sign in sign up
<<
Home , Carbon dioxide cleaning, Cleaning, Cleaning agent, Parts cleaning

GOING THROUGH PHASES

A Panel discussion of CO 2Cleaning Technology

26 PC July/August 1999 A panel recently convened at CleanTech ‘99 to examine the current state of

carbon dioxide (CO 2)-based cleaning technology. Panel members discussed the

typical applications appropriate to each system, the benefits of C02-based sys-

tems over other cleaning techniques, and the complexities of system operation

and maintenance. Panelists also reviewed performance levels, relative costs,

and specific application strategies.

iquid Carbon Dioxide Systems to 25 minutes, depending on the size of the Panelist: Liz Hill cleaning vessel, the size of the pumps, the Research Triangle Institute nature of the contaminants, and the number of carbon dioxide (LCO2 ) technol- required cycles (ie, washes and rinses). ogy is one that can be classified as a Efficacy is specific to the contaminants being "maturing" technology. At high pres- addressed: however, pretreatment or additives sures, CO 2 takes on a liquid form that (eg, or ) improve the range is sustainable at, above, and below of cleaning. Some contaminants that respond room temperatures (see Figure 1, well to LCO 2removal include: light , hydro- page 28). carbons, machining fluids, and chips (due The equipment used in LCO2 appli- to agitation). cations includes a vessel by which a To date. reliable submicron particle removal L constant pressure is maintained. (to this speaker’s knowledge) has not yet been Some form of agitation (eg, spinning, achieved by LCO 2 processes, but this limitation spraying, ultrasonics, etc) is generally required is likely a temporary one. Heavy hydrocarbon as well. To recycle and clean LCO2 it must be greases do not respond well; however, additives transferred to another chamber, where pressure can remedy this shortcoming. Salts and many is released, allowing the CO2 to expand back to other inorganic soils continue to present a chal- a gas. This leaves behind only the contaminants lenge to LCO 2systems. Paint. , and carbon that were removed from the parts, which are residues will likely never be ideal soils for this collected for disposal or re-use. The CO2 is then type of technology. returned and repressurized in the cleaning Liquid CO 2systems have demonstrated suc- chamber. In many systems, such additives as cess over a range of cleaning applications, organic solvents and surfactants are mixed with including: the LCO improve performance. s Fiber optics 2 Pure LCO2 systems do not require any per- s Machined metal parts mits, but users must be aware that some addi- s Hydrocarbon residues from electrical tives, like certain organic solvents, may have components permit issues. No or sewer connections s Rag cleaning (even paper) are required for these systems, and waste is s Dry cleaning collected at the recycling point. Carbon dioxide In LCO 2systems, the dimensions of the clean- is a greenhouse gas, but it is not a net addi- ing vessel will dictate the type of parts that can tion because it is gathered from other be cleaned efficiently. Parts must also be com- processes. Carbon dioxide is not an ozone- patible with the high pressures of the cleaning depleting compound (ODC) nor is it considered chamber. is a possibility, and a volatile organic compound (VOC); it is non- parts must be warmed properly before removal flammable and is not considered a hazardous from the chamber to prevent this. Obviously, air pollutant (HAP). this step can increase cycle time. CO2 has been Unlike many traditional systems that perform known to strip certain plasticizers and pene- to limitations set by bath life and the effective- trate some elastomers. When the pressure is ness of the , LCO 2 systems fea- dropped for removal. bubbles can form inside ture cleaning cycles that employ fresh, clean the elastomer and cause deformities. CO introduced from a recycling loop. It is a Operation of LCO equipment is fairly simple 2 2 batch process, and cycle times can be long-10 due to its high level of automation. Health and

BASED ON A PANEL PRESENTATION AT

July/August 1999 PC 27 phases become one (this is unlike the commonly held theory that a vapor becomes so dense that it acts like a liquid). As such, the fluid has the penetrating power of a gas, but the cleaning ability of a liquid. As Figure 1 shows, begins at 73 atm and 31°C. Supercritical CO 2(SCCO 2 ) possesses physical characteristics that allow it to relate with other more familiar substances. One of the attractive fea- tures of SCCO 2is its reputation as a ‘tunable” -this is, as you change the pressure of the system (and thus the density of the SCCO 2 you change its solvating characteris- tics. In principle, SCCO 2systems Figure 1. A phase diagram of carbon work in much the same manner as dioxide. those for LCO. 2 All of the environmental and safety safety concerns include proper issues that go along with LCO 2apply instruction in operating pressurized to SCCO 2The performance benefits of replacing traditional cleaning sys- systems, close monitoring of CO2 lev- els in the air, attention to possible tems with SCCO2 are also similar to overexposure to cold gas and liquid those of LCO2 Some of the more sig- (resulting in possible skin bums), and nificant beneilts include the absence temperature of parts being removed of cleaning agent residue, no required from the system. drying step, and high-temperature Prices range from $100,000 and operation, which expedites solvency higher, with the CO 2running $0.15 to action. Applications for the technol- $0.20 per gallon for bulk and about ogy are concentrated mostly in $0.85 per gallon in bottles. smaller, niche markets. LCO2 systems have proven very suc- SCCO2 is actually quite effective at cessful in cleaning a variety of sub- removing particulate matter. Research strates. Even an intricate material like has been carried out on integrated cir- aluminum honeycomb has been cuit wafers, where a concept dubbed cleaned down to practical standards “turbulent flow” has proved effective in without evidence of substrate damage. removing micron to sub-micron parti- Results also were favorable when cles. This is a significant benefit given the technology was employed on that SCCO 2will remove any organic brass hydraulic filters from a heli- residue helping the particles to adhere copter; the filters were covered with a to the surface of the substrate being gritty/oily substance (Figure 2). Prior cleaned. Lower temperatures facilitate to investigating CO 2 technologies, the removal of smaller particles due to manufacturer originally cleaned these the greater density of the fluid at parts by soaking them for 24 hours in these temperatures. a hydrocarbon mixture, scrubbing by For one manufacturer, an SCCO2 hand, soaking them for another 24 system solved the challenging task of hours, and scrubbing again. With the cleaning silicon wafers. The challenge new LCO 2 process, the parts were came in cleaning the wafers after they first soaked in a warm hydrocarbon were cut from the parent ingot with a / mixture for about 10 wire saw. Contamination resulted minutes, hand scrubbed, and then from a machine containing silicon cleaned in LCO for 20 minutes. This carbide, which was used as the abra- 2 turned a P-day process into a 45- sive to cut the individual wafers. The minute task. cut wafers had a tendency to lean The technique offers manufacturers against one another, and the oil at a way to minimize waste streams, that contact point was difficult to clean parts that are incompatible with remove again, at least with traditional water, and eliminate hazardous cleaning agents. Correction of the cleaning agents. wafer positions was not possible because they had to remain glued to Cleaning With Supercritical CO2 a substrate from the point of Panelist: Yale West, cutting through cleaning. Applied Separations Primary testing demonstrated Supercritical fluid is a very unique immediate success and was accom- state of matter, almost a fourth state plished by sandwiching two wafers per se. It essentially refers to the with oil between them and then point where the liquid and vapor attempting to clean them with

July/August 1999 Figure 2. Helicopter hydraulic filters before and after cleaning with liquid carbon dioxide.

SCCO2 Full implementation of an SCCO2 system followed, and the wafer cleaning process showed considerably improved results. SCCO2 systems can be either fully manual or automatic. Another use for SCCO2 is as a solvent/carrier for sur- face treatments.

Snow CO, Cleaning Panelist: Dr. Robert Sherman, Applied Surface Technologies Snow CO2 (SCO2 ) is generated by expanding liquid or gaseous CO, through an orifice. It removes both organic soils and particles as small as several hundred angstroms (1.0 X 10-10 meters). It will not, however, remove substances that are chemically bonded to a surface (eg, paint, epoxies, evapo- rated metals, rust. etc). SCO2 is non- toxic and nonflammable: it is not listed as an ODC or a VOC. To safely apply the technology, users should avoid skin contact, consider room oxygen displacement, and be aware of exces- sive CO, buildup in the room. According to the phase diagram (Figure 1). SCO 2 formation takes place along the phase boundary lines separating gas, liquid, and solid. The thermodynamics of snow formation should be a constant enthalpy process. The expansion within and after the orifice leads to lower pres- sures and temperatures, creating a gaseous and solid-phase mix that can travel at high velocities. Exposing a CO, gas source to the dropping pressure inside an orifice causes fluid droplets to nucleate: as a result, the percentage of liquid increases. The liquid converts to solid at the interface between the liquid-gas and gas-solid regions (near 80 psi), and this yields about 6% . Beginning with a liquid CO2 source, the pressure drop in the ori- fice generates gas bubbles. The per- centage of gas increases until the gas- solid boundary is met. Here, the remaining liquid is transformed into solid, yielding about 45% dry ice. Gas-fed systems tend to be (it is easier to filter a gas than a liq- uid), have less heavy hydrocarbon Figure 3. Particle removal by carbon dioxide snow based on Figure 4. Removal of organic contaminants by solid-state momentum transfer. carbon dioxide per the theory of liquid-phase solvency.

contamination, and have less con- drag force of the CO 2gas stream adds similar to high-pressure solvent spray sumption per unit of time. Liquid-fed to this removal effort. Organic mater- methods. Furthermore, there is strong systems produce more snow, allow for ial is removed presumably by a liquid evidence that SCO2 may freeze- faster cleaning, but involve a higher CO2 phase that can form on the sur- fracture certain organic layers from consumption rate. The source state, face as the surface pressures increase substrates. as well as orifice and nozzle designs, between the dry ice and substrate. The basic SCO 2cleaning setup is will dictate the dry ice size, velocity, Figure 3 demonstrates the particle simple and includes: a source of CO2 and percentage formed. removal mechanism following the (in a cylinder); a hose to deliver the The actual cleaning mechanism of momentum-transfer concept, while CO2 from the cylinder to the on/off SCO2 is not fully understood. The Figure 4 illustrates organic removal control device (ie, a valve or hand- most popular theory is that the high based on the theory of liquid-phase gun); and a nozzle attached to or exit velocity from the nozzle leads to solvent removal. within the handgun. momentum transfer from the solid An additional concept centers SCO2 cleaning technology has CO 2 particles to the surface contami- around the suggestion that particle proven itself successful across many nation, which essentially “knocks it removal is accomplished by a surface industries, including: off’ the substrate. The aerodynamic liquid phase. This process might be s Optics s Disk drives s Microelectronics s Precision parts and assemblies s Instrumentation An example of a successful applica- tion is shown in Figure 5 (page 33), which shows a silicon wafer that was intentionally scratched to generate particulate contamination in the micron-submicron range. After SCO2 cleaning, all particles were removed.

Pellet and Supersonic Snow CO, Panelist: Jeff Sloan. Va-Tran Systems Snow and pellet CO2 (PC02) cleaning is driven largely by kinetic energy which can be defined by the equation E = ½mvz² (where E = kinetic energy, m = mass, and v = velocity). The energy that is imparted by these CO 2clean- ing mechanisms can thus be increased by increasing either the mass of the dry ice particle or the velocity at which it is delivered. Pellet CO2 cleaning is based on the concept of increasing mass. The pri- mary dynamic of this mechanism is an impact/flushing action. The shock wave-or mass-energy coefficient of the initial impact-dislodges or fractures contaminants. The near instantaneous transformation of the pellets to a gas then increases its vol- ume over 900 times to aerodynami- cally flush contaminants away. Sup- plementally, the thermal effects of the

30 July/August 1999 Highly compressed air acts as the ‘accelerator” for the system and pushes an air/CO2 mixture through an orifice to produce snow. Applica- tions for this technology include small , graffiti removal, and layer selective removal.

Questions From the Floor Ql: What is the best technology involv- ing CO 2 for cleaning sub-micron parti- cles (0.2 µm)? I am involved with preci- sion cleaning of disk drive components and would like to explore CO, in greater detail? Hill: An effective technology to remove O.2-micron particles is snow. Both high-pressure and low-pressure snow can be used to remove particles. You will need to verify that the CO, source and gun do not deposit organ- ics on the parts when you clean them, and protect the parts from moisture during cleaning. Test the effect of the process on static-sensitive parts. West: Turbulent-flow SCCO2 has been demonstrated to be effective here. It can dissolve and remove organic contaminants and thus elimi- nate adhesion of particles. Also. remember that due to its nature, it will be effective in blind holes cold pellets can help embrittle films, allows the user to use a heated Sherman: SCO2 is highly effective at making them easier-to break off. The source of air to minimize cooling removing submicron particulates solvent characteristics of CO2 can effects on the surface being cleaned. from surfaces. The general limit provide additional help in removing The Table lists some advantages expressed by the manufacturers of some hydrocarbons. that PCOz blasting offers over tradi- CO, snow cleaning equipment is gen- To perform PCOz cleaning, a source tional [grit, sand, etc) blasting as well erally about 0.1 microns. but new of dry ice pellets is required. Unlike as some limitations of the technology. data show particle removal down to SC02. pellets must be generated prior Typical applications for PCO2 are pri- 0.03 microns. The addition of the to reaching the delivery nozzle. This marily industrial in nature. They solid dry-ice phase to the flowing gas can be accomplished via a pelletizer include: has made momentum transfer an or some device that chips or grates l Molds (tires, foundry, ) effective means for overcoming strong blocks of dry ice (see Figure 6, page l Decontamination (nuclear, heavy particulate surface adhesion forces. 34). Given the larger size of pelletized metals) Sloan: Snow cleaning is very effec- dry ice, it is often most efficient at l “Hot” systems (ie, switchgears, tive at removing submicron particles. removing thick contaminants faster, insulators) It may be helpful to use ionized air or while the chipped pellets offer a gen- l Overhaul (motors, jet engines, nitrogen to help overcome the elec- tler mode of cleaning for more delicate robots) trostatic attraction of the particles to substrates. The cost of equipment has a fairly the surface. The pellets are driven through noz- wide range, from $8000 to $35,000 zles by high-pressure air, which can for portable units to $90,000 to Q2: Please comment on the removal of be accomplished via two methods: $250,000 for fixed, automated ones. trace metal impurities (eg, Al, Fe, Na, A single-hose nozzle. In this Average operational costs run from K) from surfaces (eg, A12 03, Si, method, a high-pressure stream of air $0.50 to $1.00 per minute and glass) by any of these systems. Please passes by an auger, which feeds pel- include the cost of dry ice and com- recommend a feasible approach. lets into it and thus generates a sonic pressed air. Hill: The phrasing of the question velocity stream of dry ice pellets. It is Supersonic CO2 (SSCO2 ) blasting suggests that the problem is molecu- slightly more aggressive, has lower demonstrates higher levels of kinetic lar contamination by metallic species, noise levels, and delivers less dry ice energy by increasing the velocity not particles or organic films contain- per cubic foot than the dual hose sys- parameter of the kinetic energy ing metallic elements. Liquid and tem. The lower level of dry ice is due equation. This is accomplished by supercritical CO, will not remove to abrasive loss inside the hose. velocities up to Mach 3.5 (2100 metallic impurities bonded to a sub- A dual-hose nozzle. In this method, miles/hr or 938.8 meters/sec). The strate. You need chemical dissolution dry ice is delivered using an eduction mass of the dry ice particles is very of molecular traces of these metals. If system in which the air passes by a small, from 0.1 to 1.0 microns. the metals are part of particles. snow Venturi and sucks dry ice pellets from These small, highly energetic parti- may work. another hose into its stream. This cles are capable of deep penetration West: With the appropriate modi- system uses less source dry ice, but into layers of contamination without fiers, like chelating agents, SCCO2 delivers more to the surface. It also damaging substrate. can be very effective here. Successful

32 PC July/August1999 applications include environmental extraction of metals from soil samples for contamination analysis and, in , the recovery of precious met- als from waste slag. Sherman: If these contaminants are particulate-based contamination, CO 2 snow cleaning should remove them. Recent work has shown particle removal from several such as sapphire, MgO. and a thermally sensi- tive substrate. Testing is necessary. Figure 8. Photomicrograph of a silicone wafer demonstrating removal of micron-level Sloan: Removal of trace metal impu- particulate by carbon dioxide snow (A, prior to cleaning; B, postcleaning). rities from ceramic substrates is not likely using dry ice blasting, as the surface is not directly modified.

Q3: What types of additives are being used in LCO 2 systems to achieve the following? A. Improve ability of solvent to carry insoluble soils away from the sub- strate B. Increase the range of nonpolar soils that are dissolved C. Form oil-in-water Hill: I know of nothing going on about item c. A and b are being pur- sued by several companies involved in snow, liquid, and supercritical CO2 Dr. DeSimone is working on fluori- nated surfactants, mainly for applications. He has published some work on cleaning metal coupons with LCO 2 and surfactants. Some work has been done using cosolvents, both in snow and L- or SCCO2. West: A. There are two modes of processing with SCCO2 static and dynamic. When the process is dynamic, the SCCO2 is continuously pumped through the cleaning vessel and collected in the separator. This continuous flow is what enables removal of insoluble contaminants. If the contaminant is merely insolu- ble in SCCO.2 the addition of an appropriate modifier may provide a solution. B. SCCO 2is nonpolar. so nonpolar soils should not be a problem. For polar ones, the addition of a small percentage of polar cosolvent or a modifier, such as methanol, has proven successful. C. Due to its nonpolar nature, SCCO2 will form a water-in-oil emul- sion. There are suitable surfactants for accomplishing this.

Q4: How does one generate ultrasonic “bubbles” in an LCO 2 vessel under 1000 psi? Hill: As long as the pressure and temperature maintain the CO, in the liquid state, cavitation is possible. Cavitation has been verified using low frequency ultrasonics. Los Alamos National Lab did some of the early work in cavitating LCO2 .

July/August 1999 33

PC Q9: How do you determine which type of CO 2 (liquid, snow, etc) is right for your application? Hill: Read the available literature and talk to manufacturers. Before buying any system, watch parts being cleaned in it. Very roughly. pellets are good for tightly adhered gross surface contamination, like rust and carbon. Snow is good for loosely bound parti- cles and some organics. LCO2 and SCCO2 are good for some organics. West: Experimentation! Sherman: Understand the nature of Figure 6. Dry ice pellets produced by a pelletizer (A) and created via a chipping process (6). your contamination and test. If par- ticulate or thin organic layers, CO2 Q5: How is turbulent flow achieved in tems can start at about $1,600 and snow cleaning should be considered, the vessel? How is redeposition pre- about $10,000 for manual pellet sys- especially when one part at a time vented? How are particulates trans- tems. The LCO2 and SCCO2 systems can be cleaned. If overlayer removal is ferred from the cleaning vessel to the are much more expensive because of needed for parts that can survive the depressurization tank; would they not the high-pressure design require- pellet impact, then pellet systems are tend to accumulate? ments. As Jeff said, automated pellet considered. For greases, oils, batch West: First, designing or specifying systems cost above $30,000 and can processing situations, LCO2 has ini- the appropriate pump to achieve the easily go higher. Automation for noz- tial consideration factors, while flow velocities (200-1000 cm/sec) is zle or part movements can increase smaller parts and special removal required. Second, designing or speci- costs. Try snow cleaning first if there needs can be addressed by SCCO.2 fying the flow pattern or path will is a chance of it working. determine the level of turbulence in Sloan: Dry-ice snow is a low-cost About the Panelists the flow. Redeposition is prevented alternative. A number of entry-level Liz Hill is senior research engineer at because dynamic flow of SCCO2 systems exist at a price level below Research Triangle Institute, Research deposits contamination in the separa- $2000. Dry-ice blasting systems have Triangle Park, NC. tor vessel and returns clean to the many applications in the sub- Yale West is national sales manager cleaning vessel. Particulates only $10,000 price range, and the mainte- at Applied Separations, Allentown. Pa. accumulate in the separator, from nance costs of both of these alterna- Dr. Robert Sherman is head of which they are removed periodically. tives is very low. Applied Surface Technologies, New Providence, NJ. Q6: Could dry-ice blasting remove alu- Q8: I have heard that CO, is a very Jeff Sloan is director of minum “solder” from tool molds? time-consuming process compared to at VA-Tran Systems, Chula Vista, In an aluminum die-casting operation, traditional aqueous and solvent clean- Calif. the aluminum will build up on mold ing. 1s that so? surfaces through mechanical bonding Hill: I have seen that aqueous is Suggested Reading to mold imperfections and microcracks. slower than snow but faster than Jackson J, Carver B. Today’s forecast: Would there be enough momentum or LCO2 or SCCO 2mainly because of it looks like snow. Precision Cleaning. thermal cracking to remove solid alu- the time required to pressurize and May 1999: 17-29. minum without damaging the tooling? depressurize the latter systems. Jackson J, Carver B. Liquid CO, Sloan: Dry-ice blasting will not West: SCCO2 for precision cleaning immersion cleaning: the user’s point remove the metal-metal adhesion is a process to be used when other of view. Parts Cleaning. April 1999: described here. It will remove the old processes don’t work. If it is the only 32-37. mold release so that more can be process that works, processing time is applied to the mold, however. probably not going to be an issue. Editor’s Note: For a comprehensive Drawing from other SCCO 2applica- list of vendors who specialize in Q7: Many potential end-users feel hin- tions, one of the major advantages of CO,-based technologies, visit the dered by the initial cost of carbon diox- using SCCO2 is that it is a faster Knowledge of www.Precision ide systems. Could you please process because of its low , CleaningWeb.com and select such address this concern and also discuss variable density. and higher diffusivity. Product Categories as Liquid Carbon maintenance costs compared to tradi- Sherman: This question must be Dioxide, Carbon Dioxide Pellets, Car- tional cleaning methods? examined for each method. With CO2 bon Dioxide Snow, and Supercritical West: For SCCO2 capital costs can snow cleaning, part size is a determin- Carbon Dioxide. be a concern, but operating costs are ing factor in cleaning time. For small usually significantly lower than other parts, CO, snow can be quicker than New EAP Member processes due to no energy input to solvent or ultrasonic cleaning; a sim- Not only did Dr. Robert Sherman serve isolate the contaminants from the ple, fast on-off cycle will work if the on this CO, panel, he has now cleaning fluid. Maintenance costs are part is supported and amenable to an become a member of the Precision dependent on level of automation (the in-line process. For larger parts, clear- Cleaning Editorial Advisory Panel. The more automated a system is, the ing times per part can be comparable author of over 30 papers and holder higher the maintenance costs) and for in-line individual situations. Even- of several patents, Dr. Sherman the design and construction of the tually, part size can dictate the need obtained a BS in Physics from The system (build a robust system, main- for the aqueous cleaning methods Cooper Union and an MS and PhD in tenance costs will be low). unless there are special circumstances Material Science from the University Sherman: Costs for CO2 snow sys- for one-time parts of unique geometry. of Illinois.

July/August 1999