WO1987005701A1

WO1987005701A1 - Method for the examination of electrically active impurities of semiconductor materials or structures and measuring arrangement for carrying out the method

Info

Publication number: WO1987005701A1
Application number: PCT/HU1987/000016
Authority: WO
Inventors: Wolfgang Jantsch; György Ferenczi
Original assignee: Magyar Tudomanyos Akademia Mueszaki Fizikai Kutato; Magyar Tudomanyos Akademia
Current assignee: Magyar Tudomanyos Akademia Mueszaki Fizikai Kutato; Magyar Tudomanyos Akademia
Priority date: 1986-03-17
Filing date: 1987-03-17
Publication date: 1987-09-24
Anticipated expiration: 1988-09-17
Also published as: HUT43735A; US4839588A; JPH01500613A; CA1269762A; SU1669407A3; EP0260321B1; HU196262B; EP0260321A1

Abstract

A method for the examination of electrically active impurities of semiconductor materials or semiconductor structures which comprises the steps of providing a junction in a sample taken from the semiconductor to be tested, inserting the sample in a microwave field, providing a space charge layer in the junction by applying a reverse bias thereto, filling the electrically active defects of the space charge layer, examining the thermal emission process aiming at reaching a thermal equilibrium state that takes place following said filling step by measuring the change of the microwave field that takes place due to changes in the microwave absorption in the sample during the thermal emission process. The microwave field should be present at least during the examination of the transient of the microwave absorption. In a measuring arrangement for carrying out the method a sample (24) of the semiconductor comprises a junction, the sample is provided with a pair of electrical contacts, and the measuring arrangement comprises a biasing means (26) coupled to the contacts for reverse biasing the junction to provide a space charge layer therein, a means (26) for filling the electrically active defects in the layer during a predetermined period or periods, and transient detecting means (27) for detecting transient changes in the junction after termination of said periods, and the arrangement comprises furthermore a microwave generator (21), a microwave means (23) coupled to the generator which defines a microwave field, and the sample is arranged in the field of the microwave means with a contact coupled to earth, the transient detecting means is a microwave detector arranged to detect transient changes in the microwave absorption due to the changes in the junction.

Description

METH00 FOR THE EXAMINATION OF ELECTRICALLY ACTIVE IMPURITIES OF SEMICONDUCTOR MATERIALS OR STRUCTURES AND MEASURING ARRANGEMENT

FOR CARRYYING OUT THE METHOD

FIELD OF THE INVENTION

The invention relates to a method for the examination of electrically active impurities (deep levels) of semiconductor materials and to a measuring arrangement for carrying out the method . BACKGROUND OF THE INVENTION

Electrically active impurities such as foreign atoms, na¬ tive defects of the crystal or the complexes thereof can all exert a significant influence on the electrical and optical properties of semiconductor materials as well as of structures and devices made of such materials, therefore the examination of such impurities forms an indispensable method both of the research of semiconductor materials and of the quality control of manufacturing processes of active elements ^• in the field of microelectronics.

For investigating semiconductor materials and controlling active elements the sensitivity limit for detecting electrical¬ ly active impurities should be at least 10 atoms/cm . At present such a high sensitivity can be obtained by measuring a particular single process only.. In such a process a space charge layer is provided in the semiconductor under test which can be realized by depositing a suitable metal layer (i.e. form ing a Schott y diode), by forming a p-n junction or by estab¬ lishing a M0S structure, etc. In case of a reverse bias the space charge layer has insulating properties i.e. it does not comprise free carriers. Those portion of electrically active defects which fall within the space charge layer can be either in an electrically active or neutral state. In the process in question the active defects are filled with free charge car- riers and the subsequent thermal emission recovery process is examined . In a known examination method of this process the structure is electrically short-circuited and cooled down from room tem¬ perature to the temperature of liquid nitrogen, and in this lat ter temperature a reverse bias is applied to the sample. The electrically active defects remain saturated with free charge carriers that correspond to a noπ-equilibrum thermal state. The time constant of the recovery process to the thermal equilibrum is _n ^"1 = e_R = N_c$^" exp(-E_T/kT) (1) wherein is the time constant of the thermal emission process eπ is the probability ^J of the thermal emission

Nc is the density of the state in the valence band is the capture cross section v is the thermal drift velocity

Ey is the activation energy of the electrically active defects in electroπvolt units k is the Boltzmaπ constant T is the temperature expressed in Kelvin. In low temperature this time constant can be even some years long. It follows from equation (1) that the time constant decreases exponentially with increasing sample temperatures, and the thermal emission takes place when the characteristic temperature associated with the electrically active defects has been reached. The free charge carriers released during such emission can be detected by means of conventional current measurement or by detecting the changes of the capacitance of the sample. Corresponding known experimental methods are as follows: - Thermally Stimulated Current i.e. TSC is described e.g. by R.H. Bube: ' Phαtoelectroπic Materials and Devices' Ed. S. Larach pp. 100-139, 1965 (D. Van Nostard Comp.). Thermally Stimulated Capacitance, i.e. TSCap is described e.g. by Carabelles at al: Solid-St Communication 6_, 167, 1968).

Another widely used way of examining the thermal emission process is represented by the transient measurement technique. In such technique the sample under test is reverse biased a predetermined constant temperature and short circuited during perαdically repeated short intervals. During the short -circuiting intervals the defects are filled with free charg carriers and during re-establishment of the reverse bias thermal emission recovery process is started which has the cha racteristic time constant defined by the equation (1).

The filling of defects can be made not only in an electric al way by short-circuiting the sample but also by optica excitation, by an electron beam or by means of other kinds o ionizing radiation. The excitation should be, however periodically repetitive.

The thermal emission can be detected from the transien changes in the capacitance or current of the sample under test see e.g. R. Williams, 3 . Appl. Phys. 21* ³⁴¹¹ (1966).

The automatic detection and evaluation of the transient have been solved by means of Deep Level Transient Spectroscop (OLTS) technique. Such a method is described e.g. by Miller et al: Rev. os Sci. Instrum _, pp. 237-239, 1977 or in Hungaria patent No. 181.136. Owing to the possibility of their automate performance DLTS measurements have become the most widely use methods of examining thermical emission processes.

The use of DLTS technique is connected with a number o factors influencing or limiting the sensitivity, accuracy an conditions of the measurements. In the following discussio these limiting factors will be analysed in a more detailed wa because the understanding of these facts is thought to be in evitable for the correct evaluation of the prior art.

In the methods based on transient capacitance measurement the capacitance represented by the sample under test is con nected in a measuring bridge and excited by a high frequenc signal. The examined transient capacitance is represented by component of the high frequency output signal of the bridg which has a predetermined phase. The first limit of such mea surements is formed by the upper limit of the applicable fre quency. Owing to the serial resistance of the sample which i higher than zero, the following equation exists between the measured and actual capacitance of the sample:

C

in which

Cm is the cap ^racitance value detected byJ the capr-acitanc meter C, is the actual capacitance of the sample R. is the serial resistance of the sample ( j is the frequency of the capacitance measurement. In everyday practice there are samples which have seria resistances higher than 100 ohms and the capacitance of suc samples cannot be measured if the measuring frequency is highe than about 1 MHz. For that reason the operational frequency o the capacitance bridge in commercially used instruments is no higher than 1 MHz. In the practice, however, there exists number of semiconductor samples with serial ohmic resistance much higher than 100 ohm and one can even meet as high resis ance values as 1 Mohm. Such samples cannot be measured by DLT methods or if still measured in that way, the sensitivity o the measurement will drop well below the required level.

In addition to the limitations caused by the presence o the series resistance the shunting effects of the leakag currents flowing in the samples represent a further limitatio by decreasing the sensitivity. Owing to non-ideal surface o the semiconductor samples a certain amount of leakage curren is always present when being reverse biased. If as high sensi tivity is requested as mentioned hereiπabove then the maximu permitted leakage current should be about 1 ,uA. This conditio cannot be satisfied easily therefore it represents a furthe limitation regarding the types of samples that can be examine and/or the maximum sensitivity.

Of these reasons the maximum of the measuring freueπcy can not be higher than about 1 MHz which, however, limits the maxi mum frequency of the control pulses which alternatively excit and reverse bias the sample. In case of a measuring frequenc of 1 MHz, the response time of the capacitance meter is a least 5 ,us long and the actual measurement cannot start before the treble of the response time has elapsed. Even if the dura¬ tion of the measurement periods is chosen to be just as short as the combined duration of the exciting pulses and the sub- sequent dead periods, the maximum of the repetition frequency cannot be higher than about 25 kHz. This theoretical upper li¬ mit is substantially higher than the highest one of the repeti¬ tion frequencies used in the practice (see Hungarian patent 182.777). It is well known in the art that DLTS measurements are car¬ ried out generally when the temperature is varied. The changing of the temperature is disadvantageous because in addition to the comparatively long time of measurements (which can be typically between 20 minutes and two hours) it can result in the thermal treatment of the sample under test and can rearrange the structure of the defects. According to the DLTS measurement with constant phase position lock-in amplifier as disclosed e.g.. in Hungarian patent 181.136 frequency-scan DLTS measure¬ ments with constant temperature can be carried out, however, the attainable frequency range is limited by the maximum fre¬ quency of capacitance measurement which is about 25 kHz and of practical considerations the lowest frequency cannot fall below 0.25 Hz. With such upper and lower limits the frequency range cannot exceed more than 5 decimal orders of magnitude. In the paper of G. Fereπczi : ' The Examination of electrical¬ ly active impurities of semiconductor materials and structures' (Hiradastechπika XXXVI. 1985. 10. pp. 451-454) the frequency scan DLTS measurement is described which can offer a deep level spectrum that has an extreme value proportional with the emis- sion time constant characteristic to the type of the particular impurity. With the attainable maximum frequency coverage of

4 5 10 -10 and in case of a given temperature, it is possible to detect deep levels falling in the range of activation energies between E=0.2 - 0.3 eV. Of that reason the practically sig- nificant range of activation energies between 0.05 eV to 0.7 eV can be covered by the frequency scan method if the measurements are repeatedly carried out in different temperatures. The cor- responding temperature range is typically between 240K and 330K. For enhancing the range of frequency scan measurements in a predetermined single temperature to cover the energies of 0.05 to 0.7 eV it is required that the frequency coverage be as high as 10 . Due to the practical limitations explained above such a wide range was not realizable. Of that reasons the examination of the full deep level spectrum inevitably requir¬ ed the changing of the measuring temperature that was connected with the drawbacks associated with the unwanted thermal treat- meπt (annealing) of the sample.

A further limiting feature lies in the relative character of the sensitivity of capacitance DLTS measurements. It is a well known fact that in such measurements:

N_τ ^C ~ (2)

2N_r in which

N_j is the deep level concentration

N_η is the shallow level dopant concentration -C is the capacitance change

Co is the capacitance of the sample,

The lowest practically measurable capacitance change is about 2 x 10^" pF (see e.g. the Hungarian patent 182.777). Taking this fact into consideration the maximum attainable sensitivity is minimum of N_T/N_D) = 2. 10^"6 (4) In case of typical concentration of dopants, this sensitiv¬ ity represents a detection limit level of 10 atoms/cm . One should, however, bear it in mind that in case of higher coπ- ceπtration of dopants the detection limit decreases, thus in such samples the required sensitivity cannot be reached.

The above referred paper of G. Ferenczi refers also to the fact that in DLTS measurements the capture cross section can be determined by changing the width of the exciting pulses. The accuracy of such a measurement is limited by the minimum width of the exciting pulses which is about 1-2 ns and with such data higher capture cross section than r— =10-15 cm2 caππot be measured, however, the largest value which should be

-12 2 measured is 0" =10 cm .

This limit can be derived from the fact that in case of a capacitance measurement the sample under test should be arraπg- ed in the bridge in an isolated way i.e. unearthed. Since the measurements are carried out under varying temperatures, the practically realizable measuring arrangements are using at least 30 cm long connecting cables. With such cable lengths one cannot apply shorter exciting pulses than 1-2 ns . Summarizing the above thoughts it can be stated that the practically attainable best parameters of capacitance DLTS measurements are as follows: the maximum sensitivity N_τ/N_n.- 10^" which corresponds to a detecting limit of y 10 10 atoms/cm3 , - the maximum coverage range of the repetition frequency of the exciting pulses is about 10 , the minimum width of the exciting pulses is 1-2 ns, the highest tolerable leakage current is 1 ,uA (in case of maximum sensitivity), - the highest measuring frequency is 1 MHz, the highest series resistance of the sample is about 100 ohm. It is known in the art that very small changes of microwave absorption can be measured with a high accuracy e.g. in a micro wave cavity. Such measurements are used e.g. for the examina¬ tion of the paramagnetic resonance of electrons (see e.g.: G. Feher, Bell System Technical Journal 3__L_> pp. 444-484, 1957). By means of microwave absorption measurements transient phenomena can also be detected. The measurement of lifetime of minority charge carriers by changing the microwave absorption as a function of time has been described first by Jacob et al. in the Proceedings of the IRE 4_8, pp. 229-233, 1960. During the lifetime measurements of minority charge carriers the change in microwave absorption due to the presence of πoπ-equlibrum free charge carriers is detected. This method has become a widely used technique (e.g. R. I. Desi et al. Rev. Sci. Iπstrum. 55 , pp. 1343-1347, 1984). During lifetime measurements but also during other kinds o measurements based on microwave absorption the sample unde test is not provided with contacts for electrical connections Without electrical contacts the generated πoπ-equilibru carriers will move away with thermal drift velocity in th material sample and their number changes with the recombinatio process. The presence of free charge carriers in the sample i therefore determined by the recombination process. The abov referred measurement technique, which aims at detecting th lifetime of minority carriers in the sample, is principally in appropriate for determining the time constant of thermal emis sion.

OBJECT OF THE INVENTION

The object of the invention is to provide a method and measuring arrangement for the examination of electricall active impurities of semiconductor materials and structures, i which the detection of the thermal emission process is based o an effect which is basically different from those of the aforε mentioned conventional methods and in which the factors limit ing the performance of the measurements are much less signifi cant.

SUMMARY OF THE INVENTION

The method according to the present invention is based di rectly on the recognition according to which microwave absorp tion technique can be used for detecting and measuring th thermal emission of charged electrically active impurities of sample under test which emission takes place when a spac charge has been established, such a use requires, however, tha electrical connections be made to the semicαπdutor junctio forming the sample to provide the electrical voltage that esta lishes the space charge and to periodically fill the completel or partially depleted deep levels. By chosing an appropriat measuring arrangement the presence of the electrical connec tions to the sample cannot affect the microwave field signifi cantly in the test region i.e. in a microwave cavity or wave- gu ide .

According to the invention a method has been made for the examination of electrically active impurities of semiconductor materials or semiconductor structures which comprises the steps of providing a junction in a sample taken from the semiconduct¬ or to be tested, inserting the sample in a microwave field, pro¬ viding a space charge layer in the junction by applying a re¬ verse bias thereto, filling the electrically active defects of the space charge layer, examining the thermal emission process aiming at reaching a thermal equilibrum state that takes place following said filling step by measuring the change of the microwave field that takes place due to changes in the microwave absorption in the sample during the thermal emission process. The microwave field should be present at least during the examination of the transient of the microwave absorption.

The filling of the defects can be made by a single shot, however, it is often preferable if this step and the subsequent examination are periodically repeated.

In that case the repetition frequency can be changed to cover a range of at least 6 decimal orders of magnitude and/or the duration of the filling pulses is changed and the transient response can be examined in case if the filling pulses are shorter than 1 ns.

By changing the repetition frequency and/or the width of the filing pulses pure frequency scan measurements can be made which can be carried out at a single temperature.

It is most convenient if the filling of the deep levels is provided by the application of an electrical exciting pulse lead to the sample which decreases or eliminates the space charge region.

Instead of such an electrical pulse the deep levels can be filled also by light, by an electron beam ^'or by means of an other kind of radiation.

In case of certain kinds of examinations it can be preferable if the measurements are carried out at different constant temperatures or if the temperature of the sample is continuously changed. The semiconductor to be tested may comprise already a junc tion, however, if this is not the case the junction require for the measurements can be provided by forming a Schottky bar rier, a MOS capacitor or a p-π junction in the sample. It is preferable if the sample is exposed to the microwav field in a microwave resonator or in a microwave reflectometer. The transient measurements can be facilitated if the micro wave field is substantially eliminated during the filling step. This can be achieved not only by switching off the microwav field for the time of the filling pulses but also by de-tunin the microwave frequency to an extent which cannot disturb (ove drive) the sensitive microwave detection.

According to the invention a measuring arrangement has als been provided for the examination of electrically active impur ties of semiconductor materials or semiconductor structures, i which a sample of the semiconductor comprises a junction, th sample is provided with a pair of electrical contacts, and th measuring arrangement comprises a biasing means coupled to th contacts for reverse biasing the junction to provide a spac charge layer therein, a means for filling the electrically ac tive defects in the layer during a peredeter ined period o periods, and transient detecting means for detecting transien changes in the junction after termination of said periods, an the arrangement comprises furthermore a microwave generator, microwave means coupled to the generator which defines a micro wave field, and the sample is arranged in the field of th microwave means with a contact coupled to earth, the transien detecting means is a microwave detector arranged to detect tra sient changes in the microwave absorption due to the changes i the junction.

In a preferable embodiment the biasing and filling mean are implemented by a pulse generator that establishes periodic al filling pulses and provides a reverse bias to the junctio during intervals of the pulses. In a further preferable emodiment a transmission line con nects the pulse generator to the sample.

In an other preferable embodiment the microwave means is cavity resonator and the sample is arranged in the resonator at the maximum of the elecrical field. ,

In an alternative embodiment the microwave means is a micro¬ wave reflecto eter.

It is also preferable if the temperature of the sample can be changed and to this end the arrangement comprises -a tempera¬ ture controlling means in which the sample and the microwave means can be arranged.

DESCRIPTION OF THE DRAWINGS

The invention will now be described in connection with preferable embodiments thereof, in which reference will be made to the accompanying drawings. In the drawing:

FIG. 1 is a general schematic diagram of a first embodiment of the measuring arrangement according to the inven¬ tion;

FIG. 2 is the block diagram of a further embodiment of the arrangement using a microwave bridge;

FIG. 3 shows a third emodi eπt of the arrangement;

FIG. 4 shows the Arrhenius plot curve derived from the results of measurements made by the arrangement of FIG. 2;

FIG. 5 shows the results of isothermal frequency scan mea¬ surements made at 4 different temperatures by the arrangement of FIG. 3; and

FIG. 6 shows the Arrhenius plot corresponding to the mea¬ surements of FIG.5

DETAILED DESCRIPTION OF THE PREFERABLE EMBODIMENTS

For using the method according to the invention an experimental setup is required which allows the time-resolved detection of the microwave absorption of a Schottky diode or a p-π junction or a MOS structure, etc. with a maximum sensitivity. Fig. 1 shows schematically an arrangement usable for such measurements. A frequency and amplitude stabilized microwave generator 2 (which can be a Klystron, Gunn oscillator, IMPATT diode, F oscillator, etc.) is connected to first port of circulator 22 A next allowed exit port of the circulator 22 leads to icro wave resonator 23 in which sample 24 is arranged, preferably i a maximum of the electrical field. The sample 24 is made of th material of investigation, within which the space charge regio can be established. The sample 24 has two electrical connec tions, one of them is earthed and the other one is connected pulse-transfer line 25 including an impedance matching networ Transfer line 25 and wiring within the resonator 23 should b arranged in such a way that the quality factor of the cavit formed by the resonator 23 is not impaired. In a cavity typ resonator this requirement can be achieved by placing wires o coaxial cables along a "knot" line, where the electric micro wave field vanishes.

By adjusting the microwave frequency and the coupling o the resonator 23, the microwave reflected by the resonator 2 (including the sample 24 and a portion of the transfer line 25 is minimized. This procedure is often referred to as 'critica tuning'. The reflectivity itself is measured by means of micr wave detector 27 coupled to the next port of the circulator 22 The microwave detector 27 can be any suitable type, e.g. a cr tal rectifier, a microwave Schottky diode with fast respons time, etc. For a critically tuned resonator, any change i absorption causes a finite reflectivity, which is proportiona in magnitude to the change in absorption.

The detected signals of the microwave detector 27 ar amplified by selective amplifier 28, and the amplified signa are evaluated by computer 29 pogrammεd to carry out su measurements. Computer 29 is capable of controlling the oper tion of pulse generator 26 that transmits via the transfer li 25 the dc bias pulses to the sample 24.

The emission of carriers from deep levels, which forms t phenomenon to be investigated by the present invention, triggered periodically by first filling these deep levels the application of a short electric pulse, which reduces the width of the space charge layer by reducing the reverse bias in the sample 24. By this filling pulse free carriers are swept into the space charge layer and a fraction of them is captured by deep level states. After the end of this filling pulse these captured charge carriers are emitted again with an emission rate which is characteristic to the trap under investigation. Both capture and emission change the number of free carriers in the sample and hence the microwave losses, i.e. the absorption. In the measuring arrangement shown in Fig. 1 the microwave ab- sorption is obtained from the detector 27 as a function of time .

In order to determine the emission rate, which is the in¬ verse of the time constant for the recovery of the microwave reflectivity following the filling pulse, the selective ampli- fier 28 is designed to selectively amplify the detected signal with respect to unavoidable background noises, whereby increas¬ ed sensitivity can be reached.

Since a leadout contact of the sample 24 is earthed, both proper impedance matching and short transmission lines can be realized, and such an arrangement allows the application of narrow exciting pulses which can be as short as 100 ps.

A further arrangement for measuring the microwave absorp¬ tion is shown in Fig. 2, which uses a microwave bridge. The microwave circuitry comprises stabilized microwave generator 1, isolator 2 coupled to the output of the generator 1, a controlled switch 3 for disconnecting the passage of microwave signals in respose to control pulses applied to its control input and a microwave bridge coupled to the output of the switch 3. A first branch of the bridge comprises a first attenuator 4a with adjustable attenuation, a circulator 5 with a first port connected to the output of the first attenuator 4a, a cavity resonator 6 connected to a further port of the cir culator 5 and the end of the first branch is coupled to a third port of the circulator 5. The second branch of the bridge comp- rises a second attenuator 4b with adjustable attenuation coupl¬ ed also to the output of the switch 3, and an adjustable phase shifting means 7. The two ends of the two branches are commoπed and connected to microwave detector 8. The output of detector is coupled via amplifier 9 to input of a phase-sensitive ampli fier 11. An automatic frequency adjusting circuit 10 is fe back from the output of the amplifier 9 to the generator 1, th operation of which can be switched off for the time o transient measurements not to interfere therewith or whic should have a response time much longer than that of th measured transient.

The output of the phase-sensitive amplifier 11 is connecte to a signal processing unit 15 which in turn is coupled t computer 16 that evaluates and processes data obtained from th measurements .

The phase-sensitive amplifier 11 comprises a synchronizin input coupled to output of a pulse generator 14 which latte has two further outputs coupled respectively to trigger inpu of a further pulse generator 13 and to the control input of th microwave switch 3. The output of the second pulse generator 1 is connected via transmission line 12 to sample 17 arranged i the cavity 6. The transmission line 12, the sample 17 and th cavity 6 can be similar to those shown in Fig. 1. It is wort mentioning that a connection of the sample 17 is earthed an the transmission line 12 is capable of passing dc bias to th sample from the generator 13. The transmission line ha preferably a 50 ohm characteristic impedance. The cavity 17 can be of any suitable type, e.g. one of th standard cavities produced by the company Bruker Analytisch Mestechnik GmbH, Silberstreifen, Germany and its loaded qualit factor should be between about 1000 and 10.000. Further re quirements that can be imposed on the resonator 6 lie in th possibility of providing electrical connections to the sampl therein, the possibility of irradiating the sample by means o light or any other kind of radiation and finally that it shoul be iπsertable in a temperature regulator.

The second pulse generator 13 can be any suitable type such as type AVP-AV of the Canadian company AVTECH, Ottawa This pulse generator can generate pulses superimposed on an ad justable dc level, thus it is appropriate for reverse biasin the diode junction forming the sample and periodically short- -circuitiπg or forward biasing the same.

The microwave bridge can be realized by a number of com¬ mercially available types, and in the following example a stan- dard bridge of the Bruker company was used with an operational frequency of 9.6 GHz. The microwave frequency can be chosen to any suitable value between about 1 and 100 GHz and this selec¬ tion depends primarily on the type of the available bridge and on the requirements imposed on the measurements. The duration of the shortest detectable transient signals is limited primarily by the loaded quality factor Q, of the resonator 6. The resonance bandwidth of the cavity resonator 6 can be defined as: ω f = (5)

3 Q, wherein : ^ is the frequency of the microwave generator 1 and Q, is the loaded quality factor of the cavity resonator 6. The multiplier 3 in the denominator of equation (5) comes from the definition of the signal drop from 90% to 10% as the bandwidth. With the above referred typical Q, values varying between about 1000 and 10.000, the own response time of a transient measurement will be L = 1/ £_ f which is 5 ns or longer. A corresponding response time in transient capacitance measurements is 5 ,us or longer, thus the microwave detection of thermal emission (with a frequency of about 10 GHz) increas¬ es the measurable range of shortest thermal emission transients by a factor of at least 1000 compared to conventional tech- niques.

The timing of the measuring arrangement is controlled by the first pulse generator 14. This generator 14 delivers timing signals for the second pulse generator 13 that applies the ex¬ citing pulses to the sample 17. Generator 14 controls the phase position of the phase-sensitive amplifier 11 which receives de¬ tected signals representing the changes of microwave absorp¬ tion in time, amplifies these signals and provides the average of the amplified signals phase-locked to a predetermined phas position defined by the control pulses from the pulse generato 14. The microwave signals applied to the microwave bridge ar switched off by means of switch 3 for the duration of the exci ing pulses coupled to the sample, and the corresponding gatin pulses are generated by means of the pulse generator 14 for th switch 3.

The processing of the averaged transient signals at th output of the phase-sensitive amplifier 11 can be made in number of ways. According to a suitable embodiment the signa processing unit 15 can be a fast transient recorder capable of digitizing and averaging the input signals representing the transients of the microwave absorption. The computer 16 is used for evaluating the output data of the transient recorder. A further embodiment of a measuring arrangement for detect¬ ing transients of microwave absorption is shown in Fig. 3. The arrangement comprises microwave generator 30 coupled to cavit 31 in which a sample (not shown) is arranged as in case of th previous embodiments. Microwave absorption in the cavity 31 i sensed by microwave detector 32, and the sample is controlle by pulse generator 33 just as in case of the previous embodi ments. Preamplifier 34 amplifies the detected signals to a sui able level, and sample and hold circuit 35 receives th samples. The output of the sample and hold circuit 35 is cαupl ed to input of a lock-in amplifier 36 synchronized by outpu pulses of the pulse generator 33 in such a way that an averag ing takes place which is in a predetermined phase-position rel tive to the exciting pulses applied to the sample. The pulse g nerator 33 controls the sample and hold operation, too. Th output of the lock-in amplifier 36 is coupled to computer 3 for processing and evaluation. The temperature of the cavity 3 with the sample therein can be set and adjusted by means o temperature controller 38 reporting the actual temperature t the computer 37. In an alternative embodiment the microwave cavity 31 can b replaced by a microwave reflectometer. The sample is arrange at an end portion of a waveguide excited by the generator 30 aπd a reflector of suitable size is placed behind the sample with a predetermined and adjustable spacing. The arrangement is adjusted in such a way that the reflected microwaves from the sample and from the reflector fully offset each other at the location of the detector 32. When the absorption changes in the sample, the compensation becomes less effective and the detect¬ or 32 detects the change in microwave level.

It is also preferable if the microwave level reaching the detector is substantially eliminated during the filling pulses. This can be achieved rather by de-tuning the generator 30 than physically disconnecting the microwave path. In that case the switch 3 in the Fig. 2 embodiment can be spared and an appropri¬ ate de-tuning signal should be applied to the generator. Obviously, the control of this function is identical with that of the switch 3.

The method according to the invention will now be illust¬ rated by means of two examples.

Example 1 The measuring arrangement shown in Fig. 2 was used. The microwave bridge was realized with an X band bridge of the aforementioned Bruker company operating at 9.6 GHz. The cavity resonator 6 was placed in a cryostat capable of circulating liquid helium made by the British company Oxford Instruments. The semiconductor sample 17 under test was placed on a cylindrical quarz rod with a diameter of 4 mm, and electrical contacts to the sample were made by means of respective gold wires of 20 urn diameter having lengths of 2 cm. This very short and thin connection was required for minimizing the load transformed in the cavity. Outside the cavity 6 the connection to the pulse generator 13 was made via a short coaxial cable of 50 ohms characteristic impedance. The resistance of the semi¬ conductor sample was 10 ohmcm and the sample itself was an type silicon crystal doped with phosphor. The crystal was heat treated in a nitrogen atmosphere at a temperature of 400 °C, whereafter a wafer of 3mm x 2mm x 0.4 mm was cut from the crystal. A gold spot of 0.8 mm diameter was vacuum-deposited on a surface of the wafer and this gold layer formed the firs electrode of a Schottky diode. The ohmic contact on the othe side of the wafer was realized by building an indium allo thereon. Under the effect of the heat treatment thermal donor were created in the sample with a concentration of 5 .10 atom/cm . During the measurement af thermal emission the +/+ transition of the filled thermal donors were detected by mean of detecting the changes of microwave absorption.

The sample in the cavity 6 was measured in seven differen constant temperatures in the temperature range between 55.5 K and 80.4 K°. The sample was reverse biased by -10 V and it wa periodically excited by pulses of 9 V amplitude and 20 ,u width. The period time of the exciting pulses was continuousl varied between .1 sec and 10 sec. The microwave power wa switched off by means of switch 3 during the existence of th exciting pulses. The changes of the transient microwave absorp tion detected by the detector 8 was averaged by means of th signal processing unit 15. This operation was carried out b means of a lock-in amplifier controlled and synchronized as de scribed in the Hungarian patent 182.777. The results of the measurements are illustrated in Fig. 4 in which the points indicated by x correspond to the frequen cies associated with the signal maximums in the respectiv temperatures. The Figure is an Arrhenius plot of the tempe rature-dependency of the signal maximum as a function of fre queπcy which is used for determining the activation energy o the deep level under test. The measured activation energy wa Eγ= 0.112 eV which value fully coincides with the valu measured by conventional DLTS technique for the +/++ transitio of the thermal donor (see Kimerling et al. ' Appl. Phys. Let ters 21 PP- 410-412, 1981').

Example 2 The measuring arrangement of Fig. 3 was used to detect th microwave absorption transients of π-type FZ Si wafers with free carrier concentration of 8 x 10 12 cm-3 , doped with 1 10 11 cm-3 Au. 2 cm2 evaporated gold Schottky contacts an soldered In ohmic contacts were prepared. The wafer was mounte with 20 ,um gold wires on a temperature controllable stage of modified minority carrier lifetime measuring equipment, origin¬ ally manufactured by Leo-Gikkeπ Co. of Japan. Microwave gene¬ rator 31 of Gunπ diode type was used at a frequency of 9.6 GHz, the cavity 31 was open ended which functions as an antenna for the reflected microwave power, a temperature controllable wafer stage and a reflector were placed above it. The microwave de¬ tector 32 was a crystal detector and the microwave preamplifier 34 both had a frequency response from dc to 10 MHz. The sample was reverse biased to 20 V, and majority carrier filling pulses of 20 /us duration and 20 V amplitude were applied periodical¬ ly-

The amplified microwave signal was fed into the lock-in amplifier 36 via the sample and hold circuit 35 which gated off the perturbations caused by the filling pulses. The whole sys- tern was controlled via an IBM PC controller i.e. computer 37. The lock-in amplifier 36 and the associated stages were part of a deep level spectrometer DLS-82E manufactured by Semitrap of Hungary. The sample was kept at different constant tempratures and frequency scan measurements (e.g. as described in " G. Fe renczi et . al, Phys. stat. sol. (a), 1986, 9_4_, K119") were per formed between 2 Hz and 2.5 kHz. The resulting spectra are il lustrated in Fig. 5. The negative going peak at 2 kHz is a artifact of the measuring system (due to the capacitive coupl ing of the filling pulse via the stray capacitances). The four positive going peaks measured at 241, 245, 255 an 267 K correspond to the emission of free carriers from the A acceptor levels. The measured absorption signal originatin from the change of the widths of the space charge layer due t the recharging of the Au acceptor level corresponds to a chang of the microwave reflection coefficient caused by microwav absorption on 10 electrons.

It has to be noted that the microwave absorption signal am litude remains constant independent of the temperature in cont rast to capacitance DLTS measurements where the change i sample capacitance with temperature influences the signal amp litude. The Arrhenius plot corresponding to the results of Fig. 5 is shown in Fig. 6. The determined activation energy 0.56 e is in agreement with the traditional space-charge spectroscop data (O.V. Lang et al., Phys. Rev, 1980, B22, 3917).

Based on the above examples and the preceding considera tions, it can be determined that compared with the transien capacitance DLTS measurements the substantially higher measur ing frequency of the microwave signal (10 to 100 GHz compare to 1 MHz of the capacitance measuring signal) and the resultin shortening of the response time, which is at least 3 decimal o ders of magnitude, have substantially enhanced the time rang of the measurable transients. The pulses exciting the sampl can be varied continuously practically between .01 Hz and 10 MHz i.e. the time coverage is about 10 compared to the 10 value in case of capacitance DLTS measurements. With such high time coverage the frequency scan method is capable of exa iniπg activation energies between .1 and .7 eV at room tem perature. The shortest applicable width of the exciting puls is 100 ps due to the reasons discussed above and this shor pulse provides also for the possibility that the capture cros section be measured up too = 10 -12 cm2 by means of changing th width of the exciting pulses. This upper limit is sufficient t the measurement of the practically highest deep level cros sections, therefore such measurements can also be carried ou without changing the temperature of the sample.

By saving the necessity of changing the temperature, no only the time required for carrying out the measurements i reduced, but also the complexity of the equipment.

The smallest concentration detectable by the microwav absorption technique can be determined from the value of th smallest detectable change in absorption. If the detecting seπ sitivity of a Bruker type measuring system is considered t represent the highest available sensitivity, then according t the data given in the SRC Series Technical Manual of Bruker Co the smallest detec is:

wherein

£_, E is the change of the dectric field and

Eo is the electric field in the cavity '. It can be calculated from the equation (6) that in case of a highly conductive and/or large sample

<W mln ' ⁶ ' ^10~7' ⁽⁷⁾ while in case of a less conductive and/or smaller sample (i.e. if the presence of the sample does not change the quality factor of the cavity to a significant extent):

From this comparison it follows that even in the least favour¬ able case microwave absorption measurements can provide the sensitivity comparable with that of capacitance DLTS technique (see equation (4)) and in other cases the sensitivity is signi¬ ficantly higher and, what is even more significant, the thresh¬ old of detection is independent from the concentration of the free electrons . Finally, it should be mentioned that by means of microwave absorption technique the changes in the concentration of the free charge carriers are measured, thus the measurement is not affected by the electrical parameters of the sample. In cont rast to conventional capacitance or current DLTS measurement neither the series resistance nor the leakage current of th sample can limit the preformance of the measurements unti space charge can be established in the sample.

Although a few number of methods of measuring microwave ab sorption have been shown, it must be appreciated that a numbe of other ways are also available for such measurements. Of suc possibilities the application of a heterodyne detection syse worth mentioning. In such a system a second microwave source i introduced whose frequency differs from that of the first on by a predetermined value. The detector can be replaced by mixer which mixes the reflected microwave frequency f, with th frequency f of the additional source. The output of the mixe contains both the sum and the difference of the two frequencie of which the component fi-f_? ^^s selectively amplified by a intermediate frequency amplifier of suitable bandwidth. Thi component is also modulated by the periodic capture an emission processes. Such a principle is widely used i microwave signal reception and other fields of application. Th invention can not be limited to the way how the microwave ab^¬ sorption is sensed.

Claims

C la ims :

1. Method for the examination of electrically active impurities of semicondutor materials or semiconductor struc¬ tures, comprising the steps of providing a junction in said sample, providing a space charge layer in said junction by re- verse biasing said junction, filling the electrically active defects thereof, examining the thermal emission process aiming at reaching a thermal equilibrum state that takes place following said filling step, characterized by the steps of inserting said sample in a microwave field at least during said examination step and in said examination step measuring the change of said field that takes place due to changes in the microwave absorption in said sample during said process.

2. The method as claimed in claim 1, further comprising the step of periodically repeating said filling and examination steps.

3. The method as claimed in claim 2, wherein in association with the examination of a particular sample changing the fre¬ quency of said repetition to cover a range of at least 6 decim¬ al orders of magnitude.

4. The method as claimed in claim 2 or 3, wherein in asso¬ ciation with the examination of a particular sample changing the time of duration of said filling step and carrying out said examination even if said duration is shorter than 1 ns .

5. The method as claimed in claim 4, wherein said examina- tioπ is carried out at a single temperature.

6. The method as claimed in claim 1, wherein said fillin step is provided by the application of an electrical excitin pulse lead to said sample decreasing or eliminating said spac charge region.

7. The method as claimed in claim 1, wherein said fillin step is provided by light, by an electron beam or by means o an other radiation.

8. The method as claimed in claim 1, wherein said examina tions are carried out at different constant temperatures.

9. The method as claimed in claim 1, comprising the step o contiπuously changing the temperature of said sample.

10. The method as claimed in claim 1, wherein said junctio is provided by forming a Schottky barrier, a MOS capacitor or p-π junction in said sample.

11. The method as claimed in claim 2, wherein said sampl is exposed to said microwave field in a microwave resonator o in a microwave reflectometer.

12. The method as claimed in claim 11, further comprisin the step of substantially eliminating said microwave fiel during said filling step.

13. Measuring arrangement for the examination of electri cally active impurities of semiconductor materials or semicon ductor structures, in which a sample (17, 24) of said semicon ductor comprises a junction, said sample being provided with pair of electrical contacts, said arrangement comprising biasing means coupled to said contacts for reverse biasing sai junction to provide a space charge layer therein, a means fo filling the electrically active defects in said layer during peredetermiπed period or periods, and transient detecting mean for detecting transient changes in said junction after termina tion of said periods, characterized by comprising a microwav generator (1,21,30), a microwave means (6,23,31), coupled t said generator and defining a microwave field, said sample be ing arranged in the field of said microwave means with a con tact coupled to earth, said transient detecting means being microwave detector (8,27,32) arranged to detect transien changes in the microwave absorption due to said changes in sai junction.

14. The measuring arrangement as claimed in claim 13, wher in said biasing and filling means being a pulse generator (13

26, 33) establishing periodical filling pulses and revers biasing voltage during intervals of said pulses.

15. The measuring arrangement as claimed in claim 14 characterized by comprising a transmission line (25) connectin said pulse generator (26) to said sample (24).

16. The measuring arrangement as claimed in claim 13, i which said microwave means is a cavity resonator (6,23,31) an said sample (17, 24) is arranged in said resonator at the maximum of the elecrical field.

17. The measuring arrangement as claimed in claim 13, in which said microwave means is a microwave reflectometer .

18. The measuring arrangement as claimed in claim 13, fur¬ ther comprising a temperature controlling means (38) adjusting the temperature of said sample to predetermined values.