JP6689494B2 - Method for detecting carbon in silicon - Google Patents

Description

  The present invention relates to a method for detecting carbon in silicon.

2. Description of the Related Art As devices such as semiconductor integrated circuits have become higher in density and higher in integration, stabilization of device operation has been eagerly desired. In particular, improvement of characteristic values such as leak current and oxide film breakdown voltage is an important issue. However, if impurities are mixed into a silicon wafer which is an integrated circuit substrate, stable operation of a device manufactured thereafter cannot be expected. For example, in the manufacturing process of semiconductor integrated circuits, it is widely known that carbon in a wafer adversely affects device characteristics even at a concentration of 1 × 10 ¹⁵ atms / cm ³ or less.

  Therefore, accurate concentration measurement techniques are needed to keep the carbon in the wafer low. Usually, for this purpose, carbon concentration measurement is performed by Fourier transform infrared spectroscopy (FT-IR method). This method is a method for indirectly obtaining the interstitial carbon concentration from the infrared absorption spectrum of a silicon wafer, and is widely used because it can be easily measured.

However, in the carbon concentration measurement by the FT-IR method, it is actually difficult to accurately measure the concentration of 10 ¹⁴ atms / cm ³ unit. The same applies to SIMS (Secondary Ion Mass Spectroscopy), which is another measurement method.

In that respect, the DLTS (Deep Level Transient Spectroscopy) method is a method having a possibility of measuring a concentration of 10 ¹³ atms / cm ³ units. It is to be noted that the DLTS method is a method in which a reverse bias voltage applied to a Schottky junction or a pn junction formed on a measurement target is manipulated, and a deep impurity quasi-state is obtained from temperature dependence of a change in capacitance of a depletion layer generated at the junction. It is a way to get information about rank. The measurement result of the DLTS method is shown by, for example, a graph of DLTS signal strength and measurement temperature. The peak formed on the graph indicates the presence of some deep impurity levels. The energy of the deep impurity level is found from the temperature of the peak, and the height of the peak theoretically indicates the density of the deep impurity level.

  Here, for example, in the method disclosed in Non-Patent Document 1, the carbon-related levels E1, E2, and E3 present in the silicon crystal are deep impurity levels formed by the H—C and H—C—O composites. And in particular, the level of E3 is said to be unaffected by oxygen because of the level resulting from H—C.

However, it cannot be said that the carbon concentration on the order of 10 ¹³ atms / cm ³ can be measured with high sensitivity under this condition. Therefore, some ideas have been proposed. For example, Patent Document 1 shows that it is effective to add E1, E2, and E3, which are carbon-related impurity levels.

Minoru Yoneta, Yoichi Kamiura, and Fumio Hashimoto, "Chemical Etching-induced Defects in Phosphorus-Doped Silicon", J. Am. Appl. Phys. 70 (3), 1 August 1991, p. 1295-1308

JP, 2016-108159, A

  However, when the carbon-related impurity level in silicon is measured by the DLTS method, there is a problem that the DLTS signal intensity is not stable. In this regard, none of the conventional documents mentions the cause of unstable levels as an essential solution for highly sensitive and stable measurement of carbon-related levels existing at low concentrations. . As a result, no real countermeasures have been taken and effective methods that have taken essential solutions have not been reached.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a method capable of detecting carbon in silicon with high sensitivity or stably.

In order to solve the above-mentioned problems, the method for detecting carbon in silicon according to the present invention comprises etching silicon with a mixed acid containing HNO ₃ and HF, and leaving the silicon for 72 hours to 120 hours after the etching. It is characterized in that the impurity level caused by the complex contained at least in carbon and hydrogen contained therein is measured by the DLTS method. According to the present invention, carbon in silicon can be detected with higher sensitivity than conventional methods.

Further, the carbon detection method in the silicon of the present invention, etched by mixed acid containing HNO ₃ and HF in the silicon, after standing for more than 144 hours after performing the etching, contained in the silicon, carbon and hydrogen The impurity level resulting from the complex containing at least is measured by the DLTS method. According to the present invention, carbon in silicon can be detected more stably than conventional methods.

  Further, it is preferable that the above-mentioned complex is a complex containing only carbon and hydrogen. As a result, it is possible to suppress variations in DLTS measurement values due to the influence of the concentration of elements other than carbon and hydrogen.

FIG. 3 is a diagram showing changes in carbon-related level density obtained by the DLTS method with respect to the standing time after the wet treatment performed prior to DLTS measurement.

  Hereinafter, embodiments will be described. In the DLTS method, a metal electrode forming a Schottky junction is usually formed on the surface of the measurement sample, a metal electrode having an ohmic junction is formed on the back surface, a reverse bias is applied between the two electrodes, and the resulting depletion layer is formed. When the temperature dependence of the capacitance change in the inside is acquired, the capacitance change forms a peak at a temperature according to the energy level formed by the impurities. The impurity level density can be calculated from the capacitance change at the peak position. Specifically, in an n-type silicon wafer, gold is often used for the Schottky electrode formed on the front surface, and gallium is often used for the ohmic electrode formed on the back surface. Both electrodes are preferably formed on a clean silicon surface without an oxide film, and the surface oxide film is removed by HF to obtain the clean surface.

One feature of the present embodiment is that etching (wet treatment) with an acid is performed prior to the HF treatment, and hydrogen is deeply introduced into the silicon wafer during the reaction. Specifically, it is known that it is often effective that the acid used is a mixed acid containing HNO ₃ and HF.

  By this wet treatment, hydrogen is introduced into silicon. This is because the diffusion of hydrogen is so fast that it can be sufficiently introduced into silicon even in a wet process at room temperature.

  However, this indicates the instability of the introduced hydrogen, and the present inventors have conceived that investigating the stability of hydrogen is the key to the stable measurement of carbon-related levels.

  Figure 1 shows the results of the confirmation experiment. Of the three carbon-related levels E1, E2, and E3, the E3 level that is not affected by oxygen is left to stand for the time (hour) shown on the horizontal axis of the graph after the wet treatment with mixed acid, and then the surface with HF. After removal and Schottky electrode formation processing, DLTS measurement was performed. As shown in FIG. 1, roughly, the direction in which the level density gradually decreases immediately after the wet treatment with mixed acid is shown. This large tendency is considered to indicate the instability of hydrogen introduced by the wet treatment. On the other hand, a peak of the level density is observed when the standing time after the wet treatment is 72 to 96 hours. At present, the reason why this peak is formed is not clear, but it is considered that this peak has the highest overall efficiency in the reaction in which hydrogen and carbon are bound and converted into a complex. Further, this peak is higher than the level density immediately after the wet treatment (that is, the level density when the standing time is 0 hour). Moreover, after the peak, the level density gradually decreases.

  Therefore, carbon is present at a high concentration to some extent, and in order to quantitatively and accurately determine the concentration, the level density is stable beyond the peak, and 144 hours or more have passed since the wet treatment with mixed acid was performed. It can be seen that it is preferable to wait for the measurement to be performed and then perform the DLTS measurement. In order to obtain the true carbon concentration in the wafer by this method, a calibration curve is created between the carbon concentration in silicon previously measured by another method and the level density obtained by this method. You can leave it (details will be described later).

  On the other hand, for the qualitative purpose of detecting a trace amount of carbon in terms of quantitative accuracy, it is effective to use a peak that appears at a standing time of 72 to 120 hours after performing wet treatment with a mixed acid. This peak value has a concentration twice or more the density of carbon-related levels that can be stably measured after 144 hours or more. Therefore, if the qualitative analysis confirms the presence of carbon by the appearance of this peak, It can be an effective measurement method. However, using this peak concentration as a quantitative value is considered to be a technique that should be avoided because of the instability of the peak height.

  Hereinafter, details of the method for measuring the carbon concentration in silicon according to the present embodiment will be described. First, a method of measuring DLTS for the purpose of quantitatively evaluating the carbon concentration in silicon will be described.

(Quantitative evaluation of carbon concentration in silicon)
First, a plurality of first silicon crystals having different carbon concentrations are prepared. Specifically, for example, an n-type silicon crystal ingot pulled up by the FZ method is cut into a predetermined thickness, and the cut wafer is subjected to rough polishing, etching, polishing, etc., and a substrate (polished Wafer). It should be noted that this substrate can be a substrate manufactured for forming an electronic device such as a transistor or a diode. Next, a silicon crystal is cut out from the substrate to produce a first silicon crystal. The carbon concentration of the first silicon crystal may be adjusted to a range measurable by the FT-IR method or SIMS (for example, 1 × 10 ¹⁵ to 1 × 10 ¹⁶ atms / cm ³ ). The first silicon crystal may be formed by the CZ method (Czochralski method) or may be formed by the FZ method (floating zone method). Same as law. Further, the carbon concentration of the first silicon crystal is measured in advance by a method other than the DLTS method such as SIMS and FT-IR method, that is, is known in advance.

Next, prior to the DLTS measurement, an acid etching process (wet process) is performed on the front surface, the back surface, or both surfaces of each first silicon crystal. Examples of the wet treatment include mixed acid liquid treatment containing HNO ₃ and HF. In the mixed acid solution treatment, the acid solution is heated to a predetermined temperature to etch the surface of the first silicon crystal, and then the first silicon crystal is rinsed with pure water. As the mixed acid solution treatment, for example, the surface of the first silicon crystal is etched to 50 μm or less at a liquid temperature of 30 ° C., and then the first silicon crystal is rinsed with pure water for 3 minutes. The concentration of each acid (HNO ₃ , HF, etc.) that constitutes the mixed acid solution and the mixing ratio (volume ratio or mass ratio) may be set appropriately. Further, the time of the mixed acid solution treatment and the etching amount may be set appropriately.

  After carrying out this wet treatment, the first silicon crystal is left standing for a certain time. Specifically, it is left for 144 hours or more after performing wet treatment with mixed acid. By leaving it for 144 hours or more, the level density can be obtained by the DLTS measurement performed later while suppressing the influence of the peak of the level density that appears during 72 to 120 hours after the wet treatment ( (See FIG. 1). That is, a stable level density can be obtained. In order to obtain a more stable level density, it is preferable to leave it for 192 hours or more after the wet treatment. This is because, as shown in FIG. 1, when the wet treatment is left for 192 hours or more, the degree of change in level density with time is further reduced. At this time, there is no particular upper limit for the standing time after the wet treatment, but it may be appropriately determined in consideration of productivity and the like.

  Here, with regard to leaving for 144 hours or more after performing the wet treatment, for example, when the wet treatment is performed at 12:00 pm on February 1, it is left for 144 hours after the wet treatment at 12:00 pm on February 7. Becomes Therefore, when the wet treatment is performed at 12:00 pm on February 1, the first silicon crystal is left at least until 12:00 pm on February 7.

  Next, the density of carbon-related impurity levels is measured by the DLTS method for each first silicon crystal left for 144 hours or more after the wet treatment. Specifically, the density of impurity levels of a complex containing at least carbon and hydrogen is measured. Among the composites containing at least carbon and hydrogen, a composite containing no elements (oxygen, etc.) other than carbon and hydrogen (that is, a composite containing only carbon and hydrogen (HC composite)) It is preferable to measure the unit density. This is to prevent the DLTS signal from varying due to the effect of the concentration of other elements such as oxygen. When the conductivity type of the first silicon crystal is n-type, of the carbon-related levels E1, E2, and E3 shown in Non-Patent Document 1, the level E3 due to the H—C complex is measured. Is preferred. The three impurity levels E1, E2, and E3 are carbon-related impurity levels formed in the range of about 0.11 to 0.15 eV detected by measuring an n-type silicon crystal by the DLTS method. Therefore, the energy of the level E1 is 0.11 eV, the energy of the level E2 is 0.13 eV, and the energy of the level E3 is 0.15 eV.

  In the DLTS measurement, the front surface and the back surface of the first silicon crystal are subjected to a surface oxide film removal treatment by HF, and then Au is vapor-deposited on the front surface to form a Schottky electrode, and Ga is applied on the back surface thereof. Then, an ohmic electrode is produced. Then, a reverse bias (for example, −5 V) is applied between the two electrodes, and sweeping is performed within a predetermined temperature range (for example, a range of 30 to 300 K) to measure the carbon related level density.

  Next, based on the carbon-related level density obtained from each first silicon crystal and the carbon concentration of each first silicon crystal, the correlation between the carbon-related level density and the carbon concentration in silicon A calibration curve showing is derived. When the carbon-related level density and the carbon concentration in silicon have a substantially proportional relationship, the carbon-related level density obtained from the first silicon crystal is used as the carbon concentration of the first silicon crystal. It is also possible to derive the level formation rate, which is the rate at which carbon in silicon forms an impurity level, by dividing by. This level formation rate is also an index showing the correlation between the level density related to carbon and the carbon concentration in silicon.

  Next, a second silicon crystal whose carbon concentration is unknown is prepared. The method for producing the second silicon crystal is similar to that for the first silicon crystal.

  A wet treatment similar to that performed on the first silicon crystal is performed on the second silicon crystal, and the wet treatment is left for 144 hours or more. At this time, the leaving time of the second silicon crystal after the wet treatment is the same as the leaving time of the first silicon crystal after the wet treatment. That is, for example, when the leaving time for the first silicon crystal is 192 hours, the leaving time for the second silicon crystal is also 192 hours.

  Next, the density of carbon-related impurity levels is measured by the DLTS method with respect to the second silicon crystal left for 144 hours or more. At this time, when the previously derived calibration curve or level formation rate is obtained based on the level E3 caused by the H—C complex, the density of the level E3 also with respect to the second silicon crystal. To measure. That is, the same impurity level as the impurity level constituting the calibration curve or the level formation rate is measured from the second silicon crystal. The DLTS measurement procedure for the second silicon crystal is the same as the DLTS measurement procedure for the first silicon crystal.

  Next, a carbon-related level density of the second silicon crystal and a calibration curve or level formation rate obtained from the first silicon crystal (correlation between carbon concentration in silicon and carbon-related level density) Based on the above, the carbon concentration in the second silicon crystal is determined. From the above, the carbon concentration in the silicon crystal can be obtained by the DLTS method.

(Qualitative evaluation of carbon concentration in silicon)
Next, a method for DLTS measurement for the purpose of qualitatively evaluating the carbon concentration in silicon will be described.

  First, a silicon crystal that is an object of carbon concentration evaluation is prepared. This silicon crystal may be manufactured by the CZ method or the FZ method.

  Next, the prepared silicon crystal is subjected to the same wet treatment as that performed on the first silicon crystal.

  After carrying out this wet treatment, the silicon crystal is left to stand for a certain period of time. Specifically, the wet treatment is left for 72 to 120 hours. As described with reference to FIG. 1, since the peak of the level density appears at the time of leaving 72 to 120 hours after the wet treatment, by leaving the silicon crystal for 72 to 120 hours after the wet treatment, the DLTS measurement performed later. , The level density near this peak can be measured. That is, the level density can be measured with high sensitivity.

  Here, regarding the 72 to 120 hours left after the wet treatment, for example, when the wet treatment is performed at 12:00 pm on February 1, it is left until 12:00 pm on February 4 to 12:00 on February 6 Means to do.

  Next, the density of carbon-related impurity levels is measured by the DLTS method with respect to the silicon crystal left to stand for 72 to 120 hours after the wet treatment. The measurement of the level density is similar to the above-described measurement of the level density for the first silicon crystal.

  Then, the carbon concentration in the silicon crystal is evaluated based on the obtained level density. For example, the fact that the carbon-related level density can be measured by the DLTS method means that a small amount of carbon exists in the silicon crystal. Therefore, the presence of carbon in the silicon crystal based on the level density Confirm and evaluate.

  As described above, in this embodiment, since the silicon crystal is subjected to the wet treatment with acid prior to the DLTS measurement, hydrogen can be introduced into the silicon crystal. With this introduction, of the impurity levels in the silicon crystal, a composite level containing carbon and hydrogen can be activated, and this level can be measured with high sensitivity by DLTS measurement.

  Further, since the DLTS measurement is performed after leaving the wet treatment for 144 hours or more, or after leaving the wet treatment for 72 to 120 hours, the carbon-related level density can be stably or highly sensitively measured. That is, when the DLTS measurement is performed after leaving for 144 hours or more after the wet treatment, the measurement can be performed when the change in the carbon-related level density over time is small, so that the level density with high accuracy can be obtained. Obtainable. That is, the carbon concentration in silicon can be accurately measured.

On the other hand, when DLTS measurement is carried out after being left for 72 to 120 hours after the wet treatment, the measurement can be carried out near the peak of the level density, so that carbon in silicon can be detected with high sensitivity and in the silicon. The presence of carbon can be confirmed even if the amount of carbon is very small (for example, the carbon concentration is 10 ¹³ atms / cm ³ unit).

  Hereinafter, the present invention will be specifically described with reference to Examples and Comparative Examples, but the present invention is not limited thereto.

(Comparative Example 1)
A silicon crystal rod having a diameter of 6 inches, an orientation of <100>, an n-type, a resistivity of 10 Ωcm, and a carbon content of about 0.03 ppma was pulled by the FZ method. This silicon crystal rod was processed into a silicon wafer. This silicon wafer was divided into two groups, one group did nothing, and the other group had its surface layer etched (wet treatment) with a mixed acid of HNO ₃ and HF. In this wet treatment, the temperature of the mixed acid was set to 30 ° C., the surface of the silicon crystal was etched by 3 μm, and then the silicon crystal was rinsed with pure water for 3 minutes. Immediately thereafter, the oxide film was removed by HF from the two groups of silicon wafers, and Schottky electrodes made of Au and ohmic electrodes made of Ga were formed on the front and back surfaces, respectively, and then DLTS measurement was performed. In the DLTS measurement, −5 V (reverse bias) was applied between the electrodes, the temperature was swept in the range of 30 to 300 K, and the density of the level E3 of the H—C complex was measured.

As a result, the density of the carbon-related level E3 of about 8 × 10 ¹¹ atoms / cm ³ was obtained in the sample subjected to the mixed acid treatment, whereas it was 2 × 10 ¹⁰ atoms in the sample not subjected to the mixed acid treatment. A density of carbon-related level E3 of about / cm ³ was obtained, and it was found that the mixed acid treatment had an effect of activating the E3 level. The level density (about 8 × 10 ¹¹ atoms / cm ³ ) obtained from the sample subjected to the mixed acid treatment is shown as a point when the standing time in FIG. 1 is 0 hour.

(Example 1)
A silicon crystal rod having a diameter of 6 inches, an orientation of <100>, an n-type, a resistivity of 10 Ωcm, and a carbon content of about 0.03 ppma was pulled by the FZ method. This silicon crystal rod was processed into a silicon wafer. This silicon wafer was subjected to surface layer etching with a mixed acid of HNO ₃ and HF. In this wet treatment, the temperature of the mixed acid was set to 30 ° C., the surface of the silicon crystal was etched by 3 μm, and then the silicon crystal was rinsed with pure water for 3 minutes. The silicon wafer was left for 72 hours after the wet treatment. After that, the oxide film was removed by HF, and a Schottky electrode made of Au and an ohmic electrode made of Ga were formed on the front and back surfaces, respectively, and then DLTS measurement was performed. In the DLTS measurement, −5 V (reverse bias) was applied between the electrodes, the temperature was swept in the range of 30 to 300 K, and the density of the level E3 of the H—C complex was measured.

As a result, a density of the carbon-related level E3 of about 1.8 × 10 ¹² atoms / cm ³ was obtained, and the density of the E3 level was more than twice that of the sample on which the electrode formation and the DLTS measurement were performed immediately after the mixed acid treatment. It was possible to obtain a qualitative judgment of the existence of carbon. Note that this level density (about 1.8 × 10 ¹² atoms / cm ³ ) is shown as a point when the standing time in FIG. 1 is 72 hours.

(Example 2)
The density of the carbon-related level E3 was measured in the same manner as in Example 1 except that the standing time after the wet treatment was 192 hours.

As a result, a carbon-related level E3 density of about 1.3 × 10 ¹¹ atoms / cm ³ was obtained, and the E3 level activity was about twice as low as that of the sample not subjected to the mixed acid treatment as shown in Comparative Example 1. It was possible to obtain a stable E3 level density while maintaining the efficiency of conversion, and to make a quantitative determination of the presence of carbon with certainty. Note that this level density (about 1.3 × 10 ¹¹ atoms / cm ³ ) is shown as a point when the standing time in FIG. 1 is 192 hours.

(Example 3)
The density of the carbon-related level E3 was measured in the same manner as in Example 1 except that the standing time after the wet treatment was 96 hours.

As a result, a density of carbon-related level E3 of about 1.2 × 10 ¹² atoms / cm ³ was obtained, and an E3 level density higher than that of the sample for which electrode formation and DLTS measurement were performed immediately after the mixed acid treatment was obtained. It was possible to make a qualitative judgment of the existence of carbon. The level density (about 1.2 × 10 ¹² atoms / cm ³ ) is shown as a point when the standing time in FIG. 1 is 96 hours.

(Example 4)
The density of the carbon-related level E3 was measured in the same manner as in Example 1 except that the standing time after the wet treatment was 336 hours.

As a result, a density of the carbon-related level E3 of about 1.0 × 10 ¹¹ atoms / cm ³ was obtained, and the E3 level activation efficiency higher than that of the sample not subjected to the mixed acid treatment as shown in Comparative Example 1 was obtained. While having it, a stable E3 level density could be obtained, and the quantitative determination of the presence of carbon could be reliably performed. The level density (about 1.0 × 10 ¹¹ atoms / cm ³ ) is shown as a point when the standing time in FIG. 1 is 336 hours.

Further, the difference in level densities obtained in Examples 2 and 4 is about 0.3 × 10 ¹¹ atoms / cm ³ , and after the standing time after the wet treatment of 192 hours, before 120 hours. In comparison, the change in level density over time is small. Therefore, it can be said that the carbon-related level density is stable by setting the standing time after the wet treatment to 144 hours or longer (particularly 192 hours or longer).

  The present invention is not limited to the above embodiment. The above embodiments are mere examples, and any of the embodiments having substantially the same configuration as the technical idea described in the scope of claims of the present invention and exhibiting the same operational effect It is included in the technical scope of.

  For example, in the above example, an example of DLTS measurement of the density of the level E3 of the H-C complex was shown. However, impurity levels other than the level E3 containing at least H and C (H-C-O complex It is also possible to evaluate the carbon concentration in silicon based on the measured value by DLTS measurement of the density of the H.C. composite and the H.C.O. The density of the levels E1, E2, and E3 may be measured by DLTS, and the carbon concentration in silicon may be evaluated based on the total value of the obtained densities.

Claims (3)

Etching of silicon with a mixed acid containing HNO ₃ and HF is performed, and after the etching is left for 72 hours to 120 hours, the impurity level caused by the complex containing at least carbon and hydrogen contained in the silicon is reduced. A method for detecting carbon in silicon, characterized in that the position is measured by the DLTS method. Silicon etched by mixed acid containing HNO ₃ and HF, after standing after performing the etching for 144 hours or more, contained in the silicon, the impurity level due to including at least complex carbon and hydrogen A method for detecting carbon in silicon, characterized by being measured by the DLTS method.   The method for detecting carbon in silicon according to claim 1 or 2, wherein the complex is a complex containing only carbon and hydrogen.

Images