by John Baxter, Bill Fitzgerald and James E. Gotch, P.E.
Trends in valve design for semiconductor applications
The stringent manufacturing requirements of the semiconductor industry place high demands on valve performance. Seal integrity is of the utmost importance for health and safety reasons, while contamination introduced into the system by the valve can result in reduced yields. High purity valves used in the manufacturing of semiconductor chips must be engineered with these, and other, industry requirements in mind.
A close look at valve design, as it relates to cleanliness, leads to two areas of concern; one being the potential for trapped gasses, and the other, the potential for particle generationboth a major source of system contamination. Historically, bellows valves were the valve of choice in the semiconductor industry (Figure 1). However, because of their inherent design, gases can be trapped within the many tiny convolutions typical of the bellows design. The result is a valve design that falls short of the cleanliness requirements necessary for high purity applications.
In contrast, high-purity spring-less diaphragm valves (Figure 2), introduced to the industry more than a decade ago, expose less surface area and offer fully swept flow paths and smooth transitions. These valves made a significant step forward in addressing the industry requirements for cleaner operating valves and have become the industry standard in most high purity applications. These valves have performed well, and few improvements have been made until recently. In fact, it has been a considerable challenge to make a valve cleaner.
Recent efforts of valve manufacturers to meet the increasing demand of chip fabs have resulted in the development of a new generation of ultra high-purity (UHP) diaphragm valves. Design objectives for these valves were to reduce manufacturing costs and simplify product selection by providing a valve for both low- and high-pressure applications. Additional objectives included improving flow capabilities, extending cycle life, providing integral lock-out / tag-out capabilities, higher temperature capabilities and improving the critical area of the valve seat.
Valve failures, while not common, have usually involved the valve seat. Consequently, the newest valves have been designed to improve contamination resistance, reduce particle generation, improve leak tightness and reduce the severity of other issues like valve seat swelling. From this standpoint, valve seats are a key to improving the performance of most UHP valves.
New design tools
In their quest to find a “cleaner” valve, fabs should look to valve manufacturers that enlist the latest technology in developing products that address their requirements. As stated earlier, the performance of the seat greatly impacts the overall performance of the valve. By using available technology, seat performance can be evaluated and designed to operate effectively in a semiconductor environment.
New computer-aided engineering (CAE) tools, including Computational Fluid Dynamics (CFD) and Finite Element Analysis (FEA), have helped valve manufacturers to shorten development time and improve valve performance. Computer modeling allows the engineering team to optimize the design before moving forward with a physical prototype.
Computational Fluid Dynamics is a computer modeling technique used to analyze fluid flow. CFD permits the designer to model different geometries using different system fluids. The results identify incremental pressure drops throughout the valve, fluid velocity and shear forces. Through CFD the valve's flow coefficient can be predicted very accurately. Additionally, the design can be optimized to maximize flow, minimize purge times and prevent damaging velocities over the seat.
Flow velocity is modeled in Figure 3. In this case, the different colors represent different velocities through the valve. Blue represents the slowest moving gas while yellow and orange represent the fastest moving gas.
These results indicate that this valve has a fully swept flow path as all areas in the flow path indicate movement of gas, documenting that this design offers little opportunity for entrapment and will facilitate purging. While the highest flow velocities are directly over the seat, the computer model allows engineers to ensure that the flows are not so high that they could damage or wear it away. CFD is key to optimizing the flow characteristics of new-generation valves, which today deliver flow coefficients that are 50 percent greater than previous UHP diaphragm valves.
Another CAE tool used to advance the development of UHP diaphragm valves is Finite Element Analysis (FEA). FEA can model several valve components of different materials through a range of motions and conditions. This technology allows designers to simulate complex assembly steps and model valve performance. The output results provide a time-stepped analysis of the stresses and deformation of each component. The results can be viewed as an animated movie file.
An application of FEA technology can be seen in Figure 4 where the stresses that arise from a plastic seat being staked (physically held in place) into a stainless steel valve body are illustrated. Obviously, the plastic has a much lower strength than the stainless steel, but the computer is able to monitor the stresses in both materials during this operation. The model assists in determining the final geometry of the seat and body, allowing the designers to optimize the seat design. Additionally, the geometry changes and stresses produced during the cycling of the valve can be predicted, ensuring reliable seat performance and long seat cycle life.
Another example of FEA analysis (Figure 5) illustrates how this CAE tool can be critical to ensuring long diaphragm cycle life. This still image shows a snapshot from the full analysis, which simulates stresses throughout assembly and operation of the valve. The designer can determine the minimum diaphragm stroke required to meet flow expectations, while avoiding high stress levels that can occur if the diaphragm is over-stroked. Excessive stress can result in a reduced cycle life and unwanted particle shedding. Additionally, this model can be run to simulate low and high system pressures, further allowing designers to optimize valve performance over the full operating range.
There is a direct correlation between the valve seat and the performance of the valve. Extensive research into the industry-standard poly(chlorotrifluoroethylene) (PCTFE) seat material has led to changes in the extrusion, machining, staking processes, and geometry of the valve seat. These changes resulted in dramatic improvements in valve-seat performance.
There are two types of leaks that can be encountered in diaphragm valve seats: The first is a “path leak,” which is characterized by an actual hole or gap between the sealing surfaces. This type of leak is detected as an instantaneous spike on the helium leak detector. The second type is a “permeation leak,” which occurs when helium molecules actually pass through the PCTFE material. Permeation leaks are characterized by a gradual increase in leak rate over time.
Once the distinction is made between leak types they can be addressed in a design depending on the desired performance parameters. Engineers strive to find a balance between leak-tightness and reduced stress, which in turn reduces the likelihood of particle shedding and can reduce the size of the actuator. By optimizing seat surface condition, seat geometry and seat containment, better performing, more reliable seals can be designed.
These three factors, however, must be balanced against the need to limit permeation leaks. The permeation rate of a PCTFE seal is influenced by many factors, but the most important is seal width, which is defined by the area that makes contact with the diaphragm, not just the width of the PCTFE seat. Seats with identical widths can have different permeation times depending on the force generated by the actuator and the seat's surface finishes. While a particular finish may shut off path leaks at a relatively low closing pressure, the permeation leak rate can be relatively high if only a narrow portion of the overall seat is contacted for sealing.
Like path leaks, permeation leak rates decrease as sealing force (stress) on the seat is increased. The higher the load, the more deformation occurs, thereby increasing the seal contact width and reducing the permeation leak rate. Load versus permeation testing has shown that after applying sufficient force to seal off path leaks, increasing the sealing force can continue to improve the HLT performance by increasing the sealing width. Once contact across the entire seat has been achieved, however, a further increase in the sealing force does not improve the permeation performance. Too much force generated by the actuator can actually over stress the seat leading to a reduced cycle life and or increased particle generation.
Extensive testing and CAE modeling have helped in the development of valve seats that surpass the SEMI F1-96 requirement for HLT performance, as shown in Figure 6. The SEMI standard states that a valve must pass HLT at 1 x 10-9 std cm3/s for at least 15 seconds. The best of the new generation valves deliver this performance for as much as 220 seconds. This kind of leak-tight performance can greatly simplify pump down of a system and shortening qualification times.
Other valve design issues
As we've seen, valve performance can be affected by many different factors and making clean valves cleaner requires analysis of many, sometimes inter-related, factors. Often a design change that improves performance in one area will actually diminish performance in another. Thus engineers must maintain a balance between the sometimes conflicting variables in order to achieve the optimum design. Below is a list of the typical variables that must be considered when establishing the performance requirements of a UHP valve:
Temperature. The mechanical performance of a PCTFE seat will degrade at higher temperatures, particularly at and above the glass-transition temperature range (140 degrees F to 200 degrees F). At these temperatures, the PCTFE molecules become more mobile and creep is more pronounced. Since users require PCTFE seats to function into and beyond the glass-transition range, FEA has been used to identify ways to prevent excessive seat deformation at these temperatures. By modeling seat constraints and stress levels under different conditions, designers have been able to determine more optimum combinations.
Contamination and damage. The size and location of a contaminant are the major factors that impact a PCTFE seat's resistance to contamination. If the contaminant is large enough to span the seal width, it is difficult to compensate for with additional sealing force. Smaller contaminants, however, can be “smashed” into the PCTFE seat so, while they may increase permeation to some degree, they no longer provide a leak path. Seat damage – a scratch, nick or dent – that does not cross the entire seal width, will not always provide a leak path. However, even small nicks and scratches that traverse the seal width can affect HLT performance. As the stress on the seat is increased, those smaller scratches can be deformed to the point where full width contact is again achieved. The stress required to accomplish this depends heavily on the depth, geometry and location of the scratching.
Swelling. Although PCTFE is compatible with many chemicals, some chemical agents can cause the seat to swell to the point where valve flow can be affected dramatically. Experiments have indicated that swelling is much more pronounced in the direction that the plastic rod (the material from which the seat is machined) was extruded. Incompatible fluids have been observed to swell the seat by more than 10 percent along the axis of extrusion, while swelling in the transverse direction is less than 5 percent.
New computer-assisted engineering tools have helped suppliers to make clean valves cleaner. Today's new-generation valves provide improved seat performance, cycle life and flow capabilities while reducing manufacturing costs. Research into materials, advancements in design and continuous improvement in manufacturing enables valve manufacturers to meet the semiconductor industry's evolving needs.
James Gotch, PE, has been with Swagelok for nine years and currently is a lead design engineer involved in development of high-purity diaphragm valves. He holds a bachelor's degree in mechanical engineering from Grove City College, and a master's degree in industrial engineering from Cleveland State University.
William Fitzgerald has worked at Swagelok for the pastE10 years as both a product manager and market specialist, responsible for marketing Swagelok high-purity valve products to the semiconductor market. He holds a bachelor's degree from Long Beach State University.
John Baxter has been with Swagelok 13 years. As Field Engineering Manager, he supervises a team of design engineers working directly with customers on the development and application of Swagelok fluid system components. He has a BSME degree from Cleveland State University.