JPS6189644A - Semiconductor evaluating method and device - Google Patents

Abstract

PURPOSE:To measure a deep level by a simple analysis without necessity of correcting temperature dependency by emitting a monochromatic light of special energy to a high specific resistance semiconductor crystal, and Arrhenius-plotting on the basis of discharging rate at the temperature after interrupting the light to obtain a deep level. CONSTITUTION:A spectroscope 3 is driven to obtain a monochromatic light having specific wavelength, i.e., energy from the incandescent light of a light source 2 under the control of a computer 1. The light emitting area or light interrupting area to a sample S is controlled by closing or opening a shutter 4, thermal electromotive force of thermocouple 6 is measured by a voltmeter 6A in parallel to measure the temperature of the sample. After the light is interrupted, a passing current is measured by an ammeter 8 from a time t0 to a time t0+T at time interval DELTAt, totally at points K, the K passing current values are dispersively Fourier-converted to obtain the discharge rate (e). This process is repeated by altering the temperature of the sample S. A graph of the Arrhenius-plotting exhibiting the relationship between 1/T and T<2>/3 is formed by applying the integrated data to obtain a deep level from the gradient of the line for coupling the plotted points.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a method for evaluating deep levels in a high resistivity semiconductor and an apparatus used for the evaluation.

The conventional technique (1-deep level in Tochi semiconductors) is located in the band gap, is created by impurities and lattice defects in the semiconductor crystal, and is a factor that determines the lifetime of fractional gears. or cause deterioration in the performance of semiconductor devices,
Furthermore, it functions as a recombination center. Therefore, in evaluating the quality of semiconductors, deep levels have been attracting attention in recent years, and various measurement methods have been proposed.

Among them, the depth of low resistivity semiconductor (ρ<103Ωcm)
WX! As a 67P value method at the level of
Transient Spectroscopy) and D
L F'S method (Deep Level Fouri
erSpccl-roscopy ) or other electrical evaluation methods, or light to semiconductor crystals! An optical evaluation method with l, <1 ray is used. , the DLTS method is used, for example, in "Optical Physical Properties Measurement Technology," published by Takao Fada and Motohiro Hiiragi, 1.
June 10, 983, University of Tokyo Press, pp125-
135 and is explained in detail in D.I.F.
S method: Japanese Patent Application Laid-Open No. 58-79169 (Makigansho J 6-
No. 177594).

However, when the specific resistance of the semiconductor crystal is large,
Application of the above-mentioned back electrical evaluation method becomes difficult. Therefore,
Optical evaluation methods are exclusively used to evaluate the deep levels of high resistivity semiconductor crystals.

The problem that the invention aims to solve There are two conventional optical evaluation methods described above.

The first method is to shine a single-wavelength light from a laser light source or the like onto a high-resistivity semiconductor crystal, and set the wavelength to A (nm) C
−1240/E], and excite the deep levels separately according to their activation energy, g! ! i.

However, even if the laser light source is changed, it is difficult to continuously change the wavelength in the vicinity of the wavelength corresponding to the handgear beam energy of the semiconductor. For this reason, when the irradiation energy is fixed and there are M number 1 deep levels in the gear tube, it is not possible to activate the deep levels separately for each activation energy, There was a drawback that detailed evaluation was not possible. In other words, it was not possible to detect all existing deep levels. Further, it is possible in principle to use a tunable semiconductor laser, but there is a problem in that the apparatus becomes large-scale.

The second optical evaluation method is also known as the optical DL and TS method, and instead of applying a voltage, it irradiates a high resistivity semiconductor crystal with light at a wavelength of 100%. It is the same as the DLTS method except that the transient current after tJ+' is measured at two times L1 and 41) at one constant. Therefore, for details, please refer to Sentence 1, Beast above.

2) In short, a semiconductor crystal is irradiated with light of a single skin length, the transient current is measured at two times after the light is cut off, and the difference in market value (capacitance) at those two times is determined. and 5) repeat the measurement by changing the temperature of the semiconductor crystal, and find the temperature 'r' at which the current difference is maximum.

Find the emission rate e of electrons (regular notes) at that temperature T1, and draw an Arrhenius plot on the graph of the relationship between T,2 and (1/T,) from the relational expression between e, T1, and activation energy. . The above measurement warpage calculation was carried out by changing the two running times t1 and t2 after the light cutoff, and the activation energy (i.e., the level) was calculated from the slope of the line connecting the points plotted on the graph. It is something to be measured.

As is clear from the above explanation, in the optical DLTS method, based on the analysis principle, only the temperature at which the current difference peaks is subjected to analysis, and the efficiency is low compared to the number of data points of the signal curve. bad.

Furthermore, the transient flow change after monochromatic light is applied to seat 1: li L is i (t)-Ic-> )-c foe x p(-
e t ), but in its transient response, the coefficients e and 1° of the exponential function make the temperature dependence 1′1′f′′. Therefore, it is necessary to correct the temperature dependence, but even though e can be theoretically analysed, I 11 cannot be analyzed, so the reliability of measurement accuracy is not sufficient.

Therefore, in the optical evaluation method of the deep level of a semiconductor described above, the present invention makes it possible to measure the deep level by simple analysis without the need to correct the temperature dependence of the coefficient of the exponential function in the transient response. Semiconductor meter: We aim to provide a positive method.

Furthermore, it is an object of the present invention to provide a semiconductor evaluation method capable of detecting all existing deep levels as much as possible.

Further, the present invention aims to provide an apparatus used when implementing the semiconductor evaluation method according to the present invention so that the semiconductor evaluation method according to the present invention described above can be efficiently implemented.

According to the present invention, a means for solving the problem is to irradiate a high resistivity semiconductor crystal with monochromatic light having a specific energy in a pulsed manner, and after the light is cut off, the semiconductor crystal changes based on the light energy. The transient current is measured at multiple sampling intervals at appropriate intervals, and applied to the Fourier-transformed equation of the equation representing the transient current after light cutoff to obtain the deep level emission ratio e, and the above operation is repeated depending on the temperature of the semiconductor crystal. There is provided a semiconductor evaluation method, characterized in that the deep level in the semiconductor crystal is determined by performing an Arrhenius plot while changing the value e at each temperature.

Furthermore, according to the present invention, a white light source, a spectroscope that receives light from the white light source, selects light of a specific wavelength, and irradiates the semiconductor sample with the selected light, and a spectrometer that is placed between the white light source and the spectroscope. a shutter that has been opened, a temperature control means for varying the temperature of the semiconductor sample, a temperature measurement means for measuring the temperature of the semiconductor sample, an ammeter for measuring the current of the semiconductor sample, and a temperature control means for controlling the spectrometer. Light of a specific wavelength is selected and the shutter is opened and closed to irradiate the semiconductor sample with a light pulse of the specific wavelength. control for setting the sample to a specific temperature and sampling the output of the ammeter after the light pulse at that temperature in multiple numbers at appropriate intervals;
There is provided a semiconductor evaluation device characterized by comprising a calculation means.

In the semiconductor evaluation method according to the present invention described above, by irradiating a semiconductor sample with light of a specific wavelength, only carriers in deep levels below a specific activation energy in the semiconductor sample are excited, and their As a result, the transient current change of the semiconductor sample after light interruption is 1(t)-Ioo-'-,e
1. exp (-et) However, ■ is a steady current.

e is the emission rate of deep level electrons (holes) and is temperature dependent.

1o is the capture cross section, electrons (holes) excited by light
It is determined by density, electron (hole) thermal velocity, etc., and is a factor that has light intensity dependence and temperature dependence.

and the sampling interval CtoSt,+T)(1
, is the first sampler time after the light, a, and T is the sampling period), then 1(t)-a, /
2 +b, sin (2πn(t - t,)/T)':l
It is expressed as (1).

The Fourier coefficients an, 1)n (n≧1) in the above equation are respectively expressed as follows.

an=2eT(Ioe/2)exp(-eta)・(e
xp(-e T)-1) / ((e T)2+(2r
r n)2) bn=lrn(1,e/2)exp(-e
t,)-(exp(-e T)-1)/((eT
)2+(2;r h)2)・'2) From this, it can be seen that the Fourier coefficient an1bn is a function of e. Fourier coefficient ratio (bn/bm
, a n / b m, a , , / k) n) Alternatively, if the temperature dependence of Io can be ignored, then from a comparison with the theoretical value of the Fourier coefficient (the right-hand side of equation (2) above), at each temperature Find the value of e. Then, based on the temperature dependence of e, we used the same method as before, namely Arrhenius plot.
Analyze the energy level. In the above method, two IJ.

(compared to '), since Il, which has temperature dependence and cannot be analyzed theoretically, is eliminated, the deep level can be found without correction processing due to the warm real dependence, which was a conventional problem. be able to.

EXAMPLE Hereinafter, semiconductor according to the present invention 1;-1 will be described with reference to the accompanying drawings.
An example of a zeta value method and an apparatus for implementing the method will be described.

1 is a block diagram of an apparatus for carrying out the semiconductor evaluation method according to the present invention, and reference number 1 is a computer that functions as a control/arithmetic means, and the computer 1 is a light source such as a white light source. A spectroscope 3 placed in front of the spectrometer 2 is controlled to select and output light of a specific wavelength. Then, an annotator 4 placed in front of the spectrometer 3 opens and closes at predetermined intervals and with a predetermined open period under the control of the computer 1, producing light pulses as shown in FIG. 2(a). is irradiated onto the half-quadruple sample S in the Talaios cut 5 forming the sample chamber.

The cryostat 5 is equipped with six thermocouples, and a voltmeter 6△ connected to the thermocouple 6 measures the temperature inside the cryostat 5, that is, the semiconductor sample S, and the temperature data is supplied to the computer 1. Ru. Furthermore, a heater 7 is placed in the cryostand 5, and a heater power source 7A connected to the heater 7 is operated under the control of the computer 1, and the heater 7 is activated to raise the temperature of the semiconductor sample S to a required temperature. It is designed to heat the T. Furthermore, an ammeter 8 for detecting the current of the V-conductor sample S is provided,
The detected current ';+'j is supplied to the computer '1.The computer '1 then processes the collected data and causes the plotter 9 to plot the necessary data.

Using the above-described apparatus, the semiconductor evaluation method according to the present invention can be carried out as follows.

That is, under the control of calculations (1), (1) drive the spectroscope 3 from the white light of the light source 2 to obtain monochromatic light that captures a specific wavelength, that is, energy; (2) turn on/off the light source 4; The light irradiation or light interruption to the semiconductor sample S is controlled as shown in FIG.
Inside the cryostat 5;=Measure the temperature of the sample. (
4) After cutting off the light, as shown in Figure 2 (b), the ammeter 8
The transient current is measured in total at time intervals Δt from time t0 to time to, -IT. Note that T= (K−
1) X△to(5) Perform a discrete Fourier transform on the K transient current values, that is, the above equations (1) and (2)
) to determine the above release rate e. (6) Repeat the above process by changing the temperature of the semiconductor test tube 4S. (7)
Then, for A, the term l'j:1 holds true, which is the term in which the temperature component is removed from the term that includes the captured 11 square squares, so by applying the data collected in the above manner, 1/ Create a rough diagram showing the relationship between T and T2/e using Arrhenius sp0.7), and connect the plotted points using demand.

(]8) Furthermore, measurement i! , in order to increase the 7th degree and detect as deep a numb position as possible, from (1) to (7) described above.
) Processing up to 1)! is carried out by changing the wavelength selected by the spectrometer 3.

Thus, the activation energy of the deep level of the semiconductor sample,
The captured cross section +i'1 can be determined.

FIG. 3 shows high resistivity undoped [+1. This figure shows an example of measuring the deep level of a ball crystal (ρ-108 Ωcm). The horizontal axis is the bubble sealing temperature;
The vertical axis is the first-order discrete Fourier transform (optical DLFS signal)
, and the parameter is the energy of the irradiated monochromatic light.

(The shape showing the peak in No. 3 indicates the existence of a deep level.

Then, from the relationship between the measured temperature and the value of , we created a relationship between 1/T, specifically 1000/T, and T'/e for some deep points using an Arrhenius plot. The resulting graph is shown in Figure 4. In addition, the third
The figure shows 6 detected deep grasses.

A symbol is attached to the position. From FIG. 3, it can be seen that the signal strength of the deep level changes systematically as the energy changes.

Examining FIG. 3 and FIG. 71, it can be seen that: (1) According to the method of the present invention, the coefficient I of the transient response is obtained by applying the Fourier transform. It can be seen that the transient response can be analyzed without correcting the temperature dependence of . In particular, in FIG. 4, E+ is b1/Io, ignoring the temperature dependence of IO.
E2 and E are calculated from 1. The temperature dependence of is calculated from b1/Io, and Io
The value obtained from b1/b2 and a+/a2, which can eliminate
The value obtained from /Io is the result of assuming a value that ignores the temperature dependence of Io, and as a result, the assumption was actually approximate, but as a measured value, bz/b.

and al/a2 are reliable.

■The deep level of E1 has an irradiation energy of 0.78 e
It is not excited at energies below V, but can be excited at energies above 0.85 e V, and the deep levels of E3 and E6 are not excited at energies below 0.70 e V.
Excitation is possible above 8 eV. This makes it possible to excite carriers in specific deep levels by changing the energy of the irradiated light, making it possible to analyze deep levels without considering gross interactions between deep levels. becomes easier.

■The deep levels of E3 and E4 have an irradiation energy of 1.11
At 0eV, the signal is weak and difficult to analyze, but at 1,0
Analysis is possible with irradiation light of about 0 eV.

This makes it possible to analyze even deep levels that cannot be analyzed with single-wavelength irradiation by changing the energy of the irradiated light.

(2) In FIG. 3, points other than the peak of the signal curve can also be used for analysis of deep levels due to the principle of the present invention of analyzing the transient response at each temperature. This increases measurement efficiency and improves analysis accuracy. Another advantage is that the determination of peak temperature, which is essential in conventional methods, is not necessary.

■Furthermore, in the optical DLTS method, two sampling times t
1 and t2 must be at least a predetermined ratio, but in the present invention, any sampling interval and sampling number can be adopted without such restrictions.

Furthermore, any combination of Fourier coefficients can be used.

As described in detail, according to the method of the present invention, a sample is irradiated with monochromatic light of a specific energy, and the transient response after the light is cut off is measured using a Fourier transform method, so that high resistivity can be obtained. In deep level evaluation in semiconductors, there is no need to correct the temperature dependence of the coefficient of the exponential function in the transient response, making analysis easy. This simplifies and facilitates the analysis of deep levels.

Furthermore, by continuously changing the energy of the irradiation light and obtaining data at different wavelengths, the amount of data that can be used for analysis in the optical DLFS signal curve is enriched, making it possible to detect the deepest levels possible. This improves the accuracy of deep level analysis.

Therefore, if the present invention is applied to the detection and analysis of deep levels in compound semiconductor crystals such as high resistivity GaAs crystals that serve as substrates for high-speed semiconductor devices, it will contribute to improving the quality of the substrates.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an apparatus according to the present invention that implements a semiconductor evaluation method according to the present invention. FIGS. 2(a) and 2(b) are waveform diagrams illustrating the light irradiation method and current sampling method used in the semiconductor evaluation method according to the present invention. FIGS. 3 and 4 are graphs showing an example of data obtained by the semiconductor evaluation method according to the present invention for a high resistivity undoped GaAs crystal. [Main reference numbers] 1° computer, 2 light source, 3 spectrometer, 4 shutter, 5 cryostat, 6 thermocouple, 6A voltmeter, 7 heater, 7A heater power supply, 8 ammeter, 9 plotter S semiconductor sample patent applicant Nippon Telegraph and Telephone Public Corporation representative Patent attorney Masahiko Arai Figure 2 T= (K-1)

Claims (4)

[Claims] (1) A high resistivity semiconductor crystal is irradiated with monochromatic light having a specific energy in a pulsed manner, and after the light is cut off, the transient current of the semiconductor crystal based on the light energy is measured in multiple samples at appropriate intervals, and the light is cut off. By applying the Fourier-transformed expression of the expression for the later transient current, the deep level emission fraction e
A method for evaluating a semiconductor, characterized in that the above operation is performed while changing the temperature of the semiconductor crystal, and an Arrhenius plot is performed based on the e at each temperature to determine a deep level in the semiconductor crystal. (2) The semiconductor evaluation method according to claim (1), wherein the measurement is performed by continuously changing the wavelength of the monochromatic light. (3) a white light source, a spectrometer that receives light from the white light source, selects light of a specific wavelength, and irradiates the semiconductor sample with the selected light, and a shutter placed between the white light source and the spectrometer; temperature control means for varying the temperature of the semiconductor sample, temperature measurement means for measuring the temperature of the semiconductor sample, ammeter for measuring the current of the semiconductor sample, and controlling the spectrometer to generate light of a specific wavelength. while opening and closing the shutter to irradiate the semiconductor sample with a light pulse of the specific wavelength;
Controlling the temperature control means based on the output of the temperature measurement means to set the semiconductor sample at a specific temperature and sample the output of the ammeter after the light pulse at that temperature in multiple numbers at appropriate intervals. 1. A semiconductor evaluation device comprising: calculation means. (4) The control/calculation means controls the spectrometer,
Continuously changing the wavelength of light irradiated to the semiconductor sample,
A semiconductor evaluation device according to claim 3, characterized in that the above processing is executed.