Imaging

There are a number of options and technologies available for digital imaging in transmission electron microscopy (TEM) applications today. Traditionally, high-energy electrons could not be directly exposed to a sensor without excessively damaging the detector. As a consequence, conventional TEM cameras first expose the incoming electron beam to a scintillating film that converts the electrons into light (photons). These photons are then transferred to the sensor, either through a series of optical lenses or a coupled fiber optic plate. Finally, the light is collected by a sensor where the image is created pixel-by-pixel based on the amount of light detected at each position in the sensor.

Conventional TEM image detection architecture

There are four basic steps in TEM imaging to address incoming electrons:

1. Convert electrons to signal
2. Transfer signal
3. Detect signal with sensor
4. Electronically transfer signal and read-out to form an image

What is different with direct detection?

There are only two steps in TEM imaging with direct detection:

1. Convert electrons to signal – not applicable
2. Transfer signal – not applicable
3. Detect signal with sensor
4. Electronically transfer signal and read-out to form an image

One key difference between conventional and direct detection is a custom CMOS sensor that utilizes the only radiation-hard architecture that can tolerate direct exposure to high-energy particles. In addition, extremely high-speed electronics for data transfer and processing enable low-dose counting and super-resolution capabilities. Combined, this allows frame rates (4k x 4k) of 400 frames per second (fps) to be processed in real time to achieve optimal results.

 

Step 1) Convert electrons to signal

Gatan uses proprietary phosphor scintillators to optimize signal conversion, enhancing detector sensitivity (SENS) and resolution. When selecting a scintillator, it is appropriate to know the performance trade-offs between SENS and resolution.

• Sensitivity (signal): Ideal for dose-sensitive use cases where you need to generate more photons per incoming electron (e.g., cryo-tomography, beam-sensitive materials)
• Resolution (spatial detail): Favorable for applications where you require more information to resolve details, but you can increase the dose (signal) without harming the sample (e.g., semiconductors and other less-sensitive materials)

Step 2) Transfer signal

Various coupling (lens- and fiber-coupled) mechanisms are available to optimize signal transfer and meet cost or performance targets for a given detector.

Lens-coupled: Lens optics transmit light to the sensor where it is converted into sensor electrons (signal)

• Pros: (Gatan) Uses real transmission scintillator; can be less expensive than higher-performance fibers
• Cons: Light (information) loss with lensing, angular dependencies (<10% efficient); vignetting (light fall-off); image distortion for higher magnification

Fiber-coupled: Scintillator creates photons that subsequently create sensor electrons; fiber directly transmits light to the sensor with high efficiency

• Pros: Most efficient transmission of light information to the sensor (1:1 coupling of scintillator:sensor (>50% efficient) with no image distortion); can trade-off SENS verses resolution (fiber configuration detail)
• Cons: Fiber optics are slightly higher cost; requires process optimization including cladding, sintering, and bonding of fibers (Gatan proprietary)

Step 3) Detect the signal with the sensor

Sensor type (CCD vs. CMOS) offers significant trade-offs for TEM camera performance as there are fundamental differences in architecture.

• Charge-coupled device (CCD): Charge transfers between neighboring cells, and read-out (e.g., noise) is seen at the final stage; binning minimizes the impact of read-out noise
• Complementary metal–oxide–semiconductor (CMOS): Charge immediately converts to voltage (read-out with digital output); supports high frame rates, low overall electronics noise

Both technologies possess inherent advantages, so the question arises about what unique performance characteristics arise from each choice. CCDs can have a 100% fill factor that captures all incoming light, whereas part of the CMOS sensor is occupied by transistors and metal wiring associated with each pixel. Historically, CCDs provided higher-quality images with low noise at affordable prices. Recent design advancements and processing techniques now advance CMOS sensor performance so it is a viable choice for some applications. Note that CCDs still maintain an advantage for binning in terms of signal-to-noise. However, CMOS chips can scale the number of read-out ports and achieve very high frame rates.

Step 4) Transfer signal and read-out image

When a charge converts to voltage, you typically generate noise

• CCD: Transfer data out of a serial register
• CMOS: Converts to voltage per pixel

It is very important to optimize read-out noise (higher voltages) and speed (multi-port and fast read times) for CCDs.

• Optimize controller for low read-noise; leverage multi-port read-outs for faster speed
• Interline CCDs with binning have the fastest readouts (fps) – up to 30 fps due to 100% duty cycle

CMOS is typically seen as a fast sensor because it can run in rolling shutter mode rather than the slow global shutter mode.

Nyquist frequency
Dose fractionation and motion correction 
Improving DQE with counting and super-resolution