Infrared Laser Scattering Defects in Silicon Heavily Doped With Boron

Sponsored Links

Infrared Laser Scattering Defects in Silicon Heavily Doped With Boron

Ed Wijaranakula, Ph.D.
Chief Technology Officer, Infotix Systems, Inc. - August 11, 2003

Page 1 of 1

It is well recognized that the crystallographic defects associated with oxygen precipitation in the bulk region of silicon wafers are important for the integrated circuit (IC) device manufacturing process because the bulk defects are effective gettering sites for metallic impurities. Transition metals which are not properly gettered from the active device regions could precipitate at the gate oxide region and contribute to IC device manufacturing yield loss.

In p/p+ epitaxial wafers having substrate heavily doped in boron, transition metals such as nickel, iron and copper can be effectively gettered from the epitaxial layer via either a segregation-induced gettering [1-2] or a relaxation-induced gettering in which the bulk defects act as precipitation sites for stable metal silicides [3-4]. A segregation-induced gettering, which

is driven by an enhanced solubility of metal impurities in a highly doped substrate region, is well understood [5]. Control of relaxation-induced gettering utilizing the bulk defects as the gettering sites, on the other hand, is a complex process because the oxygen precipitation in highly doped silicon behaves in an anomalous manner during the multistep IC device annealing process [6-9].

As pointed out by Brown et al [10], a relaxation-induced gettering in device wafers may still be needed in order to keep the transition metals from migrating back to the device active regions because the partitioning of transition metals such as iron to regions of highly doped boron becomes insignificant at temperatures > 400°C. Therefore, a reliable monitoring technique of the bulk defects in the epitaxial wafers during the IC manufacturing process is essential to ensure high yield, the number of good chips from each wafer.

IC device manufacturers, such as Intel (NASDAQ:INTC), use epitaxial silicon wafers in their microprocessor chip design because the epitaxial structure provides good protection against the latch-up. The latch-up is an undesirable phenomenon in which portions of an IC can be bypassed or shorts out during its operation. In analog or mixed signal IC designs, epitaxial substrates with a

FIGURE 1

FIGURE 2

low bulk resistivity underneath the high-resistivity epitaxial layer provides good isolation between circuits which cuts down on noise.

The bulk defects in epitaxial wafers can be determined using the transmission electron microscope (TEM), but the technique is time consuming and expensive. Infrared laser scattering tomography (IR-LST) is widely used for monitoring bulk defects in lightly doped silicon because the technique is rapid and highly sensitive to small defects [11-12]. Even though defects having sizes of several hundred Å can be detected by IR laser scattering due to a large difference in the refractive index between silicon and the defect, inherent problems still exist, such as free-carrier absorption resulting from heavily doping in epitaxial substrate wafer [13].

We examined the bulk defect density in epitaxial wafers, using TEM and laser scattering techniques, from two different p/p+ epitaxial wafer lots after an IC device manufacturing process. Figure 1 shows a plan-view TEM micrograph of the disk-shaped oxide platelets (DP) and dot-like defects (DD) in

epitaxail wafers from the first lot. From the TEM analysis, the average size and density of the DP was 200 nm and 3.5x10¹²defects/cm³, respectively. The density of the DD is ≈ 10¹³defects/cm³. Figure 2 shows a high resolution image of a dot-like defect with an average size of 10 nm. In addition to isolated DP, a low density of precipitate colonies are also observed. See Figure 3.

IR laser scattering tomography was carried out at locations adjacent to those examined by the TEM using a Mitsui MO-401 BMD analyzer with a Nd-YAG laser beam focused perpendicular to the polished surface. The result from the LST analysis shows that the sample contains only 1.3x10⁹ defects/cm³.

Figure 4 shows a plan-view TEM micrograph in epitaxial wafers from the second lot after an identical IC device manufacturing process. The average size and density of the DP are 80 nm and 5.8x10¹² defects/cm³, respectively. Because wafers from both lots used in this analysis have almost identical material specifications, the difference in the defect density and size is related to the substrate wafer origin. The IR-LST analysis indicates that the epitaxial wafer from the second lot contains only 2.4x10⁸ defects/cm³ which is four orders of magnitude lower than the precipitate density determined by the TEM.

FIGURE 3

FIGURE 4

Generally, the defect density measured from the laser scattering is at least three orders of magnitude lower than that from the TEM direct observation. Compressive strain introduced by the defects, could give rise to boron segregation in the matrix surrounding the defect and in turn decrease the laser scattering intensity. Dot-like (DD) defects having an average size of ≈ 10 nm may be too small to be detected by the IR-LST analysis. In order to achieve a high-yield IC device manufacturing process, the origin of epitaxial wafers becomes a critical parameter in controlling the defect density.

REFERENCES

[1] M. Sano, S. Sumita, T. Shigematsu, and N. Fujino, in Semiconductor Silicon 1994,
H.R. Huff, W. Bergholz and K. Sumino (The Electrochem. Soc. Softbound Proceeding Series, Pennington, NJ, 1994) p.784.
[2] H. Mikoshiba, N, Nishio, T. Matsumoto, H. Kikuchi, T. Kitano, and H. Kaneko, in
Defect Engineering in Semiconductor Growth, Processing and Device Technology, S.
Ashok, J. Chevallier, K. Sumino, and E. Weber, Editors (Materials Research Society,
Pittsburgh, PA, 1992) p.629.
[3] H. Ewe, D. Gilles, S. Hahn, M. Seibt and W. Schröter, in Semiconductor Silicon
1994, H.R. Huff, W. Bergholz and K. Sumino (The Electrochem. Soc. Softbound Proceeding Series, Pennington, 1994) p.796.
[4] W. Wijaranakula, X. Gao, K. Curtis, and H. Haddad, in Defect and Impurity
Engineered Semiconductors and Devices, S. Ashok, J. Chevallier, I. Akasaki, N.M. Johnson, and B.L. Sopori, Editors (Materials Research Society, Pittsburgh, PA, 1995) p.309.
[5] D. Gilles, in Defect Engineering in Semiconductor Growth, Processing and Device Technology, S. Ashok, J. Chevallier, K, Sumino, and E. Weber, Editors (Materials Research Society, Pittsburgh, PA, 1992) p.917.
[6] W. Wijaranakula, J.Appl.Phys, 72, 4026(1992).
[7] W. Wijaranakula, J.Appl.Phys, 74, 3783(1993).
[8] W. Wijaranakula, J. Electronic Matr., 22, 105(1993).
[9] M. Hourai, E. Asayama, T. Ono, M, Sano and H. Tsuya, in High Purity Silicon IV, C.L. Claeys, P. Rai-Choundhury, P. Stallhofer, and J.E. Maurits (The Electrochem. Soc. Softbound Proceeding Series, Pennington, NJ, 1996) p.214.
[10] R.A. Brown, O. Kononchuk, I. Bondarenko, A. Romanowski, Z. Radzimski, and G.A. Rozgonyi, J.Electrochem.Soc., 144, 2872(1997).
[11] K. Moriya, K. Hirai, K. Kashima, and S. Takasu, J.Appl.Phys, 66, 5267(1989).
[12] K. Moriya and T. Ogawa, Philos.Mag, 44, 1085(1981).
[13] W. Spitzer and H.Y. Fan, Phys.Rev., 108, 268(1957).

Page 1 of 1

SPONSORED LINKS

About the Author: Dr. Ed Wijaranakula is presently the Chief Technology Officer at Infotix Systems, Inc. Prior to Infotix Systems, he has worked with Intel, Hewlett-Packard, Micron, Motorola and Texas Instruments and has held senior as well as managerial positions in semiconductor manufacturing companies. He has published over 80 technical papers and holds more than 12 U.S. and foreign patents.