100 keV vacuum sealed field emission gun for high resolution electron microscopy

The use of high brightness electron sources such as cold field electron emitters (CFE) and the thermally assisted field emission cathodes (TFE) has allowed lower beam voltages to be used in both scanning electron microscopy (SEM) as well as in transmission electron microscopy (TEM) with optimum (high) resolution. In the TEM, the perceived optimum operational energy range for high spatial resolution, particularly for material science applications, is now in the range of 200–300 keV. However, this energy regime is not suitable to use across all disciplines. For example, the optimum electron beam energy in electron cryo-microscopy (CryoEM) has been a subject of debate for some time. Recent extensive studies have found that an electron beam energy of 100 keV is preferable to higher beam energies in terms of the useful information obtainable from single-particle biological specimens for the same amount of radiation damage.^1,2 At 100 keV, this is of the order of 25% more than at 300 keV. This electron beam voltage has also been called for in electron beam lithography and some materials’ science applications where knock-on damage is less at lower electron energies.^3,4 Egerton⁴ has extensively reviewed the suitability of the incident electron beam energy in transmission electron microscopy depending on the type of sample to be imaged and its thickness. He found that the electron beam energy for TEM covers a range spanning 50–300 keV. In spite of such different electron beam energy requirements, most FEG based TEM instruments available in the market continue to be designed and hence optimized in terms of the highest spatial resolution for 200–300 kV operation. Of course, such instruments are always possible to operate at lower than their maximum voltages to suit the specimen being studied but such lower beam voltages may not represent the optimum electron optical characteristics of the instrument in question and certainly may not be as cost-effective. Such a difference in price would allow a larger number of potential users, particularly in the cryoEM field, from acquiring a FEG operated TEM. The results reported here further demonstrate the possibility of upgrading a TEM that has a thermionic source that is inadequate to use for the type of samples imaged in to a FEG featuring higher brightness and spatial coherence; both of which are essential in this application.

The use of high voltage (>100 kV) electron guns, however, also comes with its own set of construction challenges. It is, for example, customary to enclose the FEG chamber including the source's electrical vacuum feedthrough and source control electronics in SF₆ gas in order to reduce the occurrence of corona discharges. While this is now a standard practice for such instruments, the use of SF₆ gas has environmental costs and are subject to restrictive health and safety regulations. Furthermore, there are a number of mechanical constraints such a configuration imposes on the design of the instrument. In addition, high stability and low noise ripple for high voltage power supplies are increasingly challenging to design and build at higher voltages.

The use of the ZrO/W(100) electron cathode heated between 1750 and 1800 K has been referred to in the literature as both a thermally assisted field emission cathode (TFE) and a Schottky cathode. In the latter case, two operational regimes have been identified, the Schottky and the extended Schottky emission, where the extended Schottky reflects the increased contribution of tunneled electron emission at higher electric fields. Further increase of the applied electric field, however, causes in turn a higher proportion of the emitted electrons to be due to electron tunneling through the thinned potential barrier, which is referred to as thermally assisted field emission.^5–7 This is the cathode's operational condition in the present development and hence the abbreviation TFE is used. Operating a ZrO/W(100) emitter with an emitter radius between 0.4 and 0.6 μm and an angular current intensity of at least 300 μmA/str, i.e., in the TFE regime, is necessary to exploit the source's high brightness and spatial coherence, essential in cryoEM, in particular, but also for high spatial resolution TEM and SEM in general. It is because of the field emission aspect of operating such cathodes that the community, academic and industry, have by and large, referred to the electron guns using either cathode: CFE, the ZrO/W(100) in the extended Schottky-type and the TFE, as a field emission gun, normally abbreviated as FEG.

We report on the design and construction of a dedicated standalone FEG operational up to 100 kV to address such needs for the TEM, particularly for the cryoEM sector, as well as other novel laboratory-based applications. The constructed FEG, which uses a TFE cathode, is electron optically aligned during construction and assembly, thus avoiding the need for source alignment during installation and operation. The FEG is baked-out to 180 °C to achieve ultrahigh vacuum (UHV) and is also high voltage conditioned to 100 kV prior to installation on an electron optical column. The unit is then vacuum sealed and independently pumped with a nonevaporable getter pump (NEG) to allow transport under UHV. The prebaking and high voltage conditioning reduce the installation time and complexity, which in turn reduces the downtime of the instrument in question to generally no more than a couple of days. A specially designed high voltage plug and vacuum feedthrough have also been built together with a high stability, low ripple high voltage power supply. Furthermore, an advantageous feature of this development is that the FEG, feedthrough, and the high voltage plug assembly do not require the use of SF₆ gas, as is customarily the practice for high voltage FEGs used on TEMs. This allows an easier “hot swap” transition for emitter exchange, thus simplifying its installation, use, and maintenance and again reduces down time on the instrument.

The construction of this FEG with its standalone power supply unit makes it possible to exchange the electron gun on an instrument operating a thermionic source with a FEG. We estimate there are around 7000 100 keV TEMs employing a thermionic source that are currently operating around the world. In principle, it would be possible to install the 100 keV FEG, we describe here on any of these instruments through a retro-fit. In general, however, this FEG is also equally suitable and adaptable to use as a test-bed for current trends in electron microscopy, such as time-resolved studies, as well as for general research and development investigations.

The electron optical simulation program used is a home developed package that aids in optimizing the shape of electrostatic lenses for the minimization of their aberration coefficients in a given configuration. It is based on the finite element method for the design of electron lenses described by Munro.⁸ The program has been thoroughly checked against several published electrostatic lenses.^9–12 The developed FEG, which is schematically depicted in Fig. 1, comprising a TFE cathode used in the standard configuration with suppressor and extractor electrodes. The emitter protrudes 0.25 mm from the suppressor's electrode aperture and is 0.5 mm away from the extractor. The FEG is a three-element asymmetric electrostatic lens,^9,10 which can operate with a beam energy in the range of 20–100 keV. The upper limit of 100 kV is set by the high voltage stand-off of our design. Emphasis in modeling the present design has been to minimize the chromatic aberration contribution since the energy spread of the emitted electrons of the TFE cathode is about 0.7 eV when operated at angular current densities of the order of 300 μA/str but is larger for higher angular current density values. The space between the various electrodes is typically divided into 150 meshes each to achieve convergent chromatic and spherical aberration coefficients to within 2% or so. A full electron optical analysis of the FEG together with its combination with a magnetic column as used in TEMs will be given in Ref. 13.

The dimensions of the lens electrodes and their separations are depicted in Fig. 1. Two possible working distances measured from where the electron beam exits the FEG to form a focused spot have been simulated. The first of these is for experimental measurements of the electron optical properties of the FEG in a high vacuum test chamber. In this configuration, the working distance, where a Faraday cup to measure the electron beam current is located, is approximately 206 mm from the exit plane, located at the FEG's ground anode as shown in Fig. 1. The second geometry is to focus the primary electrons at 50 mm from the exit plane of the FEG. This latter configuration represents the geometrical arrangement of the positions of the original electron source in the transmission electron microscopes used in the present study: an FEI (now Thermo Fisher) Tecnai T12 and a JEOL JEM-1400HR. Both these TEMs are of the thermionic electron source types.

The relationship between the chromatic and spherical aberrations of the present configuration for the two working distances considered here as a function of the beam voltage is shown in Fig. 2, while Fig. 3 depicts the magnification as a function of beam voltage as well as a comparison of the simulated and measured lens voltages at various electron beam voltages.

The FEG vacuum chamber occupies a physical envelope of approximately 210 mm diameter and about 315 mm height (in addition to the high voltage plug and cable). This vacuum chamber maintains a base pressure of <2 × 10⁻¹⁰ mbar after baking the system to 180 °C for a period of 36 h using a 200 l/s nonevaporable getter (NEG) pump. The chamber can be valved-off to atmosphere using a manual isolation valve on the FEG's optical axis. This allows the FEG chamber to independently maintain UHV either during transport to facilitate a fast exchange of a used emitter or during maintenance of the lower electron column, thus reducing the likelihood of the instrument being off-line for long periods.

During operation of the FEG on the instrument, the built-in manual isolation valve is open. To maintain a safe operating pressure of <1 × 10⁻⁹ mbar for the TFE in the FEG chamber, a differential aperture in the FEG vacuum chamber allows a pressure differential of >10⁻² mbar to be achieved between the FEG and the lower interface vacuum chamber (i.e., an intermediate vacuum chamber that sits between the FEG chamber and the electron optical column).

The FEG is first operated in the electron optical test chamber, which consists of a high vacuum vessel that is operated at a base pressure in the order of 10⁻⁸ mbar. This configuration mimics that of operating the FEG on a TEM or a similar electron optical instrument. The electron source used is operated at 250–300 μA/str with an extraction voltage in the range of 4000–5000 V and a suppressor voltage of −300 V with respect to the filament.⁵ 

A further requirement in high resolution field emission electron optical systems is that all the electrodes must have stable voltage and current supplies. To drive the developed FEG, a standalone high stability 100 kV power supply unit (PSU) together with a plug and cable and a vacuum feedthrough have also been developed. The PSU outputs for the operation of the FEG electrodes are floating on the beam potential of up to 100 kV with voltage ripples of less than 1 ppm.

The developed FEG is designed to operate with a beam energy in the range of 20–100 keV. It can independently focus the beam into the lower electron optical column with a range of beam currents from 15 nA to greater than 200 nA at an angular current intensity in the range of 300 μA/str by using an appropriate aperture size at the extractor electrode to suit the intended use of the FEG.

In the electron optical test chamber, the focal point of the FEG, where a Faraday cup is mounted, is located at a working distance of 206 mm. For a TFE cathode operated with an angular current intensity of 300 μA/str, the estimated electron beam current, which has also been confirmed by measurement, is of the order of 15 nA. This is a typical configuration, which is intended for use in a TEM. The electron beam is focused into the Faraday cup at the various electron beam energies considered here by varying the lens electrode voltage. To facilitate the electron beam focussing, a standard copper electron microscope grid with bars measuring 24 μm and a pitch of 85 μm covers the Faraday cup which has an aperture of 1 mm diameter with an aspect ratio of 10. The beam current measurements are then carried out by placing the beam in the area between the grid bars.

A reasonably good agreement between the experimentally measured voltages and the simulations at this working distance from the exit plane of the FEG is obtained for beam energies in the range of 20–80 keV (see Fig. 3), which confirms the simulation results. In addition, the focussing lens electrode voltage simulated here is also compared with the FEG structure used by Gesley¹⁴ for an emitter voltage of 4 kV and a working distance of about 200 mm, which is similar to what is used here. Agreement of the lens voltages to within 25% is obtained, which is perhaps acceptable taking into account the difference in the electrode shapes and interelectrode dimensions of the two structures. The FEG design by Gesley is typical of electrostatic lenses used in high voltage FEG applications, particularly for EBL where the emphasis is on the size of the focussed beam which is used for imaging. In TEMs, however, the emphasis is on the source's spatial coherence and brightness, which are superior in FEGs than for thermionic sources. Nevertheless, the FEG of Gesley and the one presented here are similar in both being a three-element electrostatic configuration lenses. A comparison of the spherical and chromatic aberrations of both lenses at the image plane (145 mm reported by Gesley which corresponds to 50 mm for the present FEG) is as follows: C_co = 60 mm and C_so = 607 mm for the present FEG, while for Gesley, these are C_co = 59 and C_so = 151, respectively. These values are typical of electrostatic lenses in general and in the present example compare well, particularly for the chromatic aberration, which are the more significant in FEGs and taking into consideration the optimization of these lenses for different applications. However, it should be emphasized here, as the results on the two different columns used in this study show, the FEGs aberration contribution are less important in TEM applications as these are significantly demagnified by the magnetic column used. For TEM imaging, it is primarily the source's properties (brightness and spatial coherence) and the columns’ objective lens' aberration values that are often the limiting factor. This can be seen in the obtained spatial resolution shown in Figs. 4 and 5 for the same FEG operated on two different columns with different objective lens aberrations where a higher resolving power is obtained on the column with smaller objective lens aberration.

The FEG was transported under vacuum and independently installed on two separate TEMs: a JEOL JEM-1400HR TEM and a Philips (now Thermo Fisher) Tecnai T12 TEM, whereby the original thermionic hairpin electron source, the electron gun chamber, and the EHT unit were replaced for both TEMs with the FEG and its associated parts. In both cases, the sealed FEG chamber was installed on the TEM columns and base pressures in the range of low 10⁻¹⁰ mbar or better were obtained. To maintain the UHV base pressure in the FEG chamber during operation, an intermediate vacuum chamber evacuated with a 25 l/s IGP to achieve a base pressure in the range of 10⁻⁸ mbar was added to interface between the FEG chamber and the microscope column. This chamber also houses a pneumatic gate valve for the protection of the FEG against unexpected venting of the column or specimen chamber during use. It further houses a set of quadrupoles that are used to align the incident electron beam on the optical axis of the TEM's column. In both TEMs, the quadrupoles used were those of the original microscope.

Preliminary results show that an electron beam current stability of <1% drift over 24 h and <0.5% over 1 min are achieved and operation in the voltage range of 30–100 kV has been demonstrated. Gold particles on amorphous carbon have been used as test samples for the microscope's resolving power. The results obtained are shown in Figs. 4 and 5, which demonstrate that the gold (111) and (200) crystal planes at 0.235 and 0.202 nm, respectively, are clearly resolved at an electron beam energy of 90 keV. Furthermore, the detection of only the gold (111) crystal planes in Fig. 4 reflects the poorer environmental conditions for the Tecnai column installation at our laboratory as well as the higher resolving power of the JEM-1400HR, which is equipped with an objective lens having C_s of only 1.8 mm compared to that of the Tecnai, which is about 2.2 mm for TWIN objective.² It is important to note, however, that while graphitized carbon was easily resolved using the original thermionic source in both TEMs used here, the same was not possible for the gold nanoparticles. The ability to resolve the gold lattice planes clearly demonstrates the higher brightness and coherence of the FEG compared to the thermionic source.

An estimate of the source brightness (B) of the present FEG was made relative to that of the thermionic source of the TEMs used in this study. This was performed by illuminating the same column from the same source plane for both sources at a fixed electron beam energy, C1 setting, and C2 aperture size. The smallest probe diameter was then formed at the eucentric focus with the solid angle (α) of the probe formed at the eucentric focus being equal in the two cases due to the column optical geometry. The minimum probe diameter and total beam current were measured for each source to obtain the current density (J) into the solid angle, allowing the relative brightness of the FEG with respect to the thermal source to be calculated from Eq. (2),

The results obtained show that at an angular current intensity of 250–300 μA/str, the FEGs brightness is of the order of 150–200× greater relative to the tungsten hairpin source of these TEMs. This is in general agreement with those reported in the literature.¹⁵ 

A standalone 100 kV FEG has been designed and constructed. Electron optical simulations of the focusing voltages of the lens are compared with experimental data and show good agreement over a wide range of beam voltages. The FEG was attached to two different TEMs: an FEI (now Thermo Fisher) Tecnai T12 and a JEOL JEM-1400HR. In each case, the resolution of the microscope was improved such that the atomic lattice of gold nanoparticles on an amorphous carbon foil is resolved. Gold (111) and (200) lattice planes at room-temperature showing 0.235 and 0.202 nm resolution in images have been obtained. Such lattice planes are not resolvable by TEMs operating thermionic sources. Furthermore, the results obtained here show that for modest FEG aberration coefficients, it is the TEM's objective lens aberration coefficients which are the most detrimental factor regarding the overall microscope's resolution. This is demonstrated with the present FEG by resolving the gold (200) lattice planes on the JEM 1400 which has a smaller C_s than that of the Tecnai.

The present compact design lends itself to integration onto existing high voltage electron columns, such as upgrading thermionically operated TEMs or electron beam lithography (EBL) systems into FEG operation. A further advantage of the present design is the avoidance of using harmful gases such as SF₆ at the high voltage plug and feedthrough area, thus simplifying maintenance and operation of the instruments. This compact FEG design also makes it suitable for use in many R&D applications and as a platform for further development.

The present development has been partly funded with a generous grant from Innovate UK Grant No. 103806, the Medical Research Council under Grant Nos. MC_U105184322 and MC_UP_120117; Wellcome Trust Grant Nos. 220526/B/20/Z and EPSRC R122522. The authors would like to thank S. Bean for useful discussion on the electron optical modeling.

The data that support the findings of this study are available from the corresponding author upon reasonable request.