FR2814547A1 - Electrical property testing method of ICs, obtains laser and reference interference signals by enabling overlapping of primary and secondary laser signals and reference signals within given time respectively - Google Patents

Abstract

Primary laser and reference pulses generated at different time periods and secondary laser and reference pulses obtained by specific time delays of primary pulses, are irradiated onto IC for every input of test pattern signal so that primary and secondary laser and reference pulses overlap within given time to generate laser and reference interference signals, based on which electrical property is inspected. The time for generation of the laser pulses and reference signal is changed while repeating input of test pattern signals so that function of the laser interference signal and reference interference signal is determined at predetermined time periods. An Independent claim is also included for electrical property testing device using dual pulse optical interferometer.

Description

<Desc / Clms Page number 1>

 The present invention relates to the optical test of integrated circuits and more particularly relates to the test, by optical interferometry, of the electrical activity of integrated circuits.

 The electrical activity in integrated circuits can be optically controlled by means of the detection of variations in the refractive index and the absorption coefficient, which are caused by modification of polarization voltages and carrier densities at the level of semiconductor diode junctions. For example, variations in the refractive index and the absorption coefficient of the semiconductor material can be manifested by variations in the intensity of an optical beam reflected by the junction of the diode.

 Paniccia et al., In US Pat. No. 5,872,360, describes the detection of an electric field in the active regions of an integrated circuit. In one embodiment, a laser beam is produced with a wavelength close to the band gap of a semiconductor such as silicon. The laser beam is focused on a P-N junction (diode) such as for example the drain region of an MOS transistor, via the rear face of the semiconductor substrate. The beam passes through the junction, is reflected at the oxide interface and the metal behind the junction, returns through the junction and leaves the silicon surface. A modulation of the electric field in the junction causes an amplitude modulation in this reflected beam, due to the combined effects of an electro-absorption and an electro-refraction.

 Wilsher et al. describes, in US Pat. No. 5,905,577, the testing of integrated circuits using laser beams. A sounding beam is used to sample the waveform in a tested integrated circuit device (also referred to below as a DUT device) during

<Desc / Clms Page number 2>

 each cycle of a test profile signal applied to the DUT device. A reference laser beam is also used to sample the DUT device at the same physical location as that sampled by the probe beam. Each reference measurement is made at a fixed instant with respect to the test profile, while the survey measurements are scanned by means of the test profile, in a normal manner equivalent to time sampling, to reconstruct the form d 'wave. For each test cycle, the ratio of the probing measurement and the reference measurement is formed so as to reduce fluctuations in the probing measurement due to noise.

 Variations in the refractive index and the absorption coefficient due to an electrical activity in a DUT device can also appear in the form of phase modulation of an optical beam transmitted or reflected by the DUT device. Probes for detecting phase modulation may be more sensitive to electrical activity in a DUT device, than are probes for detecting reflectivity.

Heinrich et al. described in US Pat. No. 4,758,092, a method of interferometric measurement of a phase modulation of an optical beam by means of an active semiconductor device. An optical beam is divided into two beams which are focused on the DUT device. One beam is focused in an active region, in which it is modulated in phase by a modulated refractive index, and the other beam is focused on an inactive region so as to provide a reference. The reflected beams are recombined and interfere with each other. A modulation of the intensity of the beams, which interfere, is due to a modulation of the phase of the sounding beam due to an electrical activity in the sounded region.

 Typically, interferometric methods

<Desc / Clms Page number 3>

 are extremely sensitive to vibrations and temperature variations. In the method of Heinrich et al. the relative phase of the two beams, which interfere, is nominally insensitive to displacements in line with the axis of the beam, which are caused by vibrations of the DUT device due to the temperature of the DUT device in line with the axis of the beam. However, in a high integration density circuit, it is very difficult to find suitable points for placing the reference beam. Similarly, there is still a sensitivity to a movement transversely with respect to the axis of the laser beam.

 What is needed is an interferometric method for probing integrated circuits by means of waveforms, which is insensitive to vibrations and displacements and which is insensitive to displacements of the DUT device, caused by vibrations and by temperature, and which does not require finding an appropriate reference point on the DUT device near the active region in which one is interested.

 A method of detecting electrical activity in a semiconductor device when an electrical test profile is applied to the device includes producing a first pulse of sounding light at a selected time during each repetition of the electrical test profile and producing a first reference light pulse at a time during each repetition of the electrical test profile, offset from the selected time at which the first sounding light pulse is produced. The first probe light pulse and the first reference light pulse are each divided to form at least a second probe light pulse and a second reference light pulse. The first impulse of

<Desc / Clms Page number 4>

 sounding light and the first reference light pulse are directed to a region of the semiconductor device.

 After the first probe light pulse and the first reference light pulse have cooperated with the semiconductor device, the first probe light pulse is combined with the second probe light pulse so that they overlap in space and time, and the first reference light pulse is combined with the second reference light pulse so that they overlap in space and time. The overlapping probe light pulses are detected by outputting a probe interference signal, and the overlapping reference light pulses are detected by providing a reference interference signal. The selected instant is modified with the repetition of the electrical test profile. The ratio of the probe interference signal and the reference interference signal is determined in a plurality of the instants selected in the electrical test profile.

 An associated device includes a source producing a first probe light pulse, a source producing a first reference light pulse, a divider, a support for the semiconductor device, a beam combining device, a detector system and a processor for determining a ratio of the sounding interference signal and the reference interference signal.

 More precisely with regard to a method for detecting activity in a semiconductor device when a repetitive electrical test profile is applied to the device, characterized in that it consists in:

<Desc / Clms Page number 5>

 producing a first probe light pulse at a selected time during each repetition of the test profile and dividing the probe light pulse to produce at least a second probe light pulse; producing a first pulse of reference light during each repetition of the test profile at an instant which is offset from the selected instant at which the first pulse of probing light is delivered, and dividing the first pulse of reference light to delivering at least a second reference light pulse; directing the first probe light pulse and the first reference pulse to the semiconductor device; combining the first probe light pulse with the second probe light pulse and the first reference light pulse with the second reference light pulse after the first probe light pulse and the first reference light pulse have cooperated with the semiconductor device, so that the first probe light pulse and the second probe light pulse overlap in space and time and that the first reference light pulse and the second light pulse of reference overlap in space and in time; detecting overlapping probing light pulses to thereby obtain a probing interference signal; detecting the overlapping reference light pulses to thereby obtain a reference interference signal; determining a function of the probe interference signal and the reference interference signal.

<Desc / Clms Page number 6>

 According to another characteristic of the invention, the method further comprises modifying the selected instant with the repetition of the test profile so that the function of the sounding interference signal and the reference interference signal is determined in a plurality of selected moments within the test profile.

 According to another characteristic of the invention, the first probe light pulse is a laser light pulse.

 According to another characteristic of the invention, the method comprises the selection of the laser light pulse from a train of laser light pulses in locked mode.

 According to another characteristic of the invention, the method further comprises performing a phase lock of the train of pulses of laser light pulses in locked mode and of the repetitive test profile.

 According to another characteristic of the invention, the first reference light pulse is a laser light pulse.

 According to another characteristic of the invention, the method further comprises the execution of a modulation of the laser beam to obtain the laser light pulse.

 According to another characteristic of the invention, the probing light pulses and the reference light pulses have approximately the same wavelength.

 According to another characteristic of the invention, the method further comprises fixing the instant, during each repetition of the electrical test profile, at which the first reference light pulse is produced relative to the test profile.

 According to another characteristic of the invention, the

<Desc / Clms Page number 7>

 The method further includes directing the first probe light pulse and the first reference light pulse along a single path, toward the semiconductor device.

 According to another characteristic of the invention, the method further comprises directing the second probe light pulse and the second reference light pulse along a delay path before the second probe light pulse is combined with the first probe light pulse and the second reference light pulse is combined with the first reference light pulse.

 According to another characteristic of the invention, the method further comprises controlling a length of said delay path according to a feedback loop to compensate for a displacement of the semiconductor device.

 According to another characteristic of the invention, the control according to a feedback loop stabilizes the reference interference signal.

 According to another characteristic of the invention, the method further comprises amplifying the sounding interference signal, integrating the sounding interference signal and digitizing the sounding interference signal.

 According to another characteristic of the invention, the method further comprises amplifying the reference interference signal, integrating the reference interference signal and digitizing the reference interference signal.

 According to another characteristic of the invention, the detection of overlapping probing light pulses and the detection of overlapping reference light pulses includes the detection of overlapping probing light pulses and

<Desc / Clms Page number 8>

 overlapping reference light pulses for obtaining a detection signal, integrating and digitizing a portion of the detection signal for obtaining a probing interference signal and integrating and digitizing a portion of the detection signal to produce a reference interference signal.

 According to another characteristic of the invention, the method further comprises canceling an amplitude noise in the sounding interference signal and in the reference interference signal.

 According to another characteristic of the invention, the action of canceling an amplitude noise comprises the detection of part of a sounding light pulse energy, the detection of part of a reference light pulse energy, normalization of the sounding interference signal for the part of the sounding light pulse energy, and normalization of the sounding interference signal by the part of the energy reference light pulse.

 The invention also relates more specifically to a device for detecting activity in a semiconductor device when a repetitive electrical test profile is applied to the device, characterized in that it comprises: a source of a first pulse sounding light, which delivers a first pulse of sounding light at a selected time during each repetition of the electrical test profile; a source of a first reference light pulse which delivers a first reference light pulse at a time offset from the selected time at which the first probe light pulse is delivered; a divider, which the first impulse encounters

<Desc / Clms Page number 9>

 sounding light and the first reference light pulse, thereby resulting in the delivery of at least a second sounding pulse and a second reference light pulse; a support for the semiconductor device, encountered by the first probe light pulse and the first reference light pulse; a beam combining device, arranged to combine the first probe light pulse and the second probe light pulse, and to combine the first reference light pulse and the second reference light pulse, after the first probe light pulse and the first reference light pulse cooperated with the semiconductor device; a detector arranged to detect the first probe light pulse and the second probe light pulse so as to output a probe interference signal, and to detect the first reference light pulse and the second light pulse reference so as to output a reference interference signal; and a processor coupled to the detector so as to determine a function of the sounding interference signal and the reference interference signal.

 According to another characteristic of the invention, the selected instant is modified with the repetition of the electrical test profile so that the function of the sounding interference signal and the reference interference signal is determined for a plurality of times selected within the test profile.

 According to another characteristic of the invention, the source of the first probe light pulse comprises a first laser, and the source of the first

<Desc / Clms Page number 10>

 reference light pulse includes a second laser.

 According to another characteristic of the invention, the source of the first probe light pulse and the source of the first reference light pulse comprise the same laser.

 According to another characteristic of the invention, the source of the first probe light pulse is a laser.

 According to another characteristic of the invention, the laser is a locked mode laser, and the device further comprises an optical modulator positioned so as to select the first probe light pulse from a train of light pulses. laser in locked mode.

 According to another characteristic of the invention, the train of pulses of laser light in locked mode and the repetitive test profile are locked in phase.

 According to another characteristic of the invention, the source of the first reference light pulse is a laser.

 According to another characteristic of the invention, the device further comprises an optical modulator, and an output beam of the laser is modulated by the optical modulator to deliver the first reference light pulse.

 According to another characteristic of the invention, the probing light pulses and the reference light pulses have approximately the same wavelength.

 According to another characteristic of the invention, the instant during each repetition of the test profile, at which the first reference light pulse is delivered, is fixed relative to the test profile.

 According to another characteristic of the invention, the first sounding light pulse and the first

<Desc / Clms Page number 11>

 Reference light pulses are directed along the same path toward the semiconductor device.

 According to another characteristic of the invention, the device further comprises a delay path, along which the second probe light pulse and the second reference light pulse are directed.

 According to another characteristic of the invention, the device also comprises a feedback loop coupled to the delay path, which has the effect that a length of the delay path is controlled so as to compensate for the displacement of the semiconductor device.

 According to another characteristic of the invention, the feedback loop stabilizes the reference interference signal.

 According to another characteristic of the invention, the detector comprises a photodetector positioned so as to detect the first probe light pulse and the second probe light pulse and to deliver a probe interference signal, an amplifier so coupled receiving the sounding interference signal, an integrator coupled to receive the amplified sounding interference signal and a digitizer coupled so as to receive the integrated sounding interference signal.

 According to another characteristic of the invention, the detector comprises a photodetector positioned so as to detect the first reference light pulse and the second reference light pulse and to deliver a reference interference signal, an amplifier so coupled receiving the reference interference signal, an integrator coupled to receive the amplified reference interference signal and a digitizer coupled to receive the integrated reference interference signal.

 According to another characteristic of the invention, the

<Desc / Clms Page number 12>

 device further comprises a divider arranged so as to deflect part of a probe light pulse, a photodetector positioned so as to detect the part of the probe light pulse and to deliver a photodetector signal, a coupled amplifier so as to receive the signal from the photodetector, an integrator coupled so as to receive the amplified signal from the photodetector, and a digitizer coupled so as to receive the integrated signal from the photodetector.

 According to another characteristic of the invention, the device further comprises a divider positioned so as to deflect part of a reference light pulse, a photodetector positioned so as to detect the part of the reference light pulse and providing a photodetector signal and an amplifier coupled to receive the signal from the photodetector, an integrator coupled so as to receive the amplified signal from the photodetector, and a digitizer coupled so as to receive the integrated signal from the photodetector.

 Other characteristics and advantages of the present invention will emerge from the description given below, taken with reference to the appended drawings, in which: - Figure 1 is a schematic illustration of the device according to an embodiment of the present invention; Figure 2 shows in more detail part of the device of Figure 1; and FIG. 3 represents interference signals used to explain the operation of the device in FIG. 1.

 Integrated circuits are tested using optical interferometry. During each repetition cycle of an electrical test profile signal

<Desc / Clms Page number 13>

 applied repeatedly to an integrated circuit device under test (DUT device), a probe light pulse and a reference light pulse are delivered and move on the same optical path to sample the shape of electric wave on the DUT device. The probing pulses have a short duration relative to the duration of the maximum frequency signal of interest in the test profile, so as to provide a high bandwidth measurement of the waveform of the test profile . The reference pulses typically have a longer duration than the probing pulses. The reference pulses are delivered at a fixed time relative to the test profile. The probe pulses are scanned through the test profile during a series of cycles of the test profile, in a normal manner equivalent to time sampling, to reconstruct the waveform of the DUT device.

 A portion of each probe pulse and each reference pulse is divided and is detected to characterize the energy of the pulses that reach the DUT device. Furthermore, the sounding pulse is divided in an interferometer so as to form at least a first sounding pulse and a second sounding pulse, and the reference pulse is divided in the interferometer to obtain at least minus a first voltage pulse and a second reference pulse. The first sounding pulse and the first reference pulse sample the waveform on the DUT device, at the same physical location, but in a time-shifted manner with a variable delay when the sounding pulse is scanned at through the test profile. The second sounding pulse and the second reference pulse move on identical paths in a delay arm of the interferometer.

<Desc / Clms Page number 14>

 After cooperating with the DUT device, the first probe pulse is recombined with the second probe pulse so as to provide a probe interference signal at a detector. Similarly, after cooperating with the DUT device, the first reference pulse is recombined with the second reference pulse to provide a reference interference signal at a detector. The sounding interference signal varies over time when the sounding pulse scans the test profile over time, due to the time-varying electrical activity inside the DUT device, which modulates the optical phase of the first sounding pulse when the first sounding pulse cooperates with the DUT device. The sounding interference signal also varies over time due to noise caused by mechanical sliding, temperature variations, vibrational displacement of the DUT device, and amplitude noise in the incident sounding pulses caused by fluctuations in the sounding pulse light source output signal. The reference interference signal varies over time due to noise caused by mechanical slip, temperature variations, movement of the DUT device, and amplitude noise in the incident reference pulses. Since the reference pulses are provided at a fixed time with respect to the test profile, the reference interference signal does not vary over time under the effect of the electrical activity in the DUT device.

 The reference interference signal is stabilized (made almost constant) by rapidly adjusting the path length of the interferometer delay arm so as to compensate for a displacement of the DUT along the optical axis, under the effect of a vibration

<Desc / Clms Page number 15>

 or a slip. Advantageously, since the second probing pulse and the second reference pulse move on the same optical path along the delay arm of the interferometer, the adjustments made relative to the length of the path of the delay arm of the Interferometer to stabilize the reference interference signal also suppresses noise in the sounding interference signal due to the movement of the DUT device along the optical axis.

 The noise in the sounding interference signal, which is due to the amplitude noise in the incident sounding pulses, is canceled by normalizing the sounding interference signal by the measured energy of the sounding incident pulse, by example by calculating the reflection rate.

 Temperature variations and the movement of the DUT device across the optical axis similarly affects the probing reflection rate and the reference reflection rate. The noise produced by these sources is canceled by calculating the ratio of the sampling reflection rate and the reference reflection rate. The remaining modulation of the ratio between the probe reflection rate and the reference reflection rate represents the waveform of the electrical activity in the DUT device.

 Figure 1 is a schematic illustration of a two-pulse optical interferometer system 600. A locked mode laser 612 located on a conventional laser platform 604 delivers a train of laser pulses 606. In one embodiment, the locked mode laser is a locked mode Nd: YAG laser model 131-200 from the so-called company. Lightwave Electronics, which delivers pulses with a duration of -35 picoseconds (ps) at a repetition rate of 100 MHz and with a wavelength of 1064 nm. Pulse train 606 passes through the modulator

<Desc / Clms Page number 16>

 Conventional 608, which selects a probe pulse 609 from the train of pulses 606, as indicated in 610, for each cycle of a test profile applied repeatedly to a DUT device. The optical modulator 608 is for example an electro-optical modulator 360-80 from the company known as Conoptics with a 25D drive stage from Conoptics, but other optical modulators can also be used, for example acousto-optical modulators. In one embodiment, the optical modulator 608 includes two optical modulators in series for completely suppressing unwanted pulses in the pulse train 606.

 The continuous wave laser 612, which is also placed on the laser platform 604, delivers a laser beam with continuous amplitude 614 to the optical modulator 616.

In one embodiment, the continuous wave laser 612 is a continuous wave Nd: YAG laser from the company known as Coherent, Inc., Compass model 1064-500, which also operates at a wavelength of 1064 nm. The optical modulator 16 is also for example an electrooptical modulator 360-80 from the company Conoptics, but other optical regulators can be used. In one embodiment, the optical modulator 616 includes two optical modulators in series. The optical modulator 616 modulates the beam 614 so as to produce a reference pulse 618 of a duration as short as 10 ns for each cycle of a test profile applied to a DUT device. In one embodiment, the reference pulse has a duration of 150 nanoseconds (ns), and the reference pulse and the probe pulse are separated in time by a delay of about 1.5 microseconds (us ) at approximately 1 millisecond (ms), depending on the electrical test profile of the DUT device.

 The probing and reference pulses are directed to a beam combining device 620,

<Desc / Clms Page number 17>

 such as the high energy beam splitter model BS1-1064-10-1025-45P from the company called CVI Laser Corporation, which uses an optical beam deflection device such as a mirror 622 as necessary.

The beam combination device 620 directs the probe pulse and the reference pulse along collinear paths through a polarizer 23 towards the fiber optic coupler 624, which injects the probe and reference pulses into a fiber 626 singlemode optic which retains polarization. The division ratio of the beam combining device 620 is chosen such that, once the paths of a sounding pulse and a reference pulse have been made collinear by the beam combining device 620, the The energy in the reference pulse is approximately equal to the energy in the probing pulse. The polarizer 623 sets the probing and reference pulses in the same state of polarization.

 Otherwise, the probing and reference pulses are delivered by the same light source, such as for example a single mode laser in locked mode. In one embodiment, a single current pulse is divided by a beam splitter into probing and reference pulses. A variable length optical delay line produces a delay between the sounding pulse and the reference pulse. In another embodiment, the probe and reference pulses are derived from separate light pulses delivered by the same laser. Lasers other than locked mode lasers and inconsistent sources such as light emitting diodes are used as a light source in other embodiments to produce probing and reference pulses.

 An optical fiber 626 guides the pulses of

<Desc / Clms Page number 18>

 sounding and reference in the direction of the optical fiber coupler 628, which injects the pulses into a scanning system 630, where they meet the polarization beam splitter 632. The optical fiber 626 also filters in itself spatially the sounding and reference, which leads to an approximate adaptation of their structures in the spatial mode. A linearly polarized portion of each sounding pulse and each reference pulse is transmitted by the polarization beam splitter 632 to the optical beam control system 634, which drives the pulses by sending them to the interferometer 699. The path of the pulses entering the interferometer 699 is represented by the light beam 635. In one embodiment, the scanning system 630 is a confocal laser scanning microscope also used to form the image of a device under test.

 FIG. 2 shows in detail the interferometer 699 and the electronic control system which is associated with it. The interferometer 699 comprises a delay arm formed by a mirror 705, a quarter-wave plate 710, an opening 712, a lens 715 and a mirror 720, and an arm containing the DUT device and including a quarter-d plate wave 639, an objective lens 636 and the DUT 640 device. In one embodiment, the DUT 640 device is supported by a conventional xy table. In another embodiment an interferometer 699 and the scanning system 630 are supported by an xyz table and are moved relative to a fixed DUT device 640. The probing pulse 609 and the reference pulse 618 in the beam of light 635 meet a non-polarizing beam splitter 701, which divides each pulse of incident light into two pulses. For convenience, the portion of each sounding pulse 609 transmitted in the

<Desc / Clms Page number 19>

 arm containing the DUT device by the beam splitter 701 will be designated here as the first sounding pulse 609a, and the part of each sounding pulse deflected towards the delay arm of the interferometer 699 will be designated here as the second sounding pulse 609b. Similarly, the part of each reference pulse 618 transmitted in the arm containing the DUT device by the beam splitter 701 will be designated here as being the first reference pulse 618a, and the part of each reference pulse deflected towards the delay arm of the interferometer 699, will be designated here as being here the second reference pulse 618b.

 The first sounding pulse 609a and the first reference pulse 618a pass through the quarter-wave plate 639 which typically is oriented so as to transform the polarization of the pulses from linear polarization into circular polarization.

The first circularly polarized sounding pulse 609a and the first reference pulse 618a are focused by the objective 636 on the same spot in the active region 638 of the DUT device 640. A part of each pulse is reflected by reflection from the DUT device to through the objective lens 636, which realizes a new collimation of the beam, to the quarter-wave plate 639. The quarter-wave plate 639 transforms the polarization of the pulses by passing it from a circular polarization to the linear polarization orthogonal to their original linear polarization. After passing through the quarter-wave plate 639, part of each of the first sounding pulse, of the first reflected reference pulse, crosses the beam splitter 701 and returns to the beam path 635 to meet the beam system. scan 630.

 The second sounding pulse 609b and the second

<Desc / Clms Page number 20>

La plaque quart-d'onde 710 est orientée de telle sorte qu'après avoir traversé deux fois la plaque quart-d'onde les impulsions réfléchies sont polarisées linéairement avec une polarisation qui est orthogonale à leur polarisation d'origine et qui concorde avec celle de la première impulsion de sondage et de la première impulsion de référence. L'ouverture 712 règle les diamètres du faisceau de sonde et du faisceau de référence réfléchis, dans le bras de retardement de manière à adapter les diamètres du faisceau de sonde réfléchi et du faisceau de référence réfléchi dans le bras du dispositif DUT. reference pulse 618b are deflected by a beam splitter 701, via a shutter 703, along a path of the beam 704 of the delay arm of the interferometer. The shutter 703 is open when the system is used for interferometric measurements. For convenience, the beam path 704 is folded back by means of a mirror 705. The second probing pulse 609b and the second reference pulse 618b pass through the quarter-wave plate 710 and an opening 712 in the direction of the lens 715, which focuses the pulses on the mirror 720. The position of the mirror 720, which is mounted on a piezoelectric actuator 725, is adjusted so that the pulses are reflected by reflection along their path of incidence. The reflected pulses pass through the lens 715, the aperture 712 and the quarter-wave plate 710 in the opposite direction towards the mirror 705, which directs pulses via the shutter 703 towards the beam splitter 701 .

The quarter-wave plate 710 is oriented so that after having crossed the quarter-wave plate twice the reflected pulses are linearly polarized with a polarization which is orthogonal to their original polarization and which agrees with that of the first sounding pulse and the first reference pulse. The opening 712 adjusts the diameters of the reflected probe beam and the reference beam in the delay arm so as to adapt the diameters of the reflected probe beam and the reflected reference beam in the arm of the DUT device.

 A part of each of the pulses constituted by the second reflected sounding pulse and the second reflected reference pulse is deflected by the beam splitter 701 by being returned along the beam path 635 so as to be returned with the parts of the first thoughtful sounding pulse and

<Desc / Clms Page number 21>

 the first reference pulse reflected to the scanning system 630. Therefore the beam splitter 701 acts as a beam combining device. In one embodiment, the division ratio of the beam splitter 701 is chosen such that for a typical reflected signal from the DUT 640, the ratio of the energy contained in the first reflected reference pulse (reflected by the DUT 640) at the energy contained in the second reflected reference pulse (reflected by the mirror 720) is equal to approximately 1: 3 when these pulses leave the interferometer. In another embodiment, the division ratio is chosen such that the ratio of the energies of the reflected reference pulses is equal to about 1: 1. In another embodiment, the ratio of the energies of the reflected reference pulses is adjusted by inserting a power attenuator, such as a neutral density filter, in the path of the delay arm.

 The position of the mirror 720 is adjusted such that the first and second reflected sound pulses overlap in space and time when they return along the path 635 of the beam to the scanning system 630. The first and second reflected reference pulses are therefore also overlapped in space and time when they return along the beam path 635 to the scanning system 630. When the probing pulses are very short (35 ps), the length of the interferometer delay arm (the length of the delay path) should be set equal to, or almost equal to, the length of the interferometer arm containing the DUT device, to achieve overlapping pulses. In one embodiment, the mirror 720, the piezoelectric actuator 725 and the lens 715

<Desc / Clms Page number 22>

 are mounted on a slide which allows an approximate adjustment of the length of the delay path.

 Referring again to Figure 1, the reflected pulses return along the beam path 635 toward the beam driving optical system 634, which directs reflected pulses toward the polarization beam splitter 632. The reflected pulses , which have a linear polarization perpendicular to their initial linear polarization, are deflected by the polarization beam splitter 632 towards the mirror 642, which reflects the pulses reflected towards the optical fiber coupler 644.

The optical fiber coupler 644 injects the reflected pulses into a multimode optical fiber 646 which guides the reflected pulses towards the optical fiber coupler 648. The fiber optic coupler 648 injects the reflected pulses into the photodetector 650 of the reflected beam, inside the signal processing subsystem 652.

 The first and second reflected pulses, which overlap in space and time, produce in the photodetector 650, an electrical interference signal, having amplitudes determined by the optical phase difference between the two reflected probing pulses. Similarly, the first and second reflected reference pulses also produce, in the photodetector 650, an electrical interference signal having an amplitude determined by the optical phase difference between the two overlapping reflected reference pulses. The photodetector 650 produces an electric current pulse at the input terminal of a transimpedance amplifier 654 for each pair of reflected probing pulses and for each pair of reflected reference pulses.

 The 654 transimpedance amplifier converts

<Desc / Clms Page number 23>

 current pulses supplied by the photodetector 650 into output voltage pulses, which are transmitted by means of a fast-acting switch 656, to an energy analyzer 658 for analyzing the interference signals of sounding pulses and alternatively to the energy analyzer 660 for the reference pulse interference signals. The probing light pulses and the reference light pulses are sufficiently separated in time for the output voltage pulses which typically have a duration of 600 ns due to the limited bandwidth of the transimpedance amplifier 654, so that they can be distinguished from each other. The time integral of an output voltage pulse is a measure of the energy in the corresponding interference system. The energy analyzers 658 and 660 each usually contain an integrator and an analog-to-digital converter and thereby deliver 14-bit digital representations of the energy in the sounding interference signal and the interference signal respectively reference.

 Part of each incident probe pulse and each incident reference pulse meeting the scanning system 630 from the optical fiber 626 and the optical fiber coupler 628 is deflected by the polarization beam splitter 632 towards the mirror 632, which directs the pulses towards the fiber optic coupler 664. The optical fiber coupler 664 injects the pulses into the multimode optical fiber 666, which guides the pulses towards the optical fiber coupler 668. The fiber optic coupler 668 sends the incident pulses to a photodetector 670 of the incident beam, inside the signal processing subsystem 652.

 The photodetector 670 delivers a pulse of

<Desc / Clms Page number 24>

 current at the input terminal of the transimpedance amplifier 672 for each incident light pulse it detects. The transimpedance amplifier 672 converts the pulses of current supplied by the photodetector 670 into pulses of output voltage, which are transmitted by the fast-working switch 674 to an energy analyzer 676 for the signals of incident probing pulses. , and an energy analyzer 678 for incident reference pulse signals. The energy analyzers 676 and 678 each usually contain an integrator and an analog / digital converter, and delivers 14-bit digital representations of the energy contained respectively in the detected incident sound pulse and in the detected reference pulse .

 In one embodiment, the photodetectors 668 and 670 are PIN photodiodes of indium and gallium arsenide (TnGaAs model FD300 from the company called Fermionics Corporation, and the transimpedance amplifiers 664 and 672, the fast switches 656 and 674 and the pulse energy analyzers 658,676 and 678 are as described in Wilsher et al., US patent 5,905,577.

 A timing generator 680 controls the timing of the operation of the optical pulse interferometer system 600 with double pulses. The timing generator 680 is coupled to a source of test vectors 682, to a locked mode laser 602, to an optical modulator 608, to an optical modulator 616, to a signal processing subsystem 650, to an interferometer 699 and to a device 697 for processing and controlling data. In one embodiment, the timing generator 680 is a timing generator described by Wilsher et al., US Patent 5,905,577 for controlling a timing generator of a dual laser probe system.

 The device 697 for processing and controlling

<Desc / Clms Page number 25>

 Data is typically a general purpose programmable digital computer processor with a video display terminal and conventional panels for high speed data acquisition and digital signal processing.

 In one embodiment, the source of test vectors 682 is an ITS9000 logic test system from the company known as Schlumberger. We can also use other sources of test vectors, such as simple data generators if their clocks are sufficiently precise and stable. The test vector source 682 repeatedly delivers a sequence of test vectors (a test profile) to the DUT 640 in multiple lines 684. The test profile repeats continuously in cycles. That is, when the source 682 of test vectors sends the last vector in the test profile, it loops back to the starting vector and begins again to send the test profile. The test profile typically has a duration between about a few microseconds and a few milliseconds.

 In one embodiment, the pulse train 606 of the locked mode laser and the test profile loop provided by the source 682 of test vectors are phase locked. Of course this does not concern the optical phase of the individual laser pulses but rather concerns the phase of the train of periodic pulses related to the phase of the periodic test profile loop. Phase locking of the pulse train and test profile loop requires that there is an integer number of locked mode laser pulse periods in one period of the test profile loop. Consequently in the case of a repetition frequency of the locked mode laser pulses equal to 100 MHz, this requires that the loop of

<Desc / Clms Page number 26>

 test profile has a length equal to an integer period of 10 ns.

 In one embodiment, the timing generator 680 receives a clock signal 688 from the test vector source 682 and, using well known phase locked loop techniques, derives a clock signal 689 having the correct frequency for controlling the locked mode laser 602 which is locked in phase on the test profile loop. Another method is that the timing generator 680 produces and delivers both the clock signal 689 to the locked mode laser 602 and the clock signal 688 to the source 682 of test vectors.

 The source 682 of test vectors delivers a start signal in line 686 at the start of each cycle of the test profile loop. The start signal in line 686 is used to start a timing sequence in the timing generator 680. At a predetermined time after receiving the start signal in line 686, the timing generator 680 sends a control signal , via line 693, to the optical modulator 616. When receiving the control signal from line 693, the optical modulator 616 defines a reference light pulse 618 by allowing the passage of the beam at continuous amplitude 614 delivered by the continuous wave laser 612 for 150 ns. Similarly, at a predetermined time after reception of the start signal in line 686, the timing generator 680 sends a control signal, via line 695, to the optical modulator 608. When reception of the control signal circulating in line 695, the optical modulator 608 defines a sounding light pulse 610 by allowing the transmission of a pulse in the pulse train 606. Likewise, at predetermined times

<Desc / Clms Page number 27>

 after receiving the start signal present in line 686, the timing generator 681 sends control signals via line 691 to the signal processing subsystem 682 so that in response, fast switches 656 and 674 appropriately switch to direct voltage pulses from the output terminals of the transimpedance amplifiers 654 and 674 to pulse energy analyzers 658, 676 and 678, as previously described. The timing of the control signals present in lines 689, 691, 693 and 695 is controlled by the control data processing device 697, which typically is programmed by a user, and is transmitted to the timing generator 680 with timing signals. command present in line 690.

 The four 14-bit digital output signals, which are supplied by the signal processing subsystem 652, are applied to the control data processing device 697. Thus, for each cycle of the test profile, the data processing and control device 697 receives four data input signals representing the energies of the incident probing pulse and the incident reference pulse, and the energies of the sounding interference signal and the reference interference signal.

 The data processing and control device 697 acquires the electrical signal in the DUT 640 device by collecting data from the four pulse energy analyzers as reference pulses when the reference pulses are held in a position constant time relative to the start signal present in line 686, and the sounding pulses are moved in a series of fixed time positions relative to the start signal present in line 686, in a manner equivalent to a

<Desc / Clms Page number 28>

 temporal sampling. Typically, 500 different time positions are used. The ratio of the sounding reflection rate and the reference sounding rate is calculated for each of the sounding time positions, and is interpreted as representing the voltage variations in the DUT 640 device, corresponding to the profile waveform of test.

 In one embodiment, a single iteration of the probing pulses in a series of time positions is achieved. During this iteration, the sounding pulses remain in each timing position for many thousands of cycles of the test profile to allow reduction of waveform noise, by averaging, or by other processing applied to the data. connected. In another embodiment, about 10 to about 1000 iterations passing through the series of time positions are performed. During these iterations, the probing pulses remain in each timing position for about 10 to about 100 test profile cycles.

 The optical modulator 608 does not completely block unwanted pulses in a locked mode pulse train 606. Residual locked mode pulses leak from the optical modulator 608 and produce offset signals in photodetectors 658 and 670. In one form embodiment, a method described by Wilsher et al., US patent? 5,905,577, thereby determining the offset signals and subtracting them from the sounding interference signal and the reference interference signal.

 The amplitudes of the output interference signal and the reference interference signal are very sensitive to the difference in the lengths of the optical paths between the arm of the DUT device and the delay arm contained in the interferometer 699. By

<Desc / Clms Page number 29>

 Consequently, the interference signals are very sensitive to the position of the DUT 640 device along the optical axis established by the incident probing pulse and the incident reference pulse. The position of the DUT 640 device present along the optical axis is designated here as being the Z position of the DUT 640 device.

/2 1/2 2 A'-B). La puissance moyenne est A + B. Par exemple, The sensitivity of the interference signals to the Z position of the DUT 640 device is demonstrated by the following example. We consider a beam of reference pulses reflected by the DUT 640 device, i.e. first reflected interference pulses, and we designate it as the beam A. We also consider a beam of pulses of reference reflected by the mirror 701 of the delay arm of the interferometer, i.e. second reflected reference pulses, and they are designated as the beam B. It is assumed that the position of the mirror 720 is adjusted by so as to maximize the overlap of pairs of first and second reference pulses reflected, in time and in space, at the photodetector 650. Assuming that the power of the beam A does not vary with the Z position of the DUT 640 device and also that the power of the beam B does not vary with the position of the mirror 720, so when the position Z of the DUT 640 device moves over a distance of a quarter wavelength of light, the total power detected in the interference signal by the photodetector 650 varies around the sum of the powers of A and B. From the point of view of basic optical theory, the maximum and minimum powers found in this range of positions Z, which correspond to conditions of additive interference and subtractive interference, are respectively equal to (A + Buzz 2 and

/ 2 1/2 2 A'-B). The average power is A + B. For example,

<Desc / Clms Page number 30>

si la puissance dans le faisceau A est de 1 microwatt (usw) et la puissance dans le faisceau B est 30 uW, alors la puissance maximale détectée est égale à 7, 46 pW, la puissance minimale détectée est égale à 0, 636 pW et la puissance moyenne détectée est égale à 4 pW. En supposant que la longueur d'onde des impulsions de lumière est égale à 1464 nm, la variation du signal de sortie du photodétecteur avec la position Z est représentée par la course sinusoïdale 630 sur la figure 3. On peut voir qu'une variation de 0,1 micromètre de la position Z, qui correspond par exemple à une vibration de faible amplitude du dispositif DUT 640, entraîne une très grande variation de la puissance détectée.

if the power in beam A is 1 microwatt (usw) and the power in beam B is 30 uW, then the maximum detected power is equal to 7.46 pW, the minimum detected power is equal to 0.636 pW and the average power detected is equal to 4 pW. Assuming that the wavelength of the light pulses is equal to 1464 nm, the variation of the output signal from the photodetector with the Z position is represented by the sinusoidal stroke 630 in FIG. 3. It can be seen that a variation of 0.1 micrometer of the Z position, which corresponds for example to a low amplitude vibration of the DUT 640 device, results in a very large variation in the detected power.

 In the embodiment shown in FIG. 2, the extreme sensitivity of the interference signals at the Z position of the DUT 640 device is improved by providing closed-loop control of the position of the mirror 720. The position of the mirror 720 is controlled so as to keep the reference interference signal at the photodetector 650 approximately constant when the DUT device 640 is moving. In another embodiment, a rapid actuator is similarly provided in the DUT 640 device, as part of the support table 641 for example, and is controlled so as to oppose the movement of the DUT 640 device, to its source. The description given below of control devices and methods involves these two embodiments.

 With reference to FIGS. 1 and 2, an image of the DUT 640 device is formed, with the shutter 703 closed to prevent any interference diagram on the image, by raster scanning of the DUT 640 device by means of the probing beam or of reference, using the optical system 634 for controlling the beam. The incident beam of the sounding and reference pulses is then aligned

<Desc / Clms Page number 31>

 with a region, which is of interest, in the DUT device 640 by means of the optical system 634 for controlling the beam and the shutter 703 and the switch 740 are opened. The device 697 for processing command data sends signals control signals via line 685 to the digital / analog converter 745, which causes this converter 745 to produce a voltage at its output terminal a voltage which produces a current in the resistor 785. An amplifier 735 produces at its output terminal, a voltage proportional to the sum of the currents flowing in its input terminal, and coming from resistors 585 and 590. The output voltage coming from amplifier 735 is coupled to piezoelectric actuator 125 , which is fixed to the mirror 720. The actuator 725 moves the mirror 720 over a distance of approximately 6 micrometers for a variation of the amplifier output voltage 735 equal to approximately 100 V.

 When the switch 740 is open, the current passing through the resistor 790 is zero and the output signal from the amplifier 785 is controlled by the output signal from the digital / analog converter 745. The device 697 for data processing and control positions a series of voltages on the output terminal of the digital / analog converter 745, which causes the actuator 725 to move the mirror 720 and consequently modify the length of the path of the delay arm and record the interference signals of reference represented by the digital output signals of a pulse energy analyzer 660.

 For any Z position of the DUT 640, a position of the mirror 720 is found in a range of half a wavelength, which provides a detected reference interference signal, at the photodetector 650 at the center of the power range of the

<Desc / Clms Page number 32>

signal d'interférence de référence. Dans l'exemple décrit précédemment, ce niveau de puissance est égal à 4 uW. Une tension représentant cette puissance à mi-valeur est réglée sur la sortie d'un convertisseur numérique/analogique 775 par le dispositif 697 de traitement de données et de commande, et est utilisé en tant qu'entrée de référence dans un amplificateur différentiel 770. Un signal analogique 661 produit par l'analyseur d'énergie d'impulsion 660 et représentant l'amplitude du signal d'interférence de référence, est utilisé en tant qu'autre signal d'entrée envoyé à l'amplificateur différentiel 770. Le dispositif 697 de traitement de données et de commande positionne le signal de sortie du convertisseur numérique/analogique 745 au milieu de l'échelle ce qui règle la sortie de l'amplificateur 735 sur environ 50 volts.

reference interference signal. In the example described above, this power level is equal to 4 uW. A voltage representing this mid-value power is set on the output of a digital / analog converter 775 by the data processing and control device 697, and is used as a reference input in a differential amplifier 770. An analog signal 661 produced by the pulse energy analyzer 660 and representing the amplitude of the reference interference signal is used as another input signal sent to the differential amplifier 770. The device 697 data processing and control positions the output signal of the digital / analog converter 745 in the middle of the scale which sets the output of the amplifier 735 to around 50 volts.

 The data processing and control device 197 then closes the switch 740 to transmit a current from the input terminal of the integration amplifier 750, via the resistor 790 to the input terminal of the amplifier 735. The difference between the output signal from the reference digital / analog converter 775 and the analog output signal 661 delivered by the pulse energy analyzer 660 appears on the output terminal of the differential amplifier 770. The timing generator 680 sends control signals via the line 693 to a monostable circuit 780 as well as to the optical modulator 616, as described above. The monostable circuit 780 controls the switch 765 for about 5 ps at intervals greater than about 100 us. If the output signal of the differential amplifier 770 is not equal to zero volts when the switch 765 is closed, a current flows through the resistor 760 towards the amplifier

<Desc / Clms Page number 33>

 integration 750, which modifies the voltage at the output terminal of the amplifier 750 and the current passing through the resistor 790 and modifies, via the amplifier 735 and the piezoelectric actuator 725, the position of the mirror 720.

 Amplifiers 770,750 and 735 form a negative feedback loop controlling the position of the mirror 720. When the Z position of the DUT 640 device varies, the negative feedback loop changes the position of the mirror 720 to maintain the reference interference signal detected at the center of its power range (4 pW in the previous example). The mirror 720 has a low mass so that the feedback loop can follow the vibrations of the DUT 640 device in the frequency range from the DC component up to about 240 Hz or more. The position of the mirror 720 can be controlled by a piezoelectric device 725 or by an electromagnetic voice coil device. As noted earlier, the reaction could be applied to a similar actuator in the DUT 640 device to counter vibrations at their source, although with more difficulty given that the DUT 640 device and typically more massively than the mirror 720.

 The strength of the reference interference signal detected as a function of the mismatch of the optical path lengths of the interferometer arm containing the DUT device of the interferometer delay arm is also represented by curve 830 on the graph of the Figure 3. The reference level produced by the digital / analog converter 775 is indicated by line 805. We can see that with the feedback loop enabled, there are several stable (negative feedback) and unstable (positive feedback) positions of the mirror when the power 830 of the reference interference signal is equal to the reference level 805. These

<Desc / Clms Page number 34>

 positions appear at the points 810, 820, 840 and 850. In the absence of a vibration of the DUT 640 device, the position of the mirror 720 is moved away from the unstable points to be brought closer to one of the stable points, coming from position itself almost exactly on one of the stable points. The circuit can be provided to reverse the electronic gain in the feedback loop so that the position of the stable and unstable points can be reversed. It is important that the reaction system is sufficiently effective to allow the maintenance of the operation near its stable point in the presence of normal vibrations of the DUT device, or otherwise the reaction system constantly searches for a new stable operating point. As is well known in the art, the efficiency of the feedback system is determined in part by the gain in bandwidth of amplifiers 735 750 and 770, by the response time of the piezoelectric actuator 725, to which is attached the mirror 720, and by the frequency, with which the switch 765 is closed to sample the output signal from the amplifier 770 and therefore the analog signal 661.

 In one embodiment, the timing generator 680 is modified so as to deliver a control signal in line 693 more than once for each cycle of the test profile delivered by the source 682 of test vectors. Consequently, the reference pulse 618 is defined and the switch 765 is closed more than once during each test profile cycle, which increases the rate with which the reaction system samples the analog signal 661.

 Any difference in the wavelength of light between the probe laser and the reference laser causes the probe interference profile and the reference interference profile to differ. If the laser of

<Desc / Clms Page number 35>

 sounding had an optical wavelength shorter than that of the reference laser for example, then the sounding interference signal detected as a function of the mismatch between the lengths of the optical paths of the arm containing the DUT device and of the delay arm would be represented by a sinusoidal curve similar to curve 830 in Figure 3, but with a shorter wavelength. Consequently, in the case of a sufficiently large mismatch concerning the length of the optical path, variations in the sounding interference signal due to movements of the DUT device 640 are not compensated by the movements of the mirror 720, which stabilize the reference interference signal. If the lasers with a nominal wavelength of 1064 nm differ by about 0.5 min for example, the mismatch concerning the optical path length between the arm containing the DUT device and the delay arm must be less than about 50 um.

 If a confocal microscope is used to transmit beams of the probing pulse and the reference pulse to the DUT 640 device, then the vibrations of the DUT 640 device can modify the amplitude of the reflected pulses as well as their optical phase. Consequently, certain displacements of the mirror 720 are made so as to compensate for a change in amplitude in the reflected light rather than a change in phase in the reflected light. Since the DUT 640 is placed at the focal point of the 635 objective lens, it is expected that these amplitude variations will be small enough, typically less than +/- 10% in the focal range in which we are interested, so that they can be compensated by means of sufficiently small displacements of the mirror so that this effect is negligible.

 Variations in the amplitude of the pulses

<Desc / Clms Page number 36>

 incident references similarly lead to an erroneous compensation movement of the mirror 720. In one embodiment, these errors are avoided by controlling the feedback loop with the reflection rate of the reference pulses rather than with the signal d interference of reference pulses. The amplitude variations of the reference pulses are however typically less than 5% and therefore are small enough that the errors introduced by the erroneous displacements of the mirror 720 can be neglected.

 This description is given by way of illustration and without any limiting nature. Other modifications will appear to a person skilled in the art in the light of the present description and are intended to be contained within the scope of the present invention.

<Desc / Clms Page number 37>

LEGENDS TO THE FIGURES Figure 1: 602. Laser source in locked mode 608. Optical modulator 620. Beam combining device 624. Fiber optic coupler 612. Continuous wave laser source 616. Optical modulator 622. Mirror 762. Mirror 634. System beam control optic 642. Mirror 676. Probe 658. Probe 697. Control interface of the data processing operator and display device 639. Quarter-wave plate 638. Active region 636. Optical training system sound signal 680. Timing generator 686. Start 682. Test vector source a. 100 MHz pulse train b. Single pulse selected c. Continuous amplitude d. Pulse formed e. Interferometer unit and electronic control unit f. Photodetector for the incident beam g. Photodetector for the reflected beam h. Clock signal i. System diagram Figure 2: 730. Damping of the beam 701. Beam splitter 720. Mirror 715. Lens

<Desc / Clms Page number 38>

 725. Piezoelectric element 704. Mirror 639. Quarter-wave plate 636. Lens 735. Amplifier 740 and 765 Switch 750 and 770 Amplifier 780. Monostable rocker 5 us 775. Reference digital / analog converter a. Positioning of the digital / analog converter b. Adjustment c. Sample d. Reflected continuous wave signal e. Interferometer and electronic control system Figure 3: a. Detector output signal (microwatts) b. Variation of the Z position (micrometers)

Claims (37)

 1. Method for detecting activity in a semiconductor device when a repetitive electrical test profile is applied to the device, characterized in that it consists in: producing a first pulse of sounding light (609,609a) at a selected time during each repetition of the test profile and dividing the probe light pulse to produce at least a second probe light pulse (609b); producing a first reference light pulse (618,618a) during each repetition of the test profile at an instant which is offset from the selected instant at which the first probing light pulse is delivered, and dividing the first light pulse reference for delivering at least a second reference light pulse (618b); directing the first probe light pulse and the first reference pulse to the semiconductor device; combining the first probe light pulse with the second probe light pulse and the first reference light pulse with the second reference light pulse after the first probe light pulse and the first reference light pulse have cooperated with the semiconductor device, so that the first probe light pulse and the second probe light pulse overlap in space and time and that the first reference light pulse and the second light pulse of reference overlap in space and in time; detecting overlapping probing light pulses to thereby obtain a probing interference signal; <Desc / Clms Page number 40>  detecting the overlapping reference light pulses to thereby obtain a reference interference signal; determining a function of the probe interference signal and the reference interference signal.  2. Method according to claim 1, characterized in that it further comprises modifying the selected instant with the repetition of the test profile so that the function of the sounding interference signal and the interference signal of reference is determined in a plurality of instants selected inside the test profile.  3. Method according to claim 1, characterized in that the first probe light pulse is a laser light pulse.  4. Method according to claim 3, characterized in that it comprises the selection of the laser light pulse from a train of laser light pulses in locked mode.  5. Method according to claim 4, characterized in that it further comprises the execution of a phase lock of the train of pulses of laser light in locked mode and of the repetitive test profile.  6. Method according to claim 1, characterized in that the first reference light pulse is a laser light pulse.  7. Method according to claim 6, characterized in that it further comprises the execution of a modulation of the laser beam to obtain the laser light pulse.  8. Method according to claim 1, characterized in that the probing light pulses and the reference light pulses have approximately the same wavelength.  9. Method according to claim 1, characterized in <Desc / Clms Page number 41>  which it further comprises fixing the instant, during each repetition of the electrical test profile, at which the first reference light pulse is produced relative to the test profile.  10. The method of claim 1, characterized in that it further comprises directing the first probe light pulse and the first reference light pulse along the same path, towards the semiconductor device .  11. The method of claim 1, characterized in that it further comprises directing the second probe light pulse and the second reference light pulse along a delay path before the second pulse sounding light is combined with the first sounding pulse and the second reference light pulse being combined with the first reference light pulse.  12. The method of claim 11, characterized in that it further comprises controlling a length of said delay path in a feedback loop to compensate for a displacement of the semiconductor device.  13. Method according to claim 12, characterized in that the control according to a feedback loop stabilizes the reference interference signal.  14. Method according to claim 1, characterized in that it further comprises amplifying the sounding interference signal, integrating the sounding interference signal and digitizing the sounding interference signal .  15. The method of claim 1, characterized in that it further comprises amplifying the reference interference signal, integrating the reference interference signal and digitizing the reference interference signal . <Desc / Clms Page number 42>  16. The method of claim 1, characterized in that the detection of overlapping probing light pulses and the detection of overlapping reference light pulses includes the detection of overlapping probing light pulses and light pulses of overlapping reference for obtaining a detection signal, integration and digitization of part of the detection signal for obtaining a survey interference signal and integration and digitization of part of the detection signal for obtaining a reference interference signal.  17. The method of claim 1, characterized in that it further comprises canceling an amplitude noise in the sounding interference signal and in the reference interference signal.  18. The method as claimed in claim 17, characterized in that the action of canceling an amplitude noise comprises the detection of part of an energy of sounding light pulse, the detection of part of a reference light pulse energy, the normalization of the sounding interference signal for the part of the sounding light pulse energy, and the normalization of the reference interference signal by the part of the reference light pulse energy.  19. Device for detecting activity in a semiconductor device when a repetitive electrical test profile is applied to the device, characterized in that it comprises: a source (602) of a first pulse of sounding light, which delivers a first probe light pulse (609a) at a selected time during each repetition of the electrical test profile; a source (612) of a first pulse of <Desc / Clms Page number 43>  reference light which delivers a first reference light pulse (618a) at a time offset from the selected time at which the first sounding light pulse is delivered; a divider (639), encountered by the first sounding pulse and the first reference light pulse, thereby resulting in the delivery of at least a second sounding pulse (609b) and a second light pulse reference light (618b); a support (641) for the semiconductor device, encountered by the first probe light pulse and the first reference light pulse; a beam combining device (620; 701), arranged to combine the first probe light pulse and the second probe light pulse, and to combine the first reference light pulse and the second light pulse reference, after the first probe light pulse and the first reference light pulse have cooperated with the semiconductor device; a detector (652,650,670) arranged to detect the first probe light pulse and the second probe light pulse so as to output a probe interference signal, and to detect the first reference light pulse and the second reference light pulse so as to output a reference interference signal; and a processor (607) coupled to the detector so as to determine a function of the sounding interference signal and the reference interference signal.  20. Device according to claim 19, characterized in that the selected instant is modified with the repetition of the electrical test profile so <Desc / Clms Page number 44>  that the function of the probe interference signal and the reference interference signal is determined for a plurality of the instants selected within the test profile.  21. Device according to claim 19, characterized in that the source of the first sounding pulse (609a) comprises a first laser (602), and that the source of the first reference light pulse (618a) comprises a second laser.  22. Device according to claim 19, characterized in that the source of the first probe light pulse (609a) and the source of the first reference light pulse (618a) comprise the same laser.  23. Device according to claim 19, characterized in that the source of the first probe light pulse (609a) is a laser (602).  24. Device according to claim 23, characterized in that the laser is a locked mode laser (602), and that the device also comprises an optical modulator (608) positioned so as to select the first pulse of sounding light to be from a locked mode laser light pulse train (606).  25. Device according to claim 24, characterized in that the train of pulses of laser light in locked mode (606) and the repetitive test profile (614) are locked in phase.  26. Device according to claim 19, characterized in that the source of the first reference light pulse (618a) is a laser (612).  27. Device according to claim 26, characterized in that it further comprises an optical modulator (693), and that an output beam of the laser is modulated by the optical modulator to deliver the first <Desc / Clms Page number 45>  reference light pulse (618a).  28. Device according to claim 19, characterized in that the probing light pulses (609a) and the reference light pulses (618a) have approximately the same wavelength.  29. Device according to claim 19, characterized in that the instant during each repetition of the test profile, at which the first reference light pulse is delivered, is fixed relative to the test profile.  30. Device according to claim 19, characterized in that the first probe light pulse (609a) and the first reference light pulse (618a) are directed along the same path towards the semiconductor device (640).  31. Device according to claim 19, characterized in that it further comprises a delay path, along which the second probe light pulse and the second reference light pulse are directed.  32. Device according to claim 31, characterized in that it further comprises a reaction loop (735,750, 770) coupled to the delay path, which has the effect that a length of the delay path is controlled so as to compensate for the displacement of the semiconductor device.  33. Device according to claim 32, characterized in that the reaction loop (735,750, 770) stabilizes the reference interference signal.  34. Device according to claim 19, characterized in that the detector (652) comprises a photodetector (670) positioned so as to detect the first probe light pulse (609a) and the second probe light pulse (609b) and to deliver a sounding interference signal, a <Desc / Clms Page number 46>  amplifier (672) coupled to receive the survey interference signal, an integrator (676) coupled to receive the amplified survey interference signal and a digitizer coupled to receive the survey interference signal integrated.  35. Device according to claim 19, characterized in that the detector (652) comprises a photodetector (650) positioned so as to detect the first reference light pulse (618a) and the second reference light pulse (618b) and outputting a reference interference signal, an amplifier (654) coupled to receive the reference interference signal, an integrator (660) coupled to receive the amplified reference interference signal and a digitizer coupled to so as to receive the built-in reference interference signal.  36. Device according to claim 19, characterized in that it further comprises a divider (632) arranged so as to deflect part of a sounding light pulse, a photodetector (650,670) positioned so as to detect the part of the sounding light pulse and outputting a photodetector signal, an amplifier (654,673) coupled to receive the photodetector signal, an integrator (658,676) coupled to receive the amplified signal from the photodetector, and a coupled digitizer so as to receive the integrated signal from the photodetector.  37. Device according to claim 19, characterized in that it further comprises a divider (632) positioned so as to deflect a part of a reference light pulse, a photodetector (650,670) positioned so as to detect the part of the reference light pulse and outputting a photodetector signal and an amplifier coupled to receive the signal from the photodetector, an integrator (660, <Desc / Clms Page number 47>  678) coupled to receive the amplified signal from the photodetector, and a digitizer coupled to receive the integrated signal from the photodetector.