The first mandatory action in a site where a TEM/STEM is planned to be installed is to accurately measure the external AC and DC sources of stray fields.

Considering that the equipment itself is a source of noise. Zhu et al. [39] reported that, before microscope installation and operation, the measured induced AC fields in the instrument room, in all x, y, and z directions at different heights, were below 0.005 mG, whereas after the instrument and the ancillary equipment were switched on, the average AC magnetic fields at 60 Hz increased to 0.15 mG in the z-direction and 0.08 mG in x-y directions, which means an increase by a factor 30 in the z-direction and by a factor 15 in the x-y plane. 

The effects of EM noise are detrimental in many respects, as they produce scanning distortions in STEM images, aberrations in high-resolution TEM images, and loss of energy resolution in EELS experiments.

In [38], ad hoc experiments were planned and performed by the authors to measure the effects of EM noise in typical TEM/STEM experiments. By using a reference layer of SrTiO₃, grown on silicon, 5-unit cells thick (1.96 nm), they measured the distortion on STEM images, induced by an external AC field, generated by the current circulating in a coil one meter in diameter. Sensitivity factors to 0.5 Å/mG and 1.42 Å/mG were measured for a Tecnai F20 with a monochromator TEM/STEM instrument and for a VGHB501A UHV dedicated STEM, respectively.

Typical sources of AC fields are represented by currents lost towards ground, due to bad ground connections. These currents flow through metal conduits in the microscope room, causing the generation of stray fields [34]. These problems can be easily fixed but are difficult to isolate. Therefore, a simple way to overcome them is to remove all old wiring and to enclose new cables into electrical trenches far from the microscope, which, in addition to providing electromagnetic shielding and eliminating hazardous obstacles on the floor, prevents dust accumulation and facilitates floor cleaning. The floor tiles must be conductive and grounded to prevent electrostatic charging. Special attention must also be paid to lighting; for example, dimmable incandescent lighting must be chosen to eliminate the radio-frequency noise due to the electronics of fluorescent lighting, the use of which must be limited to maintenance operations [39].

Additionally, quasi-DC fields may be generated by metal objects moving close to the microscope, which are responsible for energy shifts in the alignment of the EELS spectrometer, making the interpretation of the spectra unreliable. A shift of the order of 1 eV in the EELS spectrum can be caused by moving the iron wheels of an office chair; therefore, wooden chairs are better suited for TEM laboratories. In [34], the authors reported on a series of typical moving objects which could cause these types of problems.

A simple way to mitigate the effect of EM fields, and the eventual environmental thermal fluctuation, is to host the microscope in a large room, since the intensity of EM fields decay very rapidly with distance, and the large volume of the room behaves as a thermal buffer stabilizer.

Screening strategies for electromagnetic disturbances can be passive or active: passive shielding uses high permeability metals or metal-alloys, which are commercially available in sheets or foils of different size and thickness.

Ferromagnetic alloys exhibit good behavior since their attenuation slightly drops as frequency decreases in comparison with pure metals, such as aluminum or copper.

Typical examples of branded ferromagnetic alloys include MuMetal^®, Netic^®, Finemet^® and Metglas^®, to name a few. Among these, Mu-Metal, a nickel–iron soft ferromagnetic alloy with high nickel content (80–82%), is particularly appreciated [34,38] for its attenuating properties and for its availability in a wide range of stock thicknesses from 0.36 mm to 5 mm.

The required thickness of the shielding foil is related to the skin depth, given by $\delta =\sqrt{\frac{2}{\sigma \mu \omega }}$ , where σ is the conductivity, µ the permeability, and w the frequency. The formula suggests that low frequency fields are difficult to attenuate; therefore, thicker and more expensive foils are required in this case.

Active cancellation systems consist of Helmholtz coils, running around the microscope room, and feedback wideband (including DC) sensors, which measure the magnetic field to be canceled. These systems are very effective in canceling the field at an exact point and at high frequencies, where the feedback sensor is more sensitive. They are not so efficient for inhomogeneous stray fields or stray fields produced by close sources. For example, in a small room, if the field is canceled at the gun level, it would be enhanced at the spectrometer [38]. More efficient systems, based on triaxial magnetic field compensation, have been designed for the new electron microscope at the Graz Centre for Electron Microscopy [71], by optimizing the position of the sensor and the shape of the coil, both tailored to the room geometry and interfering fields. From personal experience, sometimes it happens that strange blurring can be experienced during imaging or spectroscopic experiments. After sometimes long hunting for the noise source, it disappears by resetting the active compensation system. In Table 1, the strategies used by different TEM facilities to attenuate EM stray fields are reported, along with the limits required by the installation specifications and the issues experienced during the laboratory realization.

One of the most important requisites for TEM equipment is to maintain the surrounding environment at a fixed (around 20 °C) and stable (±0.1 °C) temperature. Temperature variations cause a drift of the specimen, of the microscope electronics, and of the mechanical tolerances in components, including microscope lenses, detectors, aberration correctors, and scan coils.

As in the case of the EM fields, the size of the room plays an important role in temperature stability, since larger spaces around the microscope will better dampen any heat spike within the laboratory. Two types of temperature control systems are commonly used, i.e., forced air systems and radiant panels; they remove heat and keep the temperature as required. Forced air systems remove heat by convection and conduction. Since the heat capacity of air is very low, large airflows are necessary to remove heat. Airflows cause air pressure on the microscope column, resulting in mechanical vibrations (see next paragraph). This is one of the reasons why, in modern laboratories, heat removal by air flow is minimized by using two more convenient strategies: the first one is passive and consists of locating all heat-generating equipment that can be separated from the microscope, such as power distribution racks in a service room, separated from the microscope room. The second one is active and is based on radiation, instead of conduction and convection, as the main mechanism for heat load reduction, by using thermal masses placed in the microscope room. Radiant panels [72,73] are installed on the room walls and chilled water is circulated through them; they can regulate the temperature to better than 0.1 °C, and, in case of thermal drifts, for example due to the entrance of a person in the microscope room, the return to equilibrium is quite rapid.

Therefore, forced air systems are minimally used to regulate the temperature and mainly to control humidity, thus avoiding water condensation on the radiant panels and on the cooled parts of the microscope (electronics, pumps).

The primary effect of using radiant panels instead of a forced air system is reduced drift of image and spectra, helpful for long acquisition times, like during the frame integration of STEM image acquisition, 4D-STEM, or during analytical experiments such as EELS and EDXS chemical mapping. Secondly, since both spectrometer and high-tension supply are sensitive to temperature changes, higher temperature stability results in more reliable spectra as a function of time.

Mechanical instabilities are crucial, especially for the microscopes of the latest generation, since the presence of aberration correctors and/or monochromators makes the column longer than in old microscopes and more sensitive to mechanical vibrations. The stiffness of the microscope’s column linearly worsens with height and roughly improves with the second power of diameter. Therefore, modern microscopes have been completely redesigned by manufacturers to achieve better stability. Nevertheless, environmental vibrations remain an important source of instabilities, especially at the level of the gun and the specimen. The gun is placed on top of the column and is, therefore, subjected to maximum sway, with a detrimental effect on probe formation. Stage movements are also negative for high resolution imaging and spectroscopy. Vertical vibration of the specimen within the objective lens pole pieces results in a spread of focus of the image, which limits the attainable resolution. Horizontal vibration, usually more in one direction, will smear out the image, also limiting the resolution.

One of the main difficulties related to the treatment of vibrations is the microscope’s sensitivity to low frequency vibrations, in the range of a few Hertz; these are the most difficult to eliminate from the microscope’s environment.

There are different sources of mechanical noise; they can be distinguished because of the medium responsible for the noise propagation, i.e., air and soil.

Mechanical noise from air is related to air movements, mainly generated by acoustic waves and thermal gradients, deriving from temperature control systems.

The temperature control systems must be designed and realized in such a way that air movements in the laboratory are minimized. Forced air systems should diffuse air as much as possible to produce laminar flow. To this aim, a technical solution is represented by a laminar ceiling made of perforated panels, which provide downward and even air distribution. This solution is quite expensive and difficult to retrofit. Another solution, much cheaper and particularly suitable for retrofit, makes use of a duct sock connected to the air inlet; the natural wave of the sock tissue lets air softly diffuse out, as reported in [38]. In [34], the authors created a home-made simple but effective test which could be easily carried out to roughly evaluate the airflow; the so-called “toilet-paper test”: 12 × 0.25 inches. Strips of single-ply paper are attached around the microscope, and, if they deflect at the bottom by more than an inch, then the airflow exceeds 20 ft/min. At the time the article was written, 15 ft/min was considered acceptable for a 0.2 nm resolution in STEM, but, today, the sub-angstrom resolution achievable with the modern state-of-the-art instrument requires much lower limits, as reported in Table 1, and highly sensitive airflow detectors for their quantitative measurements. Nevertheless, the test can be performed as a preliminary test to assess the condition of the room, or to periodically monitor time stability.

Air movement caused by acoustic waves impacts microscope performance, depending on intensity and frequency. Common sources of acoustic noise are computers, power racks, pumps, chillers, electronics, and air inlets, but external sources can also contribute. Noise attenuation can be achieved by using acoustic shielding, and since common sound damping materials like polyurethane or other foams are inefficacious at low frequency (f < 130 Hz) where the microscope is more sensitive, an effective solution for low frequency attenuation is to use fiberglass absorbers, placed in front of the laboratory walls with an air gap in between. Nevertheless, as in the case of the EM stray fields, before applying any attenuation strategies, the most reasonable and effective approach is to identify the sources of acoustic noise and remove them, or put them as far from the microscope as possible. For example, all ancillary equipment, such as power racks, pumps, and chillers, must be isolated in a separate service room [20,39]. For acoustic noise, it is difficult to define limits and thresholds, since each microscope reacts to acoustic waves depending on its own resonances. Generally, an accredited criterium is to reduce the sound intensity below 40 dB [39], as reported in Table 1.

Mechanical vibrations from the floor supporting the microscope may have different origins: one is related to microquakes, caused by movements of the Earth’s crust, sea waves, mountains, and even glaciers; these “microseisms” contribute to a background noise which cannot be eliminated. In [35], the authors showed plots of the micro-seismic activity near NCEM at LBNL in Berkeley recorded during a Pacific storm on 25 December 1996.