Several methods are known for studying the recombination lifetime of minority carriers. One of the basic methods is based on microwave reflection.

The present invention can be considered a development of the method using microwave, thus in what follows, only this method will be described.

First A.. P. Ramsa, H. Jacobs and F. A. Brand described a method in J.

Appi. Phys., No. 30 (1995), in which the recombination lifetime of minority carriers was determined from the decay of optical excited excess charge carriers so that the semiconductor was placed into a microwave field, and the change of microwave reflection in time was measured.

In the excitation period of the measurement, excess charge carriers i. e. electron-hole pairs, are generated by light pulses. After light pulses, the original equilibrium state of the semiconductor is restored, the excess in charge carrier concentration gradually disappears, charge carriers, electrons and holes recombine with each other. The presence of crystal defects or impurity atoms accelerates the recombination of charge carriers. The time course of the process in simple cases is of exponential character, and its time constant is the so-called recombination lifetime. The reciprocal value of lifetime is a probability parameter, which is proportional to the concentration of defects in the material, if only one kind of defects is present. Microwave reflection measurement is the most appropriate method for following the time course of this recombination process. The specimen is placed into a microwave field by using a suitable arrangement, the typical frequency of the microwave source (e. g. Gunn oscillator) is 10 GHz. To the microwave source.

characteristically a circulator is also connected, which transfers the microwave energy to an antenna. The microwave is then radiated onto the sample via the antenna. In such a reflection arrangement, the microwave reflected from the sample will be received in a detector via the antenna and then in the circulator. The intensity of microwave reflected from the semiconductor sample depends, among other factors, on the conductivity of the semiconductor wafer, i. e. on the concentration of charge carriers present in the sample.

If the concentration of charge carriers shows dependence on time, this is reflected also in the intensity variation of the reflected microwave. In case of small excitations, i. e. small number of excess charge carriers in relation to the original equilibrium charge carrier concentration the intensity of the reflected microwave signal is directly proportional to the change of the conductivity or of the charge carrier concentration. Due to this, the time course of microwave intensity sensed by the detector reflects accurately the change in charge carrier concentration. Thus, on the basis of the reflected microwave intensity detected, the lifetime of minority carriers can be determined.

More commercially available apparatuses are functioning according to the above principle, which are protected by patents, e. g. by USA patents: US 4,704,576 and US 5,406,214. Differences between the approaches consist usually in the arrangement of the microwave system. and especially in the antenna radiating the microwave. The common feature of these methods is that they are all contact-free, i. e. the microwave conductor and the sample to be measured are not in electric contact with each other. This is advantageous, as the measurement does not cause any change in the valable semiconductor materials or devices. At the same time, microwave is not transferred directly to the specimen, thus it is, in part, irradiated into the free atmosphere. The wavelength of the microwave signal of 10 GHz frequency is 3 cm in the air, thus this is the characteristic distance for which the microwave can be focussed in the sample. This range can be narrowed by applying microwaves of higher frequency, but even so, the resolution of 1 mm needed for a detailed study of semiconductor wafers cannot be reached. Lateral resolution of 1 mm is reached in mapping semiconductor wafers so that,

instead of the microwave, the exciting light (e. g. light pulses) is focussed to a spot of 1 mm diameter. In this case, the recombined transient detected is characteristic only for the illuminated area. However, microwave is reflected from a larger part of the specimen than this. This can be described in other words shortly so that only a small part of the microwave carries information on the recombination process. In addition, reflections are also detected to a significant extent from unexcited parts of the sample and the surrounding, which have no value from the viewpoint of the aim of the measurement. Due to this, the signal to noise ratio is significantly worse than ideal. The most favourable arrangement from the point of view of sensitivity would be, if the microwave reflection were restricted to the excited volume of the sample.

Improvement concerning sensitivity was achieved by Japanese researchers by increasing the exciting light spot to about 1-2 cm diameter (H. Hashizume, H. Sumie, Y. Nakai:"Carrier lifetime measurements by microwave photoconductivity decay method", Recombination Lifetime Measurements in Silicon, ASTM STP 1340, D. C. Gupta, F. R. Bacher and W. M. Hughes, Eds., American Society for Testing and Materials, 1998). However, in this case the resolution of the measurement decreased significantly, and mapping of lifetimes became impossible. This means that the requirement for making measurements more sensitive while keeping high resolution has not been satisfied. High sensitivity is needed since only by decreasing the excitation level, i. e. that of excess carrier concentration, becomes the study of basic characteristics of lifetime possible. In presence of certain impurities, the dependence of lifetime on the excitation level is characteristic, thus these impurities can be identified from the measurement (G. Ferenczi, T. Pavelka, P. Tutti: Injection Levei Spectroscopy: A Novel Non-Contact Contamination Analysis Technique in Silicon, J. Appi. Phys., Vol. 30, No. 12B, December 1992, pp. 3630-3633).

The importance of increasing the sensitivity is even greater in the case of the so-called epitaxial wafers. These wafers are not homogeneous. In typical cases, on a highly doped support, i. e. one of low conductivity (about 5-20 mf2 cm) the thickness of which is about 500-700 um, an extraordinarily thin layer ( 1-20 um m) with a resistivity of typically 10 Qcm (the so-called epi-layer) is to

be found. From the viewpoint of semiconductor technology, the properties among others, also lifetime of the epi-layer, are important parameters.

In a typical, conventional testing and measuring apparatus the microwave suffers partial reflection in the epi-layer, thus, in principe, it may be suitable for studying the epi-layer. However, due to the extraordinarily small layer thickness, the signal generally remains below the noise level. Another difficulty is that the process takes place in the immediate neighbourhood of a support layer with very high conductivity. Thus, in case of a typical epitaxial semiconductor wafer, this traditional measurement usually cannot be applied.

The aim of our method is to eliminate this draw-back and to increase thereby to a significant extent the sensitivity of lifetime measurements based on microwave reflection while keeping the resolution high (1 mm).

The present invention relates to a method for measuring the recombination lifetime of minority carriers in semiconductors so that the semiconductor is illuminated by pulses from a suitably chosen light source (they should have an energy higher than the forbidden band of the semiconductor), and the variation owing to illumination in microwave reflection caused by the minority carriers in the semiconductor material is detected as a function of time in such a way that the microwave conductor is directly contacted with the sample.

The essence of the present invention is that the microwave is led on a coaxial cable to the immediate neighbourhood of the sample, and the microwave is coupled in direct contact to the material. Thus microwave reflection occurs from a relatively smaller field as compared to contact-free methods. The volume sensed by microwave reflection can be restricted to the surrounding of the illuminated spot of 1 mm.

Further, the invention also relates to the arrangement for carrying out the method according to the present invention, which apparatus comprises a tunable, ajustable microwave signal generator, the output of which is connected to the first gate of a circulator, to the second gate of the circulator a microwave conductor (typicaily a coaxial cable) is connected, which is in direct contact with the material to be tested, whereas the third gate of the

circulator is connected to a signal processing device, and the semiconductor is coupled to a suitably chosen laser light source.

The essence of the arrangement consists of the fact that due to a direct contact between the microwave system and the specimen, the microwave is concentrated to the neighbourhood of the volume to be tested, thus the signal/noise ratio is significantly better than in traditional arrangements.

In a preferred embodiment of the arrangement the frequency of the signal generator can be varied (in a range of about 500 MHz, around the base frequency of 10,3 GHz), whereby the system can be tuned so that the reflection on the specimen is the highest possible.

The invention will be described in detail on the example of a preferred embodiment the block scheme of which is shown in Fig. I.

The apparatus shown is suitable for measuring the recombination lifetime of minority carriers in semiconductor material 6 as a specimen, as the time constant of the change in resistivity caused by charge carriers generated by a laser beam. The arrangement comprises microwave signal generator 2 tunable by varactor 1, the oscillation frequency of which can be changed so that it is coupled to varactor 1 through a control connection. The microwave signal generator 2 is preferably a known Gunn oscillator, its frequency range being between 10.2 and 10.45 GHz, and its output power about 50-100 mW.

Of course, other types of microwave signal generators 2 can also be used.

The output of microwave signal generator 2 is connected to the input of isolator 3, the output of which is connected to the first gate of circulator 4 via a coaxial cable. The second gate of circulator 4 is led to a contact needle also via a coaxial cable, whereas the third gate of circulator 4 is connected via a coaxial connector to detector 8, which is, in this case, a peak-value rectifier.

Contact needle 5 leads the microwave field to semiconductor material 6.

Contact needle 5 is in contact with semiconductor material 6, and semiconductor material 6 constitutes the impedance closing the microwave path, the change of which is measured. Thus, this solution, where the microwave system is in direct contact with semiconductor material 6 to be tested through contact needle 5, differs basically from the solutions known until now.

Optical excitation is led to semiconductor material 6 from laser light source 7, in this case from a laser diode, which is coupled to one output of pulse generator 12. The input of pulse generator 12 is connected to the output of central signal processing unit 13 controlling the timing of the system, e. g. to a computer-controlled signal processor. To the same output, inputs of varactor 1 and laser light source 7 are also connected. Detector 8 is connected through amplifier 9 to the input of sample and hold circuit 10, which separates and filters the D. C. component and low frequency noise from the signal measured, thus at its output, a zero level adjusting signal appears. The output of sample and hold circuit 10 is fed back to the input of amplifier 9. The output of amplifier 9 is connected to the input of a known type of transient recorder 11, it means that a signal having the restored null-level is coupled from the output of amplifier 9 to the transient recorder 11, which stores the transient signal in a form transformed into a digital signal. The control and timing input of transient recorder 11 is also connected to the output of pulse generator 12, as well as the control and timing input of sample and hold circuit 10. Signal processing is performed by central signal processing unit 13, which also controls varactor 1, laser beam source 7 and pulse generator 12.

The method according to the invention, and the functioning of the apparatus performing this method is the following: The measurement consists essentially in measuring the resistivity change of semiconductor material 6 caused by minority carriers generated by the laser light, and determining the time constant of resistivity change.

Semiconductor material 6 is contacted with contact needle 5 and then it is illuminated by laser light source 7 so that the time of illumination should be much shorter than the time constant of the recombination of minority carriers, then the signal is detected by detector 8. Detector 8 is in this case a broad band peak-value rectifier. The output signal of detector 8 is amplifie by amplifier 9 and then led to sample and hold circuit 10, which filters the D. C. component and the low frequency (typically below 100 Hz) noise present in the signal.

Transient recorder 11 transforms the signals arriving to its input into digital signals, and stores them in its memory. The stored signals are then evaluated

by central signal processing unit 13. The measurement consists thus essentially in a highly accurate measuring of the time course of resistivity change caused by charge carriers generated by the laser beam.

Contact needle 5 used leads the microwave directly into the specimen, thus only insignificant reflection originates from the surrounding. Laser light source 7 ensuring optical excitation can easily be located and focussed.

In its preferable form, contact needle 5 is the extension of the central core of a coaxial wave-guide, which extension is short enough (shorter than a quarter of the wavelength) to irradiate the microwave into free space only to a small extent, thus it leads the major part of the microwave signal into semiconductor material 6.

Owing to the light pulses emitted from laser light source 7, the electric potential at the output of detector 8 changes, proportionally with the conductivity of semiconductor material 6. The time constant of the attenuation of the signal form at the output of detector 8 to a base level is proportional to the recombination lifetime of minority carriers, which base level corresponds to the energy of the unmodulated constant or steady state reflected microwave energy, if there is no excitation from laser light source 7.

Advantages of the method and apparatus according to the present invention are: To the contrary of solutions known until now, in this arrangement microwave is carried into semiconductor material 6 via a direct contact, instead of irradiation into free space. As a consequence, higher sensitivity can be achieved as compared to traditional arrangements.

The present method, due to its increased sensitivity, is suitable for detecting signals also from thin epitaxial layers. In the case of very thin layers, the recombination process taking place at the surface may disturb the measurement, thus surface recombination is expedient to be suppressed. For this reason the surface is either passivated chemically, or covered by a thermal oxide-layer which is then charged. Both methods are suitable for minimising the effect of surface recombination on the measurement, and thus we can measure the characteristic lifetime in the thin layer.

Due to increased sensitivity, materials can also be tested in which, under traditional conditions, the signal level is very low or even if it does not exceed the noise level. Typical examples for this are materials of small conductivity, characteristically materials of specific resistivities smaller than 0.5 D. cm.

Another case important from the point of view of technology is, when a thin layer (the so-called epitaxial layer) is supporte on supports of very low resistivity (typically of mDcm).

Due to increased sensitivity, to the contrary of traditional methods, the measurement can be carried out not only in case of high excitation, i. e. when large number of excess charge carriers are generated, but can also be performed at low levels of excess carriers. The significance of measurements with low excitations is that these results can be modelled by calculations, and compared with results from different measurements (surface phototension).

Moreover, lifetime measuring in dependence of excitation level leads also to getting acquainted with the characteristic properties of dominant impurities.

For this, small signal data are also indispensable.

In the present arrangement, the contact needle is in direct mechanical contact with the specimen, which takes over the vibrations of the sample, thus no relative dislocation occurs. As a result, the vibrations of the surrounding have no significant effect on the measurement. To the contrary, in traditional arrangements the air gap between the specimen and the antenna can change due to vibrations, thus the signal to be measured may become noisy.

The advantage of the present arrangement is also that the measuring needle is very small (its diameter is smaller than 1 mm), thus it covers only a small field out of the surrounding of the specimen. Accordingly, the excitation of the specimen, e. g. its illumination by'light pulses can easily be carried out.