Cryo-FIB Milling of Chlamydia-Infected Cells

Plunged-frozen C. trachomatis-infected HeLa cells were coated with a protective organoplatinum layer and platinum sputter coated as described. Each plasma ion beam was aligned immediately prior to milling. To enable effective curtaining rate determination with nitrogen and oxygen, double spot compensation (Figure 1—figure supplement 4) was performed for each current.
As the practical usable current is determined by the apertures within the ion column, the current used for each gas for study is based on the use of the same aperture; for example, at the lowest aperture size used the currents for nitrogen, oxygen, xenon, and argon are 0.27 nA, 0.23 nA, and 0.1 and 0.2 nA, respectively. A list of currents for a given aperture is shown in Table 2. However, these values will change over time as the beam mills the aperture during usage.
A series of 5 windows of 2 × 2.5 × 2 µm³ were milled until the average grey level intensity for the window reached 10% of the initial signal. The acceleration voltage for the I-beam was kept at 30 or 20 kV. To accurately reflect the ion beam current applied on the sample, the current presented for this experiment is the measured current. The milled surface is then imaged 90° to the SEM without an electrostatic lens to avoid interaction of the magnetic field with the ion beam. This is crucial for imaging with oxygen and nitrogen, where the use of the pre-field immersion lens led to blurring of the ion beam images. The pixel size, voltage, current, electric gain, and voltage offset were kept identical. This was repeated three times per current (total of 15 windows per current).
As curtaining can vary significantly across a milled surface, a semi-automated, quantitative assessment of curtaining is necessary for un-biased comparison of ion sources and milling settings. We use a method based on the work from Hovden’s group (Schwartz et al., 2019 (link)). Using Fiji (Schindelin et al., 2012 (link)) the initial image is subject to a fast Fourier transform (FFT). A horizontal 5° mask is applied and a reverse FFT performed. In this new image the vertical lines have been removed. By subtracting the initial and inverted masked FFT images, we can produce a mask to isolate those lines using a default threshold based on minimum cross entropy (Li and Tam, 1998 (link)). A merged image where this mask is superimposed to the initial image allows precise identification of the location where the milling was performed. We can then produce a histogram where the white pixels are part of the curtain. The percentage of this white pixels is calculated and forms the curtaining score (Figure 1—figure supplement 1).