Camera improved for electron diffraction pattern analysis

Abstract

An apparatus for detecting a Kikuchi diffraction pattern is provided, the apparatus comprising: an electron column configured, in use, to provide an electron beam having an energy in the range of 2 keV to 50 keV directed toward a sample; an imaging detector for sensing and counting electrons from the sample resulting from interaction of the electron beam with the sample, the imaging detector having an array of pixels and a counting rate capability of at least 2,000 electrons per second for each pixel, the imaging detector providing electron energy filtering of the sensitive electrons for counting the sensitive electrons representing the diffraction pattern; and a particle detector having a passivating layer on a surface where the electrons enter toward the active region of the detector, the passivating layer dispersing the detected energy of 20 keV incident electrons over an energy spread having a full width at half maximum of less than 3.2 keV. A method for detecting a Kikuchi diffraction pattern is also provided.

Description

Technical Field

[0001] The present invention relates to an apparatus and method for performing microdiffraction analysis, particularly electron backscatter diffraction (EBSD) analysis, on a sample to detect a chrysanthemum diffraction pattern.

Background Art

[0002] In microdiffraction analysis, an electron beam is directed towards a crystalline sample, and different types of particles are generated by the interaction of the electrons within the electron beam with the sample. Electrons that are emitted from the source electron beam, elastically backscattered from the sample, and then diffracted by the lattice planes of the crystal are of particular interest in materials research. These electrons, having an energy close to that of the primary beam, form the basis for electron backscatter diffraction (EBSD) analysis. For EBSD analysis, a pixelated detector captures the diffraction contrast as an image (diffraction pattern), which is used to measure properties of the sample, such as crystal orientation and strain. A scanning electron microscope (SEM) is typically used to generate the primary electron beam and mount the sample and detector.

[0003] A particle counting pixel array may be used as an imaging detector for microdiffraction analysis, as described in US8890065. In such a system, the magnitude of the signal generated by each individual particle received by the detector is compared to a threshold value to distinguish the signal from system noise so that the individual particles can be counted. Further, since the signal level is governed by the energy of the incident particle, this threshold value can be used to distinguish between particles of different energies by counting only those particles having an energy higher than a settable threshold or falling within a band between two threshold values. Such particle counting pixel arrays were originally developed for use in high energy physics experiments or as X-ray detectors (which are sometimes known as hybrid photon counting detectors or HPC detectors), but are today adopted for use in EBSD.

[0004] In EBSD, the “signal” in question is the diffraction contrast within the EBSD pattern, which is mainly carried by electrons having energies that fall within a narrow energy band between the primary beam energy E0 of the SEM (where E0 is typically considered to be 20 keV) and typically something 1 - 2 keV lower than E0. Electrons having energies lower than this band do not contribute to the diffraction contrast, but rather simply contribute to the background for the measurement that reduces the accuracy when measuring the diffraction contrast. The relative component of backscattered electrons carrying the diffraction contrast increases when the electron beam is incident on the sample at a small angle with respect to the sample surface plane, and for this reason, conventional EBSD experiments are carried out with the sample tilted by an angle that allows this incident geometry.

[0005] To identify the type and orientation of crystals that are the cause of characteristic Kikuchi diffraction contrast, the EBSD pattern is processed to detect lines in the image and associate them with planes within the crystal. Typically, this is achieved using the Hough transform technique. With the lines and angular relationships measured, the results are used to find a close fit to the type of crystal. If not enough lines are detected or a sufficiently close fit to the type of crystal cannot be found, the pattern cannot be indexed. When the pattern can be indexed, the orientation of the crystal can be determined. When the pattern is acquired at grid points covering the field of view, a map showing the crystal orientation at every point within the field of view can be obtained. If there is a significant component of the pattern that cannot be indexed, the orientation map is not useful. In this case, the SEM electron beam current or the acquisition time for individual patterns must be increased in order to obtain an acceptable component of the successfully indexed patterns. There are limits as to how much beam current can be generated within the SEM, and the sample can be damaged if an overly high beam current is used. When the acquisition time per pixel is increased, the time to obtain the orientation map will become longer. Therefore, it is desirable to achieve a high percentage of successfully indexed patterns using the minimum beam current and shortest acquisition time.

[0006] The percentage of patterns that are successfully indexed at a given beam current and acquisition time depends in part on the sample and various experimental conditions. However, for a specific sample and fixed experimental conditions, the "sensitivity" measure describes the ability of the detection equipment to obtain patterns that can be successfully indexed using short acquisition times and low beam currents. One measure of sensitivity is the reciprocal of the product (beam current × pattern acquisition time) required to achieve an acceptable percentage (e.g., 95%) of successfully indexed patterns.

[0007] As shown in US8890065, it is expected that the sensitivity when measuring diffraction signals using an electron counting EBSD detector can be improved by using a discriminator that accepts only pulses with a higher level than a certain energy equivalent threshold level. The signal / background should improve when this threshold is raised, and the optimal result should be achieved using a threshold close to the primary beam energy. In fact, Vespucci et al. (S. Vespucci et al., "Digital direct electron imaging of energy-filtered electron backscatter diffraction patterns," Phys. Rev. B - Condens. Matter Mater. Phys., Vol. 92, No. 20, pp. 8 - 14, 2015) have shown that the diffraction contrast measured from the band intensity, i.e., ((maximum value - minimum value) / (maximum value + minimum value)), is improved by raising the threshold close to the beam energy, and a four-fold improvement in contrast can be achieved by raising the threshold from 5.5 keV to 19.4 keV when collecting EBSP from diamond with an incident beam energy of 20 keV.

Prior Art Documents

Patent Documents

[0008]

Patent Document 1

Non-Patent Documents

[0009]

Non-Patent Document 1

Non - Patent Document 2

Non - Patent Document 3

Non - Patent Document 4

Summary of the Invention

Problems to be Solved by the Invention

[0010] However, when diffraction patterns are processed using a similar embodiment of the apparatus shown in US8890065, the use of a higher threshold level improves the contrast of the diffraction pattern and enables the observation of higher - order diffraction features. However, the sensitivity is thought to decrease with an increase in the threshold. There is a need for an improved system for EBSD pattern detection that increases sensitivity when used for orientation mapping applications.

Means for Solving the Problems

[0011] According to a first aspect of the present invention, there is provided an apparatus for detecting a kikuchi diffraction pattern. The apparatus comprises an electron column adapted to provide an electron beam having an energy in the range of 2 keV to 50 keV and directed towards a sample, and an imaging detector for sensing and counting electrons from the sample resulting from the interaction of the electron beam with the sample. The imaging detector comprises an array of pixels and has a counting rate function of at least 2,000 electrons per second for each pixel. The imaging detector is adapted to provide electron energy filtering of the sensed electrons for counting the sensed electrons representing the diffraction pattern. The particle detector has an inactive layer on the surface where electrons enter towards the active region of the detector, and the inactive layer disperses the detection energy of 20 keV incident electrons by an energy spread having a full - width at half - maximum lower than 3.2 keV.

[0012] In this context, a "particle detector" refers to a detector that converts the energy of each incident particle into an electronic signal, as opposed to an "indirect" detector that, for example, uses a phosphor screen to convert particle energy into light and uses an intermediate optical system to focus this light onto a sensor. The inefficiencies associated with the optical elements for an indirect detector lead to a loss of detection sensitivity. Further, an indirect detector typically generates a signal current that represents the energy×velocity product for a stream of incident particles, whereas a particle detector can measure signals from individual particles.

[0013] As used herein, the term "pixel" generally refers to spatially distinct sensing regions of a detector such that particles incident on two different pixels are considered to have collided with two different regions of the detector. Thus, a "pixel" may refer to an array of conventional pixels in a direct electron detector or a CCD, as is known in the art, and also to the individual regions of, for example, a silicon strip detector where each "pixel" is an area associated with each contact strip.

[0014] Imaging detectors that are adapted to provide electron energy filtering of the sensed electrons to count or more specifically selectively count the sensed electrons representing the diffraction pattern described above have been found to be advantageous particularly in that this "thresholding" can improve the ratio of relevant diffraction information to diffuse background information that does not represent the diffraction pattern. The inventor has recognized that it is possible to achieve a significant improvement over this approach using the apparatus according to the first aspect. This is achieved using an inert or "insensitive" layer constituted by a detector that disperses the energy of the passing electrons to a lesser extent than when using conventional devices. In other words, the apparatus can reduce the dispersion of the recorded signal due to the effect of a relatively thick inert layer that has conventionally been used at the entrance to the sensor.

[0015] The "dead" layer is described in this disclosure as "inactive" in the sense that this term refers to an inactive material. As will be explained in more detail later in the method disclosure, a typical detector uses a layer of such material to make an electrical connection to the active sensor layer. This non-active, inactive, or "dead" layer can be thought of as the detector entrance window.

[0016] By reducing the spread in electron energy produced by transmission through the inert layer, the discrimination or division of diffraction-related electrons from non-related electrons by thresholding becomes more substantial, i.e., the range of the recorded signal that is attributable to electrons that indeed carry diffraction information is reduced compared to existing devices.

[0017] An inert layer adapted or shaped, or having a thickness and material properties such that it is particularly adapted to disperse the detected energy of 20 keV incident electrons over an energy spread having a full width at half maximum of less than 3.2 keV, will provide this advantageous effect, i.e. it will be understood that for 20 keV incident electrons, the layer will cause a spread in the transmitted electron energies such that the width of the energy spectrum curve of the electrons transmitted through the layer is measured between the energy peaks or energy values ​​that are half the maximum of the curve.

[0018] Preferably, the inert layer disperses the detected energy of 20 keV incident electrons over an energy spread having a full width at half maximum of less than 2 keV, more preferably less than 1 keV, and even more preferably less than 0.1 keV.

[0019] Preferably, the electron column is part of a scanning electron microscope (SEM) that provides beam energy in the range of 2 keV to 50 keV during normal operation. Typically, this arrangement results in the detector being on the same side of the sample as the electron source such that the sensed particles have an orbital component that returns along the electron beam axis with respect to the sample. However, the diffraction pattern can be detected by transmission, in which case the detector is placed on the side of the sample opposite the electron beam. In this architecture, the sensed particles do not have an orbital component that returns along the electron beam axis with respect to the sample. In transmission EBSD, the sensed "signal" is composed of elastically scattered electrons that have passed through the sample rather than being reflected from the sample. It will be understood that the above-described energy range from 2 keV to 50 keV with respect to the electron beam is a useful energy range for EBSD as an SEM technique.

[0020] The minimum detector counting rate capability is required by the data rate of the EBSD experiment. In some experiments, it is considered that the data rate can easily reach 10,000 events per second, and thus the detector has a counting rate capability of 2,000 events per second, preferably 5,000 events per second, more preferably 10,000 events per second, and even more preferably 100,000 events per second for each pixel. Such a counting rate is particularly important in the present invention when each particle is sensed without being filtered before hitting the detector and without being discriminated by the detector and then filtered and counted.

[0021] The detector has at least one particle counter for counting the sensed particles representing the diffraction pattern described above. The at least one particle counter has a counting rate of at least 2,000 events per second, preferably 5,000, more preferably 10,000 events per second, and even more preferably 100,000 events per second. Most preferably, the counter is at least 1×10 6It has a counting speed function per event per second. This function is particularly beneficial for the techniques contemplated by the disclosure of the present invention. In some implementations, the array of pixels and at least one particle counter are single components in which all features are integrated on one chip, and these are typically referred to as monolithic active pixel sensors (MAPS). However, in other implementations, the array of pixels is bump-bonded to at least one particle counter, and these are known as "hybrid" detectors or hybrid active pixel sensors (HAPS).

[0022] Preferably, each pixel has a corresponding particle counter, providing a "one-to-one" mapping between the pixel and the particle counter. In architectures where different numbers of pixels and particle counters exist, "one-to-many" or "many-to-one" relationships are created. Ideally, the particle counters correspond to and are arranged in an array having the same pitch as the array of pixels, although other arrangements are also contemplated. In a HAPS detector, the array of pixels is bump-bonded to the array of particle counters, although it is considered possible to have both arrays as a single member as in MAPS.

[0023] More preferably, each particle counter generates an individual output signal according to the energy of each sensed particle. When each pixel has a corresponding particle counter, this advantageously increases the rate at which incident events can be counted. The data rate in an EBSD experiment can be considered high or higher, up to about 10,000 events per second per pixel. When each pixel has a corresponding particle counter, each particle counter has a counting speed function of at least 2,000 events per second, preferably 10,000 events per second. Higher counting speed functions such as 100,000 events per second are also contemplated. Each pixel having a corresponding particle counter provides distinct advantages over sequential readout detectors such as CCDs in providing moderately fast particle measurements. Importantly, the detector used in the present invention must have the function of counting particles at a sufficiently fast rate.

[0024] Preferably, the electron amplifier at each pixel introduces an equivalent electronic noise energy having a full width at half maximum lower than 2 keV, preferably lower than 1 keV.

[0025] Yet another improvement that may be achieved in some embodiments relates to charge sharing, which will be described in detail later in the disclosure of the present invention. Preferably, the particle detector contains circuitry for detecting and correcting charge sharing between pixels that may occur with respect to a single incident particle. By thus reducing the effect of charge sharing across pixel boundaries, harmful variance in the recorded signal is further reduced. Typically, in such embodiments, the circuitry is configured to perform or achieve the steps of adding the electronic signal collected in a given pixel to the electronic signals collected in adjacent pixels, applying electronic energy filtering for counting the sensed particles representing the diffraction pattern to the added electronic signals, and assigning the counted particles to a single pixel.

[0026] Preferably, the particle detector is configured to output both the arrival time and magnitude of the signal captured in each pixel, and the computer algorithm is used to perform or is configured to perform the steps of identifying instances where a single incident particle generates simultaneous electronic signals in multiple pixels, adding the multiple electronic signals generated by a single incident particle and collected in multiple pixels, applying energy filtering for counting the sensed particles representing the diffraction pattern to the added electronic signals, and assigning the counted particles to a single pixel.

[0027] In some preferred embodiments, the ratio (active layer sensor thickness) / (pixel-to-pixel spacing) is less than 5.

[0028] Preferably, the apparatus is configured such that the number of electrons counted per pixel during pattern acquisition is read out as a data unit of 6 bits or less.

[0029] Preferably, the camera sensor array has a configurable pixel amplifier that is adapted to achieve more than one pulse length that meets different pixel counting speed requirements and energy resolution requirements.

[0030] In fact, it will be understood that the energy distribution of electrons representing the diffraction pattern and the energy distribution of electrons not representing it are typically not separate. The overlap between these distributions is preferably addressed by the filtering described above. Thus, electron energy filtering is preferably adapted to distinguish between sensing particles having energy representing or more faithfully representing the diffraction pattern described above and sensing particles having energy representing or more faithfully representing the background.

[0031] Typically, the incident electron beam is incident at an angle in the range of 45 - 90° with respect to the sample surface plane. However, it is also contemplated that an incident angle outside this range with respect to the sample surface can be used, for example, a shallower angle as conventionally used in EBSD as will be described later in the disclosure of the present invention.

[0032] According to a second aspect of the present invention, there is provided an apparatus for detecting a kikuchi diffraction pattern, the apparatus being adapted to provide an electron column having an energy in the range of 2 keV to 50 keV and directed towards a sample when used, and for sensing and counting electrons from the sample due to the interaction of the electron beam with the sample, the apparatus comprising an array of pixels having a counting speed function of at least 1,000 electrons per second for each pixel, the detector being adapted to provide electron energy filtering of the sensing electrons for counting the sensing electrons representing the diffraction pattern described above, and the particle detector having an inert layer on the surface where electrons enter towards the active region of the detector that disperses the detection energy of 20 keV incident electrons to be smaller than the spread of energy induced by the transmission of 1,500 nm of inert silicon.

[0033] Any of the features and characteristics described with respect to the above and below embodiments of the disclosure of the present invention may relate to the apparatus of either or both of the first and second aspects.

[0034] According to a third aspect of the present invention, a method for detecting a kikuchi diffraction pattern is provided. The method includes providing an electron beam having an energy in the range of 2 keV to 50 keV and directed towards a sample using an electron column, and detecting and counting electrons from the sample resulting from the interaction of the electron beam with the sample using an imaging detector having an array of pixels and having a counting rate function of at least 2,000 electrons per second per pixel. The detector is adapted to provide electron energy filtering of the detected electrons for counting the detected electrons representing the above diffraction pattern. The particle detector has an inactive layer on the surface where electrons enter towards the active region of the detector, and the inactive layer disperses the detection energy of 20 keV incident electrons with an energy spread having a full width at half maximum lower than 3.2 keV.

[0035] According to a fourth aspect of the present invention, a method for detecting a kikuchi diffraction pattern using the apparatus according to the first or second aspect is provided.

[0036] Embodiments of the present invention will be described below with reference to the accompanying drawings.

Brief Description of the Drawings

[0037]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10A

Figure 10B

Figure 11

DETAILED DESCRIPTION OF THE INVENTION

[0038] In the disclosure of the present invention, as shown in FIG. 1, special definitions are used for "signal" and "background". Electrons that are subject to the diffraction effect contributing to the Kikuchi band contrast are within an energy band of width ΔE diff that lies directly below the primary beam energy. Most of the electrons are not diffracted within the Kikuchi band and undergo multiple scattering events that result in continuous energy loss within the sample. These scattered electrons have energy that extends downward until they reach zero energy and are the cause of the diffuse background in the camera image. Some of these scattered electrons are within the same energy band ΔE diffwill have the energy inside. In representing the scattered electrons incident on the detector, two clearly different energy bands are defined, namely, an energy range E>E0-ΔE that includes substantially all signal electrons as well as some background electrons diff which has a "signal band", and an energy range E<E0-ΔE which contains only a negligible number of signal electrons and can be regarded as being composed only of background electrons diff which has a "background band". Using an electron counting detector with a configurable threshold TH, only electrons having an energy higher than TH can be counted.

[0039] The diffuse background changes only gradually across the image, while the intensity within the region of the diffraction band changes abruptly. Therefore, the (maximum - minimum) intensity observed within the diffraction band will be dominated only by the number N signal of the detected electrons affected by the diffraction effect, while the (maximum + minimum) intensity will be the total number N background of the detected electrons including the number of background electrons N tot , that is, N tot =N signal +N background and will be dominated by it. Therefore, the diffraction contrast ((maximum - minimum) / (maximum + minimum)) will depend only on N signal / (N signal +N background ) = 1 / (1 + N background / N signal ) and will continue to improve when TH increases as long as the signal - to - background ratio SBR = N signal / N background continues to increase.

[0040] The aim of the method described here is to improve the "sensitivity" of an electron counting EBSD detector. The "sensitivity" of an EBSD detector can be understood to be inversely proportional to the electron dose required to collect diffraction patterns that can be analyzed with a defined accuracy and precision. As discussed above, one way to define accuracy and precision is the percentage success rate for indexing patterns from a particular sample and fixed experimental conditions. The electron dose is defined as the total number of SEM primary electrons to which the sample is exposed during acquisition of the diffraction pattern, and is proportional to the product of the SEM primary beam current *

[0041]

[0042]

[0043] and the exposure time. When the exposure time is long, the pattern acquisition rate decreases. When the electron dose is high, any sample may be damaged. Therefore, it is desirable to use a detector with high sensitivity so that EBSD patterns can be indexed successfully as quickly as possible with a minimum electron dose. The emission of backscattered electrons from the sample is a random process, i.e., the energy and direction of individual emitted electrons are subject to a statistical distribution. Therefore, the number of electrons incident on a single pixel of the detector during EBSD pattern measurement varies around an average value according to a Poisson probability distribution. The partial randomization of the signal magnitude within each pixel has the effect of introducing random "Poisson" noise into the EBSD pattern measurements that obscures the kikuchi diffraction contrast and makes pattern indexing more difficult. The ease of being able to detect a signal depends on the magnitude of the signal relative to the magnitude of the noise introduced by a statistical variation that will be referred to herein as the signal-to-Poisson noise ratio (SPNR). The sensitivity of an EBSD experiment increases when the SPNR increases for fixed experimental conditions and electron dose. In an EBSD pattern, the magnitude of the signal within a pixel is proportional to the average number of detected electrons carrying the diffraction contrast, while the statistical noise is governed by Poisson counting statistics over the total number of detected electrons, and thus the following equation holds. JPEG0007695267000001.jpg 30150 SPNR and sensitivity clearly depend on both SBR and N signal and, to maximize SPNR, the system must be configured to achieve the optimal combination of SBR and N signal .

[0044] In an electron counting EBSD detector, both N signal and SBR are affected by TH, and SPNR will vary as a function of TH. Figure 2 illustrates how changing the level of TH will affect SPNR. At very low TH values, all background electrons and signal electrons are detected. As TH is raised, background electrons are excluded and SBR increases, but N signal remains the same, and thus SPNR increases as does SBR. In a theoretical "perfect" electron detector (perfect in terms of detecting all incident electrons and measuring their energy without error), SPNR continues to increase as TH is raised until TH = E0 - ΔE diff , the reason for this limit being that any TH value higher than this will exclude some signal electrons carrying diffraction contrast, and thus N signal will be reduced. When TH is raised above E0 - ΔE diff , some additional background electrons will be excluded, but the most significant effect is that a certain fraction of the signal electrons carrying diffraction contrast will no longer be detected. Thus, SPNR will decrease mainly as a result of fewer N signal . Thus, in a perfect electron detector, it is thought that optimal SPNR and sensitivity are obtained by setting TH close to the value of TH = E0 - ΔE diff .

[0045] In samples and experimental conditions where the amount of background electrons emitted from the sample is large, the relative increase in SPNR possible from energy thresholding is large. This is due to a greater improvement in SBR possible when only signal electrons are selectively detected. As an example, for several practical reasons, it is preferable to perform EBSD analysis in a state where the beam is incident at a larger angle (in the range of 45° to 90°) with respect to the sample surface plane (the "large angle" condition). However, under this condition, a significantly larger amount of backscattered electrons emitted are background electrons compared to conventional experiments performed using a state where the beam is incident at a small angle (~20°) with respect to the sample surface. Therefore, EBSD experiments are rarely performed under large angle conditions due to extremely low SPNR and sensitivity. When SPNR is improved by energy thresholding, EBSD analysis of samples using the large angle condition can result in higher speed or lower beam current.

[0046] Electronic noise and pulse pile-up Figure 2 illustrates a case where, although ideal, in an actual detector, the relationship between TH and SPNR becomes complicated by a combination of physical effects. One known problem is that the pulse amplitude resulting from a single incident electron is subject to electron noise. Therefore, when reducing the amplitude so that the electron noise is smaller than TH, any diffracted electrons having an incident energy greater than TH may not be detected. Similarly, when the change in electron noise takes an amplitude greater than TH, any diffuse background electrons having an incident energy smaller than TH may still be detected. At TH set near E0-ΔE diff these effects limit both SBR and N signal in the acquired EBSD pattern. Figure 3 illustrates how electron noise having a full width at half maximum (FWHM) of 2 keV affects the measured values of both the signal band and the background band by effectively spreading the measured values both above and below the true energy. When TH is set to E0-ΔE diff which is considered to be near optimal in a perfect detector, due to noise variations, some signal electrons fall below TH, thereby causing Nsignal and the SPNR decreases. When TH is lowered, the number of signal electrons will increase, but it will also allow more background electrons to be detected. As a result, the optimal value of SPNR is obtained using a TH slightly below E0-ΔE diff and is smaller than the SPNR that can be achieved using a perfect detector.

[0047] The effective blurring of the energy TH due to electronic noise is typically reported as the "energy resolution" of a particle counting camera, but the related effect on the EBSD "sensitivity" (defined above) suitable for pattern elucidation has not been recognized. Vespucci et al. conducted experiments using a camera with an electronic noise FWHM of 2 keV. When the magnitude of the electronic noise is reduced to an FWHM of 1 keV, the applicant's simulations predict that an improvement of up to 30% can be achieved with the optimal SPNR.

[0048] Therefore, in order to maximize the SPNR for EBSD, it is desirable to minimize the contribution of electronic noise to the measured values. Electronic noise can be improved through the design and processing of the imaging sensor, but it is also affected by the choice of electronic filtering for the readout amplifier for each pixel. When the filter time constant for each pixel amplifier is increased, the voltage noise is reduced, making it possible to set the threshold TH higher to improve the optimal SPNR and sensitivity for EBSD pattern elucidation. However, when the time constant is increased, the probability that the pulses resulting from the arrival of individual electrons will not be resolved increases. Since the pulse arrival time follows a Poisson time distribution, the probability that two pulses will arrive within the resolution time of the pixel amplifier increases with the counting rate. Most of the electrons hitting the detector are scattered background electrons, so there is a fairly high probability that any coincidence that is not resolved between two background electrons will occur. When such a "pile-up" occurs, the measured pulse height is approximately the sum of the pulse heights that would be seen from individual events, and may be greater than TH even if none of these pulses exceed the threshold if the individual pulses did not arrive together. Therefore, when the average pixel counting rate is on the order of the reciprocal of the pulse pair resolution time, pile-up allows more background events to be accepted, thus degrading the SPNR.

[0049] When an EBSD pattern is acquired by a camera, the pixel counting rate is directly affected by the beam current incident on the sample. When spatial resolution or sample damage is of concern, it is preferable to use a lower beam current to reduce the sample dose and the lateral size of the focused electron beam. In this context, it is advantageous to extend the filtered pulse duration for the pixel amplifier to lower the pixel counting rate, reduce the electronic noise, and enable higher SPNR and EBSD sensitivity to be achieved by thresholding. In situations where the sample can withstand a high dose and the SEM can be operated at a high beam current without sacrificing spatial resolution, the acquisition time to achieve each EBSD pattern can be shortened due to the high pixel counting rate. In this case, the pixel amplifier filter needs to provide a sufficiently short pulse length to avoid significant pile-up. The associated increase in electronic noise will reduce the achievable SPNR for a fixed count value in the image, but increasing the count value will improve the SPNR such that EBSD pattern elucidation can still be achieved at a higher rate than when using a lower beam current. To enable various different types of samples and spatial resolution requirements, it is advantageous for the camera sensor array to have a configurable pixel amplifier that enables achieving more than one pulse length to match different pixel counting rate and energy resolution requirements.

[0050] There are local variations in the properties of any real sensor, including the electron transfer properties of the sensor layer, or within the electronic counting circuit. Due to these variations, the response to incident electrons will be different for each pixel. Thus, in practice, when TH is set to the same reference level for all pixels, the range of incident electron energies that result in counting is not uniform for all pixels. When averaged over all pixels, this effect obscures the energy threshold in addition to the effect introduced by the electronic noise.

[0051] Some particle counting detectors incorporate features that compensate for variable pixel threshold effects in order to improve the uniformity of energy filtering characteristics across all pixels. Typically, these features apply a "shift" to the local electron threshold at each pixel in order to equalize the energy filtering characteristics of all pixels. These "threshold trimming" features improve the energy filtering resolution of the sensor.

[0052] Furthermore, the inventors have found that this camera cannot provide the advantages of using it to improve sensitivity to EBSD unless the particle counting camera includes at least two further distinct important features regarding the generation of charge sharing between the incident window and the pixels.

[0053] Sensor insensitive layer and energy dispersion The particle counting detector includes an active sensor layer in which the energy of incident particles is absorbed by a series of interactions that liberate electron-hole pairs. FIG. 4 illustrates a schematic cross-sectional view around a single pixel. A charge cloud is formed and swept towards the collection electrode 402 by the internal electric field, where the amount of liberated charge is measured, and this signal charge is proportional to the energy of the incident particle. The depth into the sensor where charge is liberated depends on the type of incident particle (e.g., X-ray or electron) and the energy of the particle. For incident electrons, the higher the energy, the deeper the penetration into the active sensor layer 401, but the electrons must pass through the dopant used to form the electrical connection to the active sensor layer. This inactive layer 403 substantially forms the "incident window" for the sensor. When incident electrons lose any energy by inelastic interactions within this layer, this energy does not contribute to the signal charge, and thus this layer may be referred to as an "insensitive layer". Furthermore, some of the charge liberated near the insensitive layer can recombine before they can be swept to the collection electrode, thereby causing yet another loss of a portion of the signal charge. As a result, energy may be lost when incident electrons cross the insensitive layer, and some of the liberated signal charge may be lost before being collected, and thus the measured energy may be lower than the incident electron energy.

[0054] As described by J.D. Segal et al. in "Thin-Entrance Window Sensors for Soft X-rays at LCLS-II" (Proceedings of the 2018 IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS / MIC), 2018, pages 1-2, doi:10.1109 / NSSMIC.2018.8824674.), fully depleted high resistivity silicon sensors require doping contacts at the entrance window to terminate the diode. Conventionally, this region is achieved by continuing high temperature annealing to activate the dopants in the ion implantation of dopant species. This annealing further pushes the dopant profile deeper and increases the depth of the inactive layer. Further, a surface metal layer is typically deposited on top of the doped surface layer and connected to the bias voltage. Thus, in existing devices, a 1 micron thick aluminum surface metal layer is typically used on top of the 2 micron thick implanted region into silicon for X-ray pixel sensors. When pixel sensors are used to detect X-rays greater than a few keV, any X-ray photons absorbed within the inactive layer do not generate any signal, and any photons reaching the active region generate a charge signal proportional to the total energy of the photon. Thus, known pixel detectors include a relatively thick insensitive layer that has little effect on X-rays (or high energy (100 keV) electrons), but this insensitive layer has a significant impact on the intended use of the present disclosure.

[0055] When a direct detection semiconductor sensor with a thresholding function is used to image electrons in a transmission electron microscope where the electron energy is typically greater than 100 keV, the thickness of the dead layer is not important. The high energy of the incident particles ensures that any energy loss due to the dead layer is relatively low, and by setting the threshold to approximately half of the incident energy, typically all particles are counted and the threshold is high enough to ensure that no false triggers are given due to electron noise offsets. However, in a SEM where the beam energy may be only 20 keV or less, the effect of the dead layer can be significant.

[0056] Figure 5 illustrates how the spectrum of an incident 20 keV monochromatic electron beam (a distinct peak at 20 keV energy) is thought to be modified upon passing through a 1 micron silicon dead layer. The electrons lose a variable amount of energy as a result of many random scattering interactions in this layer, which not only results in a reduction of the average energy of the transmitted electrons, but also disperses the energy over a wide range. The diffracted electrons contributing to N within a small band ΔE close to the beam energy are thought to undergo similar energy losses and dispersions. Figure 6 illustrates a schematic drawing comparing how the measured energy distributions of the signal band and the background band are thought to be detected by a "perfect" electron detector with the true incident distribution as it is thought to be detected following transmission through such a dead layer. The effect of the dead layer on these low energy electrons is such that the signal distribution and the background distribution significantly overlap, and this overlap will obscure both the signal distribution and the background distribution in such a way as to reduce the ability to separate contributions by thresholding. diff within N signal is considered to experience similar energy losses and dispersions. Figure 6 illustrates a schematic drawing comparing how the measured energy distributions of the signal band and the background band are thought to be detected by a "perfect" electron detector with the true incident distribution as it is thought to be detected following transmission through such a dead layer. The effect of the dead layer on these low energy electrons is such that the signal distribution and the background distribution significantly overlap, and this overlap will obscure both the signal distribution and the background distribution in such a way as to reduce the ability to separate contributions by thresholding.

[0057] When TH is set to a level low enough (e.g., TH1 in Figure 6) to capture substantially all signal electrons and maximize N within the measured EBSD pattern, a high proportion of background electrons N signal is backgroundwill still be detected, resulting in patterns having a considerably lower SPNR compared to what is considered achievable by thresholding in a perfect detector. To optimize the SPNR for a detector with a significant dead layer, a compromise between N signal and SBR is required, and thus the optimal SPNR and sensitivity achievable are considerably worse than what is possible using a theoretically "perfect" detector. This effect is illustrated in FIG. 7 showing the relationship between SPNR and TH for a theoretically perfect detector and one having an incident window corresponding to a 1 micron silicon dead layer.

[0058] The dead layer is a necessary feature of any semiconductor sensor due to the need to have an electrical contact with the depletion zone forming the active region of the device. There are special processing techniques that reduce the effective dead layer thickness sufficiently to allow low energy photons to reach and be detected in the active region. However, in EBSD where low energy electrons are involved, it is not only the simple transmission of the dead layer that is important to utilize the advantage of energy thresholding, but also the effective spreading (dispersion) of the energy that occurs to these electrons after they penetrate this layer. This spreading of energy occurs because the combined effect of inelastic scattering and incomplete charge collection in the dead layer results in a change in the signal obtained from electrons of a fixed energy. A typical incident window on a direct detection semiconductor detector consists of a metal contact and an injection layer that results in an energy dispersion of the detection signal from 20 keV incident electrons that is greater than the energy dispersion caused by the transmission of a 2 micron silicon layer. The inventor has determined that in order to utilize the advantage of thresholding to improve sensitivity for EBSD pattern elucidation, the dead layer including any metal contact must be attenuated such that the spread of the effective energy of a 20 keV electron beam is less than that caused by an inactive silicon layer of 1500 nm.

[0059] It will be well understood that as electrons pass through a thin layer of material, the spread of energy increases with the thickness of the material. This relationship between thickness and energy spread is described by a mathematical expression (as described in Mikheev, N. and Stepovich, Mikhail and Yudina, S., "Energy loss spectra for a fast charged particle beam transmitted through a material film of specified thickness," Journal of Surface Investigation - x ray Synchrotron and Neutron Techniques - J SURF INVESTIG - X - RAY SYNCHRO, 3, pp. 218 - 222, 10.1134 / S1027451009020086, 2016), or more typically by electron transport simulations (as described in Attarian Shandiz, M., Salvat, F., and Gauvin, R., "Detailed Monte Carlo Simulation of electron transport and electron energy loss spectra," Scanning, 38, pp. 475 - 491, https: / / doi.org / 10.1002 / sca.21280, 2016). As an example, the spread of energy of an initial monochromatic (single - energy) beam of 20 keV electrons after passing through silicon layers of different thicknesses is shown in FIG. 8, and this energy spread is represented as the full width at half maximum (FWHM) of the energy distribution after passing through the silicon layer. Using techniques such as these mathematical models and simulation software, the relationship between energy dispersion and incident window thickness can be calculated. Thus, these techniques can be used to calculate the energy dispersion due to the insensitive layer and, therefore, can be used to identify the appropriate physical, mathematical, and geometric properties for the inert layer according to the disclosure of the present invention.

[0060] In a silicon detector, the material in the incident part for manufacturing conductive contacts and semiconductor p-n junctions is typically modified by ion implantation. The conductive region gives no signal, and thus there is effectively an insensitive layer in the incident part to the active region of the detector. Therefore, electrons must pass through a thin inactive silicon layer that will cause the electron energy distribution to spread or disperse before reaching the active region when incident on the detector. By reducing the energy dispersion effect of the insensitive layer, the available SPNR from the detector will increase, and this reduction in the energy dispersion effect is achieved by using a processing technique to reduce the thickness of the insensitive layer. The electrical properties of silicon are modified by ion implantation, but the electron scattering properties are not affected, and the effect of the insensitive layer on the measured energy of electrons incident on the sensor is equivalent to the effect of electron transmission of a silicon layer of the same thickness as the insensitive layer. In addition to ion implantation, other processing methods can be used, and additional thin surface layers such as oxides and nitrides can be included.

[0061] When other materials are present on the incident surface, these materials will similarly increase the energy dispersion for electrons reaching the active region of the detector. However, in the present invention, the relevant aspect of the incident window is the amount of energy dispersion of electrons passing through the incident window, regardless of the layer thickness or the constituent material. In EBSD, it is advantageous to represent the energy dispersion effect of the insensitive layer using the full width at half maximum (FWHM) of the energy distribution of the initial monochromatic (single energy) beam of 20 keV electrons after passing through the insensitive layer. Given this value, the effect of the incident window on electrons of different incident energies can be predicted (for example, by using electron transfer simulations as in the case of Shandiz et al.).

[0062] When the insensitive layer induces an energy dispersion of 3.2 keV or higher in terms of FWHM for a 20 keV monochromatic electron beam (corresponding to the effect of a 1500 nm Si inactive layer), it has been calculated that it cannot significantly affect the SPNR of the pattern obtained by a realistic detector based on applying the detection energy threshold TH shown in FIG. 9. This graph illustrates SPNR simulation curves (similar to those described in FIG. 7) for detectors with different insensitive layer thicknesses. In these examples, an energy dispersion of 2 keV corresponds to a 1,000 nm thick insensitive layer, an energy dispersion of 3.2 keV corresponds to a 1,500 nm thick insensitive layer, and an energy dispersion of 4.8 keV corresponds to a 2,000 nm thick insensitive layer. In the case of an energy dispersion of 3.2 keV, the energy filtering concept brings only negligible advantages to the SPNR, while larger and smaller dispersions than that shown can be seen to achieve worse SPNR and significant SPNR improvement respectively. That is, at an energy dispersion of 3.2 keV, the SPNR increases when TH increases, but only to a lesser extent than in the case of 2 keV. For this reason, the present device advantageously includes an insensitive layer that induces an energy dispersion lower than 3.2 keV for a 20 keV monochromatic beam of electrons incident perpendicular to the sensor.

[0063] It will be understood that the results shown in FIG. 9 depend on other aspects of the detector included in the SPNR simulation. The simulation for this example is applied to a detector that, firstly, does not observe the charge sharing effect (e.g., using any form of charge summing algorithm), and secondly, includes an electron amplifier that introduces an electron noise equivalent with a full width at half maximum of ~2 keV. Instead, if we assume a simulation of the function for a detector with additional detrimental factors such as the charge sharing effect and high noise for the electron amplifier, the SPNR curve is thought to indicate the need for a considerably lower insensitive layer energy dispersion for the device to achieve the desired SPNR improvement.

[0064] In a typical exemplary device, the particle counter has pulse processing electronics, and each "event" to be counted is a pulse of charge achieved by an incident particle that accumulates energy in a pixel. The pulse processing electronics includes an amplifier, a discriminator, and a counter. An important aspect of the device is the counting speed of the detector. In an EBSD experiment, it is considered that the speed of incident particles hitting the detector can be 10,000 events per second. Therefore, in order to distinguish between particles of different energies, the particle counter of the detector should have the function of counting at a counting speed of 2,000 events per second each. Preferably, the particle counter can count at a speed of 10,000 events per second, more preferably 100,000 events per second. Importantly, regardless of whether each pixel has its own particle counter, the architecture of the detector is such that the detector has a counting speed function of at least 2,000 events per second per pixel.

[0065] The type of detector described in a typical embodiment is a "direct detector". Such a detector has the function of detecting any type of particle that satisfies an energy threshold, for example, electrons, X-rays, and photons. However, the present invention is not limited to direct detectors, and other types of detectors having the functions of imaging and particle counting at an appropriate speed per pixel can be used. It is considered that a direct detector can have a surface coating such as a scintillator that converts energy into light on the premise that the response time is short enough to resolve signals from individual particles. An example of another type of direct detector that can be used is a silicon strip detector. Detectors using sequential readout such as CCDs typically cannot count at a sufficiently fast speed, but in principle, such detectors can be used.

[0066] Exemplary fabrication steps for forming a detector according to the disclosure of the present invention are shown in FIGS. 10A and 10B. The device is depicted in stages 1001 to 1020 through multiple stages of its manufacture. The device to be obtained is an imaging detector having on its surface an inert layer with a thickness less than 100 nm that provides only the low-energy dispersion required for transmission electrons. In this example, the sensor is bonded to a Medipix3 readout chip. The technique described in US8890065 includes stages that use a Medipix2 readout chip. This Medipix2 readout chip has an array of 256×256 pixels, each with an area of 55 μm 2 and has a counting function up to ~1×10 6 counts per second. Here, a Medipix3 with additional on-chip charge sharing correction functionality and a configurable counter depth B (both described below) is preferred.

[0067] Exemplary materials for forming the illustrated components are shown in the symbol tables described in each of FIGS. 10A and 10B.

[0068] In section 1001, a SiO2 layer of 100 - 300 nm is deposited on an N-type silicon wafer. In 1002, photoresist patterning and boron implantation are carried out. The removal and activation of the photoresist by standard annealing treatment are shown in 1003. In 1004, the SiO2 layer on the incident window side of the wafer is thinned, and a protective photoresist layer is disposed on the readout side. In 1005, arsenic is implanted on the incident window side by ion implantation using an ion energy in the range of 5 - 15 keV. In 1006, activation is carried out by microwave annealing treatment. Conventionally, the annealing treatment is carried out using a temperature higher than 700 °C, which is considered to cause significant diffusion of arsenic dopants into silicon. The modulated output circuit wave annealing treatment in 1006 of this example enables the activation of dopants without increasing the temperature of the bulk silicon above 500 °C, thereby resulting in only negligible diffusion. Typically, the photoresist cannot withstand the annealing treatment process and is removed at this stage. In 1007, an electrical contact opening to the implanted region is etched. In 1008, aluminum is deposited on both sides by sputtering. In 1009, the aluminum on the pixel side is patterned by etching, and in 1010, the resist is removed. Next, in 1011, a passivation layer is deposited (for example, by plasma enhanced chemical vapor deposition (PECVD) using SiO2, SiN at a low temperature lower than 400 °C, or by atomic layer deposition (ALD) using Al2O3). In 1012, the passivation layer on the readout side is patterned by lithography and etching, and 1013 exemplifies the deposition of field metal (Ti - W + Cu or Au) by sputtering. This field metal is required for electroplating used to deposit under bump metal (Ni) as shown in 1014. Next, in 1015, the photoresist used for this deposition is removed. As shown in 1016, the passivation layer and aluminum are removed from the incident window. In 1017, an opening to the aluminum contact of the incident window is etched, and then, in 1018, the photoresist used in this etching process is removed.Next, the wafer is diced (not shown), and sensor chips and readout chips are produced. At 1019, field metal and under-bump metal are also produced for the readout chips shown in addition to the sensor chips, and solder bumps are electroplated. Finally, at 1020, the sensor chips and the readout chips are bump-bonded to each other.

[0069] Charge sharing Yet another problem with pixel detectors is that portions of the diffusive charge cloud generated by a single incident particle may reach the readout electrodes for adjacent pixels, such that the charge liberated by a single incident particle is effectively shared between a pixel and its adjacent pixels. This sharing is quite likely to occur when the incident particle strikes a sensor near a pixel boundary. When a pixelated detector is used to detect photons, this "charge sharing" effect is known to cause some degradation of the imaging resolution due to the spreading of the response away from the central pixel. However, in a pixel detector using thresholding, if the charge collected in adjacent pixels is low, the pulse for this pixel may not exceed TH. If only the central pixel pulse is greater than TH, full imaging resolution is maintained. Thus, to optimize the spatial resolution when using a single photon counter with a monochromatic beam and to avoid multiple pixels counting the same photon, the threshold is typically set at 50% of the incident photon energy.

[0070] The inventors have found that in EBSD, charge sharing can greatly limit the extent to which SPNR can be improved by thresholding. In EBSD, electrons incident on the sensor are absorbed near the incident surface, due to the fact that electrons typically only have energies of 20 keV or less, and therefore the released charge cloud must drift almost the entire depth of the sensor before reaching the readout electrode. As this charge cloud drifts, lateral diffusion increases the likelihood that some charge will cross the boundary between two pixels. The degree of charge sharing varies depending on where the incident electron hits relative to the pixel boundary. This variation results in a variable reduction in pulse amplitude, which in this case excludes some signal electrons that would normally be counted because the acquired pulse is smaller than TH. This effect is most evident when TH is set close to the primary beam energy E0.

[0071] The example in Figure 2 shows that for a theoretical perfect detector, TH is approximately E0-ΔE diff This illustrates that SPNR and sensitivity for EBSD can be optimized when E is set to . Typically, E can be 20 keV and ΔE diff can be 1 keV, and TH is set accordingly to 19 keV. Thus, in this example, a single electron incident with an energy of 20 keV would not be counted if the accumulated energy >1 keV (5%) is shared with another pixel. For a pixel detector with a depth of 300 μm and pixels of 55 μm×55 μm, the inventors have estimated that more than 60% of the 20 keV signal electrons will be incident close enough to the pixel boundary that the accumulated energy >1 keV will not be collected by the pixel where the electrons impinge. In these cases, the measured electron energy will be less than TH, and more than 60% of the signal electrons will not be detected, despite having an incident energy greater than TH.

[0072] The reduction of the pulse amplitude in a photon detector can be mitigated by lowering the TH. However, in EBSD, lowering the TH will result in a decrease in SBR, causing deterioration of the SPNR and sensitivity suitable for pattern elucidation. The variable charge loss from the pixel where the electrons are incident results in a similar outcome to the energy loss and diffusion that occur when the incident electrons are scattered within the non-active material of the insensitive layer. When charge sharing exists, similar to the case of the energy lost within the insensitive layer (Figs. 6 and 7), the optimal SPNR is reduced compared to a complete detector and is obtained at a lower value than that of a complete detector.

[0073] From the above description of the charge sharing mechanism, it will be understood that charge sharing can be reduced to some extent by appropriate selection of sensor design parameters. The pixel pitch of the sensor is an important factor as a larger pitch reduces the component of the pixel area near the boundary with another pixel. Furthermore, the depth of the active sensor layer reduces the degree of charge sharing, and a thinner layer enables the charge cloud to drift over a shorter distance before collection at the pixel electrode. The shorter drift time results in a reduction in the lateral diffusion of the charge cloud, limiting the possibility of the charge cloud crossing the boundary between adjacent pixels.

[0074] The lateral radius of the charge cloud scales linearly with depth as the charge cloud drifts from end to end across the sensor layer, which enables the lateral size of the charge cloud to be estimated as a function of the sensor layer thickness. This estimate can be related to the pixel pitch in order to calculate the effect of charge sharing on the SPNR. Simulations of the SPNR in EBSD experiments (Si sample, 20 keV beam energy) have suggested that pixelated sensors designed with a ratio (active sensor layer thickness / pixel pitch) greater than 5 will experience a significant drop due to the charge sharing effect at the optimal SPNR available at high TH. Therefore, pixelated electron counting sensors operating at high TH need to have a ratio (active sensor layer thickness / pixel pitch) smaller than 5 in order to observe a significant improvement in the SPNR from energy thresholding.

[0075] When the sensor thickness cannot be further reduced and the pixel pitch or pixel size cannot be made any larger, the charge sharing effect can be reduced by including an auxiliary circuit for performing "addition nodes" that add the signals within the pixel and its nearest pixels for all pixels. All pixels have a detection threshold TH det and the addition node has a separate threshold equivalent to TH that is used to distinguish between background electrons and signal electrons. Counting is assigned only to one or more pixels that exceed TH as long as the addition node is greater than TH det . TH det is set to a value low enough to detect pulses that are high enough to reduce due to charge sharing but reduce the likelihood that adjacent pixels will be greater than TH det . An alternative approach is to include a circuit that compares the pulses within an individual pixel to all adjacent pixels when any addition node is greater than TH. Counting is assigned to the pixel when the measured pulse amplitude is greater than all adjacent pixels and at least one of the adjacent addition nodes is greater than TH. These "charge addition" circuits significantly reduce the impact of charge sharing on the SPNR, but there is an increase in the effective electronic noise due to the combination of the electronic noise from each contributing pixel when voltages are added at the addition node. However, the applicant's SPNR simulations for EBSD applications show that the improvement in SPNR from the reduced charge sharing effect significantly outweighs the reduction caused by the increased effective electronic noise.

[0076] The charge addition algorithm can be implemented off-chip when the information output by the sensor is sufficient to identify and reproduce a single particle event from signals from multiple pixels (e.g., when the sensor provides both the arrival time and intensity of the signals captured by all pixels as in the case of using a Timepix3 sensor or when the expected average count rate is ≪ 1 / pixel / frame).

[0077] In one embodiment of the off-chip charge addition algorithm, a sensor such as Timepix3 can be configured to output the arrival time and magnitude of all electronic events recorded by the detector. In this case, a computer algorithm can be used to identify electronic events that are recorded substantially simultaneously within a cluster between neighboring pixels, and these can be regarded as candidate instances of charge sharing. In another embodiment, a sensor such as Timepix or Timepix3 is configured to measure the total energy accumulated in each pixel during a single exposure, and the beam current or exposure time is reduced such that the average number of incident electrons per pixel within a single exposure is << 1. Under this condition, there is a high probability that a cluster between adjacent pixels that measures the energy accumulated within a single exposure is caused by a single incident electron that undergoes charge sharing rather than multiple incident electrons. In both embodiments, the energies measured within these adjacent pixels are added, and when the total added energy is less than the primary electron beam energy (i.e., the added energy is likely to have originated from a single incident electron), the cluster between adjacent events is regarded as a single charge sharing event. In this case, the added energy from the electronic event is assigned to the single pixel within the cluster between adjacent pixels that contributed the largest amount of energy to this added energy, and the energy measured in all other pixels resulting from this electronic event is set to zero.

[0078] The processing steps applied to the off-sensor charge sharing correction method are almost the same as those implemented by the "charge addition" circuit on the sensor. However, this processing is applied by a computer program or integrated circuit separate from the readout chip, rather than by an integrated circuit on the readout chip itself. These algorithms can be useful when obtaining an EBSD pattern using a sensor without a charge addition circuit from a sample that is susceptible to beam damage. In this case, the EBSD pattern obtained using the smallest possible electron beam dose must be successfully indexed. Therefore, it is useful to process the acquired data to obtain the data at a low dose and SPNR, and then improve the SPNR to a level that enables the charge sharing effect to be removed and the pattern to be successfully indexed.

[0079] Data transfer speed An additional object of the disclosed apparatus and method of the present invention is to improve the speed at which an EBSD pattern can be transferred out of the sensor for processing. During one EBSD pattern acquisition, the electron count within each pixel is predefined but is typically stored on the sensor as a data unit having a programmable digital bit number B. Typical values of B used in electron counting experiments are 8, 12, 16, or 24. The maximum count value that can be recorded by the electron counter at each pixel during one acquisition is (2 B -1). If this value is exceeded within any pixel during one acquisition, the count value recorded in that pixel becomes invalid, and the acquired pattern is no longer an accurate measurement result of the sample diffraction pattern.

[0080] When pattern acquisition is complete, the electron counters at all pixels are read from the sensor at a fixed read speed R read (typically, B read = B) bits per second (bps), and this speed is determined by the sensor and data transfer electronics. Therefore, the speed at which a complete EBSD pattern can be read from the sensor per second is R when N bit is the total number of pixels on the sensor. pix ​frame = R bit / (B read · N pix ) is. Since new pattern acquisition cannot start until the electronic counter from the previously acquired pattern is fully read out from the sensor, R frame imposes a limit on the maximum pattern acquisition speed of the EBSD experiment. For a given detector, R bit and N pix are fixed, and thus, at higher EBSD acquisition speeds, it is preferable to minimize B read .

[0081] The B read value required for the sensor used in the EBSD experiment depends on the electron count value per pixel required for this pattern to successfully index the EBSD pattern. In detectors that do not distinguish between signal electrons and background electrons, most of the electron counts will correspond to background electrons, thereby significantly increasing the average count value per pixel required for successful indexing of good results. Detectors that selectively count signal electrons will obtain successfully indexed patterns with much fewer electron counts per pixel. This allows for the selection of a significantly smaller B read value for a good EBSD experiment and allows for an improvement in the readout speed R frame .

[0082] The number of electrons required per pixel to obtain an indexable pattern without using selective detection of electrons is typically at least 50. Taking into account an appropriate tolerance, this number requires that B read be greater than 6 to ensure that the acquired pattern is suitable for indexing. However, in detectors that acquire patterns with a high SBR due to energy-selective counting, successfully indexed patterns are obtained with 20 or often fewer electrons per pixel. As a result, B read can be set to 6 or often less in a good EBSD experiment. A small B readThe value is desirable for high-speed reading, but for other purposes, it is high or more, up to about 12 bits (for example) of B read It can be useful to include a detector using read . However, not all patterns will require this number of bits for analysis of favorable results. Therefore, the number of bits read from the register is advantageously a detector parameter that can be set to read B read (sometimes less than B) for high-speed data transfer. The number of read bits B read must be less than or equal to the number of bits used for storage, and it is advantageous if B read can be set to be 6 bits or less, 5 bits or less, 4 bits or less, 2 bits or less, or even 1 bit or less.

[0083] Improvement of atomic number contrast The total number of backscattered electrons emitted from the material increases with its average atomic number. As a result, the total count of backscattered electrons for an EBSD image is an indicator of the average atomic number of the material where the incident electron beam impinges. When the beam is scanned across a grid of positions on the sample surface and the total count of backscattered electrons is recorded at each position, a map can be generated showing the distribution of materials having different atomic numbers. This map provides additional information that supplements the crystallographic information obtained from diffraction patterns. Further, instead of adding all the counts within the EBSD image, when only the counts from a set of pixels that cover a sub-region of the entire image are added at each beam position, the map can be made to more faithfully represent a specific contrast mechanism for a limited angular range of the emitted electrons defined by the shape of the sub-region used for the addition.

[0084] The energy distribution of backscattered electrons is also affected by the atomic number of the sample. The distribution for samples with a high atomic number contains a higher proportion of electrons at higher energies than the distribution for samples with a low atomic number. Thus, when low-energy backscattered electrons are excluded by energy filtering, the ratio of the number of signals for high- and low-atomic number samples is greater than the ratio of the total number of signals obtained without using energy filtering. Thus, when EBSD patterns are measured using an energy thresholding detector and used to generate a map of electron backscatter intensity, as explained immediately above, these maps will exhibit a higher intensity contrast between multiple sample regions having different atomic numbers when low-energy electrons are excluded by thresholding.

[0085] Exemplary device In an exemplary apparatus corresponding to a preferred embodiment, the detector is a direct electron detector including a sensor layer bump-bonded to a pixelated array of particle counting electronics circuitry. The detector is positioned as close as possible to the sample such that it maximizes the component of electrons backscattered from the sample in an EBSD experiment and incident on the detector. The sensor layer is a monolithic semiconductor such as silicon, and the surface layer facing the sample is doped to provide an insensitive surface layer to enable electrical connection to the sensor layer. Backscattered electrons incident on the detector liberate charge clouds within the active portion of the sensor layer through the insensitive layer, where the insensitive layer disperses the energy of the incident electrons to a lesser extent than the spread of energy induced by a 1500 nm inactive silicon layer. Preferably, the insensitive layer must be 100 nm or less so as to lead to an energy spread lower than 100 eV for a 20 keV monochromatic electron beam. The total thickness of the sensor layer can typically be 300 μm.

[0086] The pixelated array of the particle counting electronic circuit can typically comprise an array of 256×256 pixels. The particle counting circuit at each pixel includes an amplifier for measuring the energy of the sensed electrons. This circuit results in a counted event when the measured energy of the sensed electrons is greater than a threshold value to distinguish between background electrons and signal electrons. Towards efficient selection of signal electrons, the particle counting circuit measures the energy of the sensed electrons having an electron noise distribution with an FWHM equivalent to less than 2 keV, preferably less than 1 keV. The count value recorded at each pixel for each pattern acquisition is stored, and in this case, this storage can be configured to provide a 12-bit counter per pixel, or configured as a 4-bit counter or read as a 4-bit counter to facilitate fast readout of the acquired pattern from the detector.

[0087] A preferred embodiment of the detector incorporates additional features to mitigate the effect of charge sharing on the SPNR. In one embodiment, the pitch of the pixelated array of the particle counting electronic circuit is large enough such that the ratio (sensor layer thickness) / (pixel pitch) is 5 or less. For example, when the sensor layer thickness is 300 μm, the pitch is greater than 60 μm, but pitches greater than 100 μm are also contemplated. In another embodiment, the particle counting circuit is supplemented by an additional summing node for each pixel that adds the charges liberated by a single incident electron and collected by the pixel and its immediate neighboring pixels. The summing node generates a counted event when the combined signal within it exceeds a threshold value, and this count is assigned to the single pixel that measured the largest charge amount compared to its adjacent pixels.

Description of the reference numerals

[0088] 401 Active sensor layer 402 Collection electrode 403 Inactive layer

Claims

1. An apparatus for detecting a chrysanthemum diffraction pattern, comprising: An electron column adapted to provide an electron beam having an energy in the range from 2 keV to 50 keV and directed towards a sample when in use; An imaging detector for sensing and counting electrons from the sample resulting from the interaction of the electron beam with the sample, the imaging detector comprising an array of pixels and having a counting rate function of at least 2,000 electrons per second for each pixel; and the imaging detector is adapted to provide electron energy filtering of the sensed electrons for counting the sensed electrons representing the diffraction pattern; the imaging detector has an inactive layer on the surface where the electrons enter towards the active region of the detector, the inactive layer dispersing the detection energy of 20 keV incident electrons with an energy spread having a full width at half maximum lower than 3.2 keV. The apparatus is characterized by this.

2. The apparatus according to claim 1, wherein the imaging detector comprises an electron amplifier for each pixel that introduces an equivalent electron noise energy having a full width at half maximum lower than 2 keV, preferably lower than 1 keV.

3. The apparatus according to claim 1 or 2, wherein the imaging detector contains a circuit for detecting and correcting charge sharing between pixels that can occur for a single incident particle.

4. The circuit adds the electron signal collected at a given pixel to the electron signal collected at an adjacent pixel, applies electron energy filtering to the added electron signals for counting the sensed particles representing the diffraction pattern, and assigns the counted particles to a single pixel, to achieve the above. The apparatus according to claim 3.

5. The imaging detector outputs both the arrival time and magnitude of the signal captured by each pixel, and a computer algorithm identifies instances where a single incident particle generates simultaneous electron signals in multiple pixels, adds the multiple electron signals generated by a single incident particle and collected by the multiple pixels, applies energy filtering to the added electron signals to count the sensed particles representing the diffraction pattern, and assigns the counted particles to a single pixel, for use in the device according to any one of claims 1 to 4. **Claim 6** The device according to any one of claims 1 to 5, wherein the ratio of (active layer sensor thickness) / (pixel-to-pixel spacing) is less than 5. **Claim 7** The device according to any one of claims 1 to 6, wherein the number of electrons counted per pixel during pattern acquisition is read out as a data unit of 6 bits or less. **Claim 8** The device according to any one of claims 1 to 7, wherein the camera sensor array has a configurable pixel amplifier that achieves more than one number of pulse lengths to suit different pixel counting speed requirements and energy resolution requirements. **Claim 9** The device according to any one of claims 1 to 8, wherein the electron energy filtering is configured to distinguish between sensed particles having energy representing the diffraction pattern and sensed particles having energy representing the background. **Claim 10** The device according to any one of claims 1 to 9, wherein the incident electron beam is incident at an angle within the range of 45 - 90° with respect to the sample surface plane. **Claim 11** The device according to any one of claims 1 to 10, wherein the inert layer disperses the detection energy of 20 keV incident electrons by less than the spread of the energy induced by transmission through 1500 nm of inert silicon. Claim 12 A method for detecting a Kikuchi diffraction pattern, comprising: providing, using an electron column, an electron beam having an energy in the range from 2 keV to 50 keV and directed towards a sample; detecting and counting electrons resulting from the interaction of the electron beam with the sample using an imaging detector having an array of pixels and having a counting rate function of at least 2,000 electrons per second per pixel; comprising; the detector being adapted to provide electron energy filtering of the detected electrons for counting the detected electrons representing the diffraction pattern; the imaging detector having an inactive layer on a surface where the electrons enter towards an active region of the detector, the inactive layer dispersing the detection energy of 20 keV incident electrons with an energy spread having a full width at half maximum lower than 3.2 keV; characterized in that. Claim 13 A method for detecting a Kikuchi diffraction pattern using the apparatus according to any one of claims 1 to 11.

Images