JPH01239863A - Evaluation of semiconductor and apparatus therefor - Google Patents

Abstract

PURPOSE:To simplify evaluation of the deep level of a semiconductor wafer without deterioration of accuracy and resolution by radiating the surface of a sample placed in an ambient temperature environment with light capable of conducting an absorption transition between the valence band and the conduction band of the sample, and analyzing photoluminescence light obtained from the sample. CONSTITUTION:A semiconductor sample 31 is irradiated with light 33 capable of conducting an absorption transition between the valence band and the conduction band of the sample in a state that the sample 31 to be evaluated with respect to a deep level is placed in an ambient temperature environment, and the deep level of the sample 31 is evaluated from wavelength and intensity information of photoluminescence light obtained from the sample 21 on the basis of the radiation of the light. For example, the surface of a semi-insulating GaAs wafer 31 placed on a XY stage 38 is so radiated with light 33 from a suitable laser light source 32 by focusing to a desired spot diameter by a converging lens system 34 and secondarily scanning. The photoluminescence light generated from the wafer 31 is collected by a spectral device having a condensing lens 35, a spectroscope 36 and a detector 37, and analyzed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method and apparatus for evaluating semiconductors, and more particularly to an evaluation method and apparatus suitable for evaluating deep levels of semiconductor wafers.

[Prior Art] When creating various electronic devices such as high-speed logic ICs and optoelectronic integrated circuits (OEIG), it is important to evaluate semiconductor wafers used as substrates in advance, regardless of the material or the type of device to be created. It is also important in general terms.

Even under these circumstances, recently there has been an increase in the creation of devices using gallium arsenide (GaAs) wafers, which are either semi-conductor or are sometimes referred to as "semi-insulating". However, requests have begun to be made for highly critical evaluation of the GaAs wafers.

In the first place, attempts to use such GaAs materials as substrate elements of electronic devices have been based on the fact that, as mentioned above, this GaAs material exhibits semi-insulating properties, which makes isolation between elements simple or completely unnecessary. This may be due to reasons such as the fact that the parasitic capacitance becomes extremely small and the device operation becomes faster. In general, the important geometrical region to be evaluated is a depth range of several micrometers at most from the surface of sea urchins, and several tens of micrometers at most.

This is because the semi-insulating characteristic of GaAs is achieved by the tiny intrinsic defects called EL2 in the wafer and deep levels such as Cr impurities compensating for residual impurities that form shallow levels. Moreover, unless the device is very special, what actually influences the device characteristics is the active region where such a device is constructed, which is a few micrometers from the surface of the sea urchin, or at most a few tens of micrometers thick. This is because it is only a depth range.

Of course, in the process of growing and creating GaAs wafers, if a manufacturing method were developed that would uniformly distribute deep levels in the crystal growth axis direction and in-plane direction, it would be especially important to evaluate the deep levels. It may no longer be necessary to evaluate the two-dimensional concentration distribution for each wafer. Evaluation of a single location on a wafer may represent the entire wafer, or at least other wafers in the same lot.

However, with current technology, such distribution is extremely non-uniform, and actually causes considerable variation in the characteristics of electronic devices built on GaAs wafers, which not only ultimately reduces the yield of devices, but also This was a major obstacle in achieving high reliability and performance.

For this reason, in this type of field, as mentioned above,
There has been a strong demand for evaluation of deep levels within the crystal of a GaAs wafer, particularly for accurately obtaining a two-dimensional concentration distribution of deep levels along the octahedron.

However, it is not limited to GaAs, but typically silicon, etc.
It is necessary to evaluate the deep levels of other semiconductor materials in order to improve the characteristics of devices fabricated on them, but as mentioned above, in particular GaA
Regarding s, the GaAs substrate does not just physically function as a support substrate for the device active region;
The semi-insulating property itself must be used effectively in the device structure and for device operation, so the measurement of deep levels that have a deep causal relationship with this is almost an essential requirement. be.

For this reason, the explanation will be continued below using GaAs as an example, but the evaluation methods that have been developed in the past in response to the above requirements are typically the following: ■Light absorption method, ■Resistivity method. , ■ Low-temperature photoluminescence method.

Therefore, these will be explained individually below.

■Light absorption method: This method measures the amount of absorption by deep levels by transmitting light through the wafer. For GaAs wafers, light in the infrared region is used, which has good absorption efficiency (so (also called infrared absorption method) is the principle of determining the type of deep level from the absorption spectrum by spectroscopically analyzing infrared light that has passed through a GaAs wafer.

Therefore, to be more specific, the concentration of the deep level of the specific type can be determined from the absorption peak intensity (absorption amount) of the determined deep level of the specific type. By determining the concentration in each part while performing two-dimensional scanning, a two-dimensional concentration distribution within the eight faces of the sea urchin regarding the deep level can be obtained.

■Resistivity method: This method uses the correlation between deep level concentration and resistivity to obtain a two-dimensional deep level concentration distribution from the resistivity distribution of the wafer. It is something to do.

■Low-temperature photoluminescence method: As described later, this method is, at least in principle, superior to the above two methods (■ and ■) among conventional evaluation methods, for various reasons. Furthermore, since this is a method that is the subject of direct improvement in the present invention, it will be explained using drawings.

First, FIG. 7 explains the mechanism itself in which the photoluminescence process occurs, regardless of whether the device is placed in a low-temperature environment or not.

This figure schematically shows the energy hand structure of GaAs, with the forbidden a between the valence band 1 and the conduction band 2 (the deep level 4 and the shallow level 5 are exemplified in the valence band 3). ing.

When such a material is irradiated with light having a photon energy larger than the width of the forbidden band 3, as shown by the thick arrow 12 extending from the valence band 1 to the conduction band 2, excess electrons are generated in the interband absorption process. - Hole pairs are generated, and this causes the electrons 1≦=--, which are knocked out during conduction A:b, to be partially trapped in the shallow shallow level 5 of the rod shown by the arrow 6; Similarly, a part of m is captured in the deep level 4 on the capture path shown by arrow 9 and 1.
1-ru.

As shown by arrow 7.10, these electrons generate light in the process of recombining with holes in the valence band 2. This phenomenon is generally called photoluminescence, and the light generated by this photoluminescence By spectroscopically analyzing it, it is possible to identify the type of level that caused the emission of the photoluminescence light, and by analyzing its intensity, it is also possible to know the concentration of each identified level.

However, in actual photoluminescence, in addition to light emission based on the existence of a shallow level due to the process shown by arrow 7 in Fig. 7, there are also exciton emission, band emission, and light emission in the photon energy region slightly smaller than the forbidden band energy. Luminescence phenomena collectively referred to as "forbidden band" photoluminescence, such as transition emission between be represented. In doing so, there is no loss of generality in the discussion of this book.

However, conventionally, in order to utilize this principle, an apparatus system as shown in FIG. 8 has been required.

To briefly explain, the Ga Yayatsu wafer 2 that is the subject of evaluation
1. It is immersed in a cryostat 22 filled with liquid helium, and is cooled to an extremely low temperature rather than a low temperature.

Under this cooling condition, the surface area of the wafer 21 is illuminated by an appropriate optical beam (not shown) using a spot light that has been narrowed down from the parallel laser beam 23 emitted by the laser light source 23' to an appropriate diameter using the lens system 24. The photoluminescence light from each specific part of the wafer 21 is scanned through a mechanical two-dimensional scanning mechanism, and during the scanning, the photoluminescence light from each specific part of the wafer 21 is transmitted through the condensing lens system 25 to the spectroscope 2.
6, and the light intensity in a predetermined wavelength range is further detected by a detector 27.

Therefore, upon completion of the scanning unit, a two-dimensional concentration distribution regarding the deep level of the wafer 21 can be obtained.

FIG. 9 shows an example of the results of evaluating a doped-free Ga wafer, which is currently commercially available for use in producing integrated circuits, using the apparatus system shown in FIG.

In this figure, the photon energy axis region below 1.4 eV is shown with the emission intensity multiplied by 20 times (20x), but the emission band has a fairly wide spectral width with a peak at 0.66 eVO on the photon channel. is photoluminescence light from a deep level due to a minute intrinsic defect in the GaAs crystal called EL2, and corresponds to the result of the light emission process previously indicated by the arrow IO in FIG.

On the other hand, the emission band having a peak at 1.49 eV is forbidden band line emission, and is photoluminescence light from a shallow acceptor level due to residual carbon impurities. The light emission process indicated by arrow 7 in FIG. 7 corresponds to this.

[Problem to be solved by the invention] The optical absorption method or infrared absorption method described at the beginning is a "non-destructive", "non-contact", and simple way to investigate deep levels at room temperature using a relatively simple equipment system. It is excellent in that it can be evaluated, and as mentioned above, GaAs wafers, etc.
Until now, this has been the most commonly used method unless the substrate itself, which is quite thin, is to be directly evaluated.

But 1. As can be understood from the above-mentioned limitation that "as long as the substrate is not thin", this method cannot be used for evaluation if the material to be evaluated that transmits light or infrared rays is not thick enough. However, it was not possible to increase the detection sensitivity.

For example, the deep level in an actual GaAs device wafer is in the concentration range of I
In response to recent demands, if we try to measure concentration fluctuations within the eight surfaces of a sea urchin using this method using a particle size of about ±2 x 10"cm-3, we will need to use a wafer with a thickness of about 5111111mm and polished on both sides. It is necessary to prepare.

On the other hand, GaAs, which is usually supplied for device fabrication,
Since the thickness of the wafer is only about 500 μm at most, and only one side is polished, it is clear that this light absorption method cannot be directly applied to such a substrate itself.

In other words, this light absorption method requires a sample with a thickness that is about 10 times or more than the normal substrate thickness just for evaluation, even if the material is of the same type. In the case of expensive GaAs wafers, this not only results in a large expense but also requires a large amount of labor. I have to say that this is a serious drawback.

In addition, as I mentioned earlier, even though the GaAs wafer itself is thin, about 500 μm thick, there are deep levels that affect the characteristics of devices actually created on it. This is the active layer of the device, and exists only within a depth of several micrometers from the surface of the wafer. However, in this light absorption method, the entire length of the light transmission path (the total thickness of the wafer) There is also a problem in that you can only obtain information that is similar to average values.

In other words, in the current crystal growth technology, variations in properties in the thickness direction of the wafer cannot be ignored in many cases, and therefore, the properties of important parts at a small depth from the surface of the sea urchin,
If it is averaged out due to the influence of the remaining thick part,
This results in a large error in the data that is actually needed.

For this reason, unless some improvement measures are developed, this optical absorption method will be of no use in the future, especially for the evaluation of GaAs wafers, especially for accurate evaluation of the active region. Appropriate.

On the other hand, the second method mentioned above, the resistivity method ■,
It is widely used as a practical method for evaluating wafer uniformity because the resistivity has a direct relationship with device characteristics and the measurement area can be made as small as the active area of the device. has been done.

However, this method also has three major drawbacks: it requires a special sample with electrodes attached for measurement, and it is impossible to directly measure deep levels.

In other words, the resistivity of the GaAs wafer for device fabrication is
In general, it reaches 10' to 108 Ωcm, so it is not possible to use methods such as the long-known four-probe method, which simply contact the probe with the wafer. In order to obtain ohmic contact, electrodes dedicated to measurement must be firmly fabricated on the wafer.

For this purpose, if you want to measure the resistivity distribution of the entire wafer with a two-dimensional spatial resolution of the same size as the device, it is necessary to measure the resistivity distribution of the entire wafer with a two-dimensional matrix of electrodes each having a micro area of the same size as the device. A large number of electrode patterns must be patterned, which requires a great deal of effort and expense, and it goes without saying that wafers with electrode patterns formed in this way can no longer be used to create devices. It's gone.

In other words, with this method, it is basically impossible to evaluate each individual device on which 15 devices are actually mounted, and in this respect, it is impossible to evaluate each individual device on which 15 devices are actually mounted. Giving up the advantage to the evaluation method, which has one characteristic, I would venture to say that this method is a so-called "destructive method" and is not desirable.

Furthermore, with this resistivity method, it is not possible to determine the type of deep level, and the correspondence relationship between resistivity and deep levels is not simple. It is also not possible to quantitatively obtain the concentration distribution of □.

In contrast, the conventionally proposed low-temperature photoluminescence method (■), as explained earlier using Figures 7 to 9, can evaluate minute areas on the eight faces of sea urchins by narrowing down the irradiation light. Considering the fact that it is a non-destructive, non-contact method, it is possible to conduct a tentative evaluation in an extremely shallow region corresponding to the active region for device fabrication in the depth direction.
It can be safely called the most excellent method among these conventional examples.

However, this was just a theoretical discussion based on principles, and in reality, it was not a method that could be adopted cheaply or easily by anyone or at any time.

As can be seen from the characteristic example shown in Figure 9, the intensity of the deep-level emission band is extremely small, ranging from several tenths to one hundredth, compared to the forbidden band line emission. The characteristic example shown in Figure 9 is the result of correcting the wavelength dependence of sensitivity in the measurement optical system, so the ratio becomes even larger before correction, and the emission intensity of the deep level is equal to the forbidden band emission intensity of 10,000 yen. It can even be less than one-fold.

In reality, photo-rays from deep levels with such infinitesimal strength are
In the measurement of luminescence spectra, there are only a dozen or so institutions around the world that can offer the characteristics shown in Figure 9.

Even if the device has the characteristics shown in Figure 9, quantitative measurements may be extremely difficult depending on the sample.

This is because, to put the above in reverse, in this low-temperature photoluminescence method, the forbidden band line emission from shallow levels appears at a relatively large intensity compared to that from deep levels. However, as is clear from the principle diagram shown in Figure 7, the light emission process 7 from the shallow level 5 and the light emission process IO from the deep level 4 compete for the electrons excited by light irradiation. are in a competitive relationship with each other.

Therefore, the emission intensity from the target deep level not only changes independently depending on the concentration of the deep level, but also is influenced by the emission intensity from competing shallow levels. Therefore, if the emission intensity from the shallow levels, which affects the emission intensity from the deep levels, is at a relatively high level as described above,
Naturally, the degree of influence will increase synergistically.

Moreover, the distribution of the emission intensity from the shallow level within the 8 planes of the sea urchin generally changes by a factor of about 10 or more, so in the end, the emission intensity distribution from the deep level is different from the actual one. When I try to evaluate using method ■, the results do not match at all.

These facts show that it is actually extremely dangerous to evaluate the concentration distribution of deep levels using only the conventional low-temperature photoluminescence method.

However, this point has been pointed out in the past, and a method that can be called a single-selective excitation low-temperature photoluminescence method has been proposed as an improved method.

This uses as irradiation light light that cannot excite shallow level 5 but can excite deep level 4 in the principle diagram of Figure 7. It should be possible to eliminate the competitive relationship caused by the existence of , and enable quantitative evaluation.

However, unlike the original low-temperature photoluminescence method, this method does not have a mechanism for absorbing the irradiated light through interband absorption transitions, so the irradiated light passes through the wafer. As a result, similar to the light absorption method described in (2) above, in the end, only the average value information along the entire length of the wafer in the thickness direction is detected, and local detection only in the device active layer part on the surface of the sea urchin is impossible. It's impossible.

In this way, even the low-temperature photoluminescence method and its modified examples, which are considered to be the most superior in principle, require improvement due to their principle, and above all, they are not practical in that they require an extremely low temperature environment. It lacked sex.

As mentioned earlier, there are currently only a few institutions that can extract the characteristics shown in Figure 9 with a certain level of accuracy, but if, for example, the detection system were to be made extremely sensitive, Even if the time comes when there will be no problem in terms of measurement know-how and anyone can make such an evaluation as long as they have the equipment, liquid helium itself is extremely expensive, Rather, it seems that the simple circumstances, which are based on reality, such as the need for skill to handle them, will not change in any way.

From this point of view, some of the current GaAs wafers for device fabrication have diameters of 50 mm or more, and when trying to evaluate such large-diameter wafers using this low-temperature photoluminescence method, it is common to use this type of wafer. Another major drawback is that the commercially available taliostat 22 (FIG. 7), which can be used for cryogenic temperature measurement technology, cannot be used, and a large dedicated one must be manufactured. In addition to the increased expense and labor involved, a large amount of time must be spent on evacuation, pre-cooling, etc. in the measurement procedure. In fact, at present, we are barely able to evaluate one sea urchin eight per day.

Furthermore, in order to irradiate and scan the surface of the sea urchin with a laser beam while the wafer is placed in the taliostat and detect the photoluminescence light from each part, a large-scale sample moving device and laser beam scanning are required. requires equipment;
Not only does the configuration of the measuring device become extremely complicated, but also the lens systems 24, 25, etc. shown in FIG. It is not possible to use a high magnification lens with a short working distance in the system, and therefore it is extremely difficult to reduce the spatial resolution to less than several tens of μm.

In view of these conventional circumstances, and for the reasons already stated, the present invention is an evaluation method for not only GaAs wafers, which are currently in the greatest need, but also numerically developed deep level semiconductor wafers based on the conventional examples described above. We have further improved the low-temperature photoluminescence method, which is considered to be the most superior method in terms of measurement principles, and have succeeded in achieving full satisfaction both in principle and on a practical level.
The objective is to provide a new evaluation method and evaluation device that can sufficiently simplify measurement procedures and equipment systems without deteriorating basic evaluation factors such as accuracy and resolution.

[Means for Solving Problems] With the above objective in mind, the present inventor first had doubts about the common sense matters related to the conventional low temperature photoluminescence method. Simply put, common sense means cooling the wafer to be evaluated to an extremely low temperature.

To begin with, in the past, it was believed that the intensity of light emitted by photoluminescence simply decreases as the temperature rises, although it has not been completely elucidated physically.

This means that, as shown in FIG. 7 above, for electrons trapped in the shallow level 5 and the deep level 4, their path 7. According to lO, before the radiative recombination, for the conduction band 2, as shown by the phantom arrow 8.11,
There is also a mechanism in which it is thermally excited and emitted, and this rate increases as the temperature increases.

Indeed, even in the past, there have been detailed reports regarding shallow level 5 or lanophotoluminescence. For example, at room temperature, the luminescence intensity increases from 1O-4 times when immersed in liquid helium to 1-row times as much as when immersed in liquid helium. was expected to decline.

For these reasons, in the past, in order to actually apply the principle of the photoluminescence method shown in Figure 7, the idea that the colder the sample to be evaluated is, the better, became established, and eventually They quickly cooled it down to extremely low temperatures by immersing it in liquid helium.

However, if we consider the above situation carefully, in reality, only the photoluminescence light related to shallow levels and its absolute intensity have been reported, and the rest is an intuitive problem. Either there was an awareness that the same thing could be said about deep levels, or there was no such awareness and cooling treatments to extremely low temperatures were common knowledge.

However, if we only consider the absolute intensity, the theory that the luminescence intensity of photoluminescence light from a deep level naturally decreases as the temperature rises is correct.

However, if we only consider absolute intensity in this way, this will be a problem that will be solved in the future if ultra-sensitive detectors are developed.

However, the reality is not that simple, and as I have described the shortcomings of the conventional low-temperature photoluminescence method in detail earlier, as long as such a photoluminescence method is adopted, photoluminescence due to shallow levels will be avoided. As long as there is a strict mechanism in which luminescence light affects the photoluminescence phenomenon caused by the target deep level, there still remains a problem that cannot be solved simply by improving detector sensitivity.

The present inventor first broke through the conventional wisdom mentioned above, and not only looked at the absolute value of the photoluminescence emission intensity based on the existence of each level, but also determined that photoluminescence light from shallow levels, etc. We came up with the idea that it would be possible to capture the intensity of photoluminescence light from a target deep level as a relative relationship with the intensity of other disturbing light.

As a result of considering the photoluminescence light intensity as a relative issue, the absolute value of the photoluminescence light intensity from deep levels is lower than the photoluminescence light intensity from shallow levels. Even if there is a problem, if there are conditions that improve the relative ratio, then as long as those conditions are followed, the rest can be processed with the help of a highly sensitive detector. To reiterate, until now, there have been no reports with a clear awareness of these facts.

As a result, the above idea by the present inventor was first developed by photo-photography.
This led to a change in the way we view Figure 7, which is shown as the principle of the luminescence method.

As mentioned above, the higher the temperature of the sample to be evaluated, the faster the photo-
The reason why the emission intensity of luminescence light decreases is that the proportion of electrons captured in each level 5.4 that is thermally excited increases, as shown in processes 8 and 11 in Figure 7. there were.

Considering this, the thermal energy required for this thermal excitation is
Although it is expressed as kT, which is the product of absolute temperature T and Boltzmann's constant k, the above-mentioned rate of thermal release is actually determined by the level energy.

This means that for shallow level 5, the thermal energy kT
At high temperatures, the level energy approaches or exceeds that level energy, so the proportion of thermal emission becomes very large, and the photoluminescence light intensity is extremely reduced.

On the other hand, in deep level 4, even if the temperature reaches a certain level, the level energy can still be sufficiently larger than the thermal energy, and the decrease in photoluminescence light intensity due to thermal emission can be compensated for. The match will be small.

These findings were obtained through experiments and demonstrations by the present inventors, and the following conclusions were drawn.

In other words, when the forbidden band photoluminescence light intensity, which is represented by the existence of a shallow level, is cooled to the liquid helium temperature as in the conventional case, the photoluminescence light intensity from the target deep level changes. In contrast, at higher temperatures, and ultimately at room temperature, which is most desirable in that no intentional cooling or heating is required, the photons from deep levels are much larger and predominant. Luminescent light can be dominant.

On the other hand, even though both the photoluminescence light intensity from shallow levels and the photoluminescence light intensity from deep levels will decrease in absolute value, there are great expectations for the future, and the high sensitivity that can be obtained even now. In terms of the detection system, it was confirmed that the degree of decrease in the intensity of photoluminescence light from the target deep level was within an acceptable range, that is, it was sufficiently detectable.

For these reasons, the present invention breaks away from the conventional wisdom and proposes room temperature as the temperature environment for the semiconductor to be evaluated.

Then, the surface of the sample to be evaluated placed in a room temperature environment is irradiated with light that can cause an interband absorption transition between the valence band and the conduction band of the sample, and based on this, the feed from the sample is determined.・We are trying to analyze the backing photoluminescence light.

In order to obtain a two-dimensional evaluation of the entire sample, the sample can be mechanically scanned, the irradiation light can be optically scanned, or both can be scanned. In any case, the light is scanned two-dimensionally within the plane relative to the sample.

Of course, the evaluation principle itself according to the present invention is no different from that already described with reference to FIG. Regarding the material of the sample, GaAs is most suitable in view of recent market demands, but other semiconductor materials may be used.

[Operations and Effects] In the present invention, the sample to be evaluated is placed in a room temperature environment.

Therefore, as mentioned earlier, in the conventional example in which the film was immersed in liquid helium, forbidden band photoluminescence was dominant, but the degree of forbidden band photoluminescence was at least reduced, and preferably, on the contrary, the evaluation Conditions can be created in which photoluminescence from deep levels of the target becomes dominant.

In other words, at such room temperature, as mentioned above,
As a result of the fact that the probability of light emission transition from a shallow level, which competes with the light emission process based on a deep level, is greatly reduced, the intensity of photoluminescence light based on the deep level is less likely to be adversely affected, and is substantially reduced. becomes proportional to the concentration of deep levels.

To put it another way, in the photoluminescence light spectrum, the emission band at the shallow level is concentrated in a range slightly smaller than the forbidden band energy, and therefore, at room temperature, the half-width of the emission band becomes wide, and each emission band is As a result of overlapping bodies, it is impossible to identify the type of each of these shallow levels, but the emission band based on the target deep level originally has a wide half-width even in a low-temperature environment, and it is difficult to identify the type of each shallow level. Since the energy levels are distributed over a wide energy range within the forbidden band, there is no particular problem in identifying the type, even if the half-width is wide when measured at room temperature. This fact also aids in the advantages of the present invention.

In addition, even in the room temperature photoluminescence method according to the present invention, the penetration depth of the irradiation light into the sample and the diffusion distance of excited electrons can both be maintained at several μm or less, as in the low temperature photoluminescence method. Of course,
Not only that, but when viewed as a device system, the taliostat, which had created various dimensional constraints, is no longer necessary, making it possible to narrow down the irradiation light to about 1 μm or less, if necessary. This naturally means that an extremely high spatial resolution of several μm can be obtained if necessary.

In the end, the present invention has succeeded in rationally solving all the problems that could be considered from the measurement principle and practical standpoint, while retaining the advantages of the conventional low-temperature photoluminescence method. It is expected that the impact that the invention will have on this type of technical field will be extremely significant, including in the Chiba development table, conference presentations, etc. that are scheduled to take place following the application procedure. Just in case, the effects of the present invention can be summarized as follows.

Deep level evaluations can be performed as many times as desired, and therefore it is possible to contribute to a more advanced evaluation system, such as changes in the awards of deep levels by repeating each process step.

(2) Furthermore, if the side surface of the grown ingot is slightly processed to provide a polished surface and the method of the present invention is applied thereto, it becomes possible to examine the concentration distribution in the direction of the crystal growth axis while the ingot is intact. Of course, according to the definition of the term, in such a case, the polished surface portion becomes the surface portion of the sample to be evaluated in the present invention, and the growth axis direction of the ingot is aligned with one direction in the two-dimensional direction of the surface portion.

[Example] FIG. 1 shows a schematic configuration of an apparatus used in an example of the semiconductor evaluation method of the present invention. The evaluation target wafer 31 to be evaluated regarding the deep level is placed on an X-Y stage 38 according to a known existing configuration, and of course, the measurement surrounding environment is left unchanged without any intentional cooling or heating. It is kept at room temperature in this sense.

■ Measurement of the concentration distribution of deep levels, such as the EL2 level and Cr level in GaAs wafers, which have been attracting attention in recent years, can be performed without the need for an intentional cooling medium such as liquid helium, while clearly distinguishing between the types. With simple equipment, inexpensively, at room temperature, non-destructive, non-contact, in the wafer state, if necessary, a few μm
It is possible to measure at high granularity with a high spatial resolution of about 100 yen.

■ Therefore, when this method is applied to quality control of semiconductor wafers, it is possible to easily select any wafer before shipping,
This is great news for both manufacturers and users of this type of wafer, as it is possible to quickly obtain the concentration distribution of deep levels in the surface region of the wafer, which directly affects device characteristics. .

■ Randomly extracted wafers can be returned to shipped products if necessary because they are subjected to non-destructive and non-contact testing and are not affected by severe thermal cycles such as exposure to cryogenic environments.

■ This also means that even after the wafer enters some device process, the irradiation light that should lead to photoluminescence light due to interband absorption transition in the wafer 31 at any step is , a laser beam 33 from a suitable laser light source 32, and a converging lens system 34.
As a result, the wafer 31 is irradiated with the beam focused to a desired spot diameter.

The reason why the wafer 31 is placed on the X-Y stage 38 is to relatively scan the laser beam 33 one-dimensionally or two-dimensionally on the wafer surface as necessary, and thus this purpose is achieved. As far as possible, instead of the illustrated configuration, scanning may be performed by optically deflecting the laser beam 33, or both the laser beam 33 and the wafer 31 may be scanned. Of course, if it is sufficient to evaluate only one or several specific locations on the wafer 31, a continuous scanning system is not required except for positioning.

Since it has already been explained in detail with reference to FIG. 7, the principle of photoluminescence used in the present invention will not be explained again in this section. The photoluminescence light emitted from the photoluminescence light is captured and analyzed by an evaluation means or photoluminescence spectrometer 39 consisting of a condenser lens 35, a spectrometer 36, and a detector 37.

As already mentioned, the spectroscopic device 3 as this evaluation means
9, it is naturally desirable to have a high-sensitivity device, but the applicant has already proposed a device suitable for this
There is one based on the configuration disclosed in No. 2141.

Of course, this is not limiting, and other suitable methods may be used. Rather, as already mentioned in the section on action, photoluminescence light is based on the existence of the target deep level rather than on absolute value issues,
Regarding the problem of the relative intensity ratio of forbidden band photoluminescence light, which hinders evaluation, in the case of the present invention, we have succeeded in obtaining more advantageous conditions than in the conventional cryogenic environment. Therefore, a spectroscopic device system designed to have a certain degree of sensitivity can be sufficiently and meaningfully used.

In the following, the present invention will be explained based on more specific examples.

The wafer 31 to be evaluated is an additive-free semi-insulating wafer λ for producing high-speed integrated circuits using Ga 8, which is normally commercially available.

The laser beam 33 uses a laser with a wavelength of 647 nm,
The spot diameter on the surface of the wafer 31 was narrowed down to about 100 μm using the converging lens system 34.

The light intensity of the laser beam 33 on the wafer surface was kept at about 5 mW to avoid deterioration of the characteristics of the wafer due to irradiation of the wafer with too strong light.

The photoluminescence light from the wafer 31 was captured using a spectroscopic device 39 configured in accordance with the invention related to the patent application mentioned above. 40r above that
Furthermore, as the light intensity detector 37, a Ge detector is used for wavelengths of 1.7 μm or less, and a PBS is used for wavelengths of 1.7 μm or less.
using a detector.

As a result of these measurements, as shown in Figure 2, we succeeded for the first time in the world in obtaining a unique photoluminescence spectrum under the intentional conditions of a room temperature environment.

And in this Figure 2, on the photon energy axis, 1.42e
The peak at V is forbidden band line emission, which is 0.6
It was also found that the emission band having a peak at 5 eV is caused by the EL2 level. - In other words, the emission band with a peak at 0.65 eV has an energy level of approximately 0 as analyzed from the spectral shape.
．． 75 eV, which corresponds to the level energy of a deep level called EL2, which is known to predominately exist in doped-free semi-insulating GaAs wafers. In the measurement with the sample immersed in liquid helium, a 0.66 eV emission band, which is accurately identified as the EL2 emission band, was observed, and as the temperature rose, the emission band gradually decreased until reaching the room temperature 0.65 eV emission band. A luminescent band with the same shape was traced.

Furthermore, as will be described later, the results of a trial measurement of the same sample using the light absorption method described above also revealed that the concentration distribution of the EL2 level within the wafer plane and the intensity distribution of the 0.65 eV emission band were quantitatively determined. It was a good match.

FIG. 3 shows the measurement results obtained under the same conditions as in the previous experimental example, using a semi-insulating GaAs wafer doped with Cr as the evaluation target sea urchin A31.

Although forbidden band line emission was not detected in this wafer, an emission band having a peak at 0.82 eV on the photon energy axis was observed.

This emission band was determined by comparison with the results of the conventional low-temperature photoluminescence method and optical absorption method mentioned above, and by the dependence of the emission intensity on the Crfi degree, which is the dominant level of the wafer. It was identified as photoluminescence from

As is clear from comparing Figures 2 and 3 above,
According to the present invention, since the room temperature conditions are rather limited, it is possible to clearly identify the type of deep level, such as whether it is an EL2 level or a C level, from the obtained photoluminescence spectral characteristics. able to discriminate.

Next, as a special case or a simple case of two-dimensional evaluation, as an example of an application to investigate the concentration distribution within the wafer surface regarding deep levels along a predetermined one-dimensional direction, we will examine a doped-free semi-insulating GaAs crystal. The effect of long-term heat treatment, which is known as a method for uniformizing electrical characteristics, was investigated, and the results are shown in FIG. 4 and explained below.

FIG. 4(A) shows a wafer before heat treatment at 950°C for 8 hours, and FIG. 4(B) shows a wafer after such heat treatment by applying the present invention. It shows the results of measuring the photoluminescence light intensity in the radial direction from the center position to each query, and the photoluminescence spectrum at each location is the same as shown in the second figure.

In this measurement, the wavelength of the spectrometer 36 is adjusted to the emission band of 0.65 eV, and the laser beam 33 with a beam diameter of 100 μm is sequentially applied in 1001 steps in the diameter direction of the wafer 31 using the apparatus shown in the first figure. -Dimension scanning to determine the photoluminescence light intensity at each location, and the measurement time for each sample point was approximately 1 second, including the travel time.

From this result, as shown in Figure 4 (8), the intensity fluctuations are very wide in the area around the sea urchin 8 before heat treatment (the EL2 level is unevenly distributed). I understand.

This is also in good agreement with measurement results obtained by optical absorption method, resistivity method, and low-temperature photoluminescence method.

On the other hand, looking at the intensity distribution after the heat treatment in FIG. 4(B), it is found that the intensity distribution is relatively flat, and the concentration distribution of the EL2 level is made uniform.

In fact, when devices are fabricated from wafers that have been subjected to such heat treatment, the device characteristics within the wafer surface are well uniformed.

From these facts, it is proved that the room temperature photoluminescence method according to the present invention can be used to evaluate the uniformity of electrical characteristics of a wafer by applying it as described above. Of course, for the sake of simplicity, the above is a -dimensional scan, and as shown in the examples below, naturally,
Two-dimensional scanning is possible as needed, and the beam diameter,
The scanning step, etc. is also a matter of arbitrary design.

Next, in order to investigate the relationship between the photoluminescence light intensity of the deep level El,2 and the concentration of the deep level, three wafers cut from different positions in the doped-free semi-insulating GaAs ingot were The room temperature photoluminescence method according to the present invention and the conventional light absorption method were applied together, and the correlation between the results obtained was determined.

Here, a light beam with a diameter of 3 mm was used for the measurement by the optical absorption method, and a light beam with a diameter of 100 μl was used for the room temperature photoluminescence method of the present invention.

The results are shown in Figure 5, and the range of intensity fluctuation in light absorption measurement using a light beam with a diameter of 3 mm is due to errors.
It is shown in bar notation, and the average value is plotted as a black circle.

However, this Figure 5 shows that a simple proportional relationship holds between the photoluminescence light intensity and the optical absorption coefficient, and therefore, the room temperature photoluminescence method of the present invention can be used to quantify deep levels. It has been proven that a similar analysis is also possible.

Finally, FIG. 6 also shows the results of an experimental example in which the two-dimensional distribution of deep levels within the wafer surface was investigated.

In this measurement, for an undoped semi-insulating GaAs wafer,
By operating the x-y stage 38 using the apparatus shown in FIG. The intensity of photoluminescence light from the EL2 level, which is a deep level, was measured. The total time required for the measurement was about - hours, which is extremely short compared to the case where the conventional method is adopted.

As a result of these measurements, we followed the principle that areas with weak intensity are shown with light gradations and areas with strong intensity are shown with dark gradations, and we changed the coating pattern for the density of each sample area as appropriate, but as shown in Figure 6, An interesting phenomenon was observed in which the EL2 level is distributed in a 4-fold symmetry, corresponding to the (100) crystal plane orientation of the wafer.

By the way, regarding this type of commercially available doped-free semi-insulating GaAs wafer, which is mainly taken up as the semiconductor to be evaluated in this section, photoluminescence light is not generated from deep levels other than the EL2 level. rare. Therefore, as far as such wafers are concerned, there is no need to identify even the types of deep levels. Therefore, in such a case, spectrum measurement is naturally unnecessary, so emphasis is placed exclusively on efficient detection of the photoluminescence light intensity, and the spectrometer 36 in the device shown in the first figure is omitted, and an appropriate one is used instead. It is better to insert an optical filter or the like to efficiently detect, for example, 0.65 eV band light.

However, even if the spectrometer 36 is omitted, care must be taken because the light intensity is still measured in consideration of the wavelength, as stated in the gist of the present invention. 0.6 concerned
The wavelength corresponding to the 5 eV band light is the target deep level EL.
This is because the photoluminescence light from 2 is identified.

Of course, from the above-described method principles and device configuration according to the present invention, it is clear that the present invention can be effectively utilized for various semiconductors other than GaAs.

[Brief explanation of the drawing]

FIG. 1 is a schematic configuration diagram of an example of a device configuration used in the room temperature photoluminescence method of the present invention. Figure 2 shows additive-free semi-insulating Ga obtained by applying the present invention.
A characteristic diagram showing a photoluminescence spectrum from an As wafer. Figure 3 shows a Cr-doped semi-insulating G obtained by applying the present invention.
A characteristic diagram showing a photoluminescence spectrum from an aAs wafer. Figure 4 shows additive-free semi-insulating Ga before and after heat treatment, obtained as a result of confirming the effect of ordinary heat treatment using the present invention.
A characteristic diagram showing the photoluminescence intensity distribution of the EL2 level along the radial direction of the As wafer. FIG. 5 is a characteristic diagram showing the relationship between the photoluminescence intensity at the EL2 level and the light absorption coefficient of an undoped semi-insulating GaAs wafer. FIG. 6 is an explanatory diagram showing a two-dimensional concentration distribution of photoluminescence intensity at the EL2 level of an undoped semi-insulating GaAs wafer. FIG. 7 is an explanatory diagram showing the principle of the photoluminescence method itself. FIG. 8 is a schematic diagram of an apparatus employed in the conventional low-temperature photoluminescence method. FIG. 9 is a characteristic diagram showing a photoluminescence spectrum from an additive-free semi-insulating GaAs wafer obtained by a conventional low-temperature photoluminescence method in which the temperature of liquid helium is limited as the temperature environment condition of the sample to be evaluated. It is. In the figure, 31 is a semiconductor to be evaluated, 32 is a laser light source, and 33
is a laser beam, 34 is a converging lens system, 35 is a condensing lens system, 36 is a spectrometer, 37 is a light intensity detector, and 38 is an X
-Y stage, 39 is an evaluation means or a spectroscopic device. 39 Spectroscopic device (review) ■ Hand-printed X-Y stage Figure 2 Figure 3 C port 11) Figure 0 t5-Work winter/L hole' (ev)

Claims (5)

[Claims] (1) Place a semiconductor sample whose deep levels are to be evaluated in a room temperature environment; irradiate light that can cause
A semiconductor evaluation method characterized by: evaluating the deep level of the semiconductor sample from wavelength information and intensity information of luminescence light. (2) Light capable of causing an interband absorption transition between the valence band and conduction band of the semiconductor sample is applied so as to scan the surface of the semiconductor sample, thereby increasing the two-dimensional concentration of the deep level of the semiconductor sample. The semiconductor evaluation method according to claim 1, characterized in that: obtaining a distribution; (3) means for maintaining a semiconductor sample whose deep levels are to be evaluated in a room temperature environment; means for irradiating light capable of causing an absorption transition;
A semiconductor evaluation device comprising: means for evaluating the deep level of the semiconductor sample from wavelength information and intensity information of luminescence light. (4) The semiconductor evaluation apparatus according to claim 3, further comprising means for relatively two-dimensionally scanning the semiconductor sample with respect to the semiconductor surface with respect to the semiconductor sample. (5) The means for evaluating the deep level includes a spectroscope for the wavelength of photoluminescence light and an intensity detector for the specific wavelength; as set forth in claim 3 or 4. semiconductor evaluation equipment.