RU2709687C1 - Method of determining concentration of electrically active donor impurity in surface layers of silicon by nondestructive method of ultra-soft x-ray emission spectroscopy - Google Patents

Abstract

FIELD: nanotechnologies.

SUBSTANCE: invention relates to nanotechnology, nanoelectronics and microelectronics. Method of determining concentration of electrically active donor impurity in surface layers of silicon, involving a procedure for recording characteristic X-ray emission Si L2.3 spectra of heavily doped silicon at concentration of electrically active donor impurity N_D≥10¹⁸ cm^-3 in the region of the valence band of silicon and in the region of the impurity subband of electrically active donors using a nondestructive ultra-soft X-ray emission spectroscopy method, characterized in that registration of X-ray emission spectra is carried out at voltage U = 3 kV on anode of demountable X-ray tube of monochromator spectrometer, at anode current density of 2 mA/cm² with determination of relative intensity I_D donor maximum lying in the X-ray emission Si L2.3 spectrum above the silicon valence band top at the energy E = 100 eV, corresponding to the concentration of the electrically active donor impurity in the surface layers with thickness ≤120 nm heavily doped silicon, determined from logarithmic dependence of relative intensity of donor maximum on concentration of electrically active donor impurity, described by following ratio: I_D=A⋅lgN_D+B, where I_D is relative intensity of donor maximum in X-ray emission Si L2.3 spectrum; N_D is concentration of electrically active donor impurity; A and B are empirical constants which are equal to 0.1 and -1.94 respectively.

EFFECT: technical result consists in development of direct experimental method for quantitative determination of electrically active donor impurity in surface layers with thickness ≤120 nm of silicon with level of doping ≥10¹⁸ cm^-3 non-destructive method of ultra-soft X-ray emission spectroscopy.

1 cl, 3 dwg, 1 ex

Description

The invention relates to the field of nanotechnology, nanoelectronics and microelectronics and can be used to determine the concentration of electrically active donor impurities in the surface layers of silicon with a thickness of ≤120 nm without destroying a sample heavily doped with diffusion or ion implantation methods, which are widely used in semiconductor technology.

Known and most common are methods for determining the concentration of electrically active dopant in silicon, based on measurements of the surface resistance Rs by the four-probe method [J.C. Irvin, Resistivity of Bulk Silicon and of Diffused Layers in Silicon / The Bell System Technical Journal, (1962) V. 41, No. 2, p. 387, R. B. Fair, Profile Estimation of High Concentration Arsenic Diffusions in Silicon / Journal of Applied Physics, (1972) V. 43, p. 1278., K. Pearce, A. Adame, L. Katz, J. Tsai, T. Seidel, D. McGillis, VLSI technology: In 2 pr. Prince 1. Persian from English. / Ed. S. Zee. - M .: Mir, 1986. - 404 pp.] And spreading resistance Rsr, using the two-probe method [K. Pierce, A. Adame, L. Katz, J. Tsai, T. Seidel, D. McGillis, VLSI technology: In 2 pr. Prince 1. Persian from English. / Ed. S. Zee. - M .: Mir, 1986.- 404 p., R.G. Mazur, D.H. Dickey, S Spreading Resistance Technique for Resistivity Measurements on Silicon, J. Electrochem. Soc, (1966) V. 113, p. 255.]. These methods are quite simple and are characterized by high sensitivity and reliability of the results, however, with a small diffusion depth and a high concentration of dopants, the measured surface resistance Rs cannot be used to estimate the surface concentration of electrically active impurities, and the disadvantage of the method for measuring the spreading resistance is that the measurement results the state of the sample surface and the contact of the probes with the substrate have a great influence.

Known methods for determining the concentration of electrically active dopant in silicon, based on the Hall effect [W.S. Jonson, Numerical corrections for hall effect measurements in silicon containing gaussian dopant distributions / Solid-State Electronics Pergamon Press, (1970) V. 13, p. 951-956., N.F. Kovtonyuk, Yu.A. Kontsevoy, Measurement of parameters of semiconductor materials, Publishing House Metallurgy, 1970, 432 pp., Pavlov LP, Methods of measuring parameters of semiconductor materials. - 2nd ed., Moscow: Vysh. Shk., 1987, 239 pp.], which are effectively used in microelectronics.

However, when these methods measure samples with a doping level ≥10 ¹⁸ cm ^{-3, the} Hall mobility in the surface layers of the sample can be significantly less than in deeper layers, and the error in determining the concentration of the dopant in heavily doped semiconductors can reach ~ 18%.

Known methods for determining the concentration of electrically active impurities in semiconductor structures based on measurements of capacitance-voltage characteristics [Pavlov LP, Methods for measuring the parameters of semiconductor materials. - 2nd ed., Moscow: Vysh. Shk., 1987, 239 pp., Patent RU 2028697, publ. 02/09/1995, "A method for determining the parameters of semiconductor materials and heterostructures." Authors: Polyakov V.I., Ermakova O.N., Ermakov M.G., Perov P.I., Patent RU 2234709, publ. 08/20/2004, "A method for local remote measurement at high frequencies of current carriers." Authors: Dulbeev VA], widely used to determine the concentration profile of dopants in a semiconductor. The disadvantage of these methods is the need for the formation of metal contacts on the surface of the semiconductor structure. In addition, the upper limit of the measured concentrations of electrically active impurities does not exceed 10 ¹⁸ cm ^-3 .

Higher concentrations of doping impurities in silicon (up to 10 ²⁰ cm ^-3 ), determined by measuring the capacitance-voltage characteristics, are achieved in methods that use a semiconductor-electrolyte system [Patent RU 2393584, publ. 06/27/2010, Bull. No. 18, "A method for determining the concentration profile of a dopant in semiconductors." Authors: Grokhotkov I.N., Yafyasov AM, Filatova E.O., Bozhevolnov VB, Patent SU 1728900, publ. 04/23/1992, Bull. No. 15, "A method for determining the concentration profile of a dopant in silicon epitaxial structures." Authors: Abramov A.A., Gurova G.A., Makeev M.A.], where the differential capacitance is additionally measured in the potential region corresponding to the degeneracy of charge carriers in the near-surface volume of the semiconductor. However, this method requires the creation of good ohmic contact on one of the sides of the sample and additional polishing of the surface of the investigated object. In addition, this method is aimed at determining the concentration profile of the dopant, therefore, the process of measuring the capacitance-voltage characteristics is accompanied by the destruction of the test sample, by dissolving the semiconductor in the electrolyte.

A known method for determining the concentration of the dopant by the method of mass spectroscopy of secondary ions (MSWI or SIMS) [K. Mogeb, D. Fraser, W. Fichtner, L. Parrillo, R. Marcus, C. Steidel, W. Bertrem, VLSI technology: In 2 pr. Prince 2. Trans. from English / Ed. S. Zee. - M: Mir, 1986.- 453 p., A.N. Pustovit, SIMS diagnostics of the structure of nanometer-sized semiconductor layers doped with impurities / Surface. X-ray, synchrotron and neutron studies, 2010, No. 9, p. 101-104.], Which is characterized by high sensitivity to many elements (for B and As, the sensitivity limit is 5⋅10 ¹⁵ cm ^-3 ) and high spatial resolution (~ 500 nm). However, unlike electrical methods that measure the concentration of an electrically active impurity, the MSWI method records the total concentration of the impurity introduced. Moreover, the presence in the system of traces of water vapor or hydrogen by SIMS is difficult to identify sufficient phosphorus in silicon due to the proximity of the ion masses ³¹ and P ⁺ ions ³¹ SiH ^+, formed as a result of interaction of hydrogen with silicon.

Known methods based on neutron activation analysis (NAA) of the concentration of the dopant in silicon [K. Mogeb, D. Fraser, W. Fichtner, L. Parrillo, R. Marcus, C. Steidel, W. Bertrem, VLSI technology: In 2 pr. Prince 2. Trans. from English / Ed. S. Zee. - M .: Mir, 1986. - 453 p., Patent RU 2208666, publ. 07/20/2003, "A method for determining the concentration of dopants in semiconductors after neutron-transmutation doping." Authors: V.I. Lebedev, O.G. Chernikov, E.K. Gorbunov, L.V. Shmakov, M.P. Kozyk, K.V. Grigoryev, A.N. Fursov.]. The NAA method is extremely sensitive to most chemical elements and allows the identification of extremely low concentrations of impurities (for As in silicon, the sensitivity limit is ~ 7⋅10 ¹¹ cm ^-3 ). The main disadvantage of this method is the need to withstand samples for 24-48 hours after neutron irradiation in order to ensure that the radiation of silicon radioactive isotopes is reduced to a level that is negligible compared to other elements.

The advantages of the above methods of MSWI and NNA are very high sensitivity and a wide working range of dopant concentrations, however, both of these methods carry information about the total amount of impurity introduced, and to take into account only electrically active impurities, these methods must be used in combination with the electrophysical methods listed above.

The proposed method for determining the concentration of electrically active donor impurities in the surface layers of silicon by the non-destructive method of ultra-soft X-ray emission spectroscopy (UMRES) is based on completely different principles compared to these analogues. X-ray emission spectroscopy provides direct information on the partial density of electronic states in the valence band of silicon [T.M. Zimkina, V.A. Fomichev, Ultra-soft X-ray spectroscopy. L .: Publishing house of Leningrad State University, 1971, 132 pp.], Which is well studied experimentally and theoretically. Moreover, in the works of [O.A. Golikova, E.P. Domashevskaya, M.M. Kazanin et al., Structural grid, Fermi level and density of states of amorphous silicon / Physics and Technology of Semiconductors, (1989) T. 23, p. 450-455., VA Terekhov, SN Trostyanskii, E.P. Domashevskaya et. al., Density of states and Photoconductivity of Hydrogenated Amorphous Silicon / Phys. Stat. Sol. B, (1986) V. 138, p. 647.] the manifestation in the X-ray emission spectra of amorphous silicon of additional intensity maxima located above the valence band ceiling corresponding to occupied localized states in the forbidden band of amorphous silicon due to a large number of dangling bonds in an amorphous semiconductor. Thus, the idea of the proposed method is to register X-ray emission Si L2.3 spectra in the region of the silicon valence band and above the ceiling of the valence band, where the impurity subband of electrically active donors is formed, which is formed at a dopant concentration ≥10 ¹⁸ cm ^-3 [I. Fistul, heavily doped semiconductors. M .: Nauka, 1967, 415 pp.]. In this case, only electrically active donors will appear in the X-ray emission Si L2.3 spectrum.

In addition, the UMRES method allows one to obtain X-ray emission spectra in the surface layers of silicon structures at various depths from 10 to 120 nm without destroying the test sample [E.P. Domashevskaya, Y.A. Peshkov, V.A. Terekhov et. al., Phase composition of the buried silicon interlayers in the amorphous multilayer nanostructures [(Co45Fe45Zr10) / a-Si: H] 41 and [(Co45Fe45Zr10) 35 (Al2O3) 65 / a-Si: H] 41 / Surf. Interface Anal. (2018) V. 50, p. 1265-1270., A.S. Shulakov, X-ray emission spectroscopy with depth resolution: application to the study of nanolayers / Journal of Structural Chemistry, (2011) V. 52, No. S7, p. 7-18.]. The analysis depth is changed by changing the energy of the electron beam exciting the X-ray Si L2.3 spectrum by varying the high voltage from 1 to 6 kV at the anode of the X-ray tube of the X-ray spectrometer of the RSM-500 type monochromator.

Thus, the UMRES method can be very useful for determining high concentrations (≥10 ¹⁸ cm ^-3 ) of electrically active doping impurities in the surface layers of highly doped silicon, both independently and in combination with electrophysical methods.

Moreover, to conduct research using this method, it is not necessary to carry out additional preparation of the test samples.

The problem to which this invention is directed is to develop direct experimental methods for determining high concentrations of electrically active donor impurities in the surface layers of silicon.

The technical result in the implementation of the task is to develop a new method for the quantitative determination of electrically active donor impurities in surface layers with a thickness of ≤120 nm silicon with a doping level of ≥10 ¹⁸ cm ^{-3 by the} non-destructive method of ultra-soft X-ray emission spectroscopy.

The technical result is achieved by the fact that in the method for determining the concentration of electrically active donor impurity in the surface layers of silicon by the non-destructive method of ultra-soft X-ray emission spectroscopy, which includes the registration of characteristic X-ray emission Si L2.3 spectra of heavily doped silicon at an electrically active donor impurity concentration of N _D ≥10 ¹⁸ cm ^-3 in the valence band of silicon and the impurity region electrically active donor subband using method ultramyag second X-ray emission spectroscopy, according to the invention, the registration of x-ray emission spectra is carried out at a voltage of U = 3 kV on the anode X-ray tube collapsible monochromator spectrometer, at an anodic current density of 2 mA / cm ^2, the concentration of electrically active donor impurity in the surface layers of thickness ≤120 nm strongly doped silicon is determined by the logarithmic dependence of the relative intensity of the donor maximum located in the x-ray emission Si L2.3 spectrum above the ceiling of the shaft tnoj silicon band at an energy E = 100 eV of the concentration of electrically active donor impurity, which is described by the following relation:

where I _D is the relative intensity of the donor maximum in the X-ray emission Si L2.3 spectrum; N _D is the concentration of an electrically active donor impurity; A and B are empirical constants that are 0.1 and -1.94, respectively.

The proposed method is illustrated by drawings, where:

in FIG. Figure 1 shows the X-ray emission Si L2.3 spectra of a KDB-10 grade silicon wafer (1) and the spectra of the same wafer after doping with diffusion by phosphorus (2), arsenic (3) and antimony (4), with the concentration of electrically active impurities according to measurements surface resistance Rs 8⋅10 ¹⁹ cm ^-3 , 2⋅10 ²⁰ cm ^-3 and 8⋅10 ¹⁹ cm ^-3, respectively; the insets to the spectra of each sample show the spectral regions near the ceiling of the valence band and donor maximum (D) recorded with a long accumulation time, ten times (x10) or four times (x4);

in FIG. Figure 2 shows a plot of the relative intensity I _{D of the} donor maximum in the X-ray emission Si L2.3 spectrum of heavily doped silicon versus the concentration N _{D of the} electrically active donor impurity, determined from the surface resistance Rs.

in FIG. Figure 3 shows the X-ray emission Si L2.3 spectrum of silicon doped with ion implantation of arsenic at a dose of 5 × 10 ³ μC / cm ² with an ion energy of 60 keV, followed by pulsed photon annealing with a light flux energy density of 18 J / cm ² . The spectrum is shown in the inset to the portion of the spectrum near the ceiling of the valence band and donor maximum (D), recorded with a ten-fold long accumulation time (x10).

The method for determining the concentration of electrically active donor impurities in the surface layers of silicon by the non-destructive method of ultra-soft x-ray spectroscopy is as follows.

A silicon sample with a doping level of ≥10 ¹⁸ cm ^{-3 is} irradiated with a monoenergetic electron beam, and the characteristic x-ray spectrum is recorded in the photon energy range of 80-102 eV corresponding to valence transitions at L2.3 silicon level and reflecting the density of occupied electronic states in the valence band silicon and donor states localized in the band gap of silicon by ~ 1 eV above the valence band ceiling. Registration of X-ray emission spectra is carried out using the method of ultra-soft X-ray emission spectroscopy, implemented on an X-ray spectrometer type RSM-500 monochromator. The spectrum is recorded in vacuum no worse than 1⋅10 ^-6 mm Hg. at a voltage of U = 3 kV at the anode of a collapsible x-ray tube of the spectrometer, which corresponds to an analysis depth of 60 nm [E.P. Domashevskaya, Y.A. Peshkov, VA Terekhov et. al., Phase composition of the buried silicon interlayers in the amorphous multilayer nanostructures [(Co ₄₅ Fe ₄₅ Zr ₁₀ ) / a-Si: H] ₄₁ and [(Co ₄₅ Fe ₄₅ Zr ₁₀ ) ₃₅ (Al ₂ O ₃ ) ₆₅ / a-Si: H] ₄₁ / Surf. Interface Anal. (2018) V. 50, p. 1265-1270.], While the density of the anode current should not exceed 2 mA / cm ² in order to avoid overheating of the sample and "volatilization" of the dopant from the silicon surface.

To determine the quantitative relationship between the intensity of the donor maximum located in the X-ray emission Si L2.3 spectrum above the ceiling of the silicon valence band at an energy of E = 100 eV, X-ray Si L2.3 spectra of a series of silicon wafers of the KDB- type were recorded on the concentration of the electrically active donor impurity 10 (10 Ohm⋅cm) doped with phosphorus, arsenic and antimony by the diffusion method, with different concentrations of dopants, which was controlled by measuring the surface resistance Rs. The obtained X-ray emission Si L2.3 spectra of several samples of KDB-10 silicon wafers (10 Ohm⋅cm) with an (111) orientation doped with phosphorus, arsenic and antimony by diffusion at a temperature of T = 900 ° C with an electrically active impurity concentration determined according to measurements of surface resistance Rs - 8⋅10 ¹⁹ cm ^-3 , 2⋅10 ²⁰ cm ^-3 and 8⋅10 ¹⁹ cm ^-3 , respectively, are presented in FIG. 1. Registration of X-ray emission Si L2.3 spectra was carried out on an X-ray spectrometer monochromator RSM-500, under the shooting modes indicated above. For comparison, the same figure shows the spectrum of the original plate brand KDB-10. From FIG. Figure 1 shows that the shape of the X-ray emission spectra of all samples of heavily doped silicon is identical to the spectrum of the initial silicon wafer and corresponds to the spectrum of crystalline silicon, while a narrow maximum is found in the spectra of strongly doped samples above the valence band ceiling at an energy of 100 eV, corresponding to transitions from the donor subband to Si L2. 3rd level. To increase the accuracy of determining the intensity of the donor maximum, the X-ray emission Si L2.3 spectra were recorded in the energy region near the ceiling of the silicon valence band and the donor maximum in the regime with a long signal accumulation time of ten times (x10) or four times (x4), which are given in the insets to the spectra of each sample in FIG. 1.

The results of the present study of heavily doped silicon samples by diffusion and ion implantation methods over a wide range of dopant doses show that the intensity of the maximum formed as a result of transitions from the donor subband to the internal Si L2.3 level depends on the concentration of the electrically active donor impurity. To determine the quantitative relationship between the intensity of the donor maximum and the concentration of electrically active donor impurities in highly doped silicon samples, the surface resistance Rs was measured and the concentration of electrically active donors N _{D was} calculated by the method of [JC Irvin, Resistivity of Bulk Silicon and of Diffused Layers in Silicon / The Bell System Technical Journal, (1962) V. 41, No. 2, p. 387. RB Fair, Profile Estimation of High Concentration Arsenic Diffusions in Silicon / Journal of Applied Physics, (1972) V. 43, p. 1278.]. Based on the calculation results, a graph is plotted of the relative intensity of the donor maximum I _D versus the concentration of electrically active donors N _D (Fig. 2).

Thus, the concentration of electrically active donor impurities in heavily doped silicon structures is determined by the relative intensity of the donor maximum in the X-ray emission Si L2.3 spectrum using the logarithmic dependence shown in FIG. 2, which is described by the following relationship:

where I _D is the relative intensity of the donor maximum in the x-ray Si L2.3 spectrum; N _D is the concentration of an electrically active donor impurity; A and B are empirical constants that are 0.1 and -1.94, respectively.

This approach opens up new direct experimental possibilities for determining high concentrations of electrically active donor impurities in the surface layers of silicon with a thickness of less than 120 nm.

Example 1

As an example in FIG. Figure 3 shows the X-ray emission Si L2.3 spectrum of a KDB-10 silicon wafer (10 Ohm⋅cm) with a (111) orientation doped with ion implantation of arsenic on a Vesuvius-5 installation with an ion energy of 60 keV at an ion current density of 10 μA / cm ² and the implantation dose of 5⋅10 ³ μC / cm ² . To activate the impurity, pulsed photon annealing was performed on a UOLP-1 device at a light flux energy density of 18 J / cm ² and a pulse duration of 1.5 s. The X-ray emission Si L2.3 spectrum was recorded in the photon energy range of 80-102 eV, corresponding to valence transitions at the L2.3 silicon level and reflecting the density of occupied electronic states in the silicon valence band and donor states localized in the band gap of silicon above the ceiling of the valence band at ~ 1 eV. Registration of the X-ray emission spectrum is carried out using the method of ultra-soft X-ray emission spectroscopy, implemented on an X-ray spectrometer monochromator RSM-500. Spectrum recording is carried out in a vacuum of 1⋅10 ^-6 mm Hg. at a voltage of U = 3 kV at the anode of a collapsible x-ray tube of the spectrometer, at an anode current density of 2 mA / cm ² in order to avoid overheating of the sample and “volatilization” of the dopant from the silicon surface.

The concentration of the electrically active donor impurity in heavily doped silicon structures is determined by the relative intensity of the donor maximum in the X-ray emission Si L2.3 spectrum using the logarithmic dependence shown in FIG. 2, which is described by the following relationship:

where I _D is the relative intensity of the donor maximum in the x-ray Si L2.3 spectrum; N _D is the concentration of an electrically active donor impurity; A and B are empirical constants that are 0.1 and -1.94, respectively.

From FIG. Figure 3 shows that in the X-ray emission Si L2.3 spectrum of this sample above the ceiling of the valence band at an energy of 100 eV, a donor maximum is observed whose intensity is I _D = 0.07 ± 0.001 rel. units, which according to the dependence presented in FIG. 2, corresponds to the concentration of an electrically active donor impurity N _D = (1.25 ± 0.03) ⋅ 10 ²⁰ cm ^-3 . This result agrees quite well with the concentration of electrically active donor impurities, determined according to the surface resistance measurements, which amounted to 1.20 × 10 ²⁰ cm ^-3 .

Claims (3)

A method for determining the concentration of an electrically active donor impurity in the surface layers of silicon, including the procedure for recording the characteristic X-ray emission Si L2.3 spectra of highly doped silicon at a concentration of electrically active donor impurity N _D ≥10 ¹⁸ cm ^-3 in the region of the silicon valence band and in the region of the impurity subband electrically active donors using the non-destructive method of ultra-soft x-ray emission spectroscopy, characterized in that the registration of x-ray emission with ektrov carried out at a voltage of U = 3 kV on the anode X-ray tube collapsible monochromator spectrometer, at an anodic current density of 2 mA / cm ² with determination of the relative intensity I _D of the maximum of the donor, located in the x-ray emission spectrum of Si L2.3 above the valence band of silicon at an energy of E = 100 eV, corresponding to the concentration of an electrically active donor impurity in surface layers with a thickness of ≤120 nm of heavily doped silicon, determined by the logarithmic dependence of the relative intensity of the donor maximum concentration of electrically active donor impurity described by the following relation: I _D = A⋅lgN _D + B, where I _D is the relative intensity of the donor maximum in the X-ray emission Si L2.3 spectrum; N _D is the concentration of an electrically active donor impurity; A and B are empirical constants that are 0.1 and -1.94, respectively.

Images