In-fab techniques for baselining implant dose, contamination
08/01/2002
Overview
Ion implantation in production wafer fabrication requires on-going qualification of several process parameters. Fortunately, as ion implantation has evolved, so have the required analytical techniques. Today's best analytical methods, particularly SIMS, TXRF, and VPD-ICPMS, are being judiciously used for baseline qualification and periodic checks of dopant quantification, cross-contamination, and metallic contamination on each production ion implanter.
Any process engineer responsible for ion implantation needs to qualify implanted dopant dose, long-term repeatability, dopant cross-contamination, metals contamination caused by sputtering from tool components, projected range and straggle, and energy contamination. Traditionally, ion implantation has been characterized using secondary ion mass spectrometry (SIMS). Today, we use SIMS along with other techniques and a combination of measurement procedures and more accurate standards to provide faster turnaround and better sensitivity.
As-implanted dose can be measured with SIMS using comparison to standards, typically ion implanted samples of the same matrix as the sample under investigation. SIMS is also used to measure cross-contamination when more than one dopant is implanted on a single tool. The area of analysis (~50 x 50μm) and depth resolution (down to <1nm) of SIMS is sufficient to provide range and straggle data, to check for energy contamination, and to measure ultra-shallow implants.
Use of surface SIMS, where a sample is exposed to oxygen in addition to the primary beam during analysis, provides a valid measurement, starting at the surface, that eliminates effects due to surface oxide and equilibration depth due to primary ion penetration into the sample.
Contaminants, typically metals, can be measured using SIMS or total reflection or low angle of incidence x-ray fluorescence (TXRF), and vapor phase decomposition inductively coupled mass spectrometry (VPD-ICPMS). TXRF does have limited sensitivity when analyzing boron (B). VPD-ICPMS can measure to 1 x 108 atoms/cm2 and can analyze an entire wafer surface. The high sensitivity and accuracy of SIMS make it an essential tool to complement TXRF by analysis of low-z elements and isotope 98 of molybdenum (98Mo) in boron difluoride (BF2+) ion-implanted wafers.
TXRF is limited to 4nm analysis depth, but contaminants Pb, Cu, Na, and Al are typically expected near the wafer surface because they originate from low-energy sputtering of implanter surfaces. TXRF should be adequate for these elements as long as correlation checks are made, especially for low atomic number elements. TXRF provides the speed to analyze the quantity of wafers in a fab that need to be examined for surface contamination.
TXRF does need improvements in wafer reference standards. Round robin studies have shown variation up to six times [1]. While TXRF reference wafers have been typically prepared using spin coating of an aqueous chemical solution, a new method uses ion implantation through a removable layer that puts the peak of the implant at the interface between the layer and substrate [4-6]. When the layer is removed, the surface concentration is reliably known. This approach provides a more repeatable quantification method.
TXRF and VPD-ICPMS can measure larger numbers of elements than SIMS, but cannot detect elements below the surface. While SIMS traditionally requires analysis of a specimen cleaved from the wafer, TXRF and VPD-ICPMS are performed on whole wafers. We have seen the latter techniques evolve so they provide the fastest turnaround time and analyze the largest area [5-9], but latest-generation SIMS tools can now analyze whole wafers. VPD-ICPMS has better throughput than SIMS and TXRF. VPD-ICPMS also provides better sensitivity and a more complete check because the entire wafer surface is analyzed. Any contamination not in the oxide is missed, however.
Dopant-dose measurements
Established sheet resistance (Rs) and Therma-wave (TW) techniques provide routine dose monitoring, fast turnaround, and good monitoring for dose drifts, but they cannot be generally applied for characterization of absolute dose or contaminants.
The current method to set a dose on an implanter compares Rs and TW values of control wafers to wafers implanted on a standard tool, thus setting an initial correction factor. These factors are further adjusted after split-lot analysis on product wafers by monitoring current-voltage parameters sensitive to small dose variations [10]. This method does not determine absolute dose for each species implanted.
Once implanters are matched to each other using this method, in-line Rs and TW dose measurements provide information about specific recipe performance over time and generally can provide relative dose repeatability performance within a few percent. For critical processes, capacitance-voltage measurements on control wafers may be used to check implanter performance more accurately, although this method is expensive and used infrequently. Initial calibration of all in-line techniques to absolute dose is typically done using SIMS.
During the initial qualification of an ion implanter into manufacturing, we may also perform an initial dose calibration check using SIMS, but industry standards for SIMS are only now becoming available. In addition, the analytical protocols needed to achieve high-precision requirements for dose measurement of typical implanted species (i.e., B, As, and P) have not been widely published.
To determine absolute dose with SIMS, standards must exist for each dopant, and analytical protocols must be specifically identified to achieve high-precision measurements. NIST standard reference materials for B (SRM 2137) and arsenic (As, SRM 2134) are available, but only a round-robin phosphorus (P) standard exists. In addition, several different factors have been shown to affect the precision of SIMS measurements, such as the sample holder, dopant and matrix species monitored, and instrument conditions [11, 12].
Strict adherence to these factors has led to analytical protocols for the best measurement conditions for B [10] as well as As and P [13]. Precision measurements can now be controlled to ±5% for these three dopants. The same protocols used to measure intended dose may also be used to determine levels of dopant contamination.
Ion implanter baselining
We have established a baselining protocol for qualification and periodic checks of dopant quantification, cross-contamination, and metallic contamination on each of Agere's production ion implanters. First, we qualified an implanter, then we followed up every six months or after major maintenance.
 |
For implanters used in production for three different dopants, our procedure begins by implanting a sequence of four wafers, with an Ar purge between each implant (Table 1). The sequence is important because it permits all cross-contamination possibilities to be checked. High dose implants (e.g., 1 x 1016 atoms/cm2) provide good detection of contaminants. We use lower doses (e.g., 5 x 1014 atoms/cm2) for medium current implanters and for dopant dose checks. While we initially made all measurements with SIMS, our present procedure uses TXRF for metals checks, except for 98Mo in BF2 implants, which is present mostly below the surface. We also use SIMS to measure dopants and dopant cross-contamination.
 |
Metal and cross-contamination measurements from two different high-current implanters are included in Tables 2 and 3. The data in Table 2 is typical for a solid aluminum-disk system, and the one in Table 3 is for a spoke-wheel system. Both sets of results were captured prior to releasing tools into manufacturing. Metal contamination results were not observed to degrade significantly over time or to change after major maintenance events. However, cross-contamination results on both solid-disk and spoke-wheel systems were clearly affected as a function of major maintenance events such as semi-annual or annual PMs.
 |
These data have shown us that for solid-disk systems, attempting to extend the disk replacement-refurbishment cycle beyond what is recommended needs to be weighed against the effect of cross-contamination values in the order of 1-10% of the implanted dose. This was particularly noticeable (~10%) for As cross-contamination on BF2 implants. We now replace solid disks yearly with grace periods of only a few weeks. With the exception of Al contamination, our data showed no major difference between several solid-disk and spoke-wheel implanters.
In general, metal contamination levels detected in these ion implanter baseline studies have not been a problem with our current 0.16μm device technology. Mo contamination in BF2 implants remains a potential concern, however [14].
 |
Reducing Al contamination
Because we observed Al in relatively high concentrations for standard solid disk implanters, we compared it with a silicon-coated solid disk (Table 4). We observed a reduction in Al by two orders of magnitude with the latter, which remained low throughout the disk's lifetime. However, the data also show that species cross-contamination levels increase with time, as in the case of standard solid disks [15, 16]. It is likely that as device dimensions shrink, the importance of minimizing exposed Al surfaces will increase; there is certainly a need for suppliers to make this technology more affordable.
Dedicated implanters
Others have shown through cycle-time studies and process simulations that implanter dedication, beyond that imposed by the nature of the implanter itself, comes at the cost of manufacturing speed [17]. Generally in the industry, species tool dedication is not favored unless process margin or device tolerances dictate it.
 Figure 1. Phosphorus cross-contamination in BF2 annealed implants before and after Ar beam cleaning. |
Until recently, we had successfully avoided species dedication with our implanters, but with the introduction of highly doped phosphorous gate implant process steps into manufacturing, we observed a steady drift in both As and B sheet resistance controls. These shifts were a direct result of phosphorus surface cross-contamination. The problem was so persistent that only extended Ar beam cleaning resulted in baseline recovery for cross-contamination results (Fig. 1) [18].
 Figure 2. BF2 sheet resistance control chart for a spoked-wheel ion implanter before and after species dedication. |
Figure 2 is a BF2 sheet resistance (Rs) control chart for a spoked-wheel ion implanter showing variability before and after we implemented species dedication (i.e., primarily routing P implants through one implanter). We also examined P, as measured with SIMS, for baseline performance and cross-contamination conditions on the same tool before and after P tool dedication (Fig. 3).
 Figure 3. Phosphorus cross-contamination before and after P tool dedication. |
While we did try a variety of options aimed at controlling P cross-contamination, including P run limits, modification of control wafer preparation to include a removable oxide layer, and Ar beam cleaning procedures, in the end we dedicated a single implanter to high dose P implants while retaining all-species capability across all other tools. This strategy has worked well in containing cross-contamination-related process check failures while meeting capacity and cycle time requirements.
Today, at Agere
We currently use TXRF analysis for surface metal contamination analysis except for 98Mo in BF2 implants. SIMS is still the technique of choice for dopant and cross-contamination quantification. Measurements can now be controlled to within 5% for As, P, and B dopants. Together, SIMS and TXRF analyses help us to understand and resolve implanter manufacturing issues, including control of Al contamination associated with disk-based implanters and P cross-contamination in B implants. We are continuing to evaluate the potential of VPD-ICPMS as a critical technique for full-wafer analysis, but work needs to be done to generate standards.
Acknowledgments
Additional authors include Ronald F. Roberts, a distinguished member of the technical staff (MTS) at Agere Systems. We thank Juan Becerro, Claudia Granger, Timothy Beatty, and Raymond Lo for their discussions about the implant matrix format and help in preparing and analyzing samples.
Therma-wave is a trademark of Therma-Wave Inc.
References
1. A. Schnieder, et al., J. Vac. Sci. Technol., B 14, 4, 1996.
2. F.A. Stevie, et al., J. Vac. Sci. Technol., B18, 483, 2000.
3. M.A. Decker, et al., U.S. Patent Number 6,121,624, Sept. 19, 2000.
4. D.F. Reich, et al., Proceedings of the Twelfth International Conference on Secondary Ion Mass Spectrometry, Elsevier, p. 425, 2000.
5. P. Chu, et al., J. Vac. Sci. Technol., B15, 1908, 1997.
6. R.S. Hockett, P.K. Chu, ICMPC Proceedings, p. 563.
7. P.K. Chu, et al., J. Vac. Sci. Technol., B14, 3321, 1996.
8. P.K. Chu, R.S. Hockett, Semiconductor Int., 6, 142, 1994.
9. M.A Douglas, P.J. Chen, Surf. Interface. Anal., 26, 984, 1998.
10. F.A. Stevie, et al., Proceedings of the Eleventh International Conference on Secondary Ion Mass Spectrometry, John Wiley and Sons, p. 1007, 1998.
11. P.K. Chu, et al., J. Vac. Sci. Technol., B14, 3321-3326, 1996.
12. C.M. Jones, F.L. Morgan, Proceedings of the Eleventh International Conference on Secondary Ion Mass Spectrometry, John Wiley and Sons, p. 719, 1998.
13. P.H. Chi, et al., submitted to J. Vac. Sci. Technol., 2001.
14. J.L. Benton, et al., Journal of the Electrochemical Society, 146, 1929, 1999.
15. Al-Bayati, et al., Conference on Ion Implantation Technology, Kyoto, Japan, June 22-26, 1998 (available through Applied Materials Inc.).
16. J. Bake, et al., Nucl. Inst. Methods in Physics Research, B96, 56, 1995.
17. H. Glawischnig, 2000 International Conference on Ion Implantation Technology Proceedings, IEEE, Sept. 17-22, 2000.
18. T. Morita, et al., 2000 International Conference on Ion Implantation Technology Proceedings, IEEE, Sept. 17-22, 2000.
Ramon Santiesteban received his BS and MS degrees in electrical engineering from the University of Texas at El Paso. He is a member of the technical staff (MTS) at Agere Systems, 9333 S. John Young Pkwy., Orlando, FL 32819; ph 407/371-6785, fax 407/371-6755, e-mail: [email protected].
Jennifer McKinley received her BS in chemistry and MS in materials science and engineering from the University of Central Florida. She is an MTS at Agere Systems.
Fred Stevie received his MS in physics from Vanderbilt University. He is a senior researcher at North Carolina State University.
Phillip Flatch received his AST from PTI. He is an MTS-1 at Agere Systems.
Osvaldo Rodriguez received his AS in electronic and instrumentation engineering from Bayamon's Regional College, University of Puerto Rico. He is an MTS-1 at Agere Systems.