Assessment of SDS3 Gas Adsorbent Increases Use and Safety

The deliverable capacity of adsorbent-based subatmospheric storage and delivery systems (SDS) for dopant gas is directly proportional to the adsorbent material’s storage capacity. The next-generation SDS3 uses a high-density adsorbent with 2-3× the deliverable capacity.
 
We will describe here how National Semiconductor Corp. (NSC) evaluated this new dopant gas delivery system and verified increased deliverable capacity, resulting in an implanter operating cost reduction of 10% and an 80-hour improvement in technician productivity, which also reduced hazardous gas handling time. Previously reported reductions in operating costs at NSC are attributed to increased conductivity of the subatmospheric gas delivery system and to prior upgrades in adsorbent technology. Together, these upgrades have resulted in a combined reduction of implanter operating cost of >30% over the past three years.

Evaluating adsorbed gas and delivery

Beta-site evaluations of new products are conducted prior to commercialization. A beta-level evaluation of phosphine (PH₃) SDS3 source gas was conducted at NSC. This included measurements of beam purity and stability, as well as cylinder lifetime. The evaluation’s purpose was to ensure consistency of the adsorbed gas and verify claims of increased deliverable capacity resulting from the new adsorbent material. Both attributes were independently evaluated and compared with historic data for SDS2, the previous generation of subatmospheric storage and delivery system.

The outcome of this successful evaluation is evidenced in continued improvement in productivity through reduced operating cost and increased technician efficiency. Calculation of this cost reduction, as well as verification of advertised claims, are based on the evaluation data and presented here.

Improving cylinder capacity

The SDS2 used 0.7 mm spherical carbon beads as an adsorbent. The use of beads was required by the physical design limitations of standard high-pressure, DOT-approved cylinders already in use. The total volume of adsorbent in each cylinder — and therefore of stored gas — was determined by the quantity of beads in the cylinder. The design goal of SDS3 was to overcome this restriction.

The SDS3 welded cylinder allows the use of adsorbent blocks of microporous carbon, similar in appearance to hockey pucks, to be stacked inside prior to the welded cap’s installation. This new method allows for more efficient cylinder use by reducing void space and increasing cylinder capacity utilization.

Before any new and different hardware or process material is used in production, it is qualified to assure compliance with manufacturing specifications. The qualification procedure is usually defined by the responsible engineer, and is specific to each manufacturer. This manufacturing assessment of SDS3 at NSC served as the qualification of this next generation of subatmospheric gas adsorbent technology. Because NSC uses several dedicated high-current implanters, phosphine was chosen for the evaluation since it would provide the most aggressive cylinder lifetime data.

Factors determining successful qualification were based on the differences between this and the previous generation — namely, the adsorbent material and gas cylinder, which together provide increased deliverable capacity. The primary concern for both the gas cylinder and adsorbent material was with the presence of impurities in the gas. The participants believe that any issues related to the safety of the new welded cylinder were addressed by the DOT certification.

Prior to performing any tests on product or test devices, preliminary data is required to ensure a high probability of success. Once the preliminary analysis is completed, production lots are run across several machines and implants for data comparison. Only after it has been determined that SDS3 is functionally identical to SDS2 can other factors (such as cylinder lifetime) be considered.

Preliminary analysis results

The easiest and most common method for verifying not only an implant dopant’s purity, but also a delivery system’s integrity is to perform a beam current spectrum at the implanter. The spectrum clearly showed that the SDS3 phosphine cylinder under test had no anomalous peaks, and matched exactly the spectrums of SDS1 and SDS2 cylinders.

SIMS was used to ensure that the dopant profile was identical to that from machines running SDS2. It also demonstrated that the dopant level was essentially the same. The comparison of SIMS profiles of SDS3 and two other machines running SDS2 and an SDS2 benchmark also showed no discernable difference (Fig. 1). TXRF was used to ensure that any wafer surface contamination was within acceptable levels. The comparison of surface contamination (Fig. 2 ) demonstrated acceptable levels comparable with those of a control wafer.

1. SIMS depth profile comparison of 31P from SDS3 to two SDS2 31P profiles and a standard.

2. TXRF results from a wafer implanted with 31P from SDS3.

Sheet resistance measurements were run to demonstrate that there was no appreciable difference in the long-term stability of the implanted dose. A comparison of 100 data points from a system running PH₃ SDS3 and one using PH₃ SDS2 revealed no difference in long-term dosing stability (Fig. 3 ).

3. Sheet resistivity results comparison of 31P from SDS2 and SDS3.

Once the preliminary analysis was completed, several production lots were split between the SDS3 and SDS2 systems. Both inline and end-of-line testing was performed and the results analyzed. Evaluation of key parameters such as poly sheet resistance and gate oxide integrity (Fig. 4 ) revealed no deleterious effects from the use of SDS3.

4. Probability plot of gate oxide breakdown comparing data from a lot split between two machines running 31P from SDS2 and one from SDS3.

Manufacturing considerations

All testing was done with cylinder connection plumbing designed and installed by NSC. The 0.25 in. diameter line size leading to the cylinder ensured efficient delivery of SDS gases.

The SDS3 welded cylinder has a different profile than the SDS2 high-pressure-style cylinder. The shoulder, rather than having a rounded profile, is a right angle caused by the welded cap. Although this presented no problem on the high-current implanter where SDS3 was initially tested, minor modifications were required for installation on medium-current systems. These included removal of a support designed to fit to the shoulder of the high-pressure cylinder and a minor repositioning of the gas cabinet latch.

To verify source stability, the ion beam energy’s standard deviation was reviewed over 40 wafer batches and compared with identical implants on a machine running SDS2; 40 batches covered ~800 minutes of implant time. The data showed that the SDS3 beam stability equaled or surpassed that of SDS2.

The SDS3’s primary goal is to provide increased gas capacity. To verify this in-creased capacity, daily pressure readings were taken over the life of each of three sequential cylinders. The three cylinders’ pressure was then averaged and compared with similar data from machines using both SDS1 and SDS2 cylinders. An excellent depiction of the evolution of SDS technology clearly shows the increases gained from each technology step (Fig. 5 ).

5. Cylinder depletion curves from SDS1, SDS2 and SDS3.

Lifetime comparison measurements of SDS1, SDS2 and SDS3 (Fig. 5) are 29, 56 and 108 days, respectively. A single system using SDS1 would have required 12 cylinder changes per year. The same system employing SDS2 required six, and with SDS3 will require three. Conversion from PH₃ SDS2 to SDS3 represents a reduction of six hours of hazardous labor per year per machine per PH₃ cylinder. Assuming similar savings for both arsine and boron trifluoride, the expected reduction is 60 hours of equipment technician time per year. Each packaging change has also brought a 10% reduction in the per-gram cost of the gas.

Since July 2003, NSC has aggressively pursued reductions in both implant gas cost and technician time. A redesign of the gas cylinder connection plumbing yielded a 20% increase in cylinder use with a corresponding decrease in cylinder changes and cost. Adsorbent changes have added an additional 20% cost reduction and a 25% reduction in cylinder changes.

At this writing, the original system used for testing produced 210,000 wafers using SDS3 source gas. The South Portland facility is converting all its systems to SDS3 phosphine, and has also qualified and installed cylinders of SDS3 boron trifluoride and arsine throughout.

The semiconductor industry has seen consistent pressure to reduce the cost of production consumables while maintaining quality and enhancing safety. This trend is expected to continue to drive similar, targeted reduction efforts.