C-ACT: Applied Chemical Technology

A Survey on the Use of
Supercritical Carbon Dioxide as a Cleaning Solvent

W. Dale Spall and Kenneth E. Laintz

SUBMITTED TO: prepared as chapter in:
Supercritical Fluid Cleaning. J. McHardy and S. Sawan, Eds.
Noyes Publications Park Ridge, NJ 

ABSTRACT

Because the physiochemical properties of supercritical carbon dioxide make it ideally suited for removing commonly encountered contaminant found in the cleaning of a wide variety of components and assemblies, an overall survey was conducted using a small scale supercritical fluid extraction system to investigate removal efficiencies of a wide variety of compounds from an assortment of surfaces using supercritical carbon dioxide. Data is presented demonstrating the successful removal of numerous oils, fluids, adhesives, and chemical compounds from a wide variety of surfaces with supercritical carbon dioxide. In total, the removal of 145 compounds from some 49 different substrates was investigated. It was found that to a first approximation, cleaning with supercritical CO₂ appears to be contaminant dependent while being surface independent, with an 85-95% removal rate for a wide variety of the compounds investigated.

INTRODUCTION

Many industrial facilities currently using chlorocarbons and chlorofluorocarbons (CFCs) for the cleaning of a variety of items are facing a difficult situation because of the U.S. amendments to the Montreal Protocol (1987) banning the use of CFCs at the end of 1995. For this reason, these companies must implement economical replacement technologies for cleaning applications. Of course, any solvent cleaning replacement technology must take into account the type of items being cleaned, the contaminant to be removed from these items, and the final cleanliness level that the items must possess. Alternate technologies such as aqueous and semi-aqueous based systems are currently being implemented. While these systems have advantages over CFC cleaning methods, these systems suffer from disadvantages that may not be desirable to many cleaning operations. In the case of aqueous systems, disadvantages include long drying times and flash rusting in addition to many parts not being amenable to water cleansing. In addition, water treatment costs may also be prohibitive. Many semi-aqueous cleaning systems employ toxic terpenes or CFC replacements, and it is only a matter of time before these compounds face regulation. A final alternative technology involves the use of supercritical fluids, which have been used in food, fragrance, and petroleum processes for years, for the extraction of many common compounds.

Ultimately, most cleaning specifications are based on the amount of specific or characteristic contaminants remaining on the surface being cleaned. Common contaminants can include machining oils and greases, hydraulic and cleaning fluids, adhesives, waxes, human contamination, and particulates. In addition, a whole host of other chemical contaminants from a variety of sources may soil a surface. Therefore, any CFC replacement solvent under consideration should be able to remove any of these commonly encountered soils to specified levels from a variety of surfaces, including printed circuit boards, plastics, metals, rubbers, composites, and glasses. For the purposes of this paper, precision cleaning will be addressed as opposed to bulk cleaning. This precision level can be defined as an organic contaminant level of Iess than 10 micrograms of contaminant per square centimeter.² This 10 µg/cm² level of cleanliness is either very desirable or required by the function of parts such as metal devices, machined parts, electronic assemblies, optical and laser components, precision mechanical parts, and computer parts.¹

While supercritical carbon dioxide may be an excellent cleaning solvent for many organic contaminants, many substances requiring removal in cleaning operations, inorganic or ionic contaminants, for example, are insoluble in carbon dioxide. In addition, many items requiring cleaning are intolerant of pressures associated with supercritical CO₂. for cleaning considerations, it should be noted that supercritical CO₂ is best suited for the removal of organic compounds with mid-to-low volatilities.¹ These types of compounds are often encountered as contaminants in precision cleaning, and it is on these compounds that our experimental studies were focused. Since the goal for most precision cleaning levels is less than 1 µg/cm² for most soils,³ the 49 substrate materials used in this survey were initially contaminated with 2 µg/cm² of the 145 contaminants investigated. It is the removal of this amount of material to below the desired 1 µg/cm² contamination level for this survey to determine the general applicability of supercritical fluid cleaning technology.

EXPERIMENTAL SECTION

The small scale supercritical CO₂ charting survey was undertaken to investigate the removal efficiency of a wide variety of contaminants and compounds from a wide assortment of substrates which could be encountered In a cleaning situation. The survey investigated the removal of six human based organic contaminants, five adhesives, seven different hydrocarbons, waxes, high molecular weight compounds, and thirteen different machining oils, fluids, and lubricants, including water miscible types, from fifteen different metal, nineteen polymeric, five rubber, five cable, three glass, and two fabric substrates. The different contaminants and substrates investigated are summarized in Tables 1-16. In addition, the removal of 114 different miscellaneous chemical compounds including polycyclic aromatic hydrocarbons (PAHs), amines, substituted phenols, substituted benzenes, phosphates, acids, and acid esters from 340 stainless steel, electrolytic grade copper sheet, glass fiber filled epoxy board, borosilicate glass, and cast magnesium. All of the different chemicals investigated in the survey are listed in Tables 16-20.

The contaminant materials were applied as a dilute solutions to 0.5 in. by 2 in. (12.9 cm₂) coupons made from the different substrate materials using a manual pipettor. The contaminant solutions were applied in such a manner so that the entire surfaces of the coupons were coated with 2 µg/cm₂ of each contaminant compound. While it is noted that a contamination level of 2 µg/cm₂ Is below the precision clean standard of 10 µg/cm₂, 2 µg/cm₂ of contamination was visible in many cases and was required to provide a reasonable detector signal for proper quantitation of the contaminant removal results. Once the application solvent had evaporated to dryness, a contaminated coupon was placed in a 10 ml extraction or cleaning vessel in a Suprex SFE/50 supercritical fluid extractor (Suprex Corp., Pittsburgh, PA). All contaminated coupons were cleaned or extracted dynamically, meaning that there was continuous solvent flow through the cell for each survey. The extractions were conducted using SFC/SFE grade CO₂ (with siphon tube and 1500 psi He head space, Scott Specialty Gases, Inc., Longmont, CO) at 300 atm and 45°C for 15 min. with a flow rate of 2.8 ml/min. After flowing through the extraction cell, the supercritical CO₂ containing dissolved contaminant was depressurized directly into the inlet of a Hewlett Packard (HP) 5971 gas chromatograph equipped with an HP 5972 series mass selective detector (GC-MS). The GC-MS was operated in the split mode with a split ratio of 150 to 1. The GC column was a 60 m x 0.25 mm i.d. DB-5 (5% crosslinked Ph-Me silicone) column programmed from 30 to 275°C with a temperature ramp of 7°C/min. Chromatographic peak areas and subsequent corresponding concentrations of the extracted compounds were calculated from the total ion chromatograms by the HP software. The concentrations obtained using this method were then compared to the initial concentrations of contaminant placed on the substrate coupons and prepared as percent of original material removed from the substrates. The extraction surveys were run in triplicate which yielded an overall average 7% relative standard deviation for all of the compounds investigated.

RESULTS AND DISCUSSION

Of the wide variety of contaminants and compounds investigated in the small scale supercritical CO₂ cleaning survey, of particular importance are compounds associated with human based contamination which is often a significant component of organic contamination found in many cleaning operations, especially those involved in precision cleaning. Human based contaminants can be found in sweat, fingerprints, and other human soils and can contain hundreds of different chemical compounds. Generally, the major constituents of this type of organic contamination are made up of fatty acids and oils found in the skin. For this study, representatives of the chemical classes found in skin lipids were used and consisted of squalene, triglycerol, diglycerol, cholesterol, and palrnitylpalmitate. In addition to skin oils, fingerprints tend to be commonly encountered contaminants on parts, components, and assemblies. In order to investigate the removal of fingerprints from surfaces, a fingerprint surrogate consisting of a mixture of skin lipids was prepared based upon previous work.^4–6 The components of the surrogate fingerprints consisted of 30% triolein, 25% oleic acid, 25% cotyl palmitate, 15% squalene, 2.5% cholesterol, and 2.5% cholesterol oleate (components obtained from the Aldrich Chemical Company, Wl). While salts are certainly components of fingerprints, these compounds were not added to the mixture since they were incompatible with the experimental detection system. In any event, the surrogate used for this study was assumed to behave in an analogous manner to actual fingerprints.

The results for the removal of human based organic contamination from the 49 different substrates investigated are summarized in Tables 1-3. The results presented in Table 1 summarize the removal of squalene, triglycerol, diglycerol, cholesterol, palmitylpalmitate, and synthetic fingerprints from 15 metal and 3 glass surfaces. These results show near quantitative removal of synthetic fingerprints, squalene, and palmitylpalmitate from most all of the metal and glass surfaces. However, using the test conditions as described, the cast metals, cast aluminum, magnesium, and iron, showed lower extraction efficiencies. For example, cast magnesium had a synthetic fingerprint removal rate of 56%, while stainless steel 306 had a removal rate of 97%. The low removal rate from the cast metals is believed to be due to the porosity of the substrate surface. Because of their high diffusivities and low viscosities, supercritical fluids are inherently capable of penetrating porous surfaces and removing contaminants, and increased removal rates from the cast metals were easily accomplished through parametric changes. For example, longer extraction times of 30 to 45 min. resulted in quantitative removal of the synthetic fingerprints from the cast magnesium surface. The removal rates of the gIycerols and cholesterol were lower than the other human contaminants due to their lower solubilities in supercritical CO₂.⁷ The removal of these compounds can be improved with a longer extraction time as in the case of the cast metals or through the use of a static extraction step where the substrate is immersed in supercritical CO₂ with no flow through the cell and then followed by a dynamic extraction.

The results of the removal of the skin lipids from the 19 polymeric materials used in this survey are summarized in Table 2. These results compare similarly with those observed for the removal of the lipids from the 3 glass surfaces shown in Table 1. Again, near quantitative removal of synthetic fingerprints, squalene, and palmitylpalmitate was observed with the same lower removal efficiencies for the glycerol and cholesterol. In general, the same results were observed for the removal of these compounds from the 5 rubber, 5 cable, and 2 fabric substrates as seen in Table 3. Palmitylpalmitate was not as effectively removed from the rubber surfaces, probably due to surface interactions with the acid moiety of the compound. While the fabric samples can be thought of as porous substrates, contaminant removal efficiencies from these surfaces were much higher than the cast metals because unlike the metals, supercritical CO₂ can flow through the fabrics thus limiting surface interactions between contaminant and substrate.

The results for the removal of common machining oils and fluids from the selected substrates are summarized in Tables 4-6. Oil removal rates from the 34 smooth surfaces investigated, metals, glasses, and plastics, were near quantitative as seen from Tables 4 and 5. The overall removal rates of the oils and fluids from all of these surfaces were quite good, averaging from about 90 to 97%. Of particular note, as seen in Table 4, is that supercritical CO₂ was quite effective in the removal of the various oils and fluids from all smooth metal surfaces, removing, for example, from about 89 to 99% of the Tapmatic® cutting fluid. These results show the applicability of supercritical CO₂ cleaning to machined end precision metal parts and components. Again, however, observed cleaning efficiencies using the described conditions were not as high for the porous metal substrates. On the other hand, quantitative removal of the investigated machining oils and fluids from the rubber, fabric, and cable substrates listed in Table 6 was observed to be near quantitative, averaging from about 85 to 99%. The two compounds that did not extract well from any of the 49 surfaces and not included in the aforementioned average removal rates were Molykote lubricant and silicone oil. Since Molykote consists primarily of inorganic particulate matter in a high molecular weight grease, it was expected to have low removal efficiencies with supercritical CO₂. Silcone oil was also not as efficiently removed as the other contaminants due to low solubility or to fractionation of the oil with the higher molecular weight, less soluble components remaining on the surface.

Other common contaminants associated with a machining environment can include water miscible machining fluids and surfactants. For this reason, the removal efficiencies of select compounds from these classes of fluids were also investigated. For the survey, the removal efficiencies of TRIM® SOL, Cimcool, Cimtap, which are water miscible machining fluids, and the nonionic surfactant Triton X-100 from the 49 substrates were studied. The results of this set of experiments are summarized in Tables 7, 8, and 9. Surprisingly, these water soluble materials had fairly high removal rates, generally averaging above 80% removal from all but the porous metal substrates using the specified conditions. This example suggests that while an aqueous cleaning process might under consideration as a cleaning system replacement, supercritical CO₂ may be a viable cleaning option in cases where components are not ideally suited to aqueous immersion.

Due to the physiochemical properties of a supercritical fluid, cleaning with supercritical CO₂ has a potential advantage over other cleaning technologies due to its ability to rapidly clean completely assembled components systems. In many instances, assembled components that are in need of cleaning contain adhesives, epoxies, and/or sealants. While in some cases it may be desirable to remove these substances from a surface, in other cases it may be desirable to clean the surface and leave these substances intact. With both of these strategies in mind, a selection of adhesives, epoxies, and sealants were applied to each of the 49 substrates and extracted with supercritical CO₂. The results from this portion of the cleaning survey are summarized in Tables 10, 11, and 12. As seen from these tables, it is clear that supercritical CO₂ was ineffective in removing the various adhesives from any of the substrates. For the specified conditions, from about 23 to 52% of the RTV-732 and 3110 Silastic Adhesive Sealant and the Loctite® 242 Threadlocker were removed from the surfaces. While, parametric variations such as longer extraction times, increased temperatures and/or pressures, or the inclusion of a static extraction step may increase removal rates, it is unlikely that these compounds would be quantitatively removed from the surfaces, thus precluding effective cleaning with supercritical CO₂. On the other hand, removal rates of Iess than 10% were observed for Devcon F-Fast Setting Epoxy and Eastman 910 Super Glue. This low removal rate could conceivably correspond to the extraction of residual solvents from the adhesives, thus demonstrating relative inertness to CO₂ exposure. Therefore, it is conceivable that components assembled with these or similar products could be cleaned with supercritical CO₂ without damage to the adhesive bonds.

The results from the last set of contaminants surveyed for removal efficiencies from the 49 different substrates are listed in Tables 13, 14 and 15. These contaminants are representative of larger classes of contaminants which may be encountered in cleaning operations. For example, hexadecane and tetracontane can be found in kerosene and diesel. Waxes, such as paraffin wax, are used as lubricants and mold releases. Carbowax and Microwax are chromatographic stationary phases, but they are forms of polyethylene glycol which is also a lubricant. Finally, methyl silicone gum and other methyl silicone resins are often used in protective coatings. The lower molecular weight materials, hexadecane, tetracontane, and paraffin, had fairly high removal rates, generally averaging above 80% removal from all but the porous metal substrates using the specified conditions. On the other hand, the high molecular weight waxes had fairly low removal efficiencies in the 13 - 39% range. This was to be expected since supercritical fluid extraction using CO₂ is known not to do well in dissolving high molecular weight compounds. Again, however, the removal of these compounds could probably be improved with a longer extraction time or through the use of a static extraction step followed by dynamic extraction, but it is unlikely that CO₂ cleaning alone would quantitatively remove such high molecular weight contaminants.

Because a wide combination of chemical contaminants from a variety of sources may soil a surface, the cleaning survey also included removal studies of 114 different organic chemicals from a variety of classes of compounds. These compounds include PAHs, amines, substituted phenols, substituted benzenes, phosphates, acids, acid esters, as well as an assortment of miscellaneous compounds. Using the aforementioned cleaning or extraction conditions, the removal of the compounds listed in Tables 16-20 from 5 surface representatives from the larger, previously investigated group was investigated. The surfaces that were contaminated consisted of coupons made from 340 stainless steel, electrolytic grade copper sheet, glass fiber filled epoxy board, borosilicate glass, and cast magnesium.

The results for the removal of PAHs from the 5 surface substrates are summarized in Table 16. In general, the 23 PAHs listed in the table averaged removal rates around 90% from the smooth surfaces and over 80% for the porous cast magnesium surface. In contrast, supercritical fluid extraction studies using CO₂ for the removal of PAHs from soils for environmental applications have shown relatively poor removal efficiencies for many of the compounds listed in the table often requiring the addition of secondary solvents to the CO₂.⁸ However, it appears that from the results on the removal of the PAHs shown in Table 10, surface contamination is much easier to extract and remove than interstitial or sorbed contamination as in the case of soils where a wide range of contaminant-substrate interactions are possible. Since surface interactions with the contaminants are expected to be minimal with the stainless steel, epoxy board, copper sheet, and borosilicate glass, the observed high removal rates were intuitively expected. While surface interactions may not be a dominant controlling factor in PAH removal, the chemical nature of individual PAHs control removal efficiencies. In this case, as substitution increased for various PAHs, removal efficiencies decreased. For example, pyrene had a removal rate of 97% from the glass surface whereas indeno(1,2,3-CD)pyrene had only an 86% removal rate.

Organic amines constitute a wide class of compounds ranging from solvents such as aniline and pyridine to familiar chemicals such as nicotine. Altogether, a selection of 23 organic amines, most of them aromatic compounds, was investigated in the cleaning survey. The results of the removal efficiencies of organic amines from stainless steel, copper sheet, epoxy board, borosilicate glass, and cast magnesium are summarized in Table 17. In this case, removal efficiencies were entirely compound dependent, based predominantly on contaminant solubilities in supercritical CO₂. For example, compounds such as N-nitrosodimethylamine, which is soluble in water, and N-nitrosophenylaniline had low removal efficiencies ranging from 30 to 40% from the smooth surfaces and only 21% removal from cast magnesium. On the other hand, 2-nitroaniline and 4-nitroaniline, which are soluble in polar organic solvents had over 97% and 90% removal rates, respectively. All in all, the organic amines had a general average removal rate near 80% which still shows fairly effective supercritical CO₂ cleaning.

In general phenols are polar organic compounds primarily soluble in polar organic solvents, and in some cases, water. For this reason, it was expected that removal rates for these types of compounds from the 5 substrates in the chemical removal survey would be rather low. In fact, for most of the 19 substituted phenols surveyed, the removal rates averaged near 60%, and these results are summarized in Table 18. However, several of the substituted phenols, the cresols, for example, had very effective removal rates, averaging above 90% removal using supercritical CO₂ at 10°C and 350 atm despite the fact that CO₂ is a nonpolar solvent. It is possible that the high removal rates of these compounds could be attributed to the fact that they are liquids at 40°C, thus facilitating extraction due to faster kinetics. This implies that higher removal efficiencies for the other phenols could be accomplished through higher temperature extractions. AIso summarized in Table 18 are the removals of substituted benzenes from the same surfaces. Again, in this case the chemical nature of individual compounds was the controlling factor governing removal efficiencies. On average, these compounds had removal rates around 85%, again showing fairly effective supercritical CO₂ cleaning.

Summarized in Table 19 are the results for the removal of organic phosphates, acids, and acid esters. Only one organic acid, benzoic acid, was investigated. Since this compound is soluble in water, it was expected to have a low removal efficiency from all 5 of the surfaces. This was indeed the case with an average removal of 42% from the smooth surfaces and 35% from cast magnesium. Once an organic acid is esterified, it is generally less polar than the precursor thus increasing lipophilicity. The acid esters investigated in this survey were all phthalates, and they averaged around 90% removal efficiencies. The three organic phosphates listed in the table show average removal efficiencies around 77% for the smooth surfaces. While trimethylphosphate is water soluble, it had the highest removal efficiency of the phosphates once again suggesting that supercritical CO₂ cleaning may be an aqueous cleaning alternative is some limited instances.

Finally, Table 20 lists the removal efficiencies of another 29 miscellaneous chemical compounds from stainless steel, copper sheet, epoxy board, borosilicate glass, and cast magnesium. Again, the important observation is that surface interactions appear not to be controlling compound removals as the smooth surfaces tend to have the same removal efficiencies. The chemical nature of individual compounds controls the removal efficiencies.

CONCLUSION

While supercritical CO₂ is not an absolute or drop-in solution to all cleaning problems, it is noted for its solvation of organic compounds having mid-to-low volatilities, and these types of compounds are common contaminants requiring removal to precision clean levels. Based upon the survey results presented in this chapter for the removal of 145 different compounds from 49 surfaces, it was shown that to a first approximation, cleaning with supercritical CO₂ is contaminant dependent and surface independent. Furthermore, in the case of PAHs, it was shown that surface contamination was much easier to extract and remove than interstitial or sorbed contamination. In addition, it was shown that supercritical CO₂ is also capable of removing many compounds traditionally removed by aqueous cleaning, thus expanding the scope of cleaning applicability. Therefore, besides the effectiveness of cleaning with CO₂, the economics of the entire cleaning process may direct the use of CO₂ in cleaning applications where other replacement technologies are under consideration as well as processes other than precision cleaning. Finally, the use of supercritical CO₂ as a cleaning solvent can reduce the overall use of organic solvents in manufacturing processes.

ACKNOWLEDGMENTS

This work was performed under funding from the Industrial Waste Program Office, Department of Energy, Office of Industrial Technologies.

REFERENCES

1. American National Standards Institute, ANSI/IPC-CH-65, 7.2.3, pp. 47-48 (1990)
2. Spall, W.D., Supercritical Carbon Dioxide Precision Cleaning For Solvent And Waste Reduction, Intntl. J. Environ. Conscious Design & Manufact., 2(1): 81-86 (1993)
3. McHardy, J., Stanford, T.B., Benjamin, L.R., Whiting, T.E., and Chao, S.C., Progress in Supercritical CO2 Cleaning, SAMPE Journal, 29(5): 20-27 (1993)
4. Downing, D.T., Strauss, J.S., and Pochi, P.E., Variability in the Chemical Composition of Human Skin Surface Lipids, J. Invest. Dermatol. 53(5):322-327 (1969)
5. Downing, D.T. and Strauss, J.S., Synthesis and Composition of Surface Lipids of Human Skin, J. Invest. Dermatol. 62(3):228-244 (1974)
6. Strauss, J.S., Pochi, P.E., .and Downing, D.T., The Sebaceous Glands: Twenty-Five Years of Progress, J. Invest. Dermatol. 67(1):90-97 (1976)
7. Bartle, K.D., Clifford, A.A., Jafar, S.A., and Shilstone, G.F., Solubilities of Solids and Liquids of Low Volatility in Supercritical Carbon Dioxide, J. Phys. Chem. Ref. Data 20(4):713-756 (1991)
8. Yang, Y., Gharaibeh, A., Hawthorne, S.B., and Miller, D.J., Combined Temperature/Modifier Effects on Supercritical CO2 Extraction Efficiencies of Polycyclic Aromatic Hydrocarbons from Environmental Samples. Anal. Chem. 67(3):641-646 (1995)