JP6200947B2 - Laser assisted device alternation using synchronized laser pulses - Google Patents

Description

-Cross-reference of related applications-
This application claims priority from US Provisional Application No. 61 / 381,023, filed Sep. 8, 2010, and US Patent Application No. 13 / 228,369, filed Sep. 8, 2011. The entire disclosure of which is incorporated herein by reference. This application also claims priority based on US Provisional Application No. 61 / 648,042, filed May 16, 2012, the entire disclosure of which is incorporated herein by reference.

-US government license-
The present invention was devised in cooperation with the US Government under the USAF Contract FA8650-11-C-7104 with the support of The Intelligence Advanced Research Projects Activity (IARPA). The US government has certain rights in this invention.

-Technical field-
The present invention relates to the field of laser defect location analysis of integrated circuits (ICs). More specifically, the present invention relates to IC design debugging and / or failure analysis using laser assisted device alternation (LADA) technology.

-Related technology-
LADA (Laser Assisted Device Alteration) technology allows continuous wave (CW) lasers to generate local photocurrents on the backside of an integrated circuit, resulting in pass / fail results for test stimuli on “sensitive” transistors. It can be varied to identify sensitive areas that contain design or process defects. The laser is used to temporarily change the operating characteristics of the transistors on the device. The current spatial resolution using a 1064 nm continuous wave laser is 240 nm.

  Examples of LADA technology include Critical Timing Analysis in Microprocessors Using Near-IR Laser Assisted Device Alteration (LADA) (Jermy A. Rowe E, T.E. Year). This document describes the possibility of using a CW laser with a wavelength of 1064 nm or 1340 nm. 1340 nm causes device alternation via local heating, while 1064 nm causes device alternation via photocurrent generation. It is also particularly noted that the 1064 nm laser has an advantageous spatial resolution. For this reason, the authors recommend using a 1064 nm laser.

  As shown in FIG. 1, in a conventional LADA, a CW laser is used to induce electron / hole pairs from the back surface to a device under test (DUT). The electron / hole pairs thus generated affect the timing of neighboring transistors and facilitate critical path analysis. The DUT 110 is coupled to a tester 115 connected to the computer 150, such as a conventional automatic test equipment (ATE). ATE is used in a conventional manner to stimulate a DUT with a test vector and analyze the response of the DUT to the test vector. The response of the DUT to the test vector may be further analyzed using LADA. For example, if a DUT fails a particular test, to determine if the DUT can pass under certain conditions, and if possible, which device, that is, the transistor that caused the failure LADA may be used. Conversely, if a DUT passes certain tests, under which conditions the DUT fails these tests, and if so, which device, that is, the transistor, is responsible for the failure. LADA may be used to check whether it has become.

  The LADA system in FIG. 1 operates as follows. Tiltable mirrors 130 and 135 and objective lens 140 are used to focus and scan the beam from CW laser 120 to DUT 110. This allows the laser 120 to generate photocarriers in the DUT silicon without locally heating the device. The electron / hole pairs thus generated affect the timing of neighboring transistors, that is, reduce or increase the switching time of the transistors. The tester is configured to limit the operating point of the DUT by applying a repetitive test loop of a selected voltage and frequency. The laser stimulation is used to change the pass / fail result of the tester. The position of the beam at each point is related to the pass / fail result of the tester, and a change is detected, i.e. if a previously passed transistor is now rejected, or vice versa, The coordinates of the laser beam at that time indicate the position of the “borderline” transistor.

  During LADA analysis, the tester (ATE) is configured to place the operating point of the DUT in a limit state. Laser stimulation is used to change the pass / fail result of the tester. The current state of the art in laser-assisted defect spatial localization is about 240 nm resolution. Further improvements in single photon LADA spatial resolution are limited by the wavelength of the laser light. As described in the Rowlett literature, spatial resolution is improved using short wavelengths. However, the light absorption of silicon at wavelengths below 1064 nm is a major obstacle in delivering light from the back to the transistor, so the use of short wavelengths is rejected. Thus, despite the reduced design rules in modern devices, the use of short wavelength lasers cannot improve the spatial resolution of LADA systems. For example, with the 22 nm design rule, it remains questionable whether the conventional LADA can analyze four transistors in close proximity.

  OBIC (light beam induced current) is another test and debug analysis where a laser beam illuminates the DUT. However, unlike LADA, OBIC is a static test in the sense that no stimulus signal is applied to the DUT. Instead, laser light is used to induce a current in the DUT that is measured using a low noise, high gain voltage or current amplifier. The OBIC is used in a single photon mode or a two-photon absorption mode called TOBIC or 2P-OBIC (two-photon light beam induced current).

  Two-photon absorption (TPA) is the absorption of two photons of the same or different frequencies together to excite a molecule from one state (usually the ground state) to a higher energy electronic state. . The wavelength is chosen such that the sum of the photon energies of the two photons that arrive together is equal to the energy difference between the low and high energy states of the molecule. Two-photon absorption is a secondary process and is several orders of magnitude smaller than linear (single photon) absorption. Two-photon absorption differs from linear absorption in that the intensity of absorption depends on the square of the intensity of light and is a nonlinear optical process.

  The following summary of the disclosure is included to provide a basic understanding of some aspects and features of the invention. This summary is not an extensive overview of the invention and is not intended to identify, in itself, any key or critical element of the invention or to delineate the scope of the invention. . Its sole purpose is to present some concepts of the invention in a simplified form as a prelude to the more detailed description that is presented later.

  In various disclosed embodiments, higher spatial resolution in defect localization is possible by utilizing a time domain that allows for improved spatial resolution. In the disclosed embodiment, a pulsed laser with sufficient energy is used instead of a continuous wave laser. The pulsed laser can improve the spatial resolution in synchronization with the clock of the device. In various embodiments, a 1064 nm wavelength laser for single photon LADA, or a longer wavelength to induce the LADA effect using a non-linear two-photon absorption mechanism is utilized. The latter technique is referred to herein as the two-photon laser assisted device alternation technique (2pLADA).

  In the disclosed embodiment, a test vector that stimulates the DUT is used to enable higher resolution of defect location and uses a femtosecond pulsed laser to scan the region of interest of the DUT, Check the response of the DUT to the test vector at. Also, the laser source is selected to provide photon energy whose wavelength is lower than the silicon bandgap and to provide pulses of femtosecond pulse width. The clock signal is obtained from the ATE and provided to a circuit that controls the DUT and the pulsed laser. The timing of the pulses may be shifted with respect to the ATE clock to check the pass / fail characteristics of various devices. Furthermore, by using proper synchronization of the laser pulse to the clock, the spatial resolution is improved and multiple devices, i.e. transistors, can be resolved in the laser beam.

  In an alternative embodiment, a constant pulse laser system is used. A constant pulse laser clock is sent to the ATE and used to guide the test signal to the DUT. Further, to obtain the effect of shifting the pulse timing with respect to the test signal, the clock signal of the constant pulse laser is shifted before being sent to the ATE. Thus, when the ATE generates a test signal based on the shifted laser clock, the test signal itself is shifted with respect to the laser pulse. The test signal is shifted with respect to the laser pulse by controlling the shift of the laser clock, allowing LADA inspection.

  In various embodiments, a laser assisted device alternation (LADA) system operable with an automated test equipment (ATE) is provided to test a device under test (DUT) of an integrated circuit, the system comprising: A timing electronic device including a controller for receiving and analyzing a test signal from the ATE, a first feedback loop for receiving a clock signal from the ATE and generating a synchronization signal for synchronization of the laser pulse with the clock signal, and a laser pulse A tunable pulse laser source having a second feedback loop that controls the tunable laser cavity to generate a laser pulse of a desired number of pulses Laser pulse from An optical device that directs the laser pulse to a desired position on the DUT, and the timing electronics synchronizes the laser pulse to reach the DUT transistor at a time synchronized to the clock time, The controller is configured to change the response of the transistor to a test signal applied from the ATE to the DUT, and the controller is configured to detect the changed transistor response. The first feedback loop and / or the second feedback loop may include a phase locked loop. The optical device may include a laser scanning microscope. The optical device may further include a solid immersion lens. The number of laser pulses may be configured as a multiple of the clock signal. The multiple may be an integer greater than 1 or a fraction. The timing electronics is configured to delay or advance the laser pulse with respect to the clock signal.

  According to a further embodiment, a laser assisted device alternation (LADA) system operable with an automatic test equipment (ATE) to test a device under test (DUT) of an integrated circuit is provided, the system comprising: , A controller that receives and analyzes a test signal from the ATE, a constant pulse laser source that generates a laser pulse and a pulse number signal indicating the number of pulses of the laser pulse, receives a pulse number signal, and transmits a clock signal to the ATE Timing electronics and an optical device that receives a laser pulse from a constant pulse laser source and directs the laser pulse to a desired position on the DUT, the timing electronics being a transistor of the DUT when synchronized to the laser pulse The timing output by the ATE to reach Configured to detect whether the laser pulse changes the response of the transistor to the test signal applied from the ATE to the DUT, and the controller is configured to detect the response of the changed transistor. The

  The timing electronics may further include a phase variable circuit configured to change the phase of the clock signal with respect to the pulse number signal. The phase variable circuit may be configured to delay or advance the test signal with respect to the pulse number signal.

  According to yet a further embodiment, a method is provided for testing an integrated circuit device under test (DUT) coupled with an automated test equipment (ATE) using laser assisted device alternation (LADA). The method includes obtaining a clock signal from the ATE, applying the clock signal to the DUT, obtaining a test loop signal, applying the test loop signal to the DUT, and a first feedback loop with the pulsed laser source. Applying a clock signal to a second feedback loop to synchronize the laser pulse to the clock signal and directing the laser pulse to a desired region of the DUT Steps. The first and / or second feedback loop may include a phase locked loop with an external reference signal. The external reference signal may include a clock signal. The laser pulse may include a picosecond to femtosecond laser pulse. The pulsed laser source may be operated to generate a laser pulse having a wavelength selected to generate a single photon laser assisted device alternation. The pulsed laser source may be operated to generate a laser pulse having a wavelength selected to generate a two-photon laser assisted device alternation. Directing the laser pulse may include scanning the laser pulse over a desired region of the DUT.

  According to another embodiment, there is provided a method for testing a device under test (DUT) in an integrated circuit coupled with an automatic test equipment (ATE) using laser assisted device alternation (LADA). The method includes the steps of generating a laser pulse at a predetermined number of pulses using a constant pulse laser source, obtaining a pulse number signal from the constant pulse laser source, and generating a clock signal therefrom, Are applied to the ATE, a test loop signal is generated by the ATE, the test loop signal is applied to the DUT, and a laser pulse is directed to a desired region of the DUT. The method may further include changing the phase of the clock signal with respect to the pulse number signal.

  According to a further embodiment, there is provided a method for measuring carrier lifetime in a semiconductor device under test (DUT) using laser-assisted device alternation (LADA) technology, and repeating electrical test signals to the DUT Applying a laser beam, irradiating the sensitive transistor in the DUT with a laser beam to obtain an optimal LADA signal and obtaining a spatial coordinate of the sensitive transistor, and using a pulsed laser source Generate a laser pulse, irradiate a sensitive transistor using spatial coordinates to direct the laser pulse, and change the timing of the laser pulse, and the LADA signal at each timing position until the intensity of the LADA signal reaches zero Recording the intensity of LADA, and LADA No. comprises the steps of measuring the carrier lifetime by calculating the time it takes to change to a minimum signal from the maximum signal. A laser beam for measuring spatial coordinates may be obtained from a continuous wave (CW) laser source. The step of calculating time may include plotting LADA signal strength versus time. The method further includes directing the laser pulse to spatial coordinates and varying the timing until an optimal LADA signal strength is obtained to measure the pulse timing of the maximum laser pulse overlap with test signal overlap. But you can.

  Other aspects and features of the present invention will be apparent from the detailed description which will be described with reference to the following drawings. It should also be understood that the detailed description and drawings provide various non-limiting examples of various embodiments of the invention as defined by the claims that follow.

  The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and, together with the detailed description, serve to explain and depict the principles of the invention. The drawings are intended to illustrate the main features of the exemplary embodiments by way of illustration. In addition, the drawings are not intended to show all features of actual embodiments or the relative sizes of the depicted elements. And the drawings are not to scale.

FIG. 1 shows a CW / LADA system in the prior art. FIG. 2 shows an embodiment of a pulsed laser LADA system. FIG. 2A shows an embodiment of two feedback loops. FIG. 2B illustrates an embodiment in which a clock signal is generated using a constant pulse laser source. FIG. 3 shows an embodiment for achieving a synchronizer. 4A-4C show in close proximity how the pulsed laser LADA improves the identification and isolation of both P- and NMOS transistor positions individually. 5A-5D show the improvement in spatial resolution as a result of precise pulse placement characteristics. FIG. 6 shows an embodiment of a laser repetition rate locking device. 7A-7C illustrate steps performed to measure minority carrier lifetime using LADA, according to one embodiment.

  FIG. 2 shows an embodiment using a pulsed laser source with sufficient energy instead of a continuous wave laser. This embodiment relates to the application of photon absorption to accurately inject carriers into an IC for the purpose of locating defects using LADA technology, and IC characterization and the discovery of ways to improve the design. Can be used. The technique is based on photons reaching the focal point in the transistor so that the photon energy transmitted is greater than the energy required to generate electron-hole pairs. (For example, for silicon,> 1.1 eV, other ICs such as GaAs, SiGe, InP, etc. have different band gap energies). Photon stimulation in this embodiment requires excitation with laser pulses in the nanosecond to femtosecond range. The signal is localized at the focal point of the laser, showing an immediate improvement in defect localization. The effective amount of generation of electron-hole pairs appears to decrease due to synchronization. In this embodiment, elaborate timing electronics are used to precisely control the timing of the laser pulses with respect to the transition of the edge of a tester clock, such as an ATE clock. This type of control can precisely change the delay or progression of the signal propagating through the transistor of interest for LADA.

  FIG. 2 illustrates one embodiment of the present invention, in which a DUT 210 is connected to an ATE 215 as in the prior art. However, in the embodiment of FIG. 2, nanosecond to femtosecond laser pulses are generated by the pulsed laser source 225 and focused on the DUT 210 using tiltable mirrors 230 and 235 and the objective lens 240. In 2pLADA, the laser source 225 provides a pulsed laser beam with a wavelength longer than the band gap of silicon, ie longer than 1107 nm. In one embodiment, a wavelength of 1550 nm is used, while in another embodiment, 1340 nm or 1250 nm is used. On the other hand, the same device may be used for single photon LADA, in which case the laser source may provide a pulsed beam with a wavelength such as 1064 nm. In this embodiment, tiltable mirrors 230 and 235 are applied as a laser scanning microscope (LSM). In some embodiments, a solid immersion lens (SIL) is used as part of the objective lens device 240.

  In the conventional LADA system, the laser is always output. However, in embodiments of the present invention, very short pulses are used. For this reason, it is important that device transitions occur when a laser pulse reaches the device. To accomplish this, a trigger signal 245 is obtained from the ATE and input to the timing electronics 260, which controls the pulse laser 225 to synchronize the laser pulse with the ATE test signal.

  First, tester (ATE) 215 is operated to apply a set of test vectors to determine the limit settings of DUT 210 using the system shown in FIG. That is, the voltage and frequency of the test vector determines the point just before the DUT fails the test, or just fails the test, for example, the DUT fails 50% of the number of test loops. To be changed. This is the boundary condition for pass / fail of DUT. The voltage and frequency settings are then used to generate a repetitive test signal and repeatedly stimulate the DUT at the pass / fail boundary conditions.

  A synchronization signal 245 is transmitted from the tester 215 to the timing electronics 260 when the DUT is stimulated at a boundary condition. Timing electronics 260 uses laser source 225 to acquire laser pulses at picosecond to femtosecond pulse widths and higher than the silicon bandgap for 2pLADA or shorter for single photon LADA. To control. In 2pLADA, the wavelength is generally about 1250 nm to 1550 nm and the pulse width is about 100 fs. For single photon LADA, the wavelength may be 1064 nm and the pulse width is about 100 fs. The laser pulse scans the region of interest of the DUT 240 to increase or decrease the DUT switching time and cause the DUT to exceed boundary conditions. That is, when the voltage / frequency of the test vector is set so that the DUT is about to fail, the laser pulse timing is determined so that the DUT fails. Conversely, if the voltage and frequency of the test vector is set so that the DUT just fails, the laser pulse is timed so that the DUT passes the test. During this time, the output of the DUT is monitored to determine the position of the defect. That is, at the moment the output signal from the DUT fails (the DUT will pass if there is no laser beam), the position of the beam on the DUT is determined and the position of the transistor causing the failure is determined. To do. On the other hand, at the moment when the output signal from the DUT shows a pass (the DUT will fail if the laser beam is not present), the position of the beam on the DUT is determined in advance, causing the failure, Determines the position of the transistor that passed.

  Since the synchronization signal is obtained from the tester, it should be understood that the timing of the laser pulses can be varied to change the amount of photo generation (single photon or two photon) action in the transistor. That is, the timing of the laser pulses can be changed to increase or decrease the amount by which the DUT switching time is increased or decreased. This ability helps determine the severity of the defect in addition to the position of the defect.

  Also, in embodiments of the present invention, timing electronics are used to precisely control the timing of the laser pulses with respect to the transition of the tester (eg, ATE) clock edges. This type of control can precisely change the delay or progression of the signal propagating through the transistor of interest. In one embodiment, as shown in FIG. 2A, two phase locked loops (PLL) are used to precisely control the pulsed laser. In FIG. 2A, ATE 215 provides a clock signal Clk and a test loop signal test loop. Both the clock signal and the test loop signal are input to the DUT, tapped and transmitted to the timing electronics 260 that forms the first PLL. Laser source 225 includes a second PLL.

  That is, the PLL of the laser source 225 ensures that the pulse frequency of the laser pulse is stable and accurately matches the desired frequency (eg, 100 MHz). On the other hand, the first PLL of the timing electronic device synchronizes the frequency of the second PLL with respect to the ATE clock signal. In particular, in the present context, synchronization does not necessarily mean that the laser pulse and the clock pulse are simultaneous, but rather the laser pulse and the clock pulse are synchronized over the test loop period. For example, the laser pulse so that the pulse appears at the center of the clock signal in every clock pulse, as represented by pulse train 227, or the pulse appears at the end of each clock pulse, as represented by pulse train 229. The timing may be shifted. That is, the laser pulse may be delayed or advanced with respect to the ATE clock signal, but synchronization to the ATE clock signal is maintained.

On the other hand, as described in detail below, the frequency of the laser pulse may be a multiple of the ATE clock signal. For example, the laser pulse train 223 has a multiplier of 7 so that seven laser pulses are generated for every single clock pulse of the ATE. A multiplier greater than 1 can be used to inspect whether the defect is at the rising edge or the falling edge. In addition, since a plurality of laser pulses of each clock pulse have a progression / delay function, there is no need to provide a delay or deviation of the pulses. Conversely, it is possible to have a multiplier less than one. For example, in the pulse train 224, the multiplier is ½ so that the laser pulse reaches only every other clock signal. Such an apparatus may be used to verify that the defect is actually due to the laser pulse. This is because if the defect is due to a laser pulse, the device will fail about 50% over time. In FIG. 3, an embodiment for obtaining a synchronizer is shown. The output pulses from the pulsed laser source (1) of nanosecond to femtosecond duration are synchronized to the clock cycle of the integrated circuit (IC) (2) via an intermediate phase locked loop (PLL) circuit (3). Can. In this configuration, the PPL circuit receives the IC clock cycle frequency and locks to an internal crystal oscillator of the same frequency. In this embodiment, the frequency of the clock and the crystal oscillator is fixed at 100 MHz. The IC clock signal may be generated by an ATE (not shown). This allows a synchronization ratio between the 1: 1 light pulse and the transistor switching event. In practice, under this condition, these values can be fixed in any of the 1 kHz to 10 GHz range before the effect of the photon absorption rate is individually attenuated.

  In general, it is specifically mentioned herein that a light source faster than 1 GHz is not desirable for non-linear inspection involving 2pLADA, etc., because the peak optical power contained in each pulse is inversely proportional to the repetition rate. Thus, a high repetition rate is equal to a low peak optical power that produces 2-photon absorption that is invalid, if any. At a 1: 1 synchronization ratio, a single photon is inspected at 1064 nm, but a light source of a few GHz may be useful because the photoelectron interaction varies in proportion to the incident light power. In addition, it is noted that the effect of two-photon absorption is directly proportional to the duration of the incident pulse, where femtosecond light pulses promote higher peak light power than picosecond or nanosecond options, thus improving nonlinear absorption. Should. As a result, it is desirable to utilize ultrafast (ie, picosecond or femtosecond) light pulses in non-linear inspection. On the other hand, in single photon inspection, the pulse duration is not a limiting parameter on the absorption rate, so it does not limit performance. If there are any, allow additional inspection parameters (eg, measurement of light pulse interaction time versus optoelectronic device stimulation). Furthermore, the silicon absorption coefficient is larger at a single photon wavelength (less than 1130 nm) compared to a two-photon absorption wavelength (ie, greater than 1250 nm).

  In order to maintain the effect, the synchronizer may be modified to match an integer number of incident light pulses to the transistor switching event (or device clock frequency). To guide this, the laser source needs to be designed to have a pulse picker module that generates repetition rates higher than 1 GHz and can be measured in place after pulse optimization to correct the synchronization ratio. is there. For example, instead of matching each incident light pulse to every transistor switching event, every second pulse may be matched to every subsequent switching event to create a 2: 1 synchronization ratio. In practice, this can be translated into the use of a 200 MHz repetition rate laser and 100 MHz device frequency, or a 1 GHz repetition rate laser and 500 MHz device frequency. Alternatively, the ratio may be adjusted to be 3: 1, 4: 1, etc., as long as the ratio corresponds to optoelectronic overlap with the clock pulse in synchronization with the test loop signal. In this synchronizer, the effect of photon absorption is not reduced, but the ratio at which absorption occurs results in a negative scale photon signal intensity according to the synchronization ratio. It is specifically mentioned herein that this is not a limiting factor for laser guided inspection of integrated circuits. Each device under test has a requirement to perform a photo scalability calibration to determine the maximum possible sync ratio. In addition, since two-photon absorption begins to dominate single photon absorption at wavelengths greater than about 1200 nm in silicon, by integrating a tunable light source (ie, 1000 nm to 1600 nm output wavelength) into such a system, Single photons and two-photon absorption methods are interchangeable.

  When these frequencies (ie clock and crystal oscillator frequencies) are locked together, the PLL circuit output signal is sent to the pulsed laser through a 100 MHz (or clock frequency) filter circuit to act as an input stimulus. Bear. The advantage here is that the PLL circuit has complete control over the phase of the output signal. Therefore, it is possible to control the repetition number of the laser light output and the pulse arrival time. This can be proved by comparing the output clock frequency from the IC with the trigger output from the pulse source of the oscillator (9). In this embodiment, the PPL circuit can electronically tolerate a phase delay of about 600 fs, but due to the electrical jitter of the substrate, the minimum phase delay is set to about 20 ps. The electrical jitter of the system is directly proportional to the accuracy with which the light pulses are placed in relation to the switching times of the individual transistors being evaluated. Therefore, since the electrical jitter of the system is 20 ps, the optical placement accuracy is also 20 ps, which corresponds one-to-one. This is an important parameter as the electronic placement error is larger, for example greater than 2pLADA femtosecond pulse duration, as the effect of the acquired timing may be negated. Femtosecond light pulses increase the local energy density as required for effective two-photon absorption. However, when the electrical jitter is above the isolated carrier generation time scale, the jitter can limit the subsequent signal generation and the temporal accuracy of the time-resolved data at this time.

  The laser pulse is connected to a laser scanning microscope (LSM) (4) and can be accurately transmitted to a specific position on the IC. LSM is controlled using a computer (6) with a graphical user interface and a custom digital signal processor (DSP) suite. In the disclosed embodiment, this set of applications provides the end user with the ability to communicate directly with the PLL circuit via a pre-configured DSP circuit (7), and the PLL circuit, for example, a pulse delay or Progression completely controls the time that the laser pulse reaches the device.

  With respect to device (2), it can be electrically stimulated to generate a pre-adjusted LADA pass / fail value using a custom application interface (5). The board compares real-time acquired values from the counter, latch, and shift register devices against a loaded reference value that is pre-inserted by selecting a reset switch. Precision control of the counter value loaded in real time may be controlled by a precision delay analog potentiometer on the application interface board that changes the timing of the latches that allow operation on the IC. With this configuration, the user can mainly adjust the pass, fail, or balance comparator output values. These pass / fail output values are then sent to a data adjustment circuit (in this example, a field programmable gate array FPGA 8) that accepts the real-time pass / fail digital stimulus. Programmed, and scaled to measure 0% to 100% failure values, transmits an averaged digital output (approximately 40 μs duration), and provides improved visualization in the graphical user interface and resulting pass The reject value of 0% to 100% is again scaled due to the fail level bias. The data conditioning circuit may also be used with the application interface board to calculate the magnitude of the laser induced timing delay by calibrating the output delay voltage of the application board.

  In the embodiment described above, a tunable pulsed laser source is used and the pulse frequency is adjusted to synchronize with the ATE clock. Although these embodiments are feasible, the tuned pulsed laser source is actually expensive and requires the phase locked loop described above. Another embodiment is shown in FIG. 2B, which enables LADA testing using a simpler constant pulse laser 255. For example, a mode-locked laser source may be used. Mode-locking is a technique in optical systems where the laser is made to produce very short duration light pulses on the order of picoseconds or femtoseconds. The laser pulse is used as a clock and transmitted to the timing electronics 265. A conventional ATE has a clock input port and is programmable to use an input clock for the purpose of generating a test loop signal for the clock Clk and DUT. Thus, in one embodiment, a clock signal from timing electronics 265 is input to the ATE, and the ATE is programmed to use the input clock signal to generate a clock and a test loop signal.

  However, as mentioned above, in order to obtain the most benefit of the pulsed laser LADA, the laser pulse is pulsed to reach the transistor at various times during the clock cycle, eg rising edge, middle, falling edge, etc. It is desirable to adjust. This is done in the embodiments of FIGS. 2, 2A, and 3 by delaying or advancing the laser pulse. However, in the embodiment of FIG. 2B, the laser pulse is constant and cannot be changed, so delaying or advancing the laser pulse is not a practical option. Thus, in one embodiment, the ATE is programmed to delay or advance the clock signal in synchronization with the clock signal received from the timing electronics 265. In this method, the arrival timing of the laser pulse to the transistor may be matched with the rising edge, falling edge, etc. of the ATE clock signal.

  On the other hand, the ATE and LADA tester are usually manufactured by different manufacturers, and the actual test is further performed by a third-party company engineer, which simplifies the operation of the test engineer and provides a signal from the ATE. It is beneficial that there is no need to delay or progress. This is done using the phase shifter 275 and is represented in the embodiment of FIG. 2B. That is, the clock signal output from the timing electronics 265 can be advanced or delayed with respect to the laser pulse using the phase shifter 275. As a result, the changed signal is transmitted to the ATE as an input clock signal. Thus, when the ATE outputs a clock and test loop signal, both signals can be shifted or delayed with respect to the laser pulse.

(Example)
The construction of a pulsed LADA system with a pulsed light source allows a new aspect of the operating device to be evaluated and measured. When a CW laser is utilized in a conventional single photon or alternative two-photon LADA, the light emission constantly interacts with individual transistors having a potentially harmful level of invasion. On the other hand, the pulse LADA method allows the switching characteristics of individual transistors to be specified to the same extent as in physical two dimensions. The extensive pulse LADA concept is described in detail below.

  In conventional CW • LADA stimulation, it can be seen from device theory and practice that the magnitude of the laser induced device perturbation from a p-type metal oxide semiconductor (PMOS) transistor dominates the nearby n-type (NMOS). Since the diameter of the laser beam corresponds to both p-type and adjacent n-type transistors, the resulting spatial resolution is insufficient to distinguish failing transistors. On the other hand, according to embodiments in the disclosed pulse device, temporal resolution is used to obtain higher spatial resolution even when larger wavelength lasers are used. That is, the incident pulses are adjusted to the exact time switching interval of the transistor being tested, and the peak power contained in each pulse is significantly higher than in the CW mode, so the PMOS and NMOS transistors are positioned closer together. Can be individually identified and separated. This is not possible under CW excitation, thus creating a new experimental field of semiconductor device design debug and characteristics to be explored, even with smaller design rules. In this regard, as state-of-the-art nodes are moving towards lower nanometer-scale geometries, there is growing interest within the semiconductor device defect analysis community where photoinduced transistor recognition and operating characteristics are very important. ing. Therefore, the synchronization pulse LADA has a higher value than the corresponding CW.

  A schematic example of such an improvement is shown in FIGS. 4A-4C. In continuous wave mode, the PMOS signal is usually dominant, resulting in the general spatial distribution of a single signal as shown in FIG. 4A. It is very difficult here to identify the physical distribution of the individual transistors and / or to adapt the display of these LADAs to a computer-aided design (CAD) layout. As shown in FIG. 4B, in theory, each transistor should generate its own LADA signal regardless of the magnitude of the laser guidance effect. These will then allow rapid physical and / or optoelectronic recognition to fully locate the physical location of the individual transistors. This can be reproduced in the pulse domain using the embodiments described above. That is, the laser pulse is timed and synchronized with the test signal so that the laser pulse arrives at the position of the PMOS and NMOS transistors as selected by the user. As shown in FIG. 4C, the pulse can be timed to switch the PMOS transistor to test the PMOS transistor or to switch the NMOS transistor to test the NMOS transistor. Therefore, regardless of the spatial application range of the laser, evaluation of single transistor switching and / or physical identification or recognition by enhancing the LADA signal with CAD are performed.

  Also, the additional benefit of increased peak power from ultrafast pulses is (in addition to the ability to generate LADA signals more effectively, ie, obtaining less image averages) in perturbing the laser-guided critical timing path ( Depending on whether a P- or N-MOS transistor is of interest, it is at least one of a generalized increase or a decrease, and an improvement in LADA signal collection. Larger incident light power increases the number of photo-injected carriers in the silicon and then increases the possibility of stimulating photoelectron variations in the device structure. This results in an excellent LADA signal response that can be measured more quickly with a reduced level of intrusion, and the pulsed light source has limited chances for heat buildup and damage, rather than actually on Long off. For example, an ultrafast laser with a 10 ps pulse duration and 100 MHz repetition rate is off for a duration of 10 ns, creating a ratio of 1: 1000 (on: off) to provide a sufficient cooling period. . However, it should be specifically mentioned that this is the final power ratio that causes heating. For example, a single light pulse containing 1 kJ of incident light energy meets the above criteria but is sufficient to potentially and permanently damage the device through some other thermal or non-thermal optoelectronic mechanism. Also contains a lot of energy.

  In addition, in a facility for injecting a significant level of optical power into a specific transistor in a non-intrusive manner, the position of a transistor that was previously unquestioned may be disturbed. Of course, the higher level of photocarrier generation near a given region of interest (collecting transistors of different sensitivities) increases the possibility of visualization of a wider range of LADA display positions. These activated regions can be stimulated with a laser induced photocurrent of about 10 μA to 100 μA. However, ultrafast laser pulses boasting peak optical powers close to 10 kW to 100 kW can inject 10 mA to 100 mA of photocurrent into the device (although maintaining an intrusive safety level). May be sufficient to perturb the transistor.

  Effective two-photon absorption is obtained in silicon with a focused laser power density greater than 10 MW / cm 2, but the single photon value is about 1/106 times smaller due to its relative absorption cross section. As the spatial geometry of the transistor being evaluated decreases, the level of incident optical power (or local power density) required for effective and non-invasive photocarrier injection also decreases. Also, the generation of two-photon absorption does not depend on a specific power density threshold, ie two-photon absorption is an instantaneous, quantum mechanically defined, non-linear process, which is a third order non-linear Sensitive to the imaginary part of the sensitivity (ie, this is not a power density dependency but an intensity square dependency).

  Even though a 2 photon wavelength of 1250 nm effectively produces 625 nm inside silicon (absorption cross section greater than 1064 nm), the original intensity dependence of the absorption process reduces the overall relative rate of absorption. Two-photon absorption is directly proportional to the square of the incident light intensity. In addition, the doping level of silicon also contributes to the present description, i.e., the increased or decreased doping concentration affects the absorption level as a function of wavelength. However, during this single photon bias, another novel laser probing and device characterization platform is possible for improved critical timing analysis of signal levels, either racing within the transistor or switching. CW LADA cannot provide this type of inspection due to intrusion limitations (ie, the laser is always on) and limited power transfer capability. Time-resolved pulse probing, on the other hand, is the first to examine transistor switching physics for defect analysis at sound, well-designed nodes and subsequent downchain device performance or interaction. Can make it possible. In order to effectively apply this type of device characteristics, an understanding of the level of incident optical power required is important. Perturbing a “healthy” transistor requires high peak power, but facilitates a minimal level of intrusion. Therefore, it is necessary to optimize the duration of the incident light pulse. Obviously, the duration of the picosecond pulse at 1064 nm provides a significant level of incident light power (ie, photocarrier generation) at the transistor level. This is because, for example, a 10 ps laser pulse at a repetition rate of 100 MHz and an average power of 4 mW produce 4 W peak power. However, this can be limited if the laser repetition rate matches a clock frequency faster than 1 GHz from the device under test. Increasing the number of repetitions results in a decrease in peak power. For this reason, a femtosecond laser source is used as a more appropriate alternative. Since the pulse duration is reduced to 1/1000 and the peak power is increased by the same magnitude (4 kW in the example above), the laser repetition rate is adjusted according to the device operating frequency while improving the peak optical power level Can be done. A further benefit of femtosecond pulse duration is improved temporal characteristics, but as noted above, this is limited by the magnitude of the synchronizer electrical jitter. Finally, femtosecond laser pulses reduce the level of light penetration and minimize the possibility of laser induced damage to the device compared to picosecond or nanosecond pulses.

  In addition, the pulse LADA system exhibits improved spatial resolution as a result of precise pulse placement capabilities. However, when in CW mode, the laser continuously stimulates a specific region of interest while guessing LADA information in real time. There is no distinction between highly ordered sequences of circuit functions (ie, propagating signal paths versus time), and a collective distribution is acquired with PMOS-dominated bias, thus spatially averaged 2 A dimensional LADA image is derived. However, in the pulse mode, these propagation velocity paths can be distinguished from each other with an accuracy of about 20 ps, and a greatly limited LADA signal display with improved lateral resolution becomes possible. This is because the spatially separated neighboring transistors that are not configured to be switched until later in the device operation cycle are individually and temporarily processed. This improves the LADA separation resolution and the physical LADA resolution.

  A schematic example is shown in FIGS. 5A-5D. In CW mode, the spatial distribution of the LADA signal is time averaged, so the resulting two-dimensional LADA map has a generalized optoelectronic structure that is biased according to the LADA signal strength of the individual transistors (this is (Usually because PMOS dominates NMOS). In these diagrams represented in FIG. 5A, the spatial resolution is low and the CAD overlay performance is limited. However, in pulse mode, the LADA image is improved and the spatial resolution is improved due to the time resolution of the acquisition process. Here is a sequence of events that control device operation, driven by the tester and dependent on the transistor, so that each transistor as a function of space and time (and with sufficient incident optical power to eliminate the PMOS bias effect) By specifying individually, the effects of adjacent transistors are effectively eliminated from interfering with LADA acquisition. Each transistor switches in a systematic, time-dependent order, allowing the incident light pulse to directly manipulate and measure each transistor in two physical dimensions (ie, X and Y) and time It is configured to be. As a result, the spatial resolution of the acquired LADA signal is improved, as shown in the sequence of FIGS. 5B-5C, which are acquired at different times, each showing a temporal and spatial separation diagram. Additional device data that could not be obtained before is extracted.

  In addition to collecting only LADA specific data with this technique, it is also possible to determine further photoelectronic phenomena. One example is laser-induced carrier lifetime measurement. Since carrier lifetime depends on the number of various optoelectronic parameters such as material composition, dimensions, geometry, electric field magnitude and direction, now it is possible to quantify the carrier lifetime within the location of a particular device Is very difficult. However, in pulsed LADA, this electronic time scale can be measured directly by a pseudo-pump-probe type technique where the creation of transistor-specific LADA events is related to the arrival time of the laser pulse. The measured carrier lifetime may require consideration (ie, subtraction) of the system's electronic response time for a more accurate display.

  When using a single photon LADA, ie, a laser pulse with a wavelength of 1064 nm, the magnitude of the measured LADA effect is directly proportional to the magnitude of the laser-induced photocurrent (in the two-photon technique, the LADA signal is quadratic. This is when using linear absorption since it responds). In one embodiment, the LADA signal is specified as a function of the arrival time of the laser pulse. Since the carrier life defines the magnitude of the resulting LADA signal, the carrier life can be extracted.

  In one embodiment, the process by which this is done is as follows. First, a laser beam (for example, a CW laser beam having a wavelength of 1064 nm) is arranged to irradiate a sensitive transistor in order to obtain an optimal LADA signal. This is shown in FIG. 7A. The spatial coordinates of the laser beam with the optimal LADA signal indicate the proper spatial coordinates of the transistor. The CW laser source is then disabled, the pulse laser source is activated, and the laser pulses are directed to the same spatial coordinates obtained from the CW laser. The timing of the laser pulse is adjusted to obtain and measure the optimal LADA signal in order to obtain an appropriate temporal overlap with the tester (eg, ATE) pulse that reaches the transistor. This is shown in FIG. 7B. At this time, an optimized spatial overlap of the laser spot of the transistor and a temporal overlap of the laser pulses of the test signal are achieved. The laser pulse arrival time can then be adjusted to measure the carrier lifetime. Specifically, the timing of the laser pulse is changed and the LADA signal strength is recorded at each timing position (eg, the amount of delay or progression) until the LADA signal reaches zero. The resulting LADA signal strength versus response time is then plotted, as shown in FIG. 7C. The time taken for the LADA signal to change from the maximum signal to the minimum signal (or vice versa) corresponds to the measured laser induced carrier lifetime. The above process occurs when applying an electrical test signal to the DUT repeatedly.

(Laser source)
A laser source with a repetition rate of several GHz is readily available and is configured with deep consideration of the length of the resonant cavity, that is, the repetition rate increases as the oscillation cavity becomes shorter. The cavity length can be controlled and locked by including a piezoelectric actuator located on the opposite side of the intracavity resonant mirror. This is an industry standard technique for repetition rate locking, but the electronic mixer circuit required to facilitate such devices can vary in design and application. As described in the above embodiment, two feedback loops are required to properly incorporate the tuned pulse laser source into the LADA tester, one of which is for controlling the number of repetitions of the laser pulse. The other is for synchronizing the timing of the pulse with the DUT clock. The first feedback loop controlling the number of repetitions includes a mixer that compares the free running repetition frequency of the laser with the input clock stimulus to generate a high voltage driven difference signal. The difference signal is input to the piezoelectric transducer to adjust the length of the resonant cavity, and the resonant cavity is adjusted to the desired length so that the number of pulses matches the supplied clock input. An example of such a setting is shown in FIG. In addition to the circuit depicted in FIG. 6, a second stabilizer may be included to continuously monitor and correct the output voltage from the proportional-integral amplifier. This ensures that the high voltage amplifier consistently obtains the correct input voltage so that the repetition rate lock is stable for a longer period of time, ie 7 days rather than 10 minutes.

  It should be understood that the processes and techniques described herein are not inherently related to a particular device and are applied by any suitable combination of components. Further, various types of general purpose devices may be used in accordance with the teachings described herein. It is also beneficial to configure special equipment to perform the methods described herein.

  Although the present invention has been discussed with exemplary embodiments of specific materials and specific steps, variations of these specific examples can be made and / or used, and The detailed structure and method are subject to an understanding from the examples described and depicted and a review of the steps that facilitate modification without departing from the scope of the invention as defined by the appended claims. It should be understood by those skilled in the art.

Claims (27)

A laser assisted device alternation (LADA) system operable with an automatic test equipment (ATE) for testing a device under test (DUT) of an integrated circuit comprising:
A controller for receiving and analyzing a test signal from the ATE;
Timing electronics including a first feedback loop that receives a clock signal from the ATE and generates a synchronization signal for synchronization of a laser pulse with the clock signal;
A tunable pulse laser source comprising a second feedback loop for generating the laser pulse, having a tunable laser cavity, and controlling the tunable laser cavity to generate the desired number of pulses of the laser pulse;
An optical device that receives the laser pulse from the tunable pulse laser source and directs the laser pulse to a desired position on the DUT;
The timing electronic device adjusts the timing of the laser pulse so as to reach the transistor of the DUT in a time synchronized with a clock time, and changes the response of the transistor to the test signal applied from the ATE to the DUT. And the controller is configured to detect a changed response of the transistor.   The system of claim 1, wherein the first feedback loop comprises a phase locked loop.   The system of claim 1, wherein the second feedback loop comprises a phase locked loop.   The system of claim 1, wherein the optical device comprises a laser scanning microscope.   The system of claim 4, wherein the optical device further comprises a solid immersion lens.   The system of claim 1, wherein the number of pulses of the laser pulse is configured as a multiple of the clock signal.   The system of claim 6, wherein the multiple is an integer greater than one.   The system of claim 6, wherein the multiple is a fraction.   The system of claim 1, wherein the timing electronics is configured to delay or advance the laser pulse with respect to the clock signal.   The controller of claim 1, wherein the controller includes a user interface that allows a user to directly control the first feedback loop to adjust the timing of the laser pulse reaching a transistor. system. A laser assisted device alternation (LADA) system operable with an automatic test equipment (ATE) for testing a device under test (DUT) of an integrated circuit comprising:
A controller for receiving and analyzing a test signal from the ATE;
A constant pulse laser source that generates a photon energy lower than the band gap of silicon and that generates a femtosecond pulse width laser pulse and a pulse number signal indicating the pulse number of the laser pulse;
Timing electronics for receiving the pulse number signal and transmitting a clock signal to the ATE;
Receiving a laser pulse from the constant pulse laser source and directing the laser pulse to a desired position on the DUT;
The timing electronic device synchronizes the timing of the test signal output by the ATE so as to reach the transistor of the DUT in a time synchronized with the laser pulse, and the laser pulse is applied from the ATE to the DUT. A system configured to detect whether to change a response of the transistor to the test signal, and wherein the controller is configured to detect a response of the transistor that has changed.   The system of claim 11, wherein the timing electronics further includes a phase variable circuit configured to change a phase of the clock signal with respect to the pulse number signal.   The system of claim 12, wherein the phase varying circuit is configured to delay or advance the test signal with respect to the pulse number signal. A method for testing an integrated device under test (DUT) coupled with an automated test equipment (ATE) using laser assisted device alternation (LADA), comprising:
Obtaining a clock signal from the ATE and applying the clock signal to the DUT;
Obtaining a test loop signal and applying the test loop signal to the DUT so as to repeatedly stimulate the DUT with a pass / fail boundary condition ;
Applying a first feedback loop to the pulsed laser source to generate a repeatable number of laser pulses;
Applying the clock signal to a second feedback loop to synchronize the laser pulse with the clock signal;
Directing the laser pulse to a desired region of the DUT so that the photon energy transmitted is greater than that required to generate electron-hole pairs and the photons reach the focal point of the transistor ;
Including methods.   The method of claim 14, wherein the first feedback loop comprises a phase locked loop with an external reference signal.   The method of claim 15, wherein the external reference signal comprises the clock signal.   The method of claim 14, wherein the second feedback loop comprises a phase locked loop.   15. The method of claim 14, wherein the laser pulse comprises a picosecond to femtosecond laser pulse.   15. The method of claim 14, wherein the pulsed laser source is operated to generate a laser pulse having a wavelength selected to generate a single photon laser assisted device alternation.   15. The method of claim 14, wherein the pulsed laser source is operated to generate a laser pulse having a wavelength selected to generate a two-photon laser assisted device alternation.   The method of claim 14, wherein directing the laser pulse includes scanning the laser pulse across a desired region of the DUT. A method for testing an integrated device under test (DUT) coupled with an automated test equipment (ATE) using laser assisted device alternation (LADA), comprising:
Generating laser pulses at a predetermined number of pulses using a constant pulse laser source;
Obtaining a pulse number signal from the constant pulse laser source and generating a clock signal;
Applying the clock signal to the ATE, generating a test loop signal by the ATE, and applying the test loop signal to the DUT so as to repeatedly stimulate the DUT with a pass / fail boundary condition ; ,
Directing the laser pulse to a desired region of the DUT so that the photon energy transmitted is greater than the energy required to generate electron-hole pairs and the photons reach the focal point of the transistor. .   23. The method of claim 22, further comprising changing the phase of the clock signal with respect to the pulse number signal. A method for measuring carrier lifetime in a semiconductor device under test (DUT) using Laser Assisted Device Alteration (LADA) technology, comprising:
Repeatedly applying an electrical test signal to the DUT;
Irradiating a sensitive transistor in the DUT with a laser beam to obtain an optimal LADA signal to obtain spatial coordinates of the sensitive transistor;
Irradiating the sensitive transistor using a pulsed laser source to generate a laser pulse and using the spatial coordinates to direct the laser pulse;
Changing the timing of the laser pulse and recording the intensity of the LADA signal at each timing position until the intensity of the LADA signal reaches zero;
Measuring the carrier lifetime by calculating the time taken for the LADA signal to change from a maximum signal to a minimum signal.   25. The method of claim 24, wherein the laser beam is obtained from a continuous wave (CW) laser source.   25. The method of claim 24, wherein calculating the time comprises plotting the intensity of the LADA signal versus time.   Further comprising directing the laser pulse to the spatial coordinates and varying the timing until an optimal LADA signal strength is obtained to measure the pulse timing of a maximum laser pulse overlap with test signal overlap. Item 25. The method according to Item 24.