Combining thermal waves and a signal-processing algorithm to characterize USJs

Alex Salnik, Lena Nicolaides, and Jon Ospal, Therma-Wave

The combination of ion implantation and thermal annealing, which are both silicon surface modification methods, is a key production process in semiconductor manufacturing. Thermal-wave (TW) technology measures surface modification (i.e., "damage") induced by ion implantation.^1–3 Based on modulated optical reflectance, which is also known as photomodulated reflectance, the technology is well established in wafer processing to characterize and control ion implant dose accuracy and uniformity.⁴

Advanced semiconductor fabrication uses low-energy ion implantation processes to achieve ultrashallow (<500-Å) implantation depths. Several types of annealing methods, including rapid thermal-spike anneals, laser annealing, and flash-lamp annealing, are known to minimize the diffusion of dopants, thus forming low-resistivity ultrashallow junctions (USJs). The implantation of B⁺ or As⁺ at low-energy, high-doping concentrations (~10²⁰ cm^–3) combined with rapid thermal-spike annealing (RTA) is currently the leading process used to fabricate USJs.

Potential process problems in forming USJ layers lie in both the implant and the anneal procedures. While it is relatively easy to create a shallow layer using implantation, keeping the USJ profile abrupt and close to the surface after anneal is challenging. Junction depth (X_j), profile abruptness, and resistivity (dopant activation) are the most critical parameters in the characterization of USJs. Nonuniformities associated with ion implantation and RTA or other types of anneal methods can result in residual damage areas on the wafer surface. Hence, monitoring is required to ensure process control.

Typically, destructive analytical methods such as secondary ion mass spectroscopy (SIMS), transmission electron microscopy (TEM), and spreading resistance depth profiling (SRP) have been used to analyze USJs. While these methods can provide detailed USJ profile information, they do so at the expense of turnaround times that are measured in days or even weeks.

The increasing need for a sensitive, rapid, and nondestructive technique for monitoring all key USJ profile parameters has led to a process that couples conventional TW methodology with a new signal-processing algorithm. This article focuses on that novel process, which uses the Therma-Probe, a TW-based metrology system with variable pump-probe beam offset from Therma-Wave (Fremont, CA). This system can be used to perform quantitative, simultaneous characterization of USJ depth and abruptness with one nondestructive measurement.

Development Approach

The use of the TW-based metrology technique for metal film and implant applications has been extensively described.^5–7 As illustrated in the schematic diagram in Figure 1, an intensity-modulated "pump" laser generates thermal waves in the surface of silicon. These waves propagate in the critically damped mode and interact with thermal features in the silicon. The laser also produces excessive electron-hole free carrier "plasma waves." The generation of both thermal and plasma waves by periodic heating causes optical parameters to vary periodically. These optical parameters are detected by measuring the modulated reflectance of an optical probe beam reflected from the sample surface.

Figure 1: Schematic diagram of the photothermal ion implant and USJ monitoring system, which is based on modulated optical reflectance and equipped with a pump-probe beam-scanning module.

The work discussed in this article used a pump-beam wavelength of 0.79 µm, a probe-beam wavelength of 0.67 µm, and a modulation frequency of 1 MHz. The beam diameter was 1 µm, and the pump-probe beam offset range was between 0 and 1 µm. The system is equipped with a mechanical stage that can perform area mapping across the wafer surface.

This study used a set of six 200-mm silicon wafers implanted with B⁺ ions at two different doses: 5 X 10¹⁴ cm^–2 and 10¹⁵ cm^–2. Implantation was followed by RTA at temperatures ranging from 950° to 1075°C. This set of wafers was studied using the TW technique. A second set, which underwent the same processing, was analyzed using SIMS. Table I compares TW with SIMS data for wafers processed using different implantation doses and anneal temperatures.

Figure 2 shows experimental SIMS concentration depth profiles obtained from all wafer samples. Different implantation doses and anneal temperatures resulted in quite dissimilar depth profiles. According to semiconductor industry standard practices, USJ depth is defined at the dopant concentration level of 10¹⁸ cm^–3. In Figure 2, the USJ depth for sample 4, represented by the dashed line labeled X_j, is ~240 Å. USJ profile abruptness is usually defined as the inverse slope of a concentration profile between the dopant concentration levels of 10¹⁹ cm^–3 and 10¹⁸ cm^–3 (represented by the two horizontal dashed lines in Figure 2). Profile abruptness is expressed in nm/decade, and lower abruptness values correspond to steeper profiles.

Figure 2: SIMS concentration profiles obtained for wafers 1 through 6. The vertical dashed line illustrates USJ depth at the dopant concentration level of 1018 cm–3 for wafer 4. USJ profile abruptness is measured between the dopant concentration levels of 1018 and 1019 cm–3.

The SIMS profiles in Figure 2 are not straight lines between the dopant concentration levels of 10¹⁹ cm^–3 and 10¹⁸ cm^–3. In addition, these profiles are noisy, especially below the 10¹⁸ cm^–3 concentration level. Therefore, experimental SIMS profiles were linearly fit between 10¹⁹ cm^–3 and 10¹⁸ cm^–3 to obtain reliable abruptness values.

USJ Depth Measurements

Previous studies have shown that the TW signal from surface-modified semiconductors has a very complicated physical origin.^8–12 Although important in a full theoretical treatment of the problem, the dynamics of thermal and plasma waves,^10–12 plasma-thermal coupling, and optical- and plasma-wave interference effects^8,9 lie outside the scope of this article. For the purpose of simplicity (and with a good degree of accuracy), the investigators assumed that, depending on sample properties (implantation dose, energy, and anneal parameters) and measurement conditions (modulation frequency and experimental geometry), the TW signal can be dominated by the thermal or plasma component and will behave differently in each case.

The detection technique used in a TW system provides two independent output components: TW in-phase (I) and quadrature (Q). The corresponding TW amplitude (A) and phase (Φ) are calculated as

It can be shown that in the case of USJ samples after anneal, the Q component of the total TW signal is driven primarily by plasma-wave-related effects, while the TW amplitude is more sensitive to damage-related phenomena. In a fully annealed USJ wafer, the plasma wave is quite sensitive to physical nonuniformities, such as the boundary between a highly doped USJ region and a relatively low-doped substrate. Therefore, it is reasonable to use the Q component for USJ depth measurements.

Figure 3: Correlation between USJ depth values obtained using the TW technique and SIMS USJ depth values obtained from the profiles in Figure 2. (Correlation R = 0.99.)

To obtain USJ depth, experimental Q data were measured at zero pump-probe separation for all six wafers and linearly scaled to USJ-depth values using a correlation table containing a set of coefficients specific for the dopant type, energy, annealing conditions, etc. Figure 3 shows TW junction depth values averaged across the surface of each wafer along with the corresponding SIMS data. The correlation between the TW and SIMS data was excellent (correlation coefficient R = 0.99) for USJ depths ranging from 250 to <500 Å. This high correlation was particularly remarkable because the TW data in Figure 3 represented USJ depths averaged across the entire wafer surface, while the SIMS analysis was performed only at a single central point on each wafer. TW and SIMS data for USJ depth are compared in Table I. It should be noted that the probe wavelength of 0.67 µm used in the TW-based metrology system provides better sensitivity to X_j below 500 Å than do alternative methods with longer wavelengths.

Figure 4 shows 2-D 21-point maps of USJ depths obtained by scanning the wafer surface (the depth scale is located to the right of each map). As can be seen in the maps, USJ depths vary significantly across the surface of each wafer. It has been found that TW system precision in USJ depth measurements is better than 1.5% (3σ), less than the difference in junction depths observed in the area maps.¹³ Among all the wafers studied, wafer 5 had the most homogeneous USJ depth distribution: more than 80% of its surface had junction depth variations of <5 Å. A radial pattern and bandlike patterns, which are indicative of nonuniform RTA heating, were clearly visible in wafer 3 and wafers 2 and 4, respectively. The nonuniform junction depths across the wafer surface observed in Figure 4 can be caused by small anneal temperature variations (amounting to a few degrees) or by local effects. Both problems should be taken into consideration.

Figure 4: USJ depth-uniformity maps obtained using the TW technique for wafers 1 through 6. (Depth values are in angstroms.)

As can be seen clearly in Figure 4, all scans exhibited maximum junction depths located close to the bottom of the wafer. This "hot-spot" phenomenon indicates that there were possible problems with spatial temperature uniformity during RTA. A single-point analysis method, such as SIMS, might not have detected this phenomenon.

At this stage in the study, the investigators concluded that TW technology measures an average USJ junction depth precisely and performs high-resolution depth distribution mapping across the wafer surface rapidly. They concluded that the method compares favorably with conventional but time-consuming and destructive single-point SIMS analysis.

USJ Abruptness Measurements

USJ profile abruptness has become an increasingly critical parameter as gate lengths have scaled toward 10 nm. Thus, the investigators developed a new technique to measure abruptness based on TW methodology.

As has been shown in previous work, the TW signal from surface-modified semiconductors is a measure of the total damage inflicted by ion implantation on the lattice of a semiconductor.¹¹ Although different species, in addition to implantation dose and energy, affect the depth and amount of damage, the TW signal detects universal behavior that is essentially based on total damage. When the Q and I components of the TW signal from a surface-modified semiconductor are plotted against each other, all data points are tightly distributed along a straight line on a Q-I plot.¹⁴ Q-I plots do not distinguish among species. In other words, different species, energies, or doses merely shift data points along the straight line, since these process conditions affect the degree of lattice damage.

Figure 5: Quadrature (Q) versus in-phase (I) component presentation of experimental results obtained for wafers 3 and 4. (For wafer 3, profile abruptness = 6.9 nm/decade, while for wafer 4, profile abruptness = 10.4 nm/decade.)

TW signals from USJ samples after anneal exhibit the same type of universal behavior as TW signals from materials damaged by ion implantation before anneal. Anneal temperature in the former case plays the same role as implantation energy or dose in the latter. It has been reported previously that all data obtained for USJ samples annealed at different temperatures fall along the same straight line.¹⁵

The investigators' new technique for measuring USJ abruptness includes two essential elements: it scans the pump-probe beam offset and analyzes the data in Q-I coordinates. Experimental Q and I data from recent work were taken at two different pump-beam offsets (0 and 1 µm) and plotted in the Q-I plane. The slope of the lines formed by the pair of Q-I points was extracted and converted into a USJ profile abruptness plot using a correlation algorithm based on the results of independent SIMS measurements.

For example, Figure 5 shows two sets of Q-I data obtained experimentally for wafers 3 and 4. It can be seen that the slopes between the pair of Q-I points measured with 0- and 1-µm offsets were quite sensitive to USJ profile abruptness. USJ profile abruptness was measured in the same way for all six wafers. TW and SIMS abruptness measurements are presented in Table I.

Figure 6: Correlation between USJ abruptness values obtained using the TW technique and SIMS USJ abruptness values obtained from the profiles in Figure 2. (Correlation R = 0.99.)

Figure 6 illustrates an excellent correlation (R = 0.99) between TW and SIMS abruptness data. Most wafers appeared to have relatively steep profiles, with abruptness values ~7.0 nm/decade.

Figure 7 presents a USJ abruptness uniformity map for wafer 3, whose profile abruptness had a gradient. The mean abruptness value of 6.84 nm/decade was in very good agreement with SIMS data taken at the center of the same wafer (6.7 nm/decade).

The TW technique's sensitivity to profile abruptness is believed to result from its lateral scanning of thermal- and plasma-wave fields. Because the unit has separate pump and probe beams, photothermal probing occurs closer to the surface of the wafer than in configurations in which the beams are together, revealing more information about USJ profiles. A full physical understanding of this phenomenon requires additional modeling studies, which are currently under way. The results of these studies, including how the TW method can be extended to measure the even steeper junctions used for 45-nm IC fabrication, will be presented soon.

It has also been found that only Q and I data presented in the Q-I plane are sensitive to USJ abruptness, while conventional TW amplitude and phase are not sensitive to this profile parameter.

Figure 7: USJ profile-abruptness uniformity map obtained using the TW technique for wafer 3. (Abruptness values are in nanometers per decade.)

Conclusion

The critical characteristics of USJs—junction depth and the abruptness of the doping profile—have become increasingly difficult to evaluate in production applications using time-consuming conventional methods such as SIMS, TEM, and SRP. Advanced IC manufacturing requires a rapid, nondestructive technique that can monitor all key USJ profile parameters. This study has shown that this goal can be achieved using the TW method and a new signal-processing algorithm that performs quantitative, simultaneous characterization of USJ depth and abruptness.

The work discussed in this article has demonstrated that TW technology measures USJ junction depth precisely and performs high-resolution depth-distribution mapping across the wafer surface in a short time, comparing favorably with destructive, single-point SIMS analysis. This study showed that data correlations between TW and SIMS are excellent for USJ depths ranging from 250 to 500 Å with a precision better than 1.5% (3σ). However, TW analysis is considerably more informative than SIMS analysis because TW data represent USJ depths across the entire wafer surface, while SIMS data typically represent a single central point on the wafer. This work also demonstrated an excellent correlation (R = 0.99) between TW and SIMS abruptness data. Using the TW technique, USJ and abruptness data can be obtained simultaneously.

Acknowledgments

The authors would like to thank Mira Bakshi and Kenneth Ritz of Therma-Wave for engaging in useful discussions and helping with sample preparation.

References

1. WL Smith, A Rosencwaig, and DL Willenborg, "Ion Implant Monitoring with Thermal Wave Technology," Applied Physics Letters 47, no. 6 (1985): 584–586.

2. WL Smith et al., "Ion Implant Monitoring with Thermal Wave Technology," Solid State Technology 29, no. 1 (1986): 85–92.

3. M Guidotti and HM van Driel, "Spatially Resolved Defect Mapping in Semiconductors Using Laser-Modulated Thermoreflectance," Applied Physics Letters 47, no. 12 (1985): 1336–1338.

4. A Rosencwaig et al., "Detection of Thermal Waves through Optical Reflectance," Applied Physics Letters 46, no. 11 (1985): 1013–1015.

5. L Chen et al., "Thermal Wave Studies of Thin Metal Films and Structures," in Proceedings of the 10th International Conference on Photoacoustic and Photothermal Phenomena, AIP conference proceedings vol. 463 (Melville, NY: American Institute of Physics, 1999), 368–371.

6. L Chen et al., "Meta-Probe: A New Generation Photothermal System for Thin Metal Films Characterization," in Proceedings of the 10th International Conference on Photoacoustic and Photothermal Phenomena, AIP conference proceedings vol. 463 (Melville, NY: American Institute of Physics, 1999), 102–104.

7. L Zhou et al., "Use of a New Thermal Wave Technology for Ultra-Shallow Junction Implant Monitoring," in Proceedings of the 1998 International Conference on Ion Implantation Technology (Piscataway, NJ: IEEE, 1999), 578–581.

8. J Opsal, "Thermal and Plasma Waves in Semiconductors," in Review of Progress in Quantitative Nondestructive Evaluation vol. 6B, ed. DO Thompson and DE Chimenti (New York: Plenum, 1987), 1339–1346.0

9. J Opsal, "Modulated Interference Effects and Thermal Wave Monitoring of High-Dose Ion Implantation in Semiconductors," in Review of Progress in Quantitative Nondestructive Evaluation vol. 8B, ed. DO Thompson and DE Chimenti (New York: Plenum, 1989), 1241–1245.

10. A Salnik et al., "Quantitative Characterization of Ion-Implanted Layers in Si," in Proceedings of the 10th International Conference on Photoacoustic and Photothermal Phenomena, AIP conference proceedings vol. 463 (Melville, NY: American Institute of Physics, 1999), 368–370.

11. A Salnik and J Opsal, "Quantitative Photothermal Characterization of Ion-Implanted Layers in Si," Journal of Applied Physics 91, no. 5 (2002): 2874–2882.

12. A Salnik and J Opsal, "Dynamics of the Plasma and Thermal Waves in Surface-Modified Semiconductors," Review of Scientific Instruments 74, no. 1 (2003): 545–549.

13. L Nicolaides, A Salnik, and J Opsal, "Therma-Wave System Precision and Repeatability in Ultra-Shallow Junction Depth Measurements" (to be published).

14. L Chen, J Opsal, and A Rosencwaig, "Characterizing Modulated Reflectance Signal from Ion-Implanted Silicon Wafers" (paper presented at the 9th International Conference on Photoacoustic and Photothermal Phenomena, Nanjing, China, June 27–30, 1996).

15. L Nicolaides, A Salnik, and J Opsal, "Nondestructive Analysis of Ultrashallow Junctions Using Thermal Wave Technology," Review of Scientific Instruments 74, no. 1 (2003): 586–588.

Alex Salnik, PhD, is a senior member of the technical staff at Therma-Wave (Fremont, CA). Before joining the company, he worked for Louis Pasteur University (Strasbourg, France), the Nuclear Research Center (Karlsruhe, Germany), and the University of Toronto. Salnik has 18 years of experience in photothermal science and the semiconductor industry and has authored more than 50 publications. He received MS and PhD degrees in solid-state physics from the Moscow Physics Engineering University in Russia. (Salnik can be reached at 510/668-2519 or asalnik@thermawave.com.)

Lena Nicolaides, PhD, is a research scientist in the R&D department of Therma-Wave. She has more than 10 years of experience in photothermal science and technology and has contributed to more than 20 publications. She received MS and PhD degrees in mechanical engineering from the University of Toronto. (Nicolaides can be reached at 510/668-2425 or lnicolai@thermawave.com.)

Jon Opsal, PhD, is vice president and chief technical officer of Therma-Wave. Before joining the company in 1982, he was a physicist at the Lawrence Livermore National Laboratory and research associate and assistant professor at Michigan State University in East Lansing. Opsal has more than 20 years of experience in the semiconductor industry and has authored more than 100 publications. In addition, he holds more than 60 patents. He received a BS in physics and mathematics from Eastern Washington University in Cheney and MS and PhD degrees in physics from Michigan State University in East Lansing. (Opsal can be reached at 510/668-2220 or jopsal@thermawave.com.)