I Introduction
With the advent of commercially available electron detectors boasting high pixel resolution designed for transmission electron microscopes, the scientific community is increasingly leaning towards utilizing these cutting-edge cameras for ultra-high vacuum (UHV) experiments, as highlighted by recent studies (Niu et al., 2023; Janoschka et al., 2021). These detectors, characterized as monolithic fiber-optically coupled CMOS-based systems, integrate seamlessly with a thin polycrystalline phosphor scintillator that serves as the electron-sensitive component. When high-energy electrons collide with the detector, they trigger a cascade of secondary electrons within an aluminum layer, leading to the emission of a multitude of photons in the scintillator. This intricate mechanism allows for the precise attribution of individual events captured by the CMOS chip to specific incoming electrons, as detailed in previous research (Stumpf et al., 2010).
In this report, we detail the significant upgrade of our ultrafast reflection high-energy electron diffraction experiment (URHEED) (Janzen et al., 2006) from a microchannel plate (MCP) detector (Burle Chevron 3040FM) to a superior fiber-optically coupled CMOS-based detector (TVIPS TemCam-XF416) (Tietz et al., 2022).
The detective quantum efficiency of the TemCam is remarkably high, rated at DQE(0) = 0.62, complemented by a filling factor of 0.72 (Shi et al., 2016). In contrast, the MCP detector exhibits a lower DQE due to its channel structure covering only 55% of the MCP surface, which inherently limits its efficiency in generating electron cascades at higher electron energies. This enhanced DQE of the TemCam renders it a more advantageous tool for capturing the rare electrons crucial for ultrafast diffraction experiments.
II Detector Integration and Handling
While the TemCam is engineered to withstand UHV conditions, achieving these conditions proves challenging because the detector must remain below 40°C to prevent irreparable damage. This goal complicates the essential prolonged bake-out process, as this typically requires heating the entire chamber above 100°C, which is incompatible with the thermal limitations of the camera. A potential workaround involves dismounting the detector before the bake-out; however, this introduces significant risks including physical damage during reassembly and exposure to ambient moisture upon reinstallation. Furthermore, the thermal cycles associated with dismounting and remounting exert thermal stress on the fragile fiber-CMOS unit, which ideally should be minimized. To address these challenges, we designed a self-built water-cooled collar that effectively keeps the detector’s housing cool, thereby enabling the bake-out of the entire chamber without risking the integrity of the TemCam.
In our experimental setup depicted in FIG. 1, the detector interfaces with the main chamber (MC) through a gate valve (VAT CF-F 100 CF-F 160). An added configuration comprises an additional CF-F 40 flange affixed to the detector side, equipped with an angle valve (VAT 28.4 CF-F 40). This assembly connects to a turbo molecular pump (Pfeiffer HiPace 80) situated in the load lock chamber (LL), ensuring a base pressure better than 3 × 10−8 mbar. The innovative design permits pumping through the MC or connecting to the LL without any operational interference.
In the event of venting the MC, the gate valve closes while maintaining the detector at a low temperature via its internal Peltier cooler, thus averting exposure to ambient conditions. During the critical bake-out phase, the gate valve is heated to secure UHV conditions, while the detector, positioned outside the bake-out oven, remains adequately protected from thermal radiation or convection. To further mitigate heating issues through thermal conductivity from the heated gate valve, we have incorporated a specialized water-cooled collar around the detector housing, which acts as a reliable heat sink.
This cooling collar is constructed from two solid stainless steel half-rings, each 30 mm wide and 10 mm thick. Connected by a stainless steel capillary, optimized for maximum thermal conductivity, this dual-ring structure is linked via PVC fiber-reinforced hoses into a single cooling water loop, maintained at 15°C and flowing at 1.5 l/min.
The inner contact area of each half-ring is lined with heat-conducting copper tape to maximize thermal contact area with the detector housing. Fastened together by two M4 screws, tightened at 4 Nm torque, the rings ensure a robust connection to the detector housing. Temperature monitoring during bake-out was achieved utilizing K-type thermocouples positioned at multiple points on the MC, as illustrated in FIG. 2. During this process, the temperature of the detector housing rises minimally to just 1 Kelvin, from 21°C to 22°C, while the gate valve temperatures ascend to a far higher 130°C, further emphasizing the efficiency of our cooling system.
The baking process proceeded for five days, followed by shutting off the heating mechanisms for the lower MC while the upper MC was switched off twelve hours later. Consequently, chamber pressures achieved an impressive decrease from 6 × 10−8 mbar to below 2 × 10−10 mbar, as highlighted in FIG. 2.
To eliminate noise and artifacts originating from the fiber bundle coupling at the vacuum junction and the CMOS sensor itself, both dark and flat-field correction techniques are employed. Specifically, dark images are captured under various exposure times to average out background noise, while flat images necessitate uniform illumination of the entire detector area for accurate correction. In conventional setups, such as TEM or LEEM, strong defocusing conditions via the electron source facilitate this uniform illumination (Janoschka et al., 2021). However, our URHEED setup, utilizing a single magnetic lens designed for small electron focal points, posed challenges for achieving such strong defocusing.
To resolve this, we installed an auxiliary electron disc emitter (Kimball Physics ES-535), which can be positioned at the sample location utilizing a linear translation feedthrough (VAb LDK40-150). This emitter operates at a current of 2.96 A, biased at a high negative voltage of 11 kV, adjustable up to 20 kV, effectively broadcasting high-energy electrons towards the detector. Given the considerable currents emitted from such devices, we implement safeguards such as lead glass covers on chamber windows to monitor and mitigate X-ray emissions.
The employed cathode features an activation layer designed to decrease its work function, leading to a purely divergent electron beam that results in a diffuse, magnified depiction of the cathode surface. Due to the granular nature of the activation layer, the beam exhibits an inhomogeneous emission profile, necessitating additional measures for improved flat image quality, such as deflection through alternating current (AC) magnetic fields. We arranged two pairs of parallel coils external to the chamber for horizontal and vertical beam deflection, provided with AC current from a function generator (Joy-IT JDS6600), amplified through a reference audio amplifier (Behringer A800). All relevant parameters for this deflection system are summarized in TABLE 1. By
manipulating exposure times variably between 100 ms and 1000 ms, we procured flat images with divergent mean intensities, notable for a maximum intensity inhomogeneity of just ±10%, which was deemed sufficient for effective imaging artifact correction.
III Detector Performance
A significant and valuable advantage of employing a fiber-coupled CMOS detector in diffraction experimentation lies in its remarkable insensitivity to overexposure from electrons. Unlike MCP systems, which suffer catastrophic damage upon overexposure, the TemCam exhibits resilience, allowing for the characterization of direct beams without obliging reductions in intensity that could adversely affect beam focus. Consequently, by utilizing a deflector within the electron gun, the beam can be directed onto the TemCam while simultaneously adjusting the integration time to prevent overexposure artifact occurrence.
To further assess the TemCam’s performance, we utilized the diffraction pattern of the Si(111)-(7×7) reconstruction as a benchmark, as seen in FIG. 3(a). The exposure for this image was 10 seconds, with appropriate dark and flat field corrections applied, achieved with a 500 μm electron spot at 20 keV, under an incidence angle of 1.74° and a sample temperature of 145 K.
In comparison, FIG. 3(b) presents an image of the same diffraction pattern captured using the MCP detector unit (Hafke et al., 2019). This image resulted from the summation of 92 individual exposures, each lasting 2 seconds, taken with a smaller 310 μm electron spot at 30 keV, at an incidence angle of 1.5° and a sample temperature of 80 K. Both diffraction patterns, normalized to their maximum intensity, facilitate direct comparison since they share the same logarithmic gray scale representation.
The superiority of the TemCam becomes evident through its 1.79× larger detection area, allowing it to cover a more extensive portion of the diffraction pattern and revealing additional diffraction spots that were not captured by the MCP.
The TemCam’s higher pixel density translates each diffraction spot into a representation across a greater number of pixels. This contrasts starkly with the MCP, where each pixel correlates to 2844 μm² compared to just 240 μm² per pixel in the TemCam. The twelve-fold resolution improvement is visually striking in FIG. 4. The (1/7 1/7¯17¯) spot is depicted for each detector system, illustrating the newfound capabilities of the TemCam in resolving greater detail.
The analysis reveals that the spot captured by the TemCam displays an elongated elliptical shape that the MCP could not resolve, attributed to the inherent pixel limitations and blooming effects characteristic of MCP technology (Richter and Ho, 1986). Line profiles taken in both radial and polar directions enable thorough evaluation of resolution in reciprocal space, showcasing the ability of the TemCam to depict a Gaussian profile with a FWHM of (0.266 ± 0.003)%, while the MCP detector profiles follow a Voigt distribution with a FWHM of (0.372 ± 0.006)%. These figures reflect the significant detractors, notably the blooming effect and reduced pixel counts present within the MCP system.
In terms of coherence lengths, the TemCam reveals a radial coherence length of ξr,TemCam = (125 ± 1) nm, in sharp contrast to the MCP’s coherence length of ξr,MCP = (89 ± 1) nm, clearly demonstrating the advantages of utilizing the TemCam.
The intensity profiles across a series of (7×7) spots further clarify the differences between the two systems. The varying intensities relate to differing scattering conditions, such as variations in angle of incidence and electron energy, which contribute to achieving a signal-to-background ratio of 530:1 for the TemCam versus 12:1 for the MCP detector.
Scrutiny of the spots along the 1/7 Laue ring under high magnification reveals intriguing findings. The (7×7) spots display a significant width in the polar direction compared to that observed in the radial direction. Ideally, one would expect a circular profile due to the diffraction from an immaculate surface, particularly for the (00)-spot in RHEED as the sample inherently functions as a reflective surface.
However, the inclusion of surface imperfections like steps and domain boundaries induces additional broadening effects (Horn-von Hoegen, 1999; Klein et al., 2011). The counterintuitive observation of greater width in the polar direction, juxtaposed with the anticipated broader radial characteristics, prompts further investigation.
The direct electron beam profile, shown in FIG. 5(b) and (c), corroborates the anticipated Gaussian-shaped circular profile aligningwith expected width parameters in the polar direction. The width arises from contributing factors such as the size of the electron source at the photocathode, the magnification involved in imaging to the detector, and the resulting spherical aberrations present within the magnetic lens (Hafke et al., 2019).
We correlate the narrower spot width in the radial direction with the suppression of off-axis beams inclined for significant spherical aberrations during image construction.
With the grazing incidence of 1° to 6° ensuring surface sensitivity, this means the incident electron beam travels at a grazing angle (αg) of 1.74° with implications for beam size on the sample surface. This results in a foreshortening factor of approximately 33, illustrating that a finite beam diameter of approximately 300 μm appears elongated on the sample, providing further insights into the effective performance of the detector in the context of diffraction analysis.
IV Conclusions
We have demonstrated that through the innovative use of a water-cooled collar and a differential pumpable gate valve, achieving successful bake-out conditions for temperature-sensitive devices like the TemCam is feasible without the need to disassemble the camera.
The inherent dead time of 1.2 seconds presents a limitation when utilizing the TemCam at full resolution; nevertheless, the detector operates efficiently at 10 Hz in continuous rolling shutter mode, maintaining responsiveness during adjustment and optimization phases. Additionally, the TemCam’s superior dynamic range permits effective simultaneous measurements of both weak and strong diffraction signals, significantly reducing overall measurement durations.
Considering its scintillator/CMOS chip design, the TemCam shows inherent sensitivity to photons exceeding the indirect band gap of silicon. This observation necessitates cautious deliberation when leveraging the TemCam for experiments infused with optical excitation. Employing wavelengths around 800 nm from a Ti:sapphire laser calls for effective shielding strategies against potential interference. Alternatively, utilizing photons with energies corresponding to hνEgap=1.12 eV can serve as viable pump pulses.
Ultimately, the TemCam’s robustness against damage stemming from overexposure in scenarios with intense diffraction spots or direct electron beams stands out as a key advantage, adeptly preventing the imprint of unique damage patterns upon the resultant diffraction images.
Acknowledgments
We deeply appreciate the contributions from H. R. Tietz and M. Oster of TVIPS, as well as R. Ernstdorfer, P. Baum, G. Sciaini, P. Dreher, A. Neuhaus, and F.-J. Meyer zu Heringdorf for their invaluable discussions. This work has been generously supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) through project C03 of the Collaborative Research Center SFB1242 “Nonequilibrium dynamics of condensed matter in the time domain” (Project-ID 278162697).
Data Availability
The data supporting the findings of this study are available from the corresponding author upon reasonable request.
How does the TemCam enhance the quality and uniformity of imaging in precision electron diffraction experiments?
And provides significant advantages over traditional MCP systems. The enhanced sensitivity to overexposure allows for more robust imaging without compromising the intensity or quality of the captured data. The larger detection area, higher pixel density, and superior resolution of the TemCam have proven critical in exploring intricate diffraction patterns, as evidenced by the detailed imaging of the Si(111)-(7×7) reconstruction.
The implementation of flat-field correction techniques and the auxiliary electron disc emitter has further enhanced the uniformity and quality of our imaging, allowing for clear and artifact-free results. Additionally, the measurements obtained regarding coherence lengths and intensity signal-to-background ratios underpin the TemCam’s advantages in precision electron diffraction settings.
our findings reaffirm the benefits of upgrading to a fiber-coupled CMOS detector like the TemCam in diffraction experiments, marking a significant step forward in our capabilities to analyze and interpret electron diffraction patterns. The continued refinement of imaging systems will undoubtedly contribute to advancements in the field of surface science and materials characterization. Future research should focus on optimizing the integration time and further quantifying the performance advantages while exploring other potential applications of the TemCam in various experimental setups.
The promising performance metrics of the TemCam in terms of resolution, coherence lengths, and signal quality set a new standard for electron diffraction techniques, and we anticipate that this detector will play a pivotal role in future investigations within the scientific community.
