Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Contents lists available at ScienceDirect
Ultramicroscopy journal homepage: www.elsevier.com/locate/ultramic
A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM M. Stöger-Pollach n University Service Center for Transmission Electron Microscopy, Technische Universität Wien, Wiedner Hauptstraße 8-10, A-1040 Wien, Austria
art ic l e i nf o
a b s t r a c t
Keywords: Low voltage TEM Low voltage EELS Electron conversion Resolution
The present work is a short note on the performance of a conventional transmission electron microscope (TEM) being operated at very low beam energies (below 20 keV). We discuss the high tension stability and resolving power of this uncorrected TEM. We find out that the theoretical lens performance can nearly be achieved in practice. We also demonstrate that electron energy loss spectra can be recorded at these low beam energies with standard equipment. The signal-to-noise ratio is sufficiently good for further data treatment like multiple scattering deconvolution and Kramers–Kronig analysis. & 2014 Elsevier B.V. All rights reserved.
1. Introduction Low voltage TEM (LVTEM) and low voltage STEM (LVSTEM) are door openers to investigate soft materials and soft/hard material interfaces. It is the discovery and design of such new classes of materials which is thought to become increasingly important in the global research for new sustainable energy technologies. The development of such new devices like organic photovoltaics, catalysts or fuel cells requires fundamental understanding of the used materials and their interfaces to metals. Consequently, this is placing new demands on research strategies also in electron microscopy. Low voltage applications of (scanning) transmission electron microscopy (S/TEM) have attracted great interest, because knock on damage of the specimens can be reduced [1–4] and relativistic energy losses can be avoided [5–7]. Future developments in hardware design of S/TEM points on one hand towards high tension flexibility [8,9] using energy ranges from 30 to 300 kV and on the other hand towards dedicated low energy electron microscopes [1,8–11]. Common in these projects is that such dedicated developments require solid funding. In the present work we are focusing on another solution: we align a conventional TEM (TECNAI T20 LaB6) and energy filter (GATAN GIF 2001) for low beam applications and investigate the technical limitations of this set up. When doing so, following aspects have to be discussed: (a) high tension range and high tension stability, (b) lens performance and achievable spatial resolution and (c) possibilities for low energetic electron detection. In the present work, we will see that a suitable low voltage TEM
n
Corresponding author. Tel.: þ 43 1 58801 45204; fax: þ43 1 58801 9 45204. E-mail address: [email protected]
and EELS can be achieved without any hardware adoption down to beam energies of 10 keV. For even lower beam energies the electron detection becomes a serious problem when using a YAG scintillator. A drawback with respect to the new developments is for sure the spatial resolution of the “conventional” low voltage TEM. We will characterize how well nano-objects can be imaged by using Au nanoparticles on a thin carbon membrane. The manuscript is structured as follows: first we discuss the hardware limitations to the beam energy including its stability and the time needed to stabilize. Then a section concerning the lens performance is given. We discuss the resolution by investigating a gauge specimen: gold on carbon. Finally we compare the electron conversion rate (ECR) of a ZnS based phosphor with the one of a YAG scintillator. We finally show EELS spectra recorded using the lowest beam energy ever in an EELS-TEM experiment. The direct band gap of Silicon can be easily determined from this data, because it is not hidden below relativistic energy looses anymore. 2. Hardware parameters 2.1. High tension range and stability When looking at the specifications of available TEMs one will immediately recognize that the high tension range is steadily increasing. Whereas in the earlier days a TEM was developed to be operated at a certain high tension, modern TEMs are designed to be fully operational within a wide beam-energy range. This is also reflected in the names for the microscopes of one of the world leading TEM manufacturers. Beside the range of beam energies for which the TEMs are optimized, there is a nearly unlimited variety of acceleration voltages available. In the user interface of our TECNAI this module is called “free high tension”. Basically any
0304-3991/$ - see front matter & 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ultramic.2014.01.008
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
M. Stöger-Pollach / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
2
beam energy with steps of 10 V can be selected down to approximately 4500 V. Usually TEM power supplies and lens drivers use DACs running signed 16bit controllers. Those DACs are optimized to deliver finest adjustment steps at the highest acceleration voltage, which is 200 kV in the present situation. Reduction of voltage is only done by reduction of total lens current within the board by applying only the lower bits of current controllers. At some voltage the possible dynamic range for the controllers is just between on or off governed by the smallest available bit within the DAC. By chance this value is reached for some controllers at 4500 V for a TECNAI and is a digital limit for low-voltage operation. The stability of the beam energy is measured in terms of energy drift. There is no need to think about high resolution specifications of beam stability, because the uncorrected TEM has not the resolving power for lattice imaging with beam energies as low as 20 keV or even lower. When reducing the beam energy from 200 keV to 20 keV it lasts for 1 h until it has stabilized. Then the high tension stability is outstanding as long as the beam energy is in the low energy regime. It takes only a few minutes when lowering the beam energy in 1 keV steps below 20 keV to have stable beam conditions. The TEM has to stabilize for 12 h when switching back to 200 keV. 2.2. Lens performance and image contrast When reducing the beam energy, the beam current is also reduced. Consequently this is leading to higher recording times for imaging and EELS. A detailed study on the performance of the thermionic gun is given in [7]. Some details of the lens performance at beam energies in the range from 200 keV to 20 keV were already discussed in [12]. Basically there is to state that the lens currents are coupled to the beam energy such that the refracting power of the lenses is kept nearly constant. This is not only true for the pre-selected beam energies, which are 20, 40, 80, 120, 160, and 200 keV, but also true for all free selected beam energies using the free high tension module. Fig. 1 shows gold nanoparticles at 15 keV under extended Scherzer defocus. The sharpness of the edges represents the resolving power of the SuperTwin lens, which has a C S of 1.2 mm for 200 keV. The measured image resolutions depending on various beam energies are summed up in Table 1. For the calculations we use C S ¼ C C ¼ 1:2 mm and dE ¼0.5 eV for all beam energies. It is also remarkable that the lens currents are adjusted such that the image rotation and the final magnification does not change significantly. There is no additional alignment or magnification calibration necessary. The values for image rotation and magnification correction compared to the 20 keV image are summed up in Table 2.
The contrast in the recorded bright field images is due to elastic scattering. For the determination of contrast enhancement we use a 28 mrad objective aperture. Despite any influences steming from the detector itself, which are considered here, the contrast is governed by two quantities. Firstly, the lower the beam energy is, the larger the Bragg angles are. Hence, less electrons pass through the objective aperture. Secondly, in parallel the elastic scattering cross section increases with lower beam energies. In the present experiment we define the contrast as the ratio between the carbon foil (bright) and gold particles (dark). Fig. 2 shows the specimen recorded with 8 keV (left) and 200 keV (right). The contrast in the 8 keV image is enhanced by a factor of E3.9. Additionally we observe off-axial aberrations. Only a limited field of view is in focus. The experimentally determined values for contrast enhancement with respect to the 200 keV experiment are summed up in Table 3. This means that the C/Au contrast ratios are normalized by the value determined from the 200 keV image.
2.3. Electron conversion rate Finally the low energetic beam electrons have to be detected. One has the possibility to look for a specialized detector or – if not the highest quality is required – one can use the standard scintillator coupled charge coupled device (CCD). In the described experimental set up two scintillators are compared: a ZnS based phosphor and a Yttrium stabilized Aluminum Garnet (YAG). The latter has the advantage that the surface is much smoother and consequently the gain is stronger correlated with the sensitivity of
Table 1 Spatial resolution of the TECNAI SuperTwin lens measured and calculated for 200 keV and several very low beam energies. Beam energy (keV) Exp. resolution (nm) Theor. resolution (nm)
200 0.24 0.24
20 0.88 0.66
17 1.07 0.72
15 0.96 0.77
12 1.07 0.86
10 1.17 0.94
8 2.3 1.05
Table 2 Image rotation and magnification correction for 200 keV and very low beam energies compared with the 20 keV experiment. Beam energy (keV) Image rotation (deg) Magnification correction (%)
200 20 17 15 12 10 8 10.43 – 1.66 0.78 2.77 4.86 3.36 111.4 100.0 97.7 103.8 100.9 94.8 107.9
Fig. 1. Left: Au nanoparticles on carbon imaged at 15 keV. Center: the line shows the position of the intensity profile. Right: intensity profile from the position marked in the image. The resolution is defined by the full width at half maximum of the curve, which is 0.96 nm.
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
M. Stöger-Pollach / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
3
Fig. 2. Left: Au nanoparticles on carbon imaged with 8 keV. Right: the same specimen area imaged at 200 keV.
Table 3 Experimentally determined contrast enhancement with respect to the 200 keV conditions. Beam energy (keV) Contrast enhancement
200 1.00
20 2.29
17 2.69
15 3.06
12 3.14
10 3.76
8 3.86
Fig. 3. Diagram of the cameras for electron detection as used in the described set up. The ZnS based phosphor is used in the imaging camera, whereas the YAG is used in the spectrometer and energy filter.
the CCD pixels. The collected light is excited in all depths of the YAG, because it is transparent. In the phosphor scintillator only the photons that are excited in the lower part are contributing to the signal, because it is opaque for visible light. Consequently the electrons have already been scattered elastically or inelastically within the scintillator, before they excite photons. Thus the excitation volume has already a certain extension widening the point spread function. Both scintillators are covered with an Aluminum layer (see Fig. 3) in order to prevent it from charging. Whereas the YAG has a very flat and completely Al coated surface, the ZnS based phosphor shows a rough surface with a porous Al coating. Therefore the ZnS based scintillator can be excited by even very slow electrons, which would be almost completely absorbed by the dense Al coating of the YAG. In order to measure the electron conversion rate (ECR) of our electron detection systems we are trying to expose them nearly up to the saturation level, which is 10,000 counts in both cases. We consequently know the required exposure time. Further we measure the beam current using the viewing screen of the TEM as a Faraday cup. Thus we can calculate the number of electrons hitting a single pixel of the detector. Although it might be that the viewing screen is not perfectly well suited for being used as a Faraday cup, we compare only data being acquired with the same microscope. We recognize that the electron conversion rate of the YAG is zero for beam
energies below 10 keV which can be lead back to the total absorbance of beam electrons within the Al coating covering the scintillator. This means that electron detection is limited with respect to the beam energy. At our GATAN GIF 2001 we have the YAG scintillator; hence EELS spectra can only be recorded down to beam energies of 10 keV. On the other hand the phosphor scintillator can still be illuminated such that the CCD becomes saturated at beam energies of less than 10 keV (Fig. 4). Consequently we are going to change the scintillator of the GIF in order to be able for recording EELS spectra at even lower beam energies.
3. EELS at very low beam energies As stated above, 10 keV is the limit for EELS at present for our setup. This is still sufficient to prove the value for Cerenkov loss excitation in Silicon, which was calculated to be 13 keV [6]. Fig. 5 shows the low loss spectrum of Silicon after zero loss peak (ZLP) deconvolution. For the higher beam energies the Cerenkov losses and light guiding modes are clearly visible. The signal onset is pointed out with a full circle for each spectrum. One sees that the onset does not change below 13 keV, because below this limit no Cerenkov losses and light guiding modes are present in the
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
4
M. Stöger-Pollach / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
4. Conclusions When using a conventional TEM with rather unconventional beam energies for low voltage applications it is found out that electron detection might become a problem. Depending on the scintillator the lower beam energy limit is either the electron absorbance of the detector (YAG) or the lens power supply (ZnS based scintillator). We also demonstrated that the Cerenkov limit in Silicon is 13 keV, because below this value no further shift of the signal onset in the low loss EELS spectrum can be detected. For higher beam energies the onset is variable, depending on the signal strength of the Cerenkov losses and the corresponding light guiding modes.
Acknowledgments Fig. 4. Electron conversion rates for ZnS based and YAG for various beam energies. The YAG has a problem for the lowest beam energies, because its Al coating absorbs all incoming electrons.
The author kindly acknowledges the USTEM facility (www. ustem.tuwien.ac.at) for providing the electron microscope.
References
Fig. 5. Silicon low loss spectra recorded with different beam energies. The signal onset is pointed out by a solid circle at each spectrum. It seems as if the onset is blue shifted, which is due to the decrease of intensity in the relativistic losses.
[1] U. Kaiser, J. Biskupek, J.C. Meyer, J. Leschner, L. Lechner, H. Rose, M. StögerPollach, A.N. Khlobystov, P. Hartel, H. Müller, Transmission electron microscopy at 20 kV for imaging and spectroscopy, Ultramicroscopy 111 (2011) 1239–1246. [2] O.L. Krivanek, T.C. Lovejoy, N. Dellby, R.W. Carpenter, Monochromated STEM with a 30 meV-wide, atom sized electron probe, J. Microsc. 61 (2013) 3–21. [3] R.F. Egerton, R. McLeod, F. Wang, M. Malac, Basic questions related to electroninduced sputtering in the TEM, Ultramicroscopy 110 (2009) 991–997. [4] R.F. Egerton, Choice of operating voltage for a transmission electron microscope, Ultramicroscopy, this issue. [5] M. Stöger-Pollach, H. Franco, P. Schattschneider, S. Lazar, B. Schaffer, W. Grogger, H.W. Zandbergen, Cerenkov losses: a limit for bandgap determination and Kramers-Kronig Analysis, Micron 37 (5) (2006) 396–402. [6] M. Stöger-Pollach, Optical properties and band gaps from low loss EELS: pitfalls and solutions, Micron 39 (2008) 1092–1110. [7] M. Stöger-Pollach, Low voltage TEM: influences on electron energy loss spectrometry experiments, Micron 41 (2010) 577–584. [8] U. Dahmen, A status report on the TEAM project, Microsc. Microanal. 13 (2007) 1150–1151. [9] P. Ercius, M. Boese, T. Duden, U. Dahmen, Operation of TEAM I in a user environment at NCEM, Microsc. Microanal. 18 (2012) 676–683. [10] H. Sawada, T. Sasaki, F. Hosokawa, S. Yuasa, M. Terao, M. Kawazoe, T. Nakamichi, T. Kaneyama, Y. Kondo, K. Kimoto, Higher-order aberration corrector for an image-forming system in a transmission electron microscope, Ultramicroscopy 110 (2010) 958–961. [11] O.L. Krivanek, N. Dellby, M.F. Murfitt, M.F. Chisholm, T.J. Pennycook, K. Suenaga, V. Nicolosi, Gentle STEM: ADF imaging and EELS at low primary energies, Ultramicroscopy 110 (2010) 935–945. [12] M. Stöger-Pollach, Low voltage EELS—how low? Ultramicroscopy. http://dx. doi.org/10.1016/j.ultramic.2013.07.004, this issue.
spectrum. At this energy the electrons in the TEM are slower than the speed of light inside the Silicon specimen. As soon as the valence EELS (VEELS) spectrum is free of relativistic losses, one can directly apply KKA in order to retrieve the optical properties of the specimen [12].
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
Contents lists available at ScienceDirect
Ultramicroscopy journal homepage: www.elsevier.com/locate/ultramic
A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM M. Stöger-Pollach n University Service Center for Transmission Electron Microscopy, Technische Universität Wien, Wiedner Hauptstraße 8-10, A-1040 Wien, Austria
art ic l e i nf o
a b s t r a c t
Keywords: Low voltage TEM Low voltage EELS Electron conversion Resolution
The present work is a short note on the performance of a conventional transmission electron microscope (TEM) being operated at very low beam energies (below 20 keV). We discuss the high tension stability and resolving power of this uncorrected TEM. We find out that the theoretical lens performance can nearly be achieved in practice. We also demonstrate that electron energy loss spectra can be recorded at these low beam energies with standard equipment. The signal-to-noise ratio is sufficiently good for further data treatment like multiple scattering deconvolution and Kramers–Kronig analysis. & 2014 Elsevier B.V. All rights reserved.
1. Introduction Low voltage TEM (LVTEM) and low voltage STEM (LVSTEM) are door openers to investigate soft materials and soft/hard material interfaces. It is the discovery and design of such new classes of materials which is thought to become increasingly important in the global research for new sustainable energy technologies. The development of such new devices like organic photovoltaics, catalysts or fuel cells requires fundamental understanding of the used materials and their interfaces to metals. Consequently, this is placing new demands on research strategies also in electron microscopy. Low voltage applications of (scanning) transmission electron microscopy (S/TEM) have attracted great interest, because knock on damage of the specimens can be reduced [1–4] and relativistic energy losses can be avoided [5–7]. Future developments in hardware design of S/TEM points on one hand towards high tension flexibility [8,9] using energy ranges from 30 to 300 kV and on the other hand towards dedicated low energy electron microscopes [1,8–11]. Common in these projects is that such dedicated developments require solid funding. In the present work we are focusing on another solution: we align a conventional TEM (TECNAI T20 LaB6) and energy filter (GATAN GIF 2001) for low beam applications and investigate the technical limitations of this set up. When doing so, following aspects have to be discussed: (a) high tension range and high tension stability, (b) lens performance and achievable spatial resolution and (c) possibilities for low energetic electron detection. In the present work, we will see that a suitable low voltage TEM
n
Corresponding author. Tel.: þ 43 1 58801 45204; fax: þ43 1 58801 9 45204. E-mail address: [email protected]
and EELS can be achieved without any hardware adoption down to beam energies of 10 keV. For even lower beam energies the electron detection becomes a serious problem when using a YAG scintillator. A drawback with respect to the new developments is for sure the spatial resolution of the “conventional” low voltage TEM. We will characterize how well nano-objects can be imaged by using Au nanoparticles on a thin carbon membrane. The manuscript is structured as follows: first we discuss the hardware limitations to the beam energy including its stability and the time needed to stabilize. Then a section concerning the lens performance is given. We discuss the resolution by investigating a gauge specimen: gold on carbon. Finally we compare the electron conversion rate (ECR) of a ZnS based phosphor with the one of a YAG scintillator. We finally show EELS spectra recorded using the lowest beam energy ever in an EELS-TEM experiment. The direct band gap of Silicon can be easily determined from this data, because it is not hidden below relativistic energy looses anymore. 2. Hardware parameters 2.1. High tension range and stability When looking at the specifications of available TEMs one will immediately recognize that the high tension range is steadily increasing. Whereas in the earlier days a TEM was developed to be operated at a certain high tension, modern TEMs are designed to be fully operational within a wide beam-energy range. This is also reflected in the names for the microscopes of one of the world leading TEM manufacturers. Beside the range of beam energies for which the TEMs are optimized, there is a nearly unlimited variety of acceleration voltages available. In the user interface of our TECNAI this module is called “free high tension”. Basically any
0304-3991/$ - see front matter & 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ultramic.2014.01.008
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
M. Stöger-Pollach / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
2
beam energy with steps of 10 V can be selected down to approximately 4500 V. Usually TEM power supplies and lens drivers use DACs running signed 16bit controllers. Those DACs are optimized to deliver finest adjustment steps at the highest acceleration voltage, which is 200 kV in the present situation. Reduction of voltage is only done by reduction of total lens current within the board by applying only the lower bits of current controllers. At some voltage the possible dynamic range for the controllers is just between on or off governed by the smallest available bit within the DAC. By chance this value is reached for some controllers at 4500 V for a TECNAI and is a digital limit for low-voltage operation. The stability of the beam energy is measured in terms of energy drift. There is no need to think about high resolution specifications of beam stability, because the uncorrected TEM has not the resolving power for lattice imaging with beam energies as low as 20 keV or even lower. When reducing the beam energy from 200 keV to 20 keV it lasts for 1 h until it has stabilized. Then the high tension stability is outstanding as long as the beam energy is in the low energy regime. It takes only a few minutes when lowering the beam energy in 1 keV steps below 20 keV to have stable beam conditions. The TEM has to stabilize for 12 h when switching back to 200 keV. 2.2. Lens performance and image contrast When reducing the beam energy, the beam current is also reduced. Consequently this is leading to higher recording times for imaging and EELS. A detailed study on the performance of the thermionic gun is given in [7]. Some details of the lens performance at beam energies in the range from 200 keV to 20 keV were already discussed in [12]. Basically there is to state that the lens currents are coupled to the beam energy such that the refracting power of the lenses is kept nearly constant. This is not only true for the pre-selected beam energies, which are 20, 40, 80, 120, 160, and 200 keV, but also true for all free selected beam energies using the free high tension module. Fig. 1 shows gold nanoparticles at 15 keV under extended Scherzer defocus. The sharpness of the edges represents the resolving power of the SuperTwin lens, which has a C S of 1.2 mm for 200 keV. The measured image resolutions depending on various beam energies are summed up in Table 1. For the calculations we use C S ¼ C C ¼ 1:2 mm and dE ¼0.5 eV for all beam energies. It is also remarkable that the lens currents are adjusted such that the image rotation and the final magnification does not change significantly. There is no additional alignment or magnification calibration necessary. The values for image rotation and magnification correction compared to the 20 keV image are summed up in Table 2.
The contrast in the recorded bright field images is due to elastic scattering. For the determination of contrast enhancement we use a 28 mrad objective aperture. Despite any influences steming from the detector itself, which are considered here, the contrast is governed by two quantities. Firstly, the lower the beam energy is, the larger the Bragg angles are. Hence, less electrons pass through the objective aperture. Secondly, in parallel the elastic scattering cross section increases with lower beam energies. In the present experiment we define the contrast as the ratio between the carbon foil (bright) and gold particles (dark). Fig. 2 shows the specimen recorded with 8 keV (left) and 200 keV (right). The contrast in the 8 keV image is enhanced by a factor of E3.9. Additionally we observe off-axial aberrations. Only a limited field of view is in focus. The experimentally determined values for contrast enhancement with respect to the 200 keV experiment are summed up in Table 3. This means that the C/Au contrast ratios are normalized by the value determined from the 200 keV image.
2.3. Electron conversion rate Finally the low energetic beam electrons have to be detected. One has the possibility to look for a specialized detector or – if not the highest quality is required – one can use the standard scintillator coupled charge coupled device (CCD). In the described experimental set up two scintillators are compared: a ZnS based phosphor and a Yttrium stabilized Aluminum Garnet (YAG). The latter has the advantage that the surface is much smoother and consequently the gain is stronger correlated with the sensitivity of
Table 1 Spatial resolution of the TECNAI SuperTwin lens measured and calculated for 200 keV and several very low beam energies. Beam energy (keV) Exp. resolution (nm) Theor. resolution (nm)
200 0.24 0.24
20 0.88 0.66
17 1.07 0.72
15 0.96 0.77
12 1.07 0.86
10 1.17 0.94
8 2.3 1.05
Table 2 Image rotation and magnification correction for 200 keV and very low beam energies compared with the 20 keV experiment. Beam energy (keV) Image rotation (deg) Magnification correction (%)
200 20 17 15 12 10 8 10.43 – 1.66 0.78 2.77 4.86 3.36 111.4 100.0 97.7 103.8 100.9 94.8 107.9
Fig. 1. Left: Au nanoparticles on carbon imaged at 15 keV. Center: the line shows the position of the intensity profile. Right: intensity profile from the position marked in the image. The resolution is defined by the full width at half maximum of the curve, which is 0.96 nm.
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
M. Stöger-Pollach / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
3
Fig. 2. Left: Au nanoparticles on carbon imaged with 8 keV. Right: the same specimen area imaged at 200 keV.
Table 3 Experimentally determined contrast enhancement with respect to the 200 keV conditions. Beam energy (keV) Contrast enhancement
200 1.00
20 2.29
17 2.69
15 3.06
12 3.14
10 3.76
8 3.86
Fig. 3. Diagram of the cameras for electron detection as used in the described set up. The ZnS based phosphor is used in the imaging camera, whereas the YAG is used in the spectrometer and energy filter.
the CCD pixels. The collected light is excited in all depths of the YAG, because it is transparent. In the phosphor scintillator only the photons that are excited in the lower part are contributing to the signal, because it is opaque for visible light. Consequently the electrons have already been scattered elastically or inelastically within the scintillator, before they excite photons. Thus the excitation volume has already a certain extension widening the point spread function. Both scintillators are covered with an Aluminum layer (see Fig. 3) in order to prevent it from charging. Whereas the YAG has a very flat and completely Al coated surface, the ZnS based phosphor shows a rough surface with a porous Al coating. Therefore the ZnS based scintillator can be excited by even very slow electrons, which would be almost completely absorbed by the dense Al coating of the YAG. In order to measure the electron conversion rate (ECR) of our electron detection systems we are trying to expose them nearly up to the saturation level, which is 10,000 counts in both cases. We consequently know the required exposure time. Further we measure the beam current using the viewing screen of the TEM as a Faraday cup. Thus we can calculate the number of electrons hitting a single pixel of the detector. Although it might be that the viewing screen is not perfectly well suited for being used as a Faraday cup, we compare only data being acquired with the same microscope. We recognize that the electron conversion rate of the YAG is zero for beam
energies below 10 keV which can be lead back to the total absorbance of beam electrons within the Al coating covering the scintillator. This means that electron detection is limited with respect to the beam energy. At our GATAN GIF 2001 we have the YAG scintillator; hence EELS spectra can only be recorded down to beam energies of 10 keV. On the other hand the phosphor scintillator can still be illuminated such that the CCD becomes saturated at beam energies of less than 10 keV (Fig. 4). Consequently we are going to change the scintillator of the GIF in order to be able for recording EELS spectra at even lower beam energies.
3. EELS at very low beam energies As stated above, 10 keV is the limit for EELS at present for our setup. This is still sufficient to prove the value for Cerenkov loss excitation in Silicon, which was calculated to be 13 keV [6]. Fig. 5 shows the low loss spectrum of Silicon after zero loss peak (ZLP) deconvolution. For the higher beam energies the Cerenkov losses and light guiding modes are clearly visible. The signal onset is pointed out with a full circle for each spectrum. One sees that the onset does not change below 13 keV, because below this limit no Cerenkov losses and light guiding modes are present in the
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
4
M. Stöger-Pollach / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
4. Conclusions When using a conventional TEM with rather unconventional beam energies for low voltage applications it is found out that electron detection might become a problem. Depending on the scintillator the lower beam energy limit is either the electron absorbance of the detector (YAG) or the lens power supply (ZnS based scintillator). We also demonstrated that the Cerenkov limit in Silicon is 13 keV, because below this value no further shift of the signal onset in the low loss EELS spectrum can be detected. For higher beam energies the onset is variable, depending on the signal strength of the Cerenkov losses and the corresponding light guiding modes.
Acknowledgments Fig. 4. Electron conversion rates for ZnS based and YAG for various beam energies. The YAG has a problem for the lowest beam energies, because its Al coating absorbs all incoming electrons.
The author kindly acknowledges the USTEM facility (www. ustem.tuwien.ac.at) for providing the electron microscope.
References
Fig. 5. Silicon low loss spectra recorded with different beam energies. The signal onset is pointed out by a solid circle at each spectrum. It seems as if the onset is blue shifted, which is due to the decrease of intensity in the relativistic losses.
[1] U. Kaiser, J. Biskupek, J.C. Meyer, J. Leschner, L. Lechner, H. Rose, M. StögerPollach, A.N. Khlobystov, P. Hartel, H. Müller, Transmission electron microscopy at 20 kV for imaging and spectroscopy, Ultramicroscopy 111 (2011) 1239–1246. [2] O.L. Krivanek, T.C. Lovejoy, N. Dellby, R.W. Carpenter, Monochromated STEM with a 30 meV-wide, atom sized electron probe, J. Microsc. 61 (2013) 3–21. [3] R.F. Egerton, R. McLeod, F. Wang, M. Malac, Basic questions related to electroninduced sputtering in the TEM, Ultramicroscopy 110 (2009) 991–997. [4] R.F. Egerton, Choice of operating voltage for a transmission electron microscope, Ultramicroscopy, this issue. [5] M. Stöger-Pollach, H. Franco, P. Schattschneider, S. Lazar, B. Schaffer, W. Grogger, H.W. Zandbergen, Cerenkov losses: a limit for bandgap determination and Kramers-Kronig Analysis, Micron 37 (5) (2006) 396–402. [6] M. Stöger-Pollach, Optical properties and band gaps from low loss EELS: pitfalls and solutions, Micron 39 (2008) 1092–1110. [7] M. Stöger-Pollach, Low voltage TEM: influences on electron energy loss spectrometry experiments, Micron 41 (2010) 577–584. [8] U. Dahmen, A status report on the TEAM project, Microsc. Microanal. 13 (2007) 1150–1151. [9] P. Ercius, M. Boese, T. Duden, U. Dahmen, Operation of TEAM I in a user environment at NCEM, Microsc. Microanal. 18 (2012) 676–683. [10] H. Sawada, T. Sasaki, F. Hosokawa, S. Yuasa, M. Terao, M. Kawazoe, T. Nakamichi, T. Kaneyama, Y. Kondo, K. Kimoto, Higher-order aberration corrector for an image-forming system in a transmission electron microscope, Ultramicroscopy 110 (2010) 958–961. [11] O.L. Krivanek, N. Dellby, M.F. Murfitt, M.F. Chisholm, T.J. Pennycook, K. Suenaga, V. Nicolosi, Gentle STEM: ADF imaging and EELS at low primary energies, Ultramicroscopy 110 (2010) 935–945. [12] M. Stöger-Pollach, Low voltage EELS—how low? Ultramicroscopy. http://dx. doi.org/10.1016/j.ultramic.2013.07.004, this issue.
spectrum. At this energy the electrons in the TEM are slower than the speed of light inside the Silicon specimen. As soon as the valence EELS (VEELS) spectrum is free of relativistic losses, one can directly apply KKA in order to retrieve the optical properties of the specimen [12].
Please cite this article as: M. Stöger-Pollach, A short note on how to convert a conventional analytical TEM into an analytical Low Voltage TEM, Ultramicroscopy (2014), http://dx.doi.org/10.1016/j.ultramic.2014.01.008i
