JP2683984B2 - Evaluation method for deep level defects in semiconductor crystals - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for evaluating deep level defects in a semiconductor crystal, and more specifically, D1-L in a photoluminescence (PL) spectrum of the semiconductor crystal.
The present invention relates to an evaluation method capable of evaluating Deep-Level defects due to oxygen precipitates in Si crystals that give Deep-Level luminescence such as ine, and thus evaluating heavy metal gettering ability.

[0002]

2. Description of the Related Art Photoluminescence (PL) in a silicon crystal is used as a means for identifying impurities and defects.
It was measured mainly at liquid helium temperature, and the PL spectrum of the Si crystal after oxygen precipitation (Fig. 1) shows that D1-Line
Deep-level luminescence is observed. In recent years, it has become possible to detect PL in silicon crystals at room temperature by increasing the sensitivity of detection technology [M. Tajima, T. Masui, T. Abe in "Se.
miconductor Silicon1990 "HR Huff, KG Barracl
ough, and J. Chikawa, Editors p994, The Electrochem
ical Society, Pennington, NZ (1990)], in the PL spectrum after the deposition of oxygen, band edge emission and deep-level emission are observed as shown in FIG. This Deep-L
It is known that evel light emission is D1-Line Deep-Level light emission under the heat treatment conditions (the final stage heat treatment temperature is 1000 ° C.) conducted by the present inventors.

Band-edge emission is emission when excess carriers excited by laser irradiation emit energy between bands and disappear, and deep-level emission is emission that occurs through deep levels between bands. Is. Therefore, Deep-L
The evel emission contains information on deep-level defects (optically active sites), and it is expected that information on the nature and concentration of deep-level defects can be obtained by analyzing the emission spectrum and intensity. To be done. Recently, it has been reported that these deep-level luminescence are caused by point defects and heavy metals acting on dislocations or precipitates. In addition, it has been reported that these emission intensities are affected by the amount of heavy metals decorated to dislocations and precipitates [V. Higgs, E.
C. Lightowlers, P. Kightley, Mat. Res. Soc. Proc.
Vol 163 p57 (1990); AR Peaker, Advanced Science
and Technology of Silicon Materials p436 (1991); A.
Berg, I. Brough, G. Lormer, AR Peaker, Semicon
d. Sci. Technol. 7A236 (1992)].

[0004]

On the other hand, internal gettering (IG) is a process for trapping heavy metal impurities in a strain field such as dislocations, so that deep-level emission causes dislocations to interact with point defects or heavy metals. Deep-Level if due to action
There is a possibility that PL can be used as an evaluation method of IG ability. However, there is little knowledge about the structure of deep level defects (deep-level defects) based on oxygen precipitates that give deep-level luminescence such as D1-Line in the PL spectrum of semiconductor crystals, and this structure is similar to the PL spectrum. Is not known as to what is relevant.

[0005] The present invention has been made in view of the above points,
Its purpose is D1-L in the PL spectrum of semiconductor crystals.
An object of the present invention is to provide an evaluation method capable of evaluating deep level defects (Deep-Level defects) based on oxygen precipitates that give Deep-Level luminescence such as ine, and by extension, gettering ability.

[0006]

The first method of evaluating deep level defects in a semiconductor crystal according to the present invention is the photoluminescence intensity (Deep-Level PL intensity) I _d due to Deep-Level in a semiconductor single crystal. And the band edge PL intensity I _b are obtained, and I _d / I _b ^x (where x is a value of 0.5 to 1.0) indicating the relative Deep-Level concentration is obtained to evaluate the deep level defect. It is characterized by

The second method of evaluating deep level defects in a semiconductor crystal according to the present invention is Deep-Leve in a semiconductor single crystal.
The photoluminescence intensity (Deep-Level PL intensity) I _d due to l, the band edge PL intensity I _b, and the number of oxygen precipitates per unit volume (BMD density: abbreviated as BMD in the following formula) are determined to obtain oxygen precipitation. Relative Deep-Level per item
The feature is that the deep level defect is evaluated by obtaining the concentration I _d / (I _b ^x · BMD).

The PL spectrum obtained by the room temperature PL measurement shows one deep-level emission and P with PL intensity I _d.
It consists of one band-edge emission with L intensity I _b , and its emission mechanism is much simpler than in the case of cryogenic measurements. Room temperature Deep-Level PL intensity I _d is not necessarily Deep-L
The evel concentration is not proportional to N _d , and generally I _d is Deep-Level
It is known that it is approximately proportional to the product of the concentration N _d and the carrier lifetime τ, and is represented by I _d ∝N _d τ (1). On the other hand, the band end PL
Is the recombination of excited carriers between the conduction band and the valence band, and its intensity I _b is proportional to the carrier lifetime τ to the n-th power (n is generally a value of 1 to 2) as in the following equation. To be done. I _b ∝τ ⁿ (2) The Deep-Level concentration N _d is actually necessary to evaluate the deep level density in the semiconductor crystal. From the above formula (1), the following formula (3) N _d ∝I _d / τ (3), and from the above formula (2), the following formula (4) τ ∝I _d ^{1 / n} = I _b ^x ( x≡1 / n) (4) (where x is a value of 0.5 to 1.0). From the formulas (3) and (4), N _d ∝I _d / I _b ^x (5) is obtained. That is, I _d / I _b ^x is the relative Deep-Lev
It can be seen that the el concentration is shown.

On the other hand, oxygen precipitate number per unit volume in the semiconductor single crystal: the amount obtained by dividing the I _d / I _b ^x in (BMD density abbreviated as BMD in the ^{_{formula) I d / (I b x}} · BMD) It can be seen that indicates the relative deep-level concentration per precipitate. Here, the BMD density is a value derived from the number of defects per unit area by microscopic observation after selectively etching a semiconductor crystal, or a value derived by an infrared tomograph method of observing defects using infrared scattering. . Relative Deep-Level Concentration I _d / I _b ^x obtained as described above
Relative Deep-Level Concentration I _d /
Since (I _b ^x · BMD) is an amount obtained from the Deep-Level emission intensity resulting from the interaction between dislocations or precipitates and point defects or heavy metals, it is an important index with which the gettering ability can be evaluated. .

The present invention was made based on the experimental facts described below. Initial oxygen concentration is about 18ppma
(JEIDA) P-type and N-type CZ-Si crystals with (450
℃ y hour) + (800 ℃ 4 hours) + (1000 ℃
Heat treatment was performed for 16 hours) (y = 0 to 16 hours). Relative to these samples by room temperature PL evaluation method
The Deep-Level concentration was obtained and compared with the average effective size of the precipitates obtained by the infrared scattering tomography method.

Under the above heat treatment conditions, in all the samples, the precipitation progressed until the interstitial oxygen concentration was close to the solid solution limit at the final heat treatment temperature of 1000 ° C. Therefore, the precipitation amount Δ [Oi] was 450 ° C. before the pre-heat treatment time. It became almost constant without depending on (y = 0 to 16 hours). However, as for the size and density of the precipitates of these samples, as shown in the infrared scattering image of FIG. 3 and FIG. 4, the relative value of the effective volume decreases with the increase of the 450 ° C. preheat treatment time, and the precipitate density is It turned out to be increasing. Here, the relative value of the effective volume was obtained as follows. Defects obtained from infrared scattering tomography 1
The scattering intensity S per piece is approximately expressed as S∝ (Δε ・ V) ² [Kazuo Moriya, Applied Physics Vol. 55]
No. 6, p542 (1986)]. Δε in the above equation is the scatterer and Si
, V is the volume of the scatterer. Under this experimental condition, the scatterers are oxygen precipitates and Δε is not expected to differ from sample to sample. Therefore, the relative volume was determined as V∝S ^1/2 .

On the other hand, as shown in FIG. 5, the result of room temperature PL measurement shows that as the preheat treatment time at 450 ° C. increases, the band edge P
L intensity I _b decreases and D1-Line Deep-Level PL intensity I _d
Increased. Therefore, the relative Deep-Level concentration I _d / (I
_b ) ^1/2 increases as the 450 ° C. pre-annealing (preheat treatment) time increases, as shown in FIG. From this result, even under the same Δ [Oi], the relative Deep-Level
It can be seen that the concentration increases as the density of the precipitate increases.

Next, in order to investigate the effect of size, the precipitate 1
The Deep-Level amount per piece was estimated as follows. P
The relative deep-level amount per unit volume obtained from L is I
_d / (I _b ) ^1/2 , which was divided by the number of precipitates per unit volume (BMD density). FIG. 7 is a result of plotting the amount of Deep-Level per precipitate thus obtained against the relative volume of the precipitate obtained by the infrared tomography method. It can be seen that as the size of the precipitate increases, the amount of deep level per precipitate increases.

Summarizing the above results, when the sizes of oxygen precipitates are different under the same oxygen precipitation amount (Δ [Oi]),
The amount of deep-level defects per individual precipitate increases with the size of the precipitate, but the total deep-level concentration increases as the size of the precipitate decreases. This fact is considered as follows. The formation of oxygen precipitates creates a strain field around them, causing defects such as dislocation loops (optically active sites) that contribute to deep-level emission. Considering the generated area as the shaded area in the schematic diagram of FIG. 8, the deep-level amount per precipitate increases as the size of the precipitate increases, and qualitatively agrees with the experimental fact of FIG. 7. on the other hand,
If Δ [Oi] is constant as in this experiment, Deep-Lev
It is considered that the total amount of el becomes larger when the size of the precipitates is smaller and the density is higher as in the schematic diagram of FIG. 8B, and the result of FIG. 6 can be qualitatively explained.

Deep-Level investigated by the evaluation method of the present invention
Defects are optically active sites such as dislocations induced by oxygen precipitates. The amount of optically active sites that induce oxygen precipitation is not only proportional to Δ [Oi], but is equal to Δ [Oi].
It has been found that the density and size of the precipitates are also affected under [Oi]. Further, it is considered that the amount of optically active sites is influenced by the form of the precipitate and dislocations generated around it.

On the other hand, considering the IG caused by oxygen precipitation, it is considered that the gettering site is more and the gettering ability is higher when the precipitate density is higher even if the Δ [Oi] is the same. It is considered that the gettering ability is also affected by the morphology of precipitates and dislocations generated around them. From the above, it is considered that the optically active site has the same property as the gettering site. Therefore, the Deep-Level evaluation method used by the present inventors is considered to be useful in evaluating the gettering ability.

[0017]

Next, the present invention will be described in more detail with reference to examples. Example 1 For P-type and N-type CZ-Si crystals having an initial oxygen concentration of about 18 ppma (JEIDA), (450 ° C. y hours) + (800
℃ 4 hours) + (1000 ℃ 16 hours) (y = 0 to 1
Heat treatment was performed for 6 hours. For these samples, the relative deep-level concentration was determined by the room temperature PL evaluation method and compared with the density and effective size average value of the precipitates determined by the infrared scattering tomography method. Under the above heat treatment conditions, in all the samples, the precipitation progressed until the interstitial oxygen concentration was close to the solid solution limit at the final heat treatment temperature of 1000 ° C. Therefore, the precipitation amount Δ [Oi] was 450 ° C. before the heat treatment time (y = 0 to 16 hours) and became almost constant.

FIG. 6 shows the pre-annealing time at 450 ° C. and the relative deep-level concentration I _d / (I _b ) ^1/2 (laser power: 20 to 30 mW on the sample surface, beam diameter: 100 μm).
In the range of m to 1 mm, x is 1/2. Further, FIG. 4 shows the relationship between the pre-annealing time at 450 ° C., the BMD density, and the relative volume. By comparing FIG. 6 and FIG. 4, the relative deep-level concentration is the same Δ
It can be seen that even under [Oi], it increases as the density of the precipitates increases. Also, the relative Deep-Lev calculated from PL
A value obtained by dividing the el concentration by the number of precipitates per unit volume (BMD density) was plotted against the relative volume obtained by the infrared scattering tomography method, and FIG. 7 was obtained. From Fig. 7, as the size of the precipitate increases, the Deep-Level per precipitate is increased.
It can be seen that the amount is increasing.

[0019]

According to the first method for evaluating deep level defects in a semiconductor crystal of the present invention, Deep-L in a semiconductor single crystal is used.
Photoluminescence intensity due to evel (Deep-Level PL
Intensity) I _d and band edge PL intensity I _b are measured, and I _d / I
By finding _b ^x , the relative deep-level as the concentration of optically active sites such as dislocations induced in the crystal
It is possible to evaluate the gettering ability by utilizing the fact that the optically active site has the same property as the gettering site by evaluating the deep level defect by obtaining the concentration. Also,
According to the second evaluation method of the present invention, the above I _d / I _b ^x is further divided by the number of oxygen precipitates per unit volume (BMD density) to obtain I _d / (I _b ^x · BMD). Since deep level defects are evaluated, optically active sites per precipitate,
The amount and characteristics of gettering sites can be evaluated.

[Brief description of the drawings]

Figure 1: Oxygen precipitation CZ- measured at liquid helium temperature
It is a typical PL spectrum figure of Si crystal.

FIG. 2 is a typical PL spectrum diagram of an oxygen precipitated CZ-Si crystal measured at room temperature.

FIG. 3 shows an infrared scattering tomographic image, (a) shows the case without pre-annealing, and (b) shows the case of pre-annealing at 450 ° C. for 16 hours.

FIG. 4 is a graph showing the relationship between pre-annealing time at 450 ° C. and BMD density and relative volume.

[Figure 5] Pre-annealing time at 450 ° C and Deep-Level P
It is a graph which shows the relationship with L intensity (I _d ) and band edge emission intensity (I _b ).

FIG. 6: Pre-annealing time at 450 ° C and relative Deep-L
It is a graph showing the relationship between evel concentration ^{_{I d / (I b) 1/2}} .

FIG. 7 is a graph showing the relationship between the relative volume of oxygen precipitates and the relative deep-level concentration per oxygen precipitate.

FIG. 8 is an explanatory diagram schematically showing a distribution state of oxygen precipitates and Deep-Level defects around the oxygen precipitates when the sizes of the oxygen precipitates are different under the same Δ [Oi].
Shows the case where the MD size is large and the defect density is low, (b)
Indicates the case where the BMD size is small and the defect density is high.

Claims (2)

(57) [Claims] 1. A photoluminescence intensity (Deep-Level PL intensity) I _d due to Deep-Level in a semiconductor single crystal.
And the band edge PL intensity I _b are obtained, and I _d / I _b ^x (where x is a value of 0.5 to 1.0) indicating the relative deep-level concentration is obtained.
A method for evaluating a deep level defect in a semiconductor crystal, characterized in that a deep level defect is evaluated in accordance with the above. 2. Photoluminescence intensity (Deep-Level PL intensity) I _d due to Deep-Level in a semiconductor single crystal.
And the band edge PL strength I _b and the number of precipitates per unit volume (BMD density: abbreviated as BMD in the following formula),
Relative Deep-Level concentration I _d / (I _b ^x
-A method for evaluating deep level defects in a semiconductor crystal, characterized in that the BMD) is evaluated to evaluate the characteristics of the precipitate.