Time-gated detection, dual-layer spad-based electron detection

Abstract

To provide a time-resolved electron microscopy.SOLUTION: A method according to the present invention comprises the steps of: in an electron microscope, directing an electron beam to a sample; detecting events based on a modulation of the electron beam in response to corresponding pump pulses directed to the sample in a first single photon detector (SPD) array; and associating the detected events with time bins corresponding to elapsed times from application of the corresponding pump beam pulses.SELECTED DRAWING: Figure 1A

Description

[Technical Field]

[0001] The present disclosure relates to time-resolved electron microscopy. [Background technology]

[0002] Traditional time-resolved electron microscopy revolves around synchronizing a pump pulse, such as a laser pulse, microwave pulse, temperature pulse, mechanical pulse, or other excitation, to initiate the sample probing process. Typically, the pump pulse is directed toward the sample, interacting with it and initiating a sample response. A probe pulse is then directed toward the pumped sample, and the response to the probe pulse is measured. The temporal response can be obtained by varying the time delay of the probe pulse relative to the pump pulse. The probe pulse can be an electron beam pulse generated in response to a pump laser, often by doubling the frequency of the pump pulse and generating an electron beam pulse in response to the frequency-doubled pump pulse. The pulse / probe pulse is repeated to allow for an increased signal-to-noise ratio, and varying the pulse / probe time delay allows for reconstruction of the sample's temporal response. These pump / probe measurements can require long acquisition times to provide an adequate signal-to-noise ratio. Additionally, electron beam modulation and the associated time delays complicate the measurement setup.

[0003] Certain electron detectors can provide high temporal resolution but exhibit high noise, resulting in a low signal-to-noise ratio. For example, single-photon avalanche detectors (SPADs) can generate an avalanche-based output signal in response to a single electron. Unfortunately, SPADs also generate an output signal (dark counts) in the absence of an electron beam input. These dark counts can be reduced by SPAD cooling, but cannot be distinguished from a real signal.

[0004] In view of the above, there is a need for alternatives to the conventional pump / probe measurement approach, as well as alternative detector configurations. Summary of the Invention

[0005] The charged particle (CP) detector includes a first single photon detector (SPD) array and a second SPD array aligned with the first SPD array along a CP beam axis. A coincidence detector is affixed to one of the first SPD array and the second SPD array and is operable to indicate detection events corresponding to the detection of charged particles at corresponding SPD array elements of the first SPD array and the second SPD array within a predetermined time window. In some examples, the first SPD array and the second SPD array are defined on a first substrate and a second substrate, respectively, and the first substrate is fixed relative to the second substrate. In a further example, the first substrate and the second substrate are semiconductor substrates on which the SPD array elements of the first SPD array and the second SPD array are defined, respectively, and the first SPD and the second SPD are single photon avalanche detectors (SPADs). In a typical example, the coincidence detector includes a corresponding coincidence detector element for each pair of corresponding SPD array elements in the first SPD array and the second SPD array. In some cases, the first SPD array and the second SPD array include different numbers of SPD elements, and the coincidence detector is configured to indicate a detection event in response to detecting a charged particle in at least one SPD element in the first SPD array and at least one SPD element in the second SPD array. The coincidence detector can be defined on a third substrate fixed to at least one of the first substrate and the second substrate. In another example, a frame is disposed to align the first SPD array and the second SPD array, and at least one spacer is disposed to separate the first SPD array and the second SPD array. The first SPD array and the second SPD array can be defined on the first and second substrates, respectively, and the first substrate can be fixed relative to the second substrate. The third substrate is coupled to the first SPD array and the second SPD array to receive detection events from the first SPD array and the second SPD array and to generate a timestamp associated with each received detection event for a plurality of corresponding SPD array elements of the first SPD array and the second SPD array.In some examples, the third substrate is coupled to the first SPD array and the second SPD array to receive detection events from the first SPD array and the second SPD array and generate a histogram of the detection events for a plurality of corresponding SPD array elements of the first SPD array and the second SPD array. The first SPD and the SPD array can be single-photon avalanche detector (SPAD) arrays.

[0006] The method includes, in an electron microscope, directing an electron beam toward a sample and detecting events based on modulation of the electron beam in response to corresponding pump beam pulses directed toward the sample at a first single-photon detector (SPD) array. The detected events are associated with time bins corresponding to elapsed times since application of the corresponding pump beam pulses. Typically, a histogram of the number of detected events is created. In some embodiments, the electron beam is repeatedly blanked for at least a portion of the time interval between corresponding pump beam pulses. The electron beam can be unblanked based on application of the pump pulses. In some examples, the electron beam is unblanked with a fixed time delay relative to the pump beam pulse, and the pump beam pulses and electron beam blanking are applied at a fixed repetition rate. The fixed repetition rate can be between 100 Hz and 300 MHz, and the electron beam can be repeatedly blanked at intervals of at least 95% of a fixed period associated with the fixed repetition rate.

[0007] According to a representative embodiment, the first SPAD array is a single-photon avalanche detector (SPAD) array, and a second SPAD array fixed to the first SPAD array is used to determine whether a coincident event corresponding to an event detected by the first SPAD array is detected. A histogram is formed based on the coincident events, and the time bin for each coincident event is based on the time relative to the associated pump beam pulse. In some examples, the first SPAD array and the second SPAD array are defined on first and second substrates, respectively, the first substrate is fixed relative to the second substrate, and a third substrate is coupled to the first SPAD array and the second SPAD array and is operable to generate a histogram of detected events for a plurality of corresponding SPAD array elements of the first SPAD array and the second SPAD array.

[0008] The foregoing and other features and advantages of the disclosed technology will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0009] [Figure 1A] A representative electron microscope system configured to use a pump beam and a single-photon avalanche detector (SPAD) array is shown. [Figure 1B] FIG. 1B is a representative timing diagram associated with the system of FIG. 1A. [Figure 2A] A representative electron microscope system configured to use a pump beam, a SPAD array, and to blank and unblank the probe electron beam is shown. [Figure 2B] FIG. 2B is a representative timing diagram associated with the system of FIG. 2A. [Figure 3A] 1 shows a representative electron microscope system configured to use a pump beam, a probe electron beam generated in response to the pump beam, and a SPAD array. [Figure 3B] FIG. 3B is a representative timing diagram associated with the system of FIG. 3A. [Figure 4A] 1 shows a stacked SPAD assembly including first and second SPAD arrays defined on respective substrates, such as wafers, that are secured together. [Figure 4B] 1 shows a stacked SPAD assembly including first and second spaced apart SPAD arrays. [Figure 5] 1 is a cross-sectional view of a stacked SPAD array including first and second SPAD arrays defined on opposite major surfaces of a common substrate. [Figure 6] 1 shows two-dimensional stacked SPAD arrays defined on first and second substrates, including associated substrate circuitry areas. [Figure 7] 1 shows a two-dimensional SPAD array defined on a first substrate and stacked in alignment with a one-dimensional SPAD array defined on a second substrate. [Figure 8] 1 shows a two-dimensional SPAD array comprising multiple subarrays each containing a circuit area in which detection event circuitry is located. [Figure 9] 1 illustrates an exemplary method in which sample electron beam modulation in response to a pump pulse is detected using a SPAD array to generate an arrival time histogram for some or all of the SPAD elements of the SPAD array. [Figure 10] 1 shows a representative method for using stacked SPAD arrays. DETAILED DESCRIPTION OF THE INVENTION

[0010] Introduction Unlike conventional pump-probe techniques used in electron microscopy, the disclosed approach does not require modulating the electron beam, but rather the SPAD array and pump pulses are synchronized with the gating signal applied to the SPAD array. Even when using scintillator-based detectors, time resolutions of 5-30 ps or better can be achieved. The electron beam, which serves as the probe beam, can be left on, as in conventional transmission electron microscopy, or alternatively, it can be modulated on a slower timescale using an electrostatic blanker or other modulator to reduce the sample dose.

[0011] SPAD arrays can operate in a histogram mode, where detected events are added to a histogram as they are acquired, eliminating the need to store and transmit raw individual arrival times. Such "in-pixel" histograms are based on event times relative to gate or other synchronization pulses, which can be provided by dedicated clock circuits, or on the repetition rate of the pump pulse. Because only summary statistics (such as a histogram) are communicated, rather than individual event times, the data rate required to get the image off-chip is significantly reduced.

[0012] For measurements where only a small fraction of the probe beam electrons are captured by detectors such as (weak) diffraction or (core loss) EELS, the disclosed approach can offer a substantial increase in throughput and significantly reduced complexity compared to conventional pump-probe techniques at similar time resolution. The increased throughput can translate into reduced total measurement time and better sample survival under multiple pump pulses.

[0013] The following examples generally involve arrays of single-photon avalanche detectors (SPADs) based on semiconductors such as Si, Ge, InGaAs, or other materials. While the disclosed arrangements are generally suitable for detecting charged particles, such as electrons, that can be transferred by one SPAD to another appropriately aligned SPAD, the disclosed approaches are also suitable for other charged particle beams. The disclosed approaches can also be used, for example, in measurements based on short pump pulses and continuous probes. The pump pulses can be optical or charged particle beam pulses, mechanical, temperature, or other excitations. While some of the examples below show one-dimensional SPAD arrays for convenience, two-dimensional SPAD arrays can also be provided, with stacked SPAD elements arranged in an N×M array, where N and M are positive integers. In typical examples, each of the stacked SPADs has the same number (N×M) of SPAD elements, although different numbers can also be used. Additionally, while SPAD elements can be semiconductor-based devices, other types of elements can be used, such as superconducting nanowire single-photon detectors, photomultiplier tubes, microchannel plate photomultiplier tubes, superconducting transition edge sensors, and single-photon quantum dot detectors. Detectors operating in Geiger mode, such as SPADs, can be used in temporal histograms to record signal waveforms, as discussed below. Detectors exhibiting analog or other or non-Geiger responses can also be used in histogramming if appropriately processed for histogramming. As used herein, a single-photon detector refers to a photon or charged particle detector that produces a Geiger-mode response to a single photon or charged particle. The Geiger-mode response is substantially independent of the number of photons or charged particles detected. Such a response is sometimes referred to as a binary response, changing from on to off.

[0014] SPAD elements generate an avalanche response in response to received charged particles. Such responses are referred to herein as events or detection events. Such events can be used to reconstruct temporal waveforms by forming a histogram of the number of events as a function of time. SPADs can be defined on a substrate, such as a semiconductor substrate, or as discrete devices that are then placed and secured on a substrate for support. Typically, a silicon substrate is used, but SPAD elements can be defined in other semiconductors, such as germanium and InGaAs.

[0015] In some examples disclosed herein, the time resolution is determined by the SPAD element, and the probe (electron) pulse can be almost arbitrarily long. However, although the probe pulse duration is typically limited to reduce dose and pile-up distortion due to dead time in the SPAD element, the duration and timing of the probe electron pulse are less critical than in conventional pump-probe measurements. In conventional pump-probe measurements, the time resolution is determined by the probe pulse duration, and measurement throughput is low.

[0016] In an example, an optical pump pulse is applied to a sample, and modulation of an electron beam directed at the sample is detected with a SPAD array. Typically, the electron beam is blocked or attenuated during at least some time intervals when electron beam modulation by the sample in response to the optical pump pulse is not of interest, generally to reduce the electron dose on the sample. As used herein, "blanking" or "beam blanking" refers to attenuating, blocking, or deflecting the electron beam in this manner. Such blanking can be provided by modulation of the electron beam source or by using one or more electron beam deflectors or other electron-optical elements.

[0017] Although the examples are generally directed to pump / probe measurements, the disclosed approach can be used for the temporal characterization of various electron beam modulation techniques, such as fast (electrostatic) beam blanking, RF chopped pulse trains, etc., and can therefore be used to demonstrate and verify the performance of such techniques. In this case, the SPAD is synchronized to the preferred output of the beam modulation technique rather than to a pump source such as a pump laser.

[0018] Example 1 Referring to FIG. 1A, a typical measurement system 100 includes a pump source 102, typically a light source that generates short pulses, such as pulses having durations of 30 ns, 1 ns, 300 ps, ​​30 ps, ​​less than 1 ps, or shorter. The light source 102 can be a mode-locked laser operating at a fixed repetition rate or a light source that generates pulses on demand at a fixed, variable, or arbitrary rate, and is arranged to deliver a light beam 103 to a sample 111. As shown in FIG. 1A, the pump source 102 can generate pulses in response to a clock signal generator 104; however, in some examples, the pump source operates at a fixed frequency, which serves as the clock signal. In response to the clock signal, a gate signal generator 106 generates a gate signal that can be coupled to one or both of an electron beam source 108 and an electron beam deflector 130 to provide beam blanking or other electron beam modulation. The electron beam source 108 generates an electron beam that propagates along a beam axis 109 directed at the sample 111 and can be blanked to reduce beam dose.

[0019] The gating signal generator 106 is also coupled to one or more SPAD arrays 112 that generate detected events based on modulation of the electron beam in response to the pump pulse. In the example of Figure 1A, the processing circuitry 114 generates a histogram of detected events in time bins corresponding to the elapsed time since application of the pump pulse associated with the detected event. The gating signal generator 106 can operate with a fixed, variable, or random delay relative to the clock signal from the clock signal generator 104.

[0020] As shown in FIG. 1B, the SPAD array 112 is gated to an on state using a gate pulse 140 to measure probe pulse modulation in response to an optical pump pulse 142. The SPAD detector array 112 can generate an output signal, such as output signal 146, that includes a detection event 147 at any element of the SPAD detector array 112. Although the SPAD detector array 112 operates in Geiger mode, so that the amplitude of the detection event 147 is substantially independent of the signal amplitude, the times at which detection events are generated relative to a series of probe pulses (referred to as "arrival times") can be used to generate a histogram of arrival times. This histogram can be used to obtain the sample's temporal response to the probe pulse. Relative signal timing can be electronically set by the gate signal generator 106, and the probe electron beam is blanked with a blanking signal 144 to reduce sample dose. As shown in FIG. 1B, events are assigned to one of time bins 160A-160P, corresponding to the detection time referenced to the timing of the associated pump pulse. The same time bin is used for all repeated pump pulses relative to the timing of the associated pump pulse. More or fewer time bins can be used, and the time bins can have the same or different sizes. The pump pulses are applied repeatedly, as the pulse sequence shown in FIG. 1B is repeated, resulting in many events being detected and assigned to time bins 160A-160P.

[0021] The beam deflector 130 can be controlled to deflect the electron beam until the pump pulse is applied. In this way, the SPAD array 112 and the sample 111 receive a reduced electron beam dose compared to using an unmodulated electron beam. Additionally, blanking the electron beam reduces pile-up distortion, which causes the histogram generated in the SPAD array output signal to show responses to the probe pulse with significantly reduced amplitude. For example, with a deblanked beam, the SPAD detector array elements are likely to respond to portions of the electron beam unrelated to the sample 110's response to the pump pulse. Thus, the arrival time histogram may be biased toward earlier histogram bins. Although not shown in Figure 1B, after electron detection, the SPAD array elements typically have a dead time of 30 ns to 1 μs. As a result, responses to a deblanked electron beam can increase the dead time and delay signal acquisition.

[0022] Pile-up distortion can be digitally corrected or partially corrected by reducing the electron current or by changing the timing of the electron beam unblanking, i.e., unblanking the electron beam for part of the time window of interest after each pump pulse, and then varying the unblanking interval between subsequent pump pulses. Varying the timing of the SPAD gates can also be used, with the histogram time bin definitions left relative to the pump pulses rather than to the SPAD gates.

[0023] Example 2 In another example, shown in Figure 2A, the pump beam and detector gating can operate asynchronously, and the probe electron beam can be unmodulated. However, the time window of interest is typically very short compared to the period between pump pulses, and such asynchronous operation tends to be inefficient. To reduce pile-up distortion, the SPAD gate can be varied within the duration of the beam blanking or otherwise generated electron pulse.

[0024] The system 200 includes an electron beam source 202 that directs an electron beam 206 to a sample 210 via an electron optical column 204. If desired, a beam blanker 208 can be used to reduce sample dose. The sample 210 can be positioned on a sample stage 212, and a portion of the electron beam from the sample is directed to a detector, such as a SPAD array 214A, which can be coupled to a signal processor board 214B to process the SPAD array signal, such as for time correlation and histogramming. A laser 216 directs a pulsed laser beam to an optical optical system 224, which directs a pump beam 222 to the sample 210. In some alternatives, the electron beam 206 can be generated in response to the pulsed laser beam, typically via a second harmonic generator 218, to increase laser photon energy, if desired. However, in typical examples, the electron beam 206 is modulated by modulation of the electron beam source 202 or with a beam deflector 208. As described above, the electron beam 206 is modulated to reduce the electron beam dose at the sample 210 outside the time window of interest.

[0025] The SPAD array 214A is coupled to a gate generator 230 that receives a gate signal from a delay generator 232, which can activate the SPAD array 214A at random or predetermined times associated with the pump pulse. As shown, the delay generator is included within a controller 250, which also provides a gate signal to one or more of the laser 216, the electron beam source 202, and / or the beam deflector 208. In some examples, the laser 216 provides pulses at a fixed frequency, and the gate signal is time based on the fixed frequency. The SPAD array output signal is directed to a processor 240 for time correlation and time histogramming, although these operations can also be performed on the signal processor board 214B. In some examples, the controller 250 is coupled to the laser 216 to establish laser pulse characteristics such as pulse duration, repetition rate, and pulse energy.

[0026] As shown in Figure 2B, the SPAD array 214A is gated to its on state using a gate pulse 240 that also serves to gate the laser 216 to generate a laser pulse 242 that can serve as a pump pulse. The sample 210 can also be stimulated in different ways. The electron beam can be gated with a variable delay T relative to the pump pulse. D A probe (electron beam) pulse 244 having a time period τ can be generated. However, the electron beam need not be pulsed, since the temporal resolution is determined by the SPAD array 214A. Each element of the SPAD array 214A can generate an output signal, such as output signal 246, which can include a representative detection event 247. As noted above, the laser 216 can operate at a fixed frequency and thus can be used to time the SPAD gating.

[0027] Example 3 Referring to FIG. 3A, a typical time-gated detection system 300 includes an electron beam source 302 and an electron beam optical column 304 with suitable lenses, a deflector such as deflector 305, and other electron-optical components that direct an electron beam 306 (probe beam) to a sample 310, which may be positioned on a stage 312. In some cases, a deflector 308 is positioned to sweep the electron beam 306 after interacting with the sample 310 to different detector elements of a detector, such as a single-photon avalanche diode (SPAD) array 314A. The deflector 305 can be controlled to deflect the electron beam 306 so that the sample 310 is not exposed except for a suitable time. This electron beam blanking can be used to reduce the dose on the sample 310 and the dose received by the SPAD array 314A, particularly during the recovery time of the SPAD array. After interacting with the electron beam 306, a portion of the electron beam 306 is received by the SPAD array 314A. In other examples, secondary emissions or other radiation responsive to the interaction, such as x-rays, are detected alternatively or in addition to the electron beam portion using one or more suitable detectors.

[0028] System 300 can be operated as follows. First, note that in conventional optical pump / probe systems using an electron beam probe, a pulsed laser is used to generate pump pulses directed at the sample, and a portion of the pulsed laser output is used to generate electron beam probe pulses directed at the sample with varying delays between the pump and probe pulses. By varying the pump / probe delay, the sample's response to the pump can be measured with a slow detector. A sufficient number of pump / probe delays can be used to measure the sample's temporal response to the pump. In contrast, in systems such as system 300, the detector (e.g., SPAD array 314A) is gated by SPAD gate generator 311 to initiate data acquisition. As shown in Figure 3B, SPAD array 314A is gated to its on state using gate pulse 340, which also serves to gate laser 316 to generate laser pulse 342, which can serve as a pump pulse. While sample 310 can also be stimulated in a different way (i.e., without a laser pulse), laser pulses are convenient for time-resolved measurements on short time scales. A portion of the pump pulse can be directed toward the electron beam source 302 to generate a probe (electron beam) pulse 344. The SPAD array 314A can generate multiple output signals, such as output signal 346, which includes detection event 347. Because the SPAD array 314A operates in Geiger mode, the amplitude of the output signal 346 is independent of the signal amplitude; however, the times at which detection event 347 is generated relative to a series of probe pulses (referred to as "arrival times") can be used to generate a histogram of arrival times. This histogram can be used to obtain the temporal response of the sample to the probe pulse. The relative signal timing can be set electronically by the controller 309 and / or a variable optical pulse delay 320, which allows for varying the probe pulse timing.

[0029] The configuration in Figure 3A is a representative example; in other examples, the generation of the probe pulse is not based on the optical pump pulse. For example, the beam deflector 305 can be controlled to deflect the electron beam 306 until the pump pulse is applied. In this way, the SPAD detector array 314A receives a reduced electron beam dose compared to using an unmodulated electron beam. In addition, blanking the electron beam 306 reduces electron pile-up distortion, which causes the histogram generated in the SPAD array output signal to show responses to the probe pulse with significantly reduced amplitude. Although not shown in Figure 3B, after electron detection, the SPAD array elements typically have a dead time of 30 ns to 1 μs. As a result, responses to the unblanked electron beam can increase the dead time and delay signal acquisition.

[0030] The controller 309 can generate a time correlation of the SPAD detector array signal (Time Correlated Single Photon Detection - TCSPD) and generate a histogram for each element of the SPAD detector array 314 A. In some cases, some of such processing is provided by an adjacent or integrated signal processing board 314 B.

[0031] In another approach, the electron beam source 302 can respond to optical pulses from a laser 316, such that the electron beam 306 is a pulsed electron beam or includes a pulsed component. In some examples, the electron beam pulses are generated by photoemission, and the laser beam output can be frequency doubled with a second harmonic generator 318 to improve photoemission efficiency based on the work function of the electron emitter. Additionally, a variable delay generator 320 can be provided to control the relative timing of the electron pulse with respect to the optical pulse (i.e., pump pulse) in the laser beam 322, which is directed to the sample 310 via an optical optical system 324. Because the electron beam can be pulsed, electron beam blanking is generally not required. The electron beam and laser beam propagation to the sample 310 are disposed within a vacuum enclosure 301.

[0032] Example 4. Stacked SPADs SPADs can operate with a high temporal resolution of 30 ps or better. In some examples, two SPAD-based detector wafers (conveniently referred to as the top and middle wafers) containing arrays of SPAD elements (also called pixels) are secured to a third wafer (conveniently referred to as the bottom wafer) on which processing circuitry is defined. In some examples, the processing circuitry is located elsewhere to avoid exposure to the electron beam. Because high-energy electrons are not blocked by the top detector, they can generate free charge carriers in the SPAD elements in both SPAD-based detector wafers. Thus, a single electron can generate a detection event in the corresponding SPAD element in both SPAD-based detectors. Using timing circuitry, detection events can be measured in each SPAD-based detector with an accuracy of approximately 30 ps or better. Correlation based on detection time allows simultaneous detection events in the top and middle layers to be distinguished from noise events. The wafers can be bonded with copper-to-copper direct hybrid bonding and interconnected with through-silicon vias (TSVs), which enable 3D interconnection between layers.

[0033] Example 5. SPAD with integrated histogram SPADs and other single-event detectors can generate a large number of detection events that can be processed to determine a temporal waveform. This can require high-speed data transfer and / or data storage of the detection events. In the examples disclosed herein, the SPAD array operates in a so-called histogram mode, in which individual detection events are processed on-chip at each SPAD array or element to generate an event histogram for each pixel. In this histogram, bins correspond to electron arrival times associated with each gate or pump pulse. Pump-probe TEM measurements can be performed in a stroboscopic mode, in which a pump-probe sequence is repeated multiple times to collect detection events for histogramming. Detection events can be acquired at predetermined times or other time intervals to generate a sufficient number of detection events. As the detection events are acquired, they are accumulated in an element-by-element (pixel-by-pixel) histogram to generate a histogram of the number of detection events in multiple time bins associated with the gating signal. After the detection events are acquired and processed to generate a histogram, the histogram can be transferred from the SPAD array, resulting in a significant reduction in overall data transfer requirements. Detection event data, such as event time, does not need to be transferred.

[0034] Example 6 Referring to FIG. 4A , a SPAD array system 400 includes a first SPAD array 402 defined on a first substrate 403 and a second SPAD array 404 defined on a second substrate 405. The SPAD arrays 402, 404 include respective SPAD elements, such as representative SPAD elements 412, 414. As shown in FIG. 4A , the SPAD detector system 400 is typically positioned to receive an electron beam or other particle beam propagating generally along a charged particle beam axis 416. The SPAD elements are positioned such that electrons directed at a SPAD element in the first SPAD array 402 are received by a corresponding SPAD element in the second SPAD array 404. For example, electrons detected by SPAD element 412 may also be received and detected by SPAD element 414. Unlike SPAD detection of optical signals, electrons are not absorbed by the SPAD elements but propagate and may be detected by multiple detectors.

[0035] In the example of FIG. 4A , a third substrate 406 is provided that can include electrical connections and circuitry for some or all of the SPAD elements of the first and second SPAD arrays 402, 404. The third substrate 406 can be coupled to one or more external circuits, such as a control, bias, and clock circuit 420 that can control the biasing of the SPAD elements and provide output signals indicative of detection events at the SPAD elements. The external circuitry can include a coincidence detector 418 that indicates simultaneous detection events at corresponding SPAD elements of the SPAD arrays 402, 404. As used herein, “coincident” detection events are detection events that occur within a predetermined time interval, generally corresponding to events occurring within a time interval typically less than 1 ns, 300 ps, ​​30 ps, ​​10 ps, ​​or shorter. Alternatively, coincidence detection can be provided in circuitry defined on the third substrate 406. Coincidence detection can be used to discard single noise events that might otherwise be mistaken for true signal events. The coincidence detector 418 indicates an electron detection only if both corresponding SPAD elements of the SPAD arrays 402, 404 indicate a detection event within a selected time interval.

[0036] Coincidence detection, SPAD array biasing, and other processing and control circuitry can be defined in the third substrate 406. In some cases, some circuit elements are generally defined in the first substrate 402 and the second substrate 404, or near or together with the SPAD elements. In this example, the substrates 402, 404, 406 can be bonded together with the SPAD elements properly aligned with respect to the axis 416.

[0037] Referring to FIG. 4B, a stacked SPAD array system 450 includes first and second SPAD arrays 452, 454 secured relative to one another by a frame 458. The frame 458 positions the SPAD arrays such that corresponding SPAD elements, such as representative SPAD elements 462, 464, are aligned along an axis 456. A circuit board 460 defining bias, control, coincidence detection, or other circuitry may also be secured by the frame 458. Some or all of such circuitry may be provided by an external SPAD control system 470. The SPAD arrays 452, 454 and the circuit board 460 are spaced apart by respective gaps 474, 475.

[0038] Although not shown in Figures 4A-4B, event histogramming can be performed using circuitry / processing elements located on some or all of the boards or provided externally.

[0039] Example 7 5, stacked SPAD array 500 includes corresponding SPAD arrays 503, 505 including SPAD elements, such as representative SPAD elements 504, 506, defined on respective surfaces 508, 510 of a substrate 502. Alternatively, separate substrates can be secured together, as shown at 520.

[0040] Example 8 Referring to FIG. 6 , a first two-dimensional stacked SPAD array includes a two-dimensional array of SPAD elements, such as representative element 604, defined on a first substrate 602. In addition, SPAD control circuitry can be provided within substrate region 606. A corresponding second two-dimensional stacked SPAD array (shown in dashed lines) can be defined on a second substrate 612 and can include SPAD elements, such as representative SPAD element 614 corresponding to SPAD element 604. The second substrate 612 also includes a substrate region 616 (shown in dashed lines), which can include control, bias, electrical connection, and other circuitry. The first and second two-dimensional stacked SPAD arrays are coupled to external analog and / or digital circuitry, which can include amplifiers, signal processors, filters, counters, match circuits, and a processor that receives match events and generates a histogram. In other examples, such circuitry is provided in one or more of substrate regions 606, 616.

[0041] Example 9 Referring to FIG. 7, a dual SPAD system includes a two-dimensional array of SPAD elements, such as representative elements 704A-704D, defined on a first substrate 702. In addition, SPAD control circuitry can be provided within substrate region 706. A one-dimensional SPAD array (shown in dashed lines) can be defined on a second substrate 712 and can include SPAD elements, such as representative SPAD elements 714A-714C. SPAD element 714A is associated with SPAD elements 704A-704D. In this example, SPAD elements 714A-714C are associated with multiple SPAD elements in the two-dimensional SPAD array. While the one-dimensional array does not confirm event coincidence with the same spatial precision as a stacked two-dimensional array, it can distinguish noise events. The second substrate 712 also includes substrate region 716, which can include control, bias, electrical connection, and other circuitry. SPAD arrays can be stacked and coupled to external analog and / or digital circuitry, such as that shown in 720, which can include amplifiers, signal processors, filters, counters, matching circuits, and a processor that receives matching events and generates a histogram. Typically, with such a SPAD system, events detected in the two-dimensional array are confirmed by matching detection in the one-dimensional array, and these detected events are used for histogramming. Stacked SPAD arrays can have different numbers and sizes of SPAD elements. For example, a first array can have multiple elements, while a second array can have fewer elements, including only a single element, which can be used to establish matching, but where a matched array is advantageous.

[0042] Example 10 8, a SPAD array 800 is defined on a substrate 801 and includes respective SPAD sub-arrays 806, 816, 826, 836, each including a SPAD element, such as representative SPAD elements 804, 814, 824, 834, and control circuitry 802, 812, 822, 832. Histogram data is provided to a serializer 840 for transmission from the SPAD array 800, although the histogram data may be stored on the substrate 801.

[0043] Example 11 Referring to FIG. 9 , an exemplary method 900 includes applying a pump pulse to a sample at 902 and starting a timer at 904. Typically, the probe beam is also unblanked, but unblanking can be provided before or after the pump pulse using a time increment or time delay, which can be fixed or variable. At 906, events are detected with one or more SPAD elements and assigned to a time bin at 908. At 910, a SPAD dead time or recovery time can be allowed to elapse, and at 912, it is determined whether a pulse interval, typically a pulse period corresponding to the pump pulse repetition rate, has elapsed. If not, method 900 returns to 906 for detection of additional events associated with the pump pulse. If the pulse interval has elapsed, it is determined at 914 whether to apply and measure additional pump pulses, and if so, the pump pulse is reapplied at 902. Without additional pulses, the histogram of time-binned event counts is provided directly from the SPAD detector assembly, including a suitable substrate on which the processing circuitry is defined, as in FIG. 4A, in some instances, but the histogram can be generated with processing external to the SPAD array. The histogram can be read when a specific number of counts is reached in one or more bins, after a total number of counts is reached, or at a fixed rate (regardless of the number of counts).

[0044] Example 12 Referring to FIG. 10 , a representative method 1000 includes aligning first and second SPAD arrays at 1002. At 1004, events are detected at one or more array elements of the first and second SPAD arrays. At 1006, the detected events are evaluated to detect matching events in corresponding elements of the first and second SPAD arrays. If there are matching events, they are identified as true detected events at 1008 and added to an event histogram. Alternatively, if there are no matches, they are identified as noise and discarded at 1007. In a typical example, a count of such noise events is maintained, and each additional noise event is associated with an increment of a noise event counter. A match can be established as the event is received, or the event can be processed after data collection if a suitable event timestamp is recorded to enable a match determination. As shown in FIG. 10 , after the event is processed (or during event processing), the method continues to detect events. In an example, the event is associated with electron beam modulation in response to a pump beam applied to the sample. The electron beam can be a continuous beam, but to reduce the sample dose and reduce electron pile-up, the electron beam can be unblanked during the period of interest, typically at a time interval after application of the pump beam to the sample.

[0045] General Terms As used in this application and the claims, the singular forms "a," "an," and "the" include the plural forms as well, unless the context clearly dictates otherwise. Additionally, the word "comprises" means "comprises." Furthermore, the word "coupled" does not exclude the presence of intermediate elements between the coupled items.

[0046] The systems, devices, and methods described herein should not be construed as limiting in any way. Instead, the present disclosure is directed to all novel and non-obvious features and aspects of the various disclosed embodiments, alone and in various combinations and subcombinations with one another. The disclosed systems, methods, and devices are not limited to any particular aspect or feature or combination thereof, nor do the disclosed systems, methods, and devices require that any one or more particular advantages be present or problems be solved. While any theory of operation is for ease of explanation, the disclosed systems, methods, and devices are not limited to such theory of operation.

[0047] Although some operations of the disclosed methods are described in a particular order for convenience of presentation, it should be understood that this description style encompasses reordering unless a specific order is required by specific language described below. For example, operations described sequentially may, in some cases, be reordered or performed simultaneously. Furthermore, for simplicity, the accompanying figures may not show the various ways in which the disclosed systems, methods, and apparatuses can be used with other systems, methods, and apparatuses. Additionally, this specification sometimes uses terms such as "generate" and "provide" to describe the disclosed methodologies. These terms are high-level abstractions of the actual operations that are performed. The actual operations corresponding to these terms will vary depending on the particular implementation and will be readily discernible to those skilled in the art.

[0048] In some instances, values, procedures, or devices are referred to as "lowest," "best," "smallest," etc. It will be understood that such descriptions are intended to indicate that choices can be made from among many functional alternatives used, and that no such choice is necessarily better, smaller, or otherwise more preferable than other choices.

[0049] The examples are described with reference to directions indicated as "above," "below," "upper," "lower," etc. These terms are used for convenience of description and do not imply any particular spatial orientation.

[0050] In view of the many possible embodiments to which the principles of the disclosed technology can be applied, it should be recognized that the illustrated embodiments are preferred examples only and should not be construed as limiting the scope of the disclosure.

Claims

1. 1. A method comprising: In an electron microscope, directing an electron beam at a sample; detecting events based on modulation of the electron beam in response to corresponding pump pulses directed at the sample at a first single photon detector (SPD) array; associating the detected event with a time bin corresponding to an elapsed time since application of the corresponding pump beam pulse; A method comprising:

2. The method of claim 1 , further comprising forming a histogram of the number of detected events.

3. The method of claim 1 , further comprising repeatedly blanking the electron beam for at least a portion of the time interval between corresponding pump pulses.

4. The method of claim 3 , wherein the electron beam is unblanked upon application of the pump pulse.

5. The method of claim 4 , wherein the electron beam is unblanked at a fixed time delay relative to the pump pulse.

6. The method of claim 4 , wherein the pump pulses and the electron beam blanking are applied at a constant repetition rate.

7. 7. The method of claim 6, wherein the constant repetition frequency is between 100 Hz and 300 MHz, and the electron beam is repeatedly blanked at intervals of at least 95% of a fixed period associated with the constant repetition frequency.

8. the first SPD array is a single photon avalanche detector (SPAD) array; and determining whether a matching event corresponding to the event detected in the first SPAD array has been detected using a second SPAD array fixed to the first SPAD array; forming a histogram based on the matching events; and The method of claim 1 , wherein the time bin for each coincidence event is based on the time relative to the associated pump beam pulse.

9. the first SPAD array and the second SPAD array are defined on first and second substrates, respectively, and the first substrate is fixed relative to the second substrate; 9. The method of claim 8, wherein a third substrate coupled to the first SPAD array and the second SPAD array is operable to form a histogram of detection events for a plurality of corresponding SPAD array elements of the first SPAD array and the second SPAD array.

Images