In-Chamber Sublimation: A Practical Approach for Mitigating Ice and Curtaining in Cryo-Electron Tomography Lamellae Preparation

This paper demonstrates that performing a controlled sublimation step within the SEM chamber prior to FIB milling effectively mitigates surface ice and curtaining artifacts in cryo-ET lamella preparation without compromising sample vitrification or structural integrity.

The Gist

Imagine you are a master sculptor trying to carve a tiny, intricate statue out of a block of ice. This isn't just any ice; it's a frozen snapshot of a living yeast cell, preserved so perfectly that you can see the tiny machines (proteins) inside it. To do this, you use a super-powerful laser (called a Focused Ion Beam) to shave off thin slices of the ice block, creating a "lamella" (a thin window) that an electron microscope can look through.

But there's a problem: Frost.

Just like your freezer door gets a layer of fuzzy frost if you leave it open too long, these frozen samples collect a layer of surface ice during the transfer process. This frost is like a dirty, uneven blanket over your sculpture. If you try to carve through it, your laser bounces off unevenly, creating jagged, messy cuts (called "curtaining") that ruin the view of the delicate structures inside.

The Old Way vs. The New Trick

The Old Way:
Scientists usually tried to fix this by:

1. Building better freezers: Installing expensive, ultra-dry rooms (costly and hard for small labs).
2. Manual scraping: Using a tiny, frozen paintbrush to physically scrape the ice off. This is risky; it's like trying to clean a diamond with a toothbrush—you might scratch the gem or break it.

The New Trick (In-Chamber Sublimation):
The authors of this paper discovered a clever, low-tech solution: Let the ice disappear into thin air.

They realized that if they gently warmed up the sample inside the vacuum chamber of their microscope, the frost wouldn't melt into water (which would ruin the sample); instead, it would sublimate. Sublimation is when ice turns directly into gas, skipping the liquid phase entirely. Think of it like dry ice vanishing into a cloud, but happening right on your sample.

How They Did It (The "Slow Cook" Method)

The team treated the frozen yeast samples like a slow-cooked meal rather than a quick sear.

• The Danger: If you heat the ice too fast or too much, the delicate "glass-like" structure of the frozen water (vitrified ice) turns into regular, crystalline ice. This is called devitrification. Imagine turning a perfect glass window into a cloudy, cracked pane. Once that happens, the high-resolution view is lost forever.
• The Solution: They carefully warmed the sample to a very specific, chilly temperature (around -104°C) and held it there for about an hour. They watched through the microscope the whole time, like a chef checking a soufflé. As the frost sublimated, the surface became smooth and clean.
• The Result: The "dirty blanket" vanished, leaving a pristine, smooth surface ready for carving.

Did It Work?

Yes, and better than expected.

1. Smoother Carving: When they used the laser to cut the thin slices, the cuts were incredibly smooth. No more jagged "curtains" of ice blocking the view.
2. No Damage: They were worried that warming the sample even slightly would ruin the "frozen glass" state. But when they looked at the final images and analyzed the tiny proteins (ribosomes) inside, everything looked perfect. The proteins were sharp and clear, proving the ice never turned into regular crystals.
3. High-Resolution Success: They were able to reconstruct the 3D shape of the yeast's protein machines with high detail, proving that this "frost-melting" trick didn't ruin the sample.

Why This Matters

This is a game-changer because it requires no new expensive equipment. Any lab with a standard cryo-microscope can do this. It turns a difficult, risky process (scraping ice) into a gentle, controlled one (letting it evaporate).

The Analogy:
Think of it like cleaning a foggy car windshield before driving.

• Old way: You wipe it with a rag (might scratch it) or buy a new car with heated glass (expensive).
• New way: You just turn on the defroster. The fog turns to gas and disappears, leaving a clear view without touching the glass.

The Bottom Line

The authors showed that by gently "defrosting" the surface of frozen biological samples just enough to remove the frost, but not enough to melt the ice, they can get much clearer, higher-quality 3D images of cells. It's a simple, practical tweak that makes the complex world of cellular imaging much easier and more accessible.

Technical Summary

1. Problem Statement

Cryo-electron tomography (cryo-ET) workflows, particularly those involving focused ion beam (FIB) milling of high-pressure frozen (HPF) samples, face a persistent challenge: surface ice contamination.

• Impact: Surface ice obscures regions of interest and, more critically, causes curtaining artifacts during FIB milling. This results in uneven lamellae with rough surfaces, compromising the quality of subsequent 3D reconstructions.
• Current Limitations: Existing mitigation strategies include expensive facility upgrades (low-humidity cleanrooms), specialized transfer systems, or manual removal of ice using cryo-compatible paintbrushes. Manual cleaning is operator-dependent, risks sample damage, and is difficult to standardize.
• The Barrier to Sublimation: While sublimation (controlled thermal desorption of amorphous ice) is a standard technique in cryo-SEM for surface cleaning, it has been largely avoided in cryo-ET workflows due to the fear that warming the sample—even under vacuum—would cause devitrification (crystallization of the vitreous ice), rendering the sample unsuitable for high-resolution imaging.

2. Methodology

The authors systematically evaluated the feasibility of performing in-chamber sublimation within a cryo-FIB-SEM (Aquilos 2) prior to lamella milling, using high-pressure frozen yeast cells as a model system.

• Sample Preparation: Yeast cells (BY4741 strain) were prepared using the "waffle method" with high-pressure freezing (2200 bar). Samples were handled in humidity-controlled rooms (10–30% RH).
• Experimental Groups:
  1. Control: Samples milled immediately after sputter coating (no sublimation).
  2. Slow-Sublimation: The cryo-stage temperature was gradually increased from -175°C to -104°C over two hours. Sublimation was monitored via SEM imaging every 5–15 minutes. The process was terminated when surface ice was visually undetectable (~55 minutes of active sublimation).
  3. Fast-Sublimation (Supplemental): A more rapid ramp to -97°C over 35 minutes to test the limits of thermal control.
• Monitoring & Quantification:
  • Surface roughness was quantified using Root-Mean-Square (RMS) roughness ($S_q$) calculated from preprocessed SEM images.
  • Vitrification Check: Post-milling, tilt-series were collected on a Titan Krios. Tomograms were analyzed for diffraction patterns indicative of crystalline ice (cubic or hexagonal).
• Downstream Analysis:
  • Subtomogram Averaging: 80S ribosomes were extracted from the tomograms.
  • Resolution Assessment: A Rosenthal-Henderson plot was generated to calculate the B-factor, and Fourier Shell Correlation (FSC) was used to determine global resolution.

3. Key Contributions

• Validation of In-Chamber Sublimation: The study demonstrates that sublimation can be safely performed directly inside the cryo-FIB-SEM chamber without additional hardware, provided the temperature ramp is carefully controlled.
• Protocol Refinement: The authors established a "slow-sublimation" protocol (gradual warming to -104°C) that effectively removes surface ice without inducing devitrification, contrasting with "fast" protocols that caused surface damage.
• Quantitative Metrics: The paper introduces a method to quantify surface ice removal via image-based roughness analysis ($S_q$) and correlates this directly with milling quality.

4. Results

• Surface Roughness & Curtaining:
  • Sublimated Samples: Showed a significant reduction in surface roughness ($S_q$) and produced lamellae with smooth surfaces. FIB milling resulted in minimal curtaining artifacts across all stages (trench, rough, and fine milling).
  • Control Samples: Retained rough surfaces and exhibited severe curtaining artifacts during milling.
• Preservation of Vitrified State:
  • Tomograms from sublimated samples showed no evidence of crystalline ice. Fast Fourier Transforms (FFTs) of tomogram slices lacked the distinct reflections associated with cubic (3.1 nm⁻¹) or hexagonal (2.5 nm⁻¹) ice.
  • Raw tilt images and aligned series showed no diffraction effects, confirming the sample remained vitreous.
• High-Resolution Imaging:
  • Subtomogram averaging of 2,738 yeast 80S ribosome particles yielded a global resolution of 13.5 Å.
  • The calculated Rosenthal-Henderson B-factor was 1502 Ų.
  • These results are comparable to previously published data for similar sample thicknesses and particle counts, proving that the sublimation process did not degrade the high-resolution signal.
• Fast-Sublimation Failure: The rapid ramp experiment (-97°C) resulted in visible pitting, rips, and holes in the lamellae, likely due to "deep-etching" effects where ice between cells was removed too aggressively. This highlighted the necessity of slow, controlled warming.

5. Significance

• Practical Workflow Improvement: This approach offers a cost-effective, hardware-free solution to a major bottleneck in cryo-ET. It requires no new instrumentation, only a modification of the standard cryo-FIB-SEM operating parameters.
• Enhanced Reproducibility: By replacing manual brushing (which is invasive and variable) with a visually guided, automated thermal process, the method significantly improves the reproducibility of lamella preparation.
• Broad Applicability: While tested on HPF yeast, the authors suggest this method is particularly valuable for:
  • High-pressure frozen tissue samples.
  • Correlative Light and Electron Microscopy (CLEM) workflows, where multiple transfers increase contamination risk.
  • Any workflow where surface ice accumulation obscures features of interest.
• Paradigm Shift: The study challenges the long-standing assumption that sublimation is incompatible with cryo-ET, demonstrating that with precise thermal control, it is a viable and essential tool for achieving high-quality lamellae.

In conclusion, the paper establishes in-situ sublimation as a robust, essential step for preparing high-quality cryo-lamellae, effectively mitigating ice contamination and curtaining while preserving the structural integrity required for high-resolution cryo-ET.