Gas Delivery Upgrade Reduces Implanter Operating Cost
David B. Turnbull, National Semiconductor Corp., South Portland, Maine; Bob Brown, ATMI Inc., Danbury, Conn. -- Semiconductor International, 8/1/2004
|
The gas delivery system in many implanters using sub-atmospheric gas source materials reaches conduction limit before complete desorption of available gas occurs, resulting in premature replacement and return of gas cylinders. To obtain the specified, expected amount of gas, and therefore to achieve maximum cylinder lifetime, the delivery system components must have adequate conduction.1,2
The pressure drop across 5 feet of 1/8 in. tubing at 10 Torr is ~5 Torr compared with 0.5 Torr for the same length of 1/4 in. tubing (Fig. 1 ). A minimum of 1/4 in. diameter tubing is recommended. The cylinder connection pigtail originally fitted with this implanter consisted of ~12 in. of 1/8 in. diameter tubing. To increase conduction, we replaced it with 1/4 in. diameter flexible tube and modified the cylinder connection to facilitate cylinder installation. An implanter using SDS1 (safe delivery system) phosphine (PH3) gas was used for this evaluation.
Experiment
Prior to replacement of the pigtail, cylinder pressure data was obtained daily over the life of three cylinders (Fig. 2a ). We replaced the standard 1/8 in. diameter pigtail cylinder connection with 1/4 in. diameter flexible stainless steel tubing.
![]() |
| 2. Daily cylinder pressure for three cylinders was obtained with a 1/8 in. pigtail (left) and a 1/4 in. pigtail (right). |
After installation of the higher-conductance connection, we performed daily pressure readings over the life of three more cylinders (Fig. 2b). A comparison of the average cylinder lifetimes (Figs. 2a and 2b) indicates a 30% increase with the higher-conductance connection (Fig. 3 ). Total cost reductions are driven by this increase in cylinder lifetime.
The number of wafers run was also evaluated to verify that production demand did not have an effect on the usage rate (Table ).
During testing, it was noted that the flexible line was too tight. To make the connection easier, the rigid plumbing was modified to move the connection point further away from the cylinder. This allowed a more forgiving fit and easier installation.
Once the design was finalized, the modified rigid pipe section was fabricated at a local shop and the custom flexible section was ordered from a major component supplier. The total upgrade cost was about $600 per cylinder module.
Since this evaluation was completed in September 2003, it has been implemented on 14 SDS gas modules in seven implanters. There have been no failures of any type related to the gas delivery system, and cylinder installation is easier than with the original configuration.
Deliverable capacityThe amount of available gas in an SDS cylinder is based on the difference between the actual full pressure and hypothetical empty pressure. For example, the amount of PH3 gas available in a JY cylinder is based upon the difference between 211 g at 650 Torr, and 32 g at 20 Torr. Therefore, an implanter whose delivery system pressure differential is limited to 50 Torr will not be able to fully deliver all available material, and the cylinder will likely be returned with usable gas.
Because of different adsorbent capacities for different species, the effect is most dramatic with BF3 SDS cylinders. At a final pressure of 50 Torr, only ~60% of the deliverable BF3 has been used (Fig. 4 ); 40% of the deliverable BF3 will be returned unused.
About 30% of all SDS BF3 cylinders returned in 2003 were at a pressure greater than 20 Torr (Fig. 5 ). Some of these cylinders may have been returned prematurely for other reasons, but it is likely that most were the result of a lack of knowledge of exact cylinder contents or conduction limitations in the implanter gas delivery system.
However, while ~30% of the cylinders returned in 2003 were not fully utilized, ~70% were returned well below the hypothetical empty cylinder pressure of 20 Torr. This indicates a greater utilization of the SDS cylinder than is specified or expected, and therefore a corresponding lower cost per gram.
High gas box temperature also contributes to a lower return pressure. The temperature of some gas boxes is as high as 90°F during operation. At 78°F and sea level, an SDS gas cylinder filled to 650 Torr may reach 760 Torr, one atmosphere.3
The non-linear adsorption isotherm shows that there is more gas available at lower cylinder pressures where the slope is steeper (Fig. 4 ).
Therefore, it is especially important that the user pay particular attention to ensure maximum conduction of the gas delivery system to achieve most economical operation.
ConclusionsSDS source gas cylinder utilization was increased ~30% by a simple and inexpensive modification of the gas delivery system, and was done without sacrifice to safety or reliability. This delivery system upgrade was performed internally, but may also be available from the implanter OEM or a third-party manufacturer.
An additional utilization increase of 6% in PH3 was obtained subsequently by the conversion from SDS1 to SDS2. SDS2 cylinder capacity is increased by the use of carbon adsorbent that provides twice as much capacity and is offered at a lower cost per gram.
Increased cylinder utilization also provides a corresponding reduction in maintenance related to cylinder changes that has not been quantified for this report.
RecommendationsTo optimize gas cylinder utilization and therefore minimize operating costs, ion implanter gas delivery systems must be compatible with the type of source gas materials used. This requires that implanter manufacturers design their machines with this in mind. Engineers specifying these machines must also pay particular attention to gas source utilization in making their selections.
A flow control device, or a complete implanter, that is SDS-certified ensures the conduction requirements that correlate with the most economical usage of the SDS source gas materials have been incorporated into the design.4
| Author Information |
| David Turnbull joined National Semiconductor in 1996 as a member of the project team that built National's first 200 mm wafer manufacturing line. He is currently a staff engineer in the Operations Engineering Department. He has been involved in ion implantation technology for nearly 25 years. |
| Bob Brown joined ATMI in 1998 after retiring from IBM Microelectronics Division, and is currently senior marketing specialist for ion implant products. He has been involved in ion implantation technology for nearly 30 years. He has a B.S. in electrical engineering from Northeastern University. |
| References |
|
| Acknowledgements | ||
| The authors would like to acknowledge the efforts of the technicians and engineers at National Semiconductor in South Portland, Maine, who were involved in the implementation of the hardware and analysis of the data for this evaluation. | ||





