New
06/01/2000
overview
A new segregated hydrofluoroether under development has a global warming potential <3% that of comparable perfluorocarbon fluids. With a boiling point of 128°C, this fluid is well suited for use in ion implanters, steppers, etchers, plasma-aided tools, and automated testers. Its heat transfer performance is shown to be superior to that of perfluorinated fluids of similar boiling point. Electrical properties permit its use in all but the most demanding dielectric applications. Material compatibility should permit "drop-in" substitution in well-designed PFC systems.
Phillip E. Tuma, Lew Tousignant, 3M Specialty Materials Laboratory, St. Paul, Minnesota
 Figure 1. The search for alternatives to PFC fluids is a balancing act. |
Perfluorocarbon (PFC) heat transfer fluids, such as Fluorinert, have been used for years in the semiconductor manufacturing industry to control the temperature of vital components in ion implanters, dry etchers, deposition tools, steppers, automated test equipment (ATE), and other machines. (They are also used to cool sensitive military electronics, supercomputers, fuel cells, lasers, x-ray targets, etc.) These "fully fluorinated" [1] fluids are most often used because they are excellent, very stable dielectrics. Many applications, particularly those involving electrostatic chucks or strong RF fields, require this property. Other applications make use of PFC fluids for the added margin of safety should the fluid accidentally leak on sensitive electronics. These fluids are colorless and odorless, evaporate cleanly, and possess excellent toxicological properties. Because they are among the most chemically inert compounds known, PFC fluids can be used in direct contact with almost any construction material, including sensitive electronics.
The same chemical inertness that makes PFC fluids so well suited for the aforementioned heat transfer applications also makes them very long-lived in the upper atmosphere. Atmospheric lifetimes for these fluids range from 400 to over 2000 years. This, coupled with their infrared absorption profiles, translates to very high global warming potentials (GWPs). Certain PFC fluids were among compounds listed in the 1998 Kyoto Protocol, the international agreement intended to reduce emissions of greenhouse gases.
The contribution of PFC heat transfer fluids (leaks, evaporation, etc.) to a fab's global warming emissions profile has been of secondary importance in recent years with attention focused instead on net PFC emissions (million metric tons of carbon equivalence) from plasma-aided manufacturing processes. While these plasma-aided processes have traditionally been the primary source of global warming emissions from fabs, proactive abatement programs and alternative chemistries have been very successful at reducing their impact. As the industry moves forward with its plans to reduce emissions from plasma-aided processes even further, the percentage of a plant's net PFC emissions attributable to perfluorinated heat transfer liquids will rise. Though fluid losses from many heat transfer systems can be reduced, use of low-GWP alternative fluids offers a more effective means of reducing the impact of these fluids.
 |
Hydrofluoroethers (HFEs) are compounds that contain hydrogen and fluorine on a carbon backbone that contains one or more ether linkages [2]. Because hydrogen is present on these molecules, they are more susceptible to breakdown by hydroxyl radicals in the upper atmosphere. As a result, HFEs generally have far shorter atmospheric lifetimes and lower GWPs than PFCs. Though the global warming potential tends to decrease as the hydrogen content is increased, other desirable properties such as dielectric strength, material compatibility, and nonflammability are compromised. An appropriate HFE, therefore, will possess a balance of properties (Fig. 1), a balance that favors the environment, but still allows it to be used safely in demanding dielectric applications.
 |
Though the stability, toxicology and atmospheric lifetimes of HFEs can vary widely [3], research has shown that segregated HFE liquids, as a class, have a desirable balance of properties. "Segregated" here refers to HFEs in which a perfluorocarbon segment of the molecule is separated from a fully hydrogenated segment by the ether oxygen (e.g., C3F7OCH3, C4F9OCH3, etc.). These molecules possess a balance of environmental, toxicological, and thermal stability properties that make them well suited for dielectric heat transfer applications.
One such HFE, C4F9OCH3, is already being used in ATE. More widespread use of this molecule has been limited by its relatively low boiling point of 61°C. Segregated HFEs with higher boiling points are being researched. The first of these to be commercialized is C7F15OC2H5. With a boiling point of 128°C and excellent low temperature properties, this molecule is well suited for "medium temperature" applications such as dry etcher cooling. As we show below, this molecule possesses thermophysical, electrical, toxicological, and environmental properties that ensure its utility for many years.
Properties
Table 1 lists properties of C7F15OC2H5, Fluorinert FC-3283, and Galden HT-135.
FC-3283 and HT-135 were chosen for comparison because they are PFCs with similar boiling points and are widely used in "medium temperature" applications in the semiconductor manufacturing industry.
We measured the liquid density of C7F15OC2H5 from 10-90°C using an Anton-Parr Model DMA58 densitometer with a DMA602 external cell. Densities below this temperature were obtained by linear extrapolation of the measured data. The specific heat was measured by differential scanning calorimetry according to ASTM E 1269-95 over a temperature range of -50 to 50°C. The kinematic viscosity was measured using a calibrated Canon-Fenske viscometer from -80 to 23°C. Viscosity data were regressed using methods outlined in ASTM Standard 341-77.
The thermal conductivity was measured at 23°C. Thermal conductivity values at other temperatures were extrapolated using this data point and the experimentally determined temperature dependence for the hydrofluoroether C4F9OC2H5. These values were determined using a transient, dual hot-wire thermal conductivity cell over the temperature range -50 to +50°C.
Safety
 |
PFC fluids like FC-3283 and HT-135 typically show no toxicological effects even at saturation levels in air. For this reason, exposure guidelines for these fluids are usually not set. Likewise, toxicity tests conducted to date on C7F15OC2H5 indicate this compound exhibits a very low order of toxicity. For instance, the acute oral LD50 for this compound is greater than 2000 mg/kg. C7F15OC2H5 is not classified as either an eye or skin irritant. The molecule tested negative in two mutagenicity screens — the chromosomal aberration assay and the Ames assay (a widely used bacterial screen that in conjunction with other screening tools ascertains the propensity of a material to cause cancer). In a 28-day feeding study, no adverse health effects were observed at 1000mg/kg body weight.
 Figure 3. Theoretical heat transfer data at a) 20°C and b) -20°C. |
All of the aforementioned fluids have similar, rather high boiling points. Their saturation levels in air are, therefore, very low (typically <2% by volume). Because the fluids are used in nearly hermetic systems, typical worker exposure levels are expected to be exceedingly low and limited to low level acute exposures during filling and draining operations.
None of the fluids discussed in this work have an open or closed cup flash point by conventional ASTM test methods.
 |
All of the fluids discussed in this work are non-ozone-depleting and do not contribute to the formation of tropospheric ozone formation. The global warming potential of C7F15OC2H5 is 210, less than 3% that of its PFC predecessors. It is less than 14% that of HFC-134a (GWP=1600 [4]), a common next-generation refrigerant. This same unique combination of properties (non-ozone-depleting, non-VOC, low GWP and safety) has earned the C4F9OCH3 molecule (GWP=390 [2]) the distinction of being approved for "use without restriction" under the US EPA's Significant New Alternatives Program (SNAP). This program regulates the use of new materials in such chlorofluorocarbon (CFC) replacement applications as solvent, refrigeration, and heat transfer. The spirit of the SNAP is reflected in a growing global trend to select engineering fluids with the lowest possible global warming potential. On this basis, HFEs such as C7F15OC2H5 should be viable products for many years to come.
 Figure 4. Theoretical and experimental pumping power data at a) 20°C and b) -20°C. |
The dielectric strength of C7F15OC2H5 is similar to those of the PFC fluids. The electrical resistivity, however, is several orders of magnitude lower. This reduction is a consequence of the hydrogen present on the molecule. As such, it can be thought of as a "price to pay" for enhanced environmental properties. However, because this resistivity is still an order of magnitude higher than the best attainable deionized water (typically 1.8x107ohm-cm), it is adequate for all but the most demanding dielectric applications.
C7F15OC2H5 is composed of a single molecule. FC-3283, like HT-135, is a fraction or "distillation cut." This means that these materials are actually composed of several molecules of differing molecular weight. Upon boiling (or evaporation) the vapor phase of such blends will be richer in the lower boiling, lower molecular weight constituents. As this phase leaves the fluid, the average molecular weight and viscosity of the remaining liquid is increased. The extent of fractionation depends largely upon the mode of leakage and width of the boiling point range. Gross, liquid phase leaks do not lead to fractionation, however. The primary mode of leakage is usually evaporation: air drawn into the expansion reservoir while the fluid cools (contracts) is later vented as the fluid warms. This vented air is typically saturated with fluid vapor. In practice, fractionation of such fluids is only important if the low temperature limit of the fluid is approached or if specifications for the process cannot tolerate slight shifts in viscosity.
Performance
PFC fluids are poor solvents for anything but other fluorochemicals. Since most of the plasticizers and additives used in the manufacture of elastomeric polymers (O-rings, etc.) are hydrocarbon-based, PFC fluids have little if any solvency for them. Because HFEs like C7F15OC2H5 are partially hydrogenated, they do have some solvency for these additives and possess an increased tendency to extract them from the elastomer during extended exposures. This, in turn, can cause shrinking or embrittlement of the elastomer. For this reason, elastomers should be chosen from among those that are not heavily plasticized. Most ethylene-propylene and butyl compounds fall into this category as do some polyurethanes. Fluorocarbon elastomers (Fluorel, Kalrez, Viton), though compatible with HFEs (e.g., no extraction), will swell just as they do with PFC fluids. All common hard plastics and metals are compatible with C7F15OC2H5, just as with PFC fluids.
The thermal stability of C7F15OC2H5 was compared to that of three PFC fluids via a fluoride ion formation rate study. In this study, samples of the fluid were aged at 150°C in Monel vessels for 16 hours. Half of the vessels contained deionized water and the others were dry. For samples containing DI water, fluoride ion was quantified using a fluoride ion specific electrode applied to the aqueous phase after aging. For the dry samples, the electrode was applied to a DI water extractant that had been equilibrated with the sample following the aging. Though the fluoride ion formation rate determined in this way cannot be related directly to the decomposition rate, it provides an indicator that is useful for comparing different fluids and temperatures.
Table 2 shows results of this testing for C7F15OC2H5 compared with three PFC fluids: Fluorinert Electronic Liquids FC-77 (bp 100°C), FC-3283, and FC-43 (bp 174°C). Fluoride ion formation rates are expressed as micro grams of fluoride ion per gram of fluid per unit time. The data show that fluoride ion formation rates for C7F15OC2H5 are very similar in magnitude to those of the PFC fluids.
The lower viscosity and higher specific heat of C7F15OC2H5 indicate that for a given temperature and heat transfer capacity, it should be a more effective heat transfer fluid. Theoretical and experimental data confirm this.
For the theoretical analyses, fully developed, steady pipe flow calculations were used to compare the four fluids flowing through a 0.477cm diameter smooth pipe. This diameter is equivalent to that used in the experimental portion of this work. An appropriate independent variable for comparing fluids in semiconductor applications is the thermal capacity (C) defined as
C = m c
Here m is the mass flow rate and c is the specific heat of the fluid. The fluid properties were evaluated at a constant temperature. This uniform temperature assumption is equivalent to a "no load" or adiabatic condition.
The heat transfer coefficient (h) was calculated from the definition of the Nusselt number (Nu),
 |
null
where d is the pipe diameter and k is the thermal conductivity of the fluid. The Nusselt number in the laminar regime (Re<2300) is constant at 4.36. In the transition and turbulent regimes (Re>2300), a correlation by Gnielinski [5] was used
 |
null
This correlation is valid for 2300<Re<5x106 and 0.5<Pr<2000 and these conditions were met in this study. The Reynolds number (Re) is defined as
 |
null
and the Prandtl number (Pr) as
 |
null
where m is the dynamic viscosity, u is the kinematic viscosity and r is the fluid density. The gradient of pressure along the pipe length (dp/dx) was calculated using relations for the smooth pipe friction factor (f)
 |
null
and the definition of the friction factor
 |
null
where V is the average fluid velocity
 |
null
The pumping power (P) per unit length (L) of pipe can then be calculated using
 Figure 2. Apparatus used for experimental pressure drop measurements. |
null
Experiments were run to verify the pressure drop predictions of the theoretical model detailed above (no heat transfer experiments were run). The experimental apparatus is shown schematically in Fig. 2. This apparatus was used to measure the pressure drop through a 0.477cm dia., insulated copper coil as a function of mass flow rate at various temperatures.
 |
null
Figures 3a and 3b show theoretical heat transfer data for 20°C and -20°C. At 20°C, heat transfer coefficients for C7F15OC2H5 are similar to those of FC-3283, but significantly higher than those of HT-135. At -20°C, heat transfer coefficients for C7F15OC2H5 are higher than those of HT-135 and FC-3283.
Figures 4a and 4b show experimental and theoretical pumping power data for 20°C and -20°C. At 20°C the pumping power requirement for HT-135 is about 30-35% higher than for C7F15OC2H5 or FC-3283. At -20°C the pumping power requirement for HT-135 is about 80-85% higher than for C7F15OC2H5 or FC-3283. Experimental data show good agreement with theory. Slight differences are attributed to secondary flows encountered in the curved coil which are not modeled by the correlation used.
Conclusion
The new HFE C7F15OC2H5 has a GWP of 210, less than 3% that of its PFC predecessors. Toxicological data obtained to date indicate that it exhibits a very low order of toxicity. Its dielectric strength and thermal stability are similar to those of PFC fluids. Its electrical resistivity is superior to the best attainable DI water. Material compatibility should permit drop-in or near drop-in substitution in well-designed PFC systems. Experimental and theoretical data show superior heat transfer and pumping power performance.
Acknowledgments
The authors thank Ben McGibbon and Takeshi Machida for their significant contributions to the experimental portion of this work.
Fluorinert and Novec are registered trademarks of 3M. Galden is a registered trademark of Ausimont Chemical. Fluorel is a registered trademark of Dynion. Kalrez and Viton are registered trademarks of DuPont. Monel is a registered trademark of Inco Alloys.
References
- PFC fluids are generally composed of a carbon backbone on which all available bonding sites are occupied by fluorine atoms. The backbone may be a simple carbon chain (i.e., C6F14), a carbon chain with one or more ether oxygen linkages (i.e. C3F7OC3F7) or an amine (i.e., N(C2F5)3).
- C4F9OCH3, sold commercially as 3M Novec Engineered Fluid HFE-7100, is an example of a hydrofluoroether with one oxygen or "ether" linkage.
- D.B. Bivens, B.H. Minor, "Fluoroethers and other Next-Generation Fluids," ASHRAE/NIST Refrigerants Conference, October 1997, pp. 122-134.
- World Meteorological Organization (WMO), Scientific Assessment of Ozone Depletion: 1998, Global Ozone Research and Monitoring Project — Report No. 44, WMO, Geneva, 1999.
- V. Gnielinski, "New Equations for Heat and Mass Transfer in Turbulent Pipe and Channel Flow," Int. Chem. Eng., 16(2), pp. 359-368, 1976.
Phillip E. Tuma received his BS and MS in mechnical engineering. He has worked for five years developing applications for fluorochemicals in the areas of semiconductor heat transfer and refrigeration. Tuma is a senior application development engineer in the 3M Specialty Materials Laboratory, 3M Center 236-2B-01, St. Paul, MN 55144-1000; ph 651/737-9895, fax 651/733-4335, email [email protected].
Lew Tousignant has over 20 years experience with fluorochemical materials. Tousignant is a senior application development engineer in the 3M Specialty Materials Laboratory.
Step Ahead with HFE
 Teradyne's J973 test system |
Finding a cooling fluid that can be used in the semiconductor industry is not a simple task. The coolant must have high dielectric strength, sufficient thermal properties, and most important, be safe for the end user. There are many traditional fluids available that meet some of these requirements, but few that satisfy all. For years, perfluorocarbon (PFC) fluids have been the coolant of choice, but with their higher global warming potentials, we believe this trend will probably change. Although not yet mandated or regulated, Teradyne's VLSI Test Div. decided to switch to the more environmental friendly hydrofluoroethers (HFE). When switching from the PFC fluid to the HFE, we discovered that the HFE required a much more robust plumbing design, especially in regard to elastomers. This design change was painful at first, but now that the engineering investment has been made and we understand how to build reliable HFE systems, we are better prepared for the future. Not only is Teradyne a step ahead if regulations do come into place in the semiconductor industry, we are glad to be doing what's best for the environment.
John Mahoney, mechanical design engineer, Teradyne, Agoura Hills, CA