Purging FOUPs that open to front-end minienvironments using an inert-gas curtain

Keyvan Keyhani, Sameer Abu-Zaid, and Haifeng Zhang, Asyst Technologies

Purging front-end unified pods (FOUPs) with an inert gas such as nitrogen or with clean dry air (CDA) removes airborne molecular contaminants that can potentially damage integrated circuits and significantly impact manufacturing yields. Immersing silicon wafers in an inert environment has been proven to enhance certain IC manufacturing processes.^1,2 That method has also been used to protect wafers in pods or FOUPs that are at rest on storage racks or in process queues.

Previous experimental and computational studies have shown that a closed FOUP can be effectively purged using a continuous flow of nitrogen through ports on the lower surface of the FOUP.^3,4 When the door of a fully purged FOUP is opened to the front-end minienvironment, airflow from the minienvironment enters, increasing oxygen and moisture concentrations inside the pod. This occurs even if the flow of inert gas into the FOUP inlet ports continues after the pod door has been opened.

One solution to this problem is to purge the entire minienvironment with inert gas. Depending on the size of the minienvironment, the purge process may require the introduction of large amounts of inert gas or CDA into the front-end environment using a single-pass or a recirculating system. However, that option is very costly and requires significant changes in the design of the entire front-end system.

An alternative solution is to introduce a downward flow of inert gas into the front-end minienvironment, in front of the opened FOUP. A small plenum can be placed on the inner surface of the minienvironment front wall, above the FOUP opening. This creates a flow of inert gas and forms an inert-gas curtain between the pod and minienvironment, reducing or preventing the flow of front-end air into the FOUP.

In the study described in this article, computational fluid dynamics (CFD) models were developed to determine the flow of air from a front-end minienvironment into a FOUP and the effectiveness of FOUP purging with and without an inert-gas curtain. In the model, the following parameters were varied: the cross-sectional area of the plenum outlet (achieved by changing the plenum depth), plenum outlet velocity, and the FOUP inlet-port flow rate. The rate at which the oxygen concentration in the pod increases after the door is opened without the presence of a nitrogen curtain was also determined.

Computational Method

Minienvironment Configuration. Configurations of the minienvironment and FOUP used in this study are shown in Figure 1. The system consisted of a single-loadport front-end minienvironment and a single FOUP that opened to the minienvironment. The only feature inside the front-end minienvironment was the robot plate, which had a porous surface and offered specified flow resistance.

Figure 1: (a) Minienvironment configuration with 2-in.-deep plenum (green area) located 2 in. above the FOUP opening on the front surface of the minienvironment (gray area). The minienvironment inlet is purple, the outlet is red, and the robot plates are yellow; and (b) configuration of the FOUP. The inlet and outlet ports on the bottom surfaces are shown in blue and green, respectively.

Three different configurations were modeled. Two of the them, as illustrated in Figure 1a, included a plenum with a width of 16.5 in. and a height of 4 in. that was attached to the front wall of the enclosure 2 in. above the FOUP opening. In one of the two models, the depth of the plenum was 1 in., providing a flow outlet area of 16.5 sq in.; in the other, the depth was 2 in., providing an outlet area of 33 sq in. The third configuration did not contain a plenum, simulating a typical front-end minienvironment.

Except for the removal of the door, the FOUP configuration used here was the same as that used in a previous study.³ As shown in Figure 1b, the 25-wafer FOUP had two inlet ports (illustrated in blue) and two outlet ports (illustrated in green) in its bottom surface. The flow of inert gas could be simulated through the pod's inlet ports or the plenum attached to the minienvironment.

To improve the cleanliness performance of 300-mm front-end systems, a 2-mm gap was included between the FOUP shell and the front wall of the minienvironment (the port plate of the front-end system).⁵ The outer boundaries of this gap were open to the ambient environment.

Process Parameters. Velocity boundary conditions were applied on the inlet (top surface) of the front-end minienvironment, the two inlet ports of the FOUP, and the outlet (bottom surface) of the inert-gas plenum. A uniform velocity of 90 ft/min was applied across the minienvironment inlet for all cases. Different values of velocity were applied on the FOUP inlet ports and the plenum outlet.

Gas flow was modeled for a binary nitrogen-oxygen mixture. On the minienvironment inlet, the oxygen mass fraction was set to 0.23 (i.e., the nitrogen mass fraction was 0.77). In the cases involving the use of a plenum, the oxygen mass fraction across the plenum outlet was set to 0 (i.e., pure nitrogen was used). In the cases in which purging took place through the inlet ports, the oxygen mass fraction was set to 0 across the FOUP inlet ports.

Differential pressure, with respect to ambient, was set to 0 at the outlet boundaries: the two pod outlet ports, the gap between the FOUP shell and the minienvironment, and the minienvironment outlet. Additionally, flow resistances were imposed on the minienvironment outlet to model pressure drops through the minienvironment louvers (corresponding to a 20% open area) and on the FOUP outlet ports to model pressure drops through the pod's port filters (the pressure drop was 0.5 in. for a flow rate of 30 L/min through each port).

A single transient case was simulated to determine the rate at which oxygen concentration in the FOUP increased when the nitrogen curtain was not present. For that case, a steady-state velocity field was first computed using a nitrogen flow rate of 30 L/min through each of the two FOUP inlet ports. That velocity field was used to solve the transient transport equation for oxygen and to determine the oxygen and nitrogen concentration fields over time. All other boundary conditions for that case were the same as in the other cases. At the onset of the test (time = 0), two oxygen mass fractions were used to solve the oxygen transport equation: 0 in the FOUP (i.e., the pod had been completely purged before the door was opened) and 0.23 in the minienvironment (i.e., the minienvironment contained air before the FOUP door was opened).

Computational Results

Steady-State Cases. For all simulated steady-state cases, the oxygen average mass fraction in the FOUP after the pod door was opened is provided in Table I. Parameters for the eight cases included plenum depth, velocity of the nitrogen flow exiting the plenum, plenum flow rate, and the flow rate of nitrogen that entered the FOUP from the pod's inlet ports (FOUP purge). The velocity of the nitrogen flow exiting the plenum was varied while the total nitrogen flow rate was kept constant. That change was accomplished by changing the plenum depth and, therefore, the plenum outlet area. For case 4, the FOUP inlet and outlet ports were closed.

In case 2, which did not use a nitrogen curtain, the flow rate through the inlet ports was 60 L/min (30 L/min through each port), with an average steady-state mass fraction of oxygen in the FOUP of 0.133. Case 8, which had the highest plenum nitrogen velocity (140 ft/min) and the highest FOUP inlet port flow rate (60 L/min), resulted in the lowest FOUP oxygen concentration (0.00489).

For cases 7 and 8, which involved the same total plenum flow rate of 452 L/min and FOUP inlet port flow rate of 60 L/min, an increase in the plenum nitrogen velocity from 70 to 140 ft/min decreased the average oxygen mass fraction in the FOUP from 0.00827 to 0.00489, a 41% reduction.

The difference in average oxygen mass fraction between cases 5 and 6, where the FOUP inlet port flow rate doubled from 30 to 60 L/min at a plenum nitrogen velocity of 70 ft/min and plenum flow rate of 226 L/min, was smaller than the average oxygen mass fraction that resulted from doubling the plenum nitrogen velocity from 35 to 70 ft/min (cases 3 and 4). In the former instance, the average oxygen mass fraction decreased from 0.0385 to 0.0353 (~8% reduction), while in the latter, the average oxygen mass fraction decreased from 0.0871 to 0.0616 (~28% reduction). In contrast, without the use of the nitrogen curtain, doubling the FOUP inlet port flow rate from 30 to 60 L/min had a large impact, decreasing the average oxygen mass fraction in the pod from 0.176 to 0.133 (~24% reduction).

Figures 2 and 3 show contour plots of oxygen on planes parallel to the side walls (y-z planes) of the system for cases 2 and 8, respectively. Oxygen passed through the center of the FOUP (Figure 2a), 8 cm from the center (Figure 2b), and 12 cm from the center (Figure 2c), toward the center of the minienvironment (away from the minienvironment side wall). The 8-cm plane passed across the center of one of the FOUP inlet ports. The plots reveal that the oxygen concentration throughout the pod was higher for case 2, which did not use a nitrogen curtain, than for case 8, which did use a curtain. However, in both cases, high oxygen concentrations were detected immediately above the FOUP inlet ports, where pure nitrogen entered the pod. Inside the FOUP, relatively high oxygen concentrations were detected at the front (near the opening), bottom, and sides (adjacent to the side walls).

Figure 4: Contours of the velocity component normal to the FOUP opening plotted on the FOUP opening plane (a) for case 2 (no nitrogen curtain), where a negative Uz value indicates flow into the FOUP; and (b) for case 8 (nitrogen curtain, plenum velocity of 140 ft/min), where a negative Uz value indicates flow into the FOUP. (The right side of the figure represents the middle of the minienvironment.)

Figures 4a and 4b present contours of the component of velocity (U_z) normal to the pod opening, plotted on the FOUP opening plane for cases 2 and 8, respectively. Negative U_z values indicate flow into the FOUP from the minienvironment side. Figures 5a and 5b are velocity vectors on a y-z plane 12 cm from the center of the pod for these two cases. These figures show that flow from the minienvironment into the FOUP occurred primarily near the side wall of the pod toward the midplane of the minienvironment (the right side in Figure 4) and in the center regions of the FOUP opening. The plenum width, therefore, must be larger than the width of the FOUP to prevent the flow of minienvironment air into the pod near its sidewalls. The U_z contour plots also show that directional flow across the FOUP opening was similar whether or not a nitrogen curtain was used. However, when a 140-ft/min curtain was used, the flow into the pod consisted only of nitrogen from the plenum and no or relatively low levels of oxygen.

Transient Case. Figure 6 shows the average mass fraction of oxygen in the FOUP plotted as a function of time after the pod door was opened under the same boundary conditions as case 2. In this transient simulation, the flow of pure nitrogen into the FOUP through the pod inlet ports continued at 60 L/min after the door was opened. At the outset of the test, the mass fraction of oxygen in the pod was set to 0 (i.e., it was assumed that the closed FOUP was fully purged before the door was opened), while the mass fraction of oxygen in the minienvironment was set to 0.23. This simulation demonstrates the rate at which the concentration of oxygen in the FOUP increased after the pod door was opened when a nitrogen curtain was not used.

The average oxygen mass fraction in the FOUP increased from 0 to 0.0794 in 15 seconds (~60% of the steady-state of 0.133), 0.105 in 30 seconds (~79% of the steady-state value), and 0.122 in 60 seconds (~92% of the steady-state value). Since a FOUP door is typically open for more than several seconds, these results show that without a nitrogen curtain, there is sufficient time for the oxygen concentration to increase significantly as a result of convection and diffusion of minienvironment air into the pod.

Figure 6: Mean mass fraction of oxygen in the FOUP as a function of time after the FOUP door has been opened (at t = 0 seconds). The FOUP is continuously purged with nitrogen through its two inlet ports, with a nitrogen flow rate of 60 L/min. The steady-state value of mean oxygen mass fraction in the FOUP is 0.133.

Conclusion

The simulations discussed in this article resulted in the following findings:

• When a nitrogen curtain is not used and a FOUP is purged via the pod inlets only, airflow in the minienvironment migrates into the FOUP, increasing the oxygen concentration in the pod significantly within several seconds.

• Placing a plenum above the FOUP opening to create a nitrogen curtain decreases the flow of front-end minienvironment air into the pod, decreasing the oxygen concentration in the FOUP.

• Higher plenum velocity results in lower average oxygen concentration in the FOUP than lower plenum velocity for the same plenum volumetric flow rate (which can be kept constant by changing the plenum outlet depth).

• For simulated curtain velocities and flow rates, increasing the nitrogen flow rate through the pod's inlet ports decreases average oxygen concentration in the FOUP, improving purge efficiency when the door opens.

In this study, the lowest average oxygen concentration in the FOUP was obtained when the plenum nitrogen velocity was 140 ft/min, the front-end minienvironment airflow velocity was 90 ft/min, and the FOUP inlet port nitrogen flow rate was 60 L/min.

References

1. C Wiebe, SA Abu-Zaid, and H Zhang, "Benefits of Purging a 200-mm SMIF Pod with Nitrogen between Wafer Processing Steps" (paper presented at SEMI Technology Symposium, Semicon China, Shanghai, March 11–14, 2003).

2. S Ito et al., "Wafer Ambient Control for Agile FAB," in Proceedings of the ISSM 2001 (Piscataway, NJ: IEEE, 2001).

3. K Keyhani, SA Abu-Zaid, and H Zhang, "Modeling of FOUP Inert Gas Purge: Comparison of a Continuous Inlet Flow and an Alternating Flow between Two Inlets," Internal Report (Fremont, CA: Asyst Technologies, 2002).

4. H Zhang and SA Abu-Zaid, "Studies of N₂ Purge at Different Wafer and FOUP Conditions," Internal Report (Fremont, CA: Asyst Technologies, 2002).

5. H Zhang and SA Abu-Zaid, "Clean Manufacturing: A Parametric Study of Airflow and Airborne Particle Performance for a 300-mm Loadport," Advance Applications in Contamination Control (A2C2) 6, no. 8 (August 2003): 19–23.

Keyvan Keyhani, PhD, is a staff engineer in the corporate contamination control group at Asyst Technologies (Fremont, CA). His field of expertise is flow and heat/mass transfer modeling. He received an MS in chemical engineering from Columbia University in New York City and a PhD in bioengineering from the University of Pennsylvania in Philadelphia. (Keyhani can be reached at kkeyhani@asyst.com.)

Sameer Abu-Zaid, PhD, is a senior manager of the corporate contamination control group at Asyst Technologies. He has authored or coauthored several papers in scientific journals in the areas of turbulence, multiphase flows, and fluid flow modeling. In addition, he has given several presentations at semiconductor-related conferences. He received a PhD in mechanical engineering from Clarkson University in Potsdam, NY. (Abu-Zaid can be reached at 510/661-5460 or sabuzaid@asyst.com.)

Haifeng Zhang, PhD, is a senior engineer in the corporate contamination control group at Asyst Technologies, where he is responsible for airborne particle testing and ISO certification, as well as CFD modeling for minienvironment design and evaluation. He received BS and MS degrees in mechanical engineering from Fudan University in Shanghai and a PhD in mechanical engineering from Clarkson University in Potsdam, NY. (Zhang can be reached at 510/661-5094 or hzhang@asyst.com.)