Demonstratioxx of Liquid C0a as an Alternative fox Metal Parte
Cleaning
By Charles H. Darvin*and Elizabeth A. Hill+
Introduc tioix
The use of carbon dioxide (C02) in its supercritical state
is an established technology for solvent extraction in processes
such as decaffeinating coffee, wastewater treatment and chemical
analysis.1"3 Several studies have been conducted with
supercritical C02 for surface cleaning.4,s However, the potential
for liquid C0a {LCQ25 as a surface cleaning agent has remained
largely unexplored. This may be due to supercritical eo2's
greater solvency for some contaminants compared to LCO^* The
difference in cleaning capability can be offset in a liquid
system with the addition of process enhancements such as
ultrasonics or raegasonics, techniques that have not been very
effective with C02 in the supercritical state.
In early 1995, a program was initiated by the Environmental
Protection Agency to investigate new and innovative surface
cleaning and degreasing technologies as alternatives to ozone-
depleting compounds including 1,1,1-trichloroethane (TCA),
methylene chloride, and Freon 113™ [1,1,2-trichloro-l,2,2-
trifluoroethane, chlorofluorocarbon (CFC)-113]. One promising
candidate identified during the investigation was LC02. The
United States Air Force at the Warner Robins Air Logistics Center
(WR-ALC), Robins AFB, GA, was requested to participate in the
demonstration and served as the demonstration site. The WR-ALC
uses several surface cleaning processes at the facility during
aircraft systems maintenance. These processes generate a broad
range of air, water and solid waste environmental contaminants.
The volume of these wastes could be reduced by the use of LC02
cleaning.
LC02is of interest as a potential solvent degreasing
substitute largely because of what it is not. First of all, it is
not an ozone-depleting compound (ODC) . Therefore, it does not
present a threat to the earth's ozone layer as found with typical
chlorof luorocarbon (CFC) solvents. It is nonflammable and has low
toxicity. Thus, it does not present a safety hazard when used
properly. Finally, LC02 is not expensive when compared to CFCs
and equivalent substitutes. This is important to its viability as
an industrial surface cleaning and degreasing solvent.
. The objective of the project was to demonstrate the
viability and efficiency of the LC02 cleaning as an alternative
to current cleaning and degreasing technologies. The study was
designed to show that LC02 cleaning, when properly integrated
into the manufacturing process, could remove various organic and
solid contaminants typically removed during vapor degreasing with
ozone-depleting solvents.
Liquid C02
Carbon dioxide is a gas at standard temperature and
pressure, 32°F (0°C) and 14.7 psi (1 bar). By increasing the
pressure and temperature of the C02, it can be converted from the
gaseous phase to the liquid and supercritical phases. As a liquid
* U.S. EPA, APrCD, MD-61, Research Triangle Park, NC 27711.
+ Research Triangle Institute, P. O. Box 12194. Research Triangle Park,
27709.

-------
or supercritical fluid, C02 has good solvent properties for oils,
greases, and other common machining contaminants. Changing the
operating pressure and temperatures within the defined state
boundaries will allow selective removal from a surface and
separation of a variety of metal finishing and fabrication
contaminants- Although solubility is not the sole property that
defines the acceptability of a potential solvent substitute, LCOs
compares favorably with other typical degreasing solvents. Two
other physical properties shown in Table l that affect cleaning
are surface tension and viscosity. LC02 has low surface tension
and very low viscosity, which improves the likelihood that the
solvent will wet the surface to be cleaned and penetrate into
small crevices and blind holes in the parts.
TABLE 1. Chemical and Physical Properties of Degreasing Solvents
Solvent
Solubility
Parameter
(MPa1/a) *
Viscosity at 20 *C
(Centipoise)*
Surface Tension
at 20 *C
(Dynes/cm)9
TCA
17.7
1.2*
25.5=
Methylene
chloride
20.3
0.5
26.5
Acetone
20.0
0.4
23.7
|i Xj x qu i d CO^
20-22'
0.07
5.0
(a)	Calculated from Giddings equation, page 224 o?Reference
(b)	From Reference 9.
(c)	Value at 25 °C, from supplier literature.
Liquid COj Cleaning and Degreasing Process
The LC02 cleaning process is primarily a degreasing process
consistent with vapor degreasing. Similar to vapor degreasing, it
has only limited capability to remove particulate matter from a
surface without additional mechanical enhancements such as
ultrasonics or sprays. The process will remove most light and
medium weight hydrocarbon oils, gross particulate contamination,
drawing compounds, and other machining fluids. A typical LO02
degreasing cycle takes approximately 20 to 25 minutes, including
loading and unloading parts. The best applications are those
where organic vapor degreasing solvents will work. Similar to
CFCs and aqueous solvent systems, LC03 will not remove rust,
paint, coatings or most adhesives. These'are typically removed by
other surface preparation techniques such as abrasive blasting or
surface stripping. An illustration of the LC02 equipment used
during this demonstration is shown in Figure 1. It consists of, a
sealed pressure vessel, associated vacuum pumps and compressors,
and a C02 capture and recycle tank. During the evaluation period,
a centrifuge and a hot oil process {HOP} tank with ultrasonic
generator were included and used to enhance the capability of the
cleaning process. HOP oil is a proprietary formulation of a
lightweight, low volatility hydrocarbon oil and surfactants. If
heavily contaminated with soils, the item to be cleaned is first
-2-

-------
submerged in the HOP oil tank. Ultrasonic agitation can be
incorporated in the HOP tank to further break the soils from the
part's surface. The HOP oil is subsequently removed during the
LCOa degreasing step.
Farts Cleaner	C0> Recyclcr
Figur® 1. illustration of LCO, Equipment.
Laboratory Feasibility Tests
Prior to conducting the site demonstration, preliminary
laboratory evaluations were conducted by cleaning parts similar
to those anticipated during the demonstration. These tests were
performed in a prototype of the equipment planned for use at
RAFB. The prototype was functional but not as fast or.effective
as the next generation of, LCOa cleaning equipment to be produced.
During these tests, contaminated aluminum fuel line tubing of
varying lengths, diameters and shapes were cleaned. The parts
were cleaned with the HOP oil first, followed by LC02. The HOP
process was used because the drawing compound contained a
combination of soaps and oils. LC02 by itself removed the oils,
but not all of the soaps. The HOF/LCOa process removed the
drawing compound as effectively as the TCA process from some of
the tubes but not all of them. Tubes with smaller diameters or
more bends were not cleaned as well as larger diameter, straight
tubes. A summary of the results for these tests are shown in
Table 2.
Because some tubes were cleaned as well as with TCA even
using the prototype equipment, it was decided that the LC02
results were good enough to warrant further testing and
demonstration at RAFB once the improved LC02 equipment was
available.
-3-

-------
TABLE 2. SUMMARY OF LABORATORY EVALUATIONS OF NVR FOR TUBES
Soiled/Cleaned
Average NVR
(g/fta)
Range of HVR
Values (g/ft2)
Standard
Deviation
(g/fta>
Contaminated tubes
2.311
1.612 - 2.808
0.623
TCA-cleaned
0.007
0.005 - 0.008
0.002*
HOP/LCO,-cleaned
0.035
0.006 - 0.181
0.050
Sit© Demonstration and Evaluation.
The demonstration and evaluation was conducted at WR-ALC, GA
during August 1995 to show the feasibility and capability of the
LC02 process. Actual aircraft parts, including scrap parts and
non-flight tools were cleaned during the demonstration*. These .
parts were not placed back in the maintenance inventory after the
cleaning tests because this cleaning process is not currently
included in the parts specifications. A great variety of parts
were selected for cleaning, including bearings, fuel system
tubing, filters, bolts, and other mechanical and structural
aircraft parts. The demonstration incorporated HOP oil and
ultrasonics to enhance and maximize the efficiency of the
process.
The evaluation procedure to define the capability of the
system was a combination of laboratory analytical testing and
visual inspections ¦ conducted by the facility maintenance
personnel. The final level.of surface cleanliness achieved was
quantified by measuring and comparing the amount of nonvolatile
residue on parts cleaned by the TCA and LCOz processes. All NVR
tests were performed by sonicating the cleaned or dirty parts in
solvents that had been proven to be effective in removing the
contaminants from the parts, followed by evaporating the solvents
and weighing the residue. Because cleaning effectiveness was to
be defined by comparison between parts cleaned by the two
processes, it was not necessary to quantitatively determine the
amount of contamination present on the surface of the parts
before they were cleaned. This was important in the test
evaluations since the many of the parts provided were of various
sizes, shapes, and diameters.
Fuel Line Tubes
The fuel line tubes cleaned were contaminated with drawing
compound with an average NVR of 5.462 g/ft2. The HOP process was
used prior to LCOj on the tubes since it was determined that the
drawing compound, was soluble in the HOP. Table 3 summarizes the
results of the fuel line samples cleaned during the on-site
evaluation. These tubes were cleaned as well by the H0P/LC02
process as those by the current TCA process.
-4-

-------
TABLE 3. AVERAGE NVR LEVELS POR FUEL LIKE CLEANING USING LCO,
Soiled/Cleaned
Average NVR
(g/ffc*)
Standard Deviation
(g/fta)
Contaminated tubes
5.462
2.471
TCA cleaned
0.009
0.005
LC02 batch 10
0.005
0.005
LCOj batch 36
0.008
0.004
LC0S batch 37
0.003
0.002
Steel Bolts,
Steel bolts are typically removed from aircraft during
maintenance and repair. The bolts normally are cleaned,
inspected, and reintroduced into the inventory. When removed,
they are encrusted with heavy grease and embedded soils. The
current cleaning process used at the depot includes a pre-soak in
a petroleum distillate similar to the hydrocarbon oil of the HOP
process. The bolts are scrubbed with a brush to remove visible
grease, then placed in a vapor degreaser containing TCA to remove
the petroleum distillate and any grease residue. Finally, the
bolts undergo a abrasive blasting process to remove ruet and
carbon deposits.
The LC02 process steps used were similar to the current
process. The bolts.were cleaned in HOP oil with ultrasonics
followed by degreasing with LC02. Table 4 shows the average NVR
level remaining on the samples after cleaning with HOP and LC02.
No bolts cleaned by the current process were available for
comparison.
TABLE 4. NVR RESULTS POR BOLT CLEANING WITH HOP/LCOa
| Part Description
Average NVR/bolt (mg)
1 Contaminated
788.08
| .Cleaned by H0P/LC02
70.17
The results of the bolt cleaning evaluation by NVR and
visual examination indicate that the bolts cleaned by the
H0P/LC02 process were clean enough to be sent on to the next
step, abrasive blasting. The H0P/LCO2 process requires
approximately the same steps as the current process, however, the
LCOj process eliminates the use of .TCA.
Brass PiIters ,
The brass filters evaluated are used in aircraft propeller
assemblies to filter hydraulic fluid. The filter contaminants
included hydraulic fluid, aircraft grease, and heavy carbon
deposits. A picture of brass filters is shown in Figure 2. The
current process to clean these parts includes soaking them
-5-

-------
overnight in a petroleum distillate (PD) formulation, then hand
scrubbing with a brush, washing in water-based general purpose
detergent, rinsing with water, and drying in an oven. Due to the
prolonged soaking and hand scrubbing, cleaning of this part is
labor intensive and time consuming. Although the cleaning of the
filters does not require the use of a CFC 'Solvent, it does result
in the generation of contaminated -water stream. In addition, the
process leaves visible black carbon deposits on the surface of
Figure 2. Brass Filters.
The HOP/LC02 cleaning process consists of three steps
compared to five for the current process. Hand scrubbing in HOP •
oil with no pre-soak replaces the overnight soak and hand
scrubbing in the PD; and LC02 cleaning replaces washing in water
and detergent, water rinsing, and oven drying. An added blow-out
technique using compressed shop air to dislodge particles trapped
within the filter mesh improved the NVR of the filters- In
addition, the filters cleaned with the HOP/LC02 process were
visually much cleaner and shinier than those cleaned with the
current process. Table 5 compares the NVR levels for the filters
cleaned by the various methods.
TABLE 5. SUMMARY OF NVR LEVELS FOR BRASS FILTERS
Filter Sample
Average NVR
(mg/filter)
Standard Deviation
(mg/filter)
Contaminated filters
289.02
210.13
Current process
17.13
4.68
LC02 without blow out
25.13
11.98
LCOa with blow out
16. 80
3.18
-6-

-------
Processing Cos tp
During the two week evaluation period a total of 37
experimental batches were run. The LC02 recovery rate over the
two week period was 94 percent. A total of 24 gallons of LC02 was
lost during this time. At an average cost of approximately $0,70
per gallon, the total LC02 cost was $16.80. The spindle oil,
which constitutes the major HOP cost, averages $490.00 for 55
gallons. The HOP oil recovered during the LCO, degreasing step is
recycled to the HOP tank. The spindle oil in the HOP tank can be
reused indefinitely with disposal required only after it becomes
saturated with contaminants to a point where it cannot complete
its function. This contamination level was not achieved during
the two week evaluation period.
The capital cost of the LC02 equipment will vary
significantly depending on its configuration. It can be
relatively high when compared to conventional vapor degreasing or
aqueous cleaning. Equipment cost will depend on the size of parts
to be cleaned, since it requires pressure vessels fabricated to
hold the range of parts to be cleaned, and the ultimate
configuration of the system. A realistic estimate of capital
costs for a LCOa cleaning system and recycler is $175,000 to
$350,000. However, the return on investment of the system should
be weighed against the savings achieved in material costs,
disposal costs, elimination of regulatory compliance costs, and
pollution abatement costs.
Conclusions
The LC02 process was shown to be equivalent to TCA in
performance for cleaning the parts evaluated during this project.
Successful introduction of the LC02 process into a facility will
require a detailed knowledge of the contaminant to be removed,
the desired surface cleanliness level, and the configuration of
the part. This knowledge will permit the LC02 process to be
effectively integrated into the production operation to achieve
the required cleanliness level.
References
1.	R. P. de Fillipi and M.E. Chung, "Laboratory Evaluation of
Critical Fluid Extraction for Environmental Applications,H
EPA Report EPA-600/2-85-045, April 1985.
2.	Katauskas, T. and H. Goldner, "SFE: Will it Solve Your Lab's
Solvent Waste Problems?" R&D Magazine. March 1991, pp. 40-
44.
3.	Stahl, E., et. al, "Extraction of Seed Oil with Liquid and
Supercritical Carbon Dioxide," J. Aoric. Food Chem.. Vol.
28, 1980, pp. 1153-1157.
4.	Bok, Edward, K. Dieter, and K.S. Schumacher, "Supercritical
Fluids for Single Wafer Cleaning," Solid State Technology.
June 1992, pp.117-120.
5.	McHardy, J., T.B. Stanford, L.R. Benjamin, T.E. Whiting, and
S.C. Chao, "Progress in Supercritical COa Cleaning," SAMPE
Journal. Vol. 29, No. 5, September/October 1993, pp.20-27.
6.	Phelps, M.R., M.O. Hogan, and L.J. Silva, "Fluid Dynamic
Effects on Precision Cleaning with Supercritical Fluids,"
-7-

-------
In: Conference Proceedings for 1994 International CFC and
Haion Alternatives Conference, October 24-26, Washington,
DC, pp.540-549.
1, Barton, A.F.M., Handbook of Solubility Parameters and other
Cohesion Parameters, CRC Press, Boca Raton, 1983, pp 153-
. 158.
8.	Determined from Figure 3-4S and Table 3-283 of; R. H. Perry
and C.H. Chilton, Chemical Engineer's Handbook. 5th Edition,
McGraw-Hill Book Company, 1973.
9.	Handbook of Chemistry and Physics. 56th Edition, CRC Press,
Boca Raton, FL, 1975. -
Notices
The information described in this paper has been funded
wholly by the Environmental Protection Agency under Cooperative
Agreement No. CR818419 to Research Triangle Institute. It has
been subjected to Agency review and approved for publication.
The use of trade names and company names in this paper does
not signify recommendation for use or endorsement by either the
EPA or Research Triangle Institute.
The LCOj cleaning and recycling equipment, the hot oil
process {HOP) material, and related processes, designs and
operations used in this investigation and described in this paper
are proprietary to DEFLEX Corporation and are the subject matter
of issued patents and pending patents and applications. DEFLEX
Corporation, Burbank, CA, a member of the technical team,
provided the LCO: cleaning equipment. Nobles Manufacturing
Incorporated provided a centrifugal spinner/drier.
-8-

-------