Source Brightness at 20–80 kV
May 21, 2018 – Researchers at Ulm University have shown a simple way to measure the brightness of an electron source and the spread of illumination angles in a transmission electron microscope: Young’s fringes tests at stepped defocus. On the SALVE microscope at 20 and 80 kV they found a reduced axial brightness of 1.8 × 108 A/(m2 sr V) for its X-FEG source with the monochromator switched off, more than twice the manufacturer’s specification. Since this value stays constant, the illumination angle for any setting can now be estimated from the dose rate.
Every image in an electron microscope is a compromise between two wishes: as many electrons as possible, and electrons that all travel in parallel. How far both can be had at once is set by one property of the electron source, its brightness. It decides how much current fits into a fine probe in scanning mode and how much signal a parallel beam delivers in conventional TEM. Yet reliable and simple ways to measure it are hard to find, and for the spread of illumination angles that goes with it, the authors note, the assessment “cannot be found in textbooks.” Felix Börrnert, Julian Renner and Ute Kaiser at Ulm University have now shown a simple route to both quantities: a series of Young’s fringes tests at stepped defocus, in which the fringe pattern shrinks as the defocus grows (Fig. 1). On the SALVE microscope, corrected for chromatic and spherical aberration, they measured at 20 and 80 kV a reduced axial brightness of 1.8 × 108 A/(m2 sr V) for its FEI X-FEG source, more than twice the value the manufacturer specifies. The monochromator built into the source was switched off.
Why brightness matters
In scanning transmission electron microscopy (STEM), the brightness sets the current in a probe of given size. A probe could in principle be made arbitrarily small, but then no signal would be left. Aberration correction in STEM did not so much improve the ultimate resolution as the usable signal at a given resolution[1]. In conventional TEM, too, the signal-to-noise ratio, and in the end the resolution, depends on the total dose in the image. Since drift, vibration and beam damage limit the exposure time, a high dose rate is needed, and the dose rate a source can deliver at a given illumination angle is again a matter of brightness.
The angles themselves matter as well. An ideal TEM illumination would be a plane wave, all electrons parallel. A real source has a finite size, and its electrons arrive at the sample from a small spread of directions, the illumination semi-angle distribution. A tilted electron passes the back focal plane of the objective lens at a shifted position and so samples the phase errors of the lens at another spot; the image points blur. This effect, the illumination damping, dominates the information transfer in microscopes without aberration correction[2]. In off-axis electron holography, the same spread governs the contrast of the interference fringes and with it how much information a hologram holds[3],[4].
The spread could be made arbitrarily small by demagnifying the source, but then hardly any current would be left for imaging. Beyond the emission process itself, effects such as the Löffler effect, often wrongly called the lateral Boersch effect[5], and in the end the fact that electrons are fermions, put a fundamental limit on how bright a source can be. Pioneering methods to determine the illumination angle date from the 1970s and 1980s[6],[7], but in most cases it is still derived from the brightness given for the source and the dose in an image. Measuring the smallest size of diffraction spots[8] overestimates the spread of modern high-brightness sources, because it is so small. Moreover, the spread changes with magnification and illumination density, so it has to be known for the very conditions of an image.
Fringes at stepped focus
The Ulm method uses an established test. For a Young’s fringes test, two micrographs of a thin amorphous sample, shifted slightly against each other, are added and Fourier transformed. The shift imprints a pattern of parallel fringes on the diffractogram[9], and how far out in spatial frequency the fringes reach marks the information limit, the finest detail the microscope still transfers. The team recorded such tests on thin amorphous tungsten at many defocus values, each in two orthogonal shift directions to reveal any anisotropy. With growing defocus the illumination damping grows, and the fringe disk shrinks visibly (Fig. 1).
The illumination semi-angle follows from fitting the information limit against defocus. In an aberration-corrected microscope this works only if all damping mechanisms enter the fit: the illumination damping, the focus spread from the energy spread of the electrons and, in a chromatically corrected instrument, the image spread caused by magnetic noise[10], as well as the spherical aberrations of third and fifth order. Then only two parameters remain free, the semi-angle distribution and a factor for the signal-to-noise ratio. The box below shows how the damping functions combine in SALVE and why a single defocus value is not enough (Figs. 2, 3 and 4).
One source, three settings
The measurements ran on the SALVE instrument, a dedicated low-voltage microscope for 20 to 80 kV based on an FEI Titan Themis3 (today Thermo Fisher Scientific) with the CEOS SALVE corrector, which corrects axial geometric aberrations up to fifth order, off-axial ones up to third order and the first-order chromatic aberration[12] (see SALVE Cc Correction Realized[13]). The corrector also made the experiment convenient: it measures the defocus automatically and shifts the image optically for the fringe test. The source is an FEI X-FEG, a Schottky emitter with a Wien-filter monochromator, which stayed switched off. For high dose rates the beam was condensed until the illuminated spot was only slightly larger than the camera area, and the images were exposed for 2 s. For low dose rates the monochromator focus was defocused until the dose rate in the image dropped to about 3000 e−/(nm2 s), a value typical for imaging biological samples, with 10 s exposures. The magnification was chosen so that the camera oversampled the finest information at least four times.
The fits gave a semi-angle distribution of 70 µrad at 20 kV and 110 µrad at 80 kV for high dose rates, and 7 µrad at 20 kV for the low dose rate (Fig. 5), with an error of at most about 10 %. For high dose rates the conventional noise level fitted well; for the low dose rate the signal-to-noise factor came out at 0.475. Nonlinear transfer, known to falsify Young’s fringes tests[14], shifts the measured points up or down but hardly changes the shape of the curve, and the shape is what fixes the semi-angle.
| Quantity | 20 kV, high dose rate | 20 kV, low dose rate | 80 kV, high dose rate |
|---|---|---|---|
| Mean signal (counts per pixel) | 13,500 | 660 | 11,000 |
| Exposure time | 2 s | 10 s | 2 s |
| Camera conversion (counts per electron) | 12 | 12 | 15 |
| Pixel size | (40 pm)2 | (40 pm)2 | (10 pm)2 |
| Illumination semi-angle distribution | 70 µrad | 7 µrad | 110 µrad |
| Reduced axial brightness, A/(m2 sr V) | 1.8 × 108 | 1.8 × 108 | 1.8 × 108 |
Table 1: Brightness of the X-FEG source of SALVE at two voltages and two dose rates (this work).
From the semi-angle the brightness follows with the same images, once the camera is calibrated. At 20 kV and high dose rate, for example, a pixel measured (40 pm)2, the camera gave 12 counts per electron, and a 2-s image held 13,500 counts per pixel on average. That is a dose rate of 3.5 × 105 e−/(nm2 s), or a current density of 5.6 × 104 A/m2. Divided by the solid angle of 1.5 × 10−8 sr and the relativistically corrected voltage, this gives a reduced axial brightness of 1.8 × 108 A/(m2 sr V). The reduced brightness is an intrinsic property of the source and independent of the voltage[15], and indeed all three settings gave the same value (Table 1).
This constancy turns the measurement into a rule of thumb. The square of the semi-angle scales with the current density: at 20 kV the dose rate fell a hundredfold, from 3.5 × 105 to 3.4 × 103 e−/(nm2 s), and the semi-angle tenfold, from 70 to 7 µrad. “Having confirmed this, a reasonable estimate of the illumination semi-angle distribution is possible for most TEM imaging conditions,” the authors write. Such values enter every contrast calculation for the low-voltage microscope (see Optimum HRTEM Contrast[16]). How the measured brightness compares with the manufacturer’s figure and with earlier measurements at FEI in Eindhoven is shown in the box below.
Challenges named by the authors
The illumination semi-angle changes with magnification and illumination density, so a realistic value has to be measured for the conditions of each image, or scaled from a measured one with the constant brightness. When the information limit is small, reading it from the fringes carries a large relative error, because the fringe disk is small compared with the uncertainty of its radius. The series must cover a wide defocus range, and the fit must include the higher-order aberrations and all other damping mechanisms. Nonlinear transfer does not falsify the semi-angle, but only in the approximation that it opens the chromatic and spatial damping by the same factor at every defocus; a sample that fulfills the linear imaging approximation is preferable.
Outlook
The method needs nothing beyond what every transmission electron microscope offers: a thin amorphous sample, a defocus series and a Fourier transform. The SALVE corrector shows that it works even where other damping effects dominate. A similar route, determining the brightness from the defocus dependence of the information limit, had been presented by Peter Tiemeijer of Thermo Fisher Scientific in Eindhoven at the SALVE workshop in 2015. For SALVE the result means that its X-FEG source delivers more than twice the brightness the manufacturer specifies, with the monochromator switched off, and that the illumination angle for any setting at 20 to 80 kV can now be estimated from the dose rate. The work was part of the SALVE project at Ulm University, funded by the DFG, the Ministry of Science, Research and the Arts of Baden-Württemberg and the DFG priority program SPP 1569.
Resource:
Börrnert, F., Renner, J., & Kaiser, U. (2018).
Electron Source Brightness and Illumination Semi-Angle Distribution Measurement in a Transmission Electron Microscope.
Microsc. Microanal. 24(3), 249–255.
https://doi.org/10.1017/S1431927618000223
Update – August 30, 2018
The news article of May 21, 2018 ends above. Later work that uses its results is listed below.
Brightness in the contrast calculations
Börrnert and Kaiser used the measured brightness and the dose-rate scaling to set the illumination semi-angle in their contrast-transfer calculations for graphene at 80 and 20 kV: 160 µrad at 80 kV and 320 µrad at 20 kV. The calculated information limits and focus bands agreed with the images (see 3D Imaging of Graphene[20]).
Download this news article as PDF – for reading on paper, taking notes or filing.
-
Müller, H., Uhlemann, S., Hartel, P., Haider, M. (2006). Advancing the Hexapole Cs-Corrector for the Scanning Transmission Electron Microscope. Microsc. Microanal., 12, 442–455. https://doi.org/10.1017/S1431927606060600
-
Erni, R. (2010). Aberration-Corrected Imaging in Transmission Electron Microscopy. London: Imperial College Press. https://doi.org/10.1142/p703
-
Lichte, H., Börrnert, F., Lenk, A., Lubk, A., Röder, F., Sickmann, J., Sturm, S., Vogel, K., Wolf, D. (2013). Electron Holography for Fields in Solids: Problems and Progress. Ultramicroscopy, 134, 126–134. https://doi.org/10.1016/j.ultramic.2013.05.014
-
Röder, F., Lubk, A., Wolf, D., Niermann, T. (2014). Noise Estimation for Off-Axis Electron Holography. Ultramicroscopy, 144, 32–42. https://doi.org/10.1016/j.ultramic.2014.04.002
-
Löffler, K. H. (1964). Erzeugung Freitragender Mikroobjekte Durch Elektronenstrahlaktivierten Kohlefolienabbau. Dissertation, Berlin.
-
Frank, J. (1976). Determination of Source Size and Energy Spread From Electron Micrographs Using the Method of Young’s Fringes. Optik, 44, 379–391.
-
Tsuji, M., St. John Manley, R. (1983). Determination of the Illuminating Angle and Defocus Spread in Transmission Electron Microscopy. J. Microsc., 130, 93–98. https://doi.org/10.1111/j.1365-2818.1983.tb04202.x
-
Hosokawa, F., Suzuki, M., Ibe, K. (1991). Determination of the Effective Source From Its Image in the Back Focal Plane of the Objective Lens. Ultramicroscopy, 36, 367–373. https://doi.org/10.1016/0304-3991(91)90128-S
-
Zemlin, F., Schiske, P. (1980). Measurement of the Phase Contrast Transfer Function and the Cross-Correlation Peak Using Young Interference Fringes. Ultramicroscopy, 5, 139–145. https://doi.org/10.1016/0304-3991(80)90019-4
-
Uhlemann, S., Müller, H., Hartel, P., Zach, J., Haider, M. (2013). Thermal Magnetic Field Noise Limits Resolution in Transmission Electron Microscopy. Phys. Rev. Lett., 111, 046101. https://doi.org/10.1103/PhysRevLett.111.046101
-
Tiemeijer, P. C., Bischoff, M., Freitag, B., Kisielowski, C. (2012). Using a Monochromator to Improve the Resolution in TEM to Below 0.5 Å. Part I: Creating highly coherent monochromated illumination. Ultramicroscopy, 114, 72–81. https://doi.org/10.1016/j.ultramic.2012.01.008
-
Linck, M., Hartel, P., Uhlemann, S., Kahl, F., Müller, H., Zach, J., Haider, M., Niestadt, M., Bischoff, M., Biskupek, J., et al. (2016). Chromatic Aberration Correction for Atomic Resolution TEM Imaging From 20 to 80 kV. Phys. Rev. Lett., 117, 076101. https://doi.org/10.1103/PhysRevLett.117.076101
-
SALVE news (2016). SALVE Cc Correction Realized. salve-project.de, August 9, 2016. https://www.salve-project.de/newspress/SALVE-CC-correction-realized.html
-
Barthel, J., Thust, A. (2008). Quantification of the Information Limit of Transmission Electron Microscopes. Phys. Rev. Lett., 101, 200801. https://doi.org/10.1103/PhysRevLett.101.200801
-
Reimer, L., Kohl, H. (2008). Transmission Electron Microscopy: Physics of Image Formation, 5th ed. New York: Springer. https://doi.org/10.1007/978-0-387-40093-8
-
SALVE news (2012). Optimum HRTEM Contrast. salve-project.de, January 1, 2012. https://www.salve-project.de/newspress/optimum-hrtem-contrast.html
-
Tiemeijer, P. C., Bischoff, M., Freitag, B., Kisielowski, C. (2012). Using a Monochromator to Improve the Resolution in TEM to Below 0.5 Å. Part II: Application to focal series reconstruction. Ultramicroscopy, 118, 35–43. https://doi.org/10.1016/j.ultramic.2012.03.019
-
Williams, D. B., Carter, C. B. (2009). Transmission Electron Microscopy: A Textbook for Materials Science, 2nd ed. New York: Springer. https://doi.org/10.1007/978-0-387-76501-3
-
van Veen, A. H. V., Hagen, C. W., Barth, J. E., Kruit, P. (2001). Reduced Brightness of the ZrO/W Schottky Electron Emitter. J. Vac. Sci. Technol. B, 19, 2038–2044. https://doi.org/10.1116/1.1409390
-
SALVE news (2018). 3D Imaging of Graphene. salve-project.de, August 30, 2018. https://www.salve-project.de/newspress/3d-imaging-of-2d-materials.html





