. Harvey AH, Kaplan SG, Burnett JH. Effect of Dissolved Air on the Density and Refractive Index of
Water. Int J Thermophys. 2005;26:1495-1514.

24. Carroll JJ, Slupsky JD, Mather AE. The Solubility of Carbon Dioxide in Water at Low Pressure. J Phys
Chem Ref Data. 1991;20:1201-1209.

25. Yang S, Schultz AJ, Kofke DA, Harvey AH. Interpreting Gas-Saturation Vapor-Pressure
Measurements Using Virial Coefficients Derived from Molecular Models. J Chem Eng Data.
2014;59:3183-3192.

26. Hyland RW, Mason EA. Third Virial Coefficient for Air-Water Vapor Mixtures. J Res Nat Bur Stand
A. 1967;71A:219-224.

27. Tee LS, Gotoh S, Stewart WE. Molecular Parameters for Normal Fluids: The Lennard-Jones 12-6
Potential. Ind Eng Chem Fundam. 1966;5:356-363.

28. Vanderzee CE, Haas, NC. Second cross virial coefficients B12 for the gas mixture (carbon dioxide +
water) from 300 to 1000 K. J Chem Thermodyn. 1981;13:203-211.

29. Patel MR, Holste JC, Hall KR, Eubank PT. Thermophysical Properties of Gaseous Carbon Dioxide –
Water Mixtures. Fluid Phase Equilib. 1987;36:279-299.

30. Hellmann R. Ab initio potential energy surface for the carbon dioxide molecule pair and
thermophysical properties of dilute carbon dioxide gas. Chem Phys Lett. 2014;613:133-138.

31. Singh JK, Kofke DA. Mayer Sampling: Calculation of Cluster Integrals using Free-Energy
Perturbation Methods. Phys Rev Lett. 2004;92:220601.

32. Schultz AJ, Kofke DA, Harvey AH. Molecular-based Virial Coefficients of CO2-H2O Mixtures. AIChE J.,
submitted.

33. Li H, Jakobsen JP, Wilhelmsen Ø, Yan J. PVTxy properties of CO2 mixtures relevant for CO2 capture,
transport and storage: Review of available experimental data and theoretical models. Appl.
Energy. 2011;88:3567-3579.

22
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 23 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

Table 1. Uncertainty elements for the dew-point measurement facility.

Contribution to ur(f) ×
Description Symbol Uncertainty
100
Saturation System
Saturator pressure p 0.0001 p 1.0 × 10−2
Saturator temperature TDP 0.005 K ≤3.5 × 10−2
Water vapor-pressure calculation ecalc 4.4 × 10−5 ecalc 4.4 × 10−3

Water Mass Measurement

Mass measurements m' 7.0 × 10−5 g 1.5 × 10−2(1−yw)
Sensitivity mass msens 1.2 × 10−4 g 6.0 × 10−3(1−yw)
Air pressure pa 25 Pa 4.0 × 10−3(1−yw)
Air temperature Ta 0.2 K 1.0 × 10−2(1−yw)
Air relative humidity RH 1% 5.0 × 10−3(1−yw)
Water escaping collection tubes mesc 7.5 × 10−8 mg 7.5 × 10−6(1−yw)/yw

Gas Mass Measurement

Gas temperature Tg 0.1 K 3.4 × 10−2(1−yw)
Gas pressure pg 13 Pa 1.3 × 10−2(1−yw)
Prover tube piston displacement Δz 4.7 × 10−3 cm 7.8 × 10−3(1−yw)
Prover tube area A 9.0 × 10−3 cm2 5.5 × 10−3(1−yw)

Relative Combined Expanded Uncertainty (k=2): Ur(f) × 100 = 0.12, where Ur(f) = U(f)/f

23
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 24 of 41

25 March 2015, revision for AIChE Journal

Table 2. Measured values of the saturator (dew-point) temperature T, saturator pressure p, saturated
water amount fraction yw, and water-vapor enhancement factor f, for the six isotherms measured in this
work.

T [K] p [MPa] yw×103 f

283.15 0.5002 2.6064 1.0614
283.15 1.0004 1.3886 1.1310
283.15 1.5007 0.9925 1.2127
283.15 2.0028 0.8016 1.3073
283.15 2.4986 0.6944 1.4128
283.15 3.0182 0.6293 1.5466
283.15 3.5009 0.5916 1.6866
283.15 3.9937 0.5812 1.8898
294.83 0.5000 5.4788 1.0557
294.83 1.0003 2.8883 1.1134
294.83 1.5005 2.0413 1.1805
294.83 2.0008 1.6239 1.2523
294.83 2.5010 1.3849 1.3348
294.83 3.0013 1.2332 1.4274
294.84 3.5016 1.1338 1.5299
294.84 4.0112 1.0738 1.6599
294.84 4.5027 1.0445 1.8125
294.84 4.9854 1.0244 1.9681
303.14 0.5002 8.8966 1.0482
303.15 1.0003 4.6867 1.1038
303.14 1.5008 3.2833 1.1608
303.15 2.0008 2.6054 1.2273
303.14 2.5013 2.2033 1.2982
303.15 3.0013 1.9371 1.3689
303.14 3.5017 1.7714 1.4611
303.15 4.0019 1.6650 1.5689
303.14 4.5020 1.5719 1.6672
303.15 5.0055 1.5022 1.7705
313.17 0.5002 15.340 1.0381
313.17 1.0003 8.0707 1.0919
313.17 1.5007 5.6219 1.1410
313.17 2.0010 4.4139 1.1949
313.17 2.5012 3.7211 1.2589
313.17 3.0013 3.2649 1.3252
313.17 3.5016 2.9370 1.3914
313.17 4.0019 2.7320 1.4787
313.16 4.5022 2.5664 1.5635
313.17 5.0024 2.4569 1.6623
333.19 0.5000 41.504 1.0386
333.18 1.0002 21.461 1.0743

24
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 25 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

333.18 1.5007 14.798 1.1118

333.18 2.0009 11.514 1.1534
333.18 2.5010 9.5623 1.1974
333.18 3.0012 8.2871 1.2451
333.18 3.5015 7.3862 1.2948
333.18 4.0018 6.7515 1.3504
333.17 4.5019 6.2355 1.4059
333.18 5.0024 5.8609 1.4681
353.15 0.4999 97.482 1.0280
353.15 1.0001 50.406 1.0633
353.14 1.5005 34.562 1.0941
353.15 2.0007 26.675 1.1257
353.14 2.5010 21.942 1.1576
353.14 3.0012 18.871 1.1948
353.15 3.5017 16.681 1.2320
353.14 4.0016 15.051 1.2705
353.15 4.5020 13.793 1.3095
353.15 5.0023 12.819 1.3523

25
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 26 of 41

25 March 2015, revision for AIChE Journal

Table 3. Values of the third cross virial coefficients C122 and C112. The estimated values of C122 are
calculated using the procedure of Hyland and Mason.26 The values of C112 are optimized as described in
the text. Our estimate for the expanded uncertainty of the determined C112 values is given in the last
column.

T C122 (calculated) C112 (optimized) U(C112)

[K] [cm6/mol2] [cm6/mol2] [cm6/mol2]
283.15 −380000 −14700 3000
294.83 −290000 −1800 2000
303.15 −240000 −500 5000
313.17 −190000 −3700 6000
333.18 −130000 1800 1500
353.15 −92000 5000 3000

26
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 27 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

Table 4. Optimized values of the second cross virial coefficients B12. Our estimate for the expanded
uncertainty of the determined B12 values is given in the last column.

T B12 (optimized) U(B12)

[K] [cm3/mol] [cm3/mol]
283.15 −203 3
294.83 −186 2
303.15 −173 6
313.17 −155 5
333.18 −132 2
353.15 −113 3

27
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 28 of 41

25 March 2015, revision for AIChE Journal

Table 5. Standard deviation of the (∆f)r values and final relative combined expanded uncertainty Ur(f) of
the water-vapor enhancement factor for each isotherm.

T [K] σ[(Δf)r × 100] Ur(f) × 100

283.15 0.06 0.17
294.83 0.07 0.18
303.15 0.37 0.75
313.17 0.25 0.51
333.18 0.08 0.20
353.15 0.15 0.32

28
AIChE Journal
This article is protected by copyright. All rights reserved.
Page 29 of 41 AIChE Journal

25 March 2015, revision for AIChE Journal

Figure Captions
1. Saturation component for the dew-point measurement facility.

2. Schematic diagram of the saturators.

3. Photograph of the final saturator.

4. Schematic diagram of the water collection tubes used in the gravimetric hygrometer.

5. Schematic diagram of the gas collection system used in the gravimetric hygrometer.

6. Comparison of measurements of the water-vapor enhancement factor f(T, p) in CO2-free air with
known values calculated using Eq. 18 of Ref. 5. The measurements were all made at t = 21.7 °C.
Here, (∆f)r = [f(measured) − f(calculated)] / f(calculated).

7. Comparison of measurements of f(T, p) in CO2 with those in CO2-free air. The values of f are
plotted as a function of p on the isotherm t = 21.7 °C. The air data are those shown in Figure 6.

8. Measurements of f(T, p) in CO2. The values of f are plotted as a function of p on the isotherms
10 °C, 21.7 °C, 30 °C, 40 °C, 60 °C and 80 °C.

9. Relative deviations of f(T, p) data to the calculated values, (∆f)r, where the calculations use
values of B12 from Ref. 6 and assume C112 = C122 = 0.

10. Relative deviations of f(T, p) data to the calculated values, (∆f)r, where the calculations use the
optimized values of B12 and C112 given in Table 4 and Table 3, respectively, and the calculated
C122 values given in Table 3.

11. Sensitivity of (∆f)r values to B12 and C112. The plot shows relative deviations of the f(T, p) data to
the calculated values, (∆f)r, for the isotherm at t = 80 °C, where the calculations use values of B12
= −113 cm3/mol and C112 = 5000 cm6/mol2 (the optimized values). Determinations of (∆f)r are
also shown for B12 = −109.7 cm3/mol and C112 = 2000 cm6/mol2. The maximum difference
between the two sets of (∆f)r × 100 values is 0.3, which is the same as the total relative
combined expanded uncertainty Ur × 100 that we have estimated for our determinations of f.

12. Comparison of the values of B12 from this work with those from the predictions of Wheatley and
Harvey6 and from other experiments.7-10,28,29

29
AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 30 of 41

1. Saturation component for the dew-point measurement facility.

168x106mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
Page 31 of 41 AIChE Journal

2. Schematic diagram of the saturators.

191x185mm (150 x 150 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 32 of 41

3. Photograph of the final saturator.

223x108mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
Page 33 of 41 AIChE Journal

4. Schematic diagram of the water collection tubes used in the gravimetric hygrometer.
183x91mm (132 x 132 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 34 of 41

5. Schematic diagram of the gas collection system used in the gravimetric hygrometer.
191x153mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
Page 35 of 41 AIChE Journal

6. Comparison of measurements of the water-vapor enhancement factor f(T, p) in CO2-free air with known
values calculated using Eq. 18 of Ref. 5. The measurements were all made at t = 21.7 °C. Here, (∆f)r =
[f(measured) − f(calculated)]/ f(calculated).
240x174mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 36 of 41

7. Comparison of measurements of f(T, p) in CO2 with those in CO2-free air. The values of f are plotted as
a function of p on the isotherm t = 21.7 °C. The air data are those shown in Figure 6.
240x174mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
Page 37 of 41 AIChE Journal

8. Measurements of f(T, p) in CO2. The values of f are plotted as a function of p on the isotherms 10 °C,
21.7 °C, 30 °C, 40 °C, 60 °C and 80 °C.
240x174mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 38 of 41

9. Relative deviations of f(T, p) data to the calculated values, (∆f)r, where the calculations use values of B12
from Ref. 6 and assuming C112 = C122 = 0.
240x174mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
Page 39 of 41 AIChE Journal

10. Relative deviations of f(T, p) data to the calculated values, (∆f)r, where the calculations use the
optimized values of B12 and C112 given in Table 4 and Table 3, respectively, and the calculated C122 values
given in Table 3.
1034x1106mm (96 x 96 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
 AIChE Journal Page 40 of 41

11. Sensitivity of (∆f)r values to B12 and C112. The plot shows relative deviations of the f(T, p) data to the
calculated values, (∆f)r, for the isotherm at t = 80 °C, where the calculations use values of B12 = −113
cm3/mol and C112 = 5000 cm6/mol2 (the optimized values). Determinations of (∆f)r are also shown for
B12 = −109.7 cm3/mol and C112 = 2000 cm6/mol2. The maximum difference between the two sets of (∆f)r
× 100 values is 0.3, which is the same as the total relative combined expanded uncertainty Ur × 100 that
we have estimated for our determinations of f.
240x174mm (120 x 120 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.
Page 41 of 41 AIChE Journal

12. Comparison of the values of B12 from this work with those from the predictions of Wheatley and Harvey
and from other experiments.
159x138mm (300 x 300 DPI)

AIChE Journal
This article is protected by copyright. All rights reserved.