Volume electron microscopy is a family of imaging techniques that builds three-dimensional pictures of cells, tissues, or materials by collecting and stacking hundreds to thousands of nanometer-resolution electron microscope images. Where a conventional electron microscope gives you a single two-dimensional slice, volume EM reconstructs an entire block of sample in 3D, revealing how organelles connect, how neurons wire together, or how the internal architecture of a battery cathode changes over time. The field has grown rapidly since the mid-2000s, driven by advances in automated sectioning hardware, faster detectors, and machine-learning tools that can make sense of the enormous image datasets these instruments produce.
Why a Single Slice Is Not Enough
Traditional electron microscopy produces spectacularly detailed images, but each image captures only one thin section of the sample, typically less than 100 nanometers thick. That is like trying to understand the layout of a building by looking at a single floor plan. For decades, researchers got around this by manually cutting serial sections and imaging them one by one on a transmission electron microscope, then painstakingly aligning the images by hand. The process was heroic but painfully slow and error-prone, and it limited the volumes that could realistically be reconstructed. Volume EM automates and scales that process, making it feasible to image entire cells or even small tissue blocks at resolutions down to a few nanometers per voxel (the 3D equivalent of a pixel).1Nature. An open-access volume electron microscopy atlas of whole cells and tissues
The Main Flavors of Volume EM
Volume EM is not a single instrument or protocol. It is an umbrella term for several related but distinct techniques, each with different trade-offs in resolution, volume, speed, and destructiveness. The choice of method depends on what question you are trying to answer and how big or small the structure of interest is.
Serial Block-Face Scanning Electron Microscopy
In serial block-face SEM (SBF-SEM), a diamond knife mounted inside the vacuum chamber of a scanning electron microscope shaves an ultrathin layer from the top of a resin-embedded sample block. The freshly exposed surface is then imaged by the SEM, and the knife removes the next layer. This cycle repeats automatically, collecting hundreds to thousands of serially registered images without any human intervention.2PubMed Central. Serial Block-Face Scanning Electron Microscopy (SBF-SEM) of Biological Tissue Samples The big advantage is throughput: you can leave the instrument running overnight and come back to a complete dataset the next day. The trade-off is that the process is destructive. Once a layer is cut away, it is gone. If something went wrong during imaging of that slice, there is no going back. Typical slice thickness ranges from roughly 25 to 50 nanometers, which is thicker than what FIB-SEM can achieve but still fine enough for many questions in cell biology and neuroscience.
Focused Ion Beam Scanning Electron Microscopy
FIB-SEM replaces the diamond knife with a focused beam of gallium ions that mills away material from the block face a few nanometers at a time. Because the ion beam can remove much thinner layers than a physical knife, FIB-SEM achieves finer z-resolution, down to roughly 3 to 4 nanometers per slice, producing near-isotropic voxels where the resolution is essentially the same in all three dimensions.3PubMed. High-resolution three-dimensional reconstruction of a whole yeast cell using focused-ion beam scanning electron microscopy That isotropic quality is a genuine asset when you need to trace delicate structures like endoplasmic reticulum tubules or mitochondrial cristae through a cell. Recent engineering improvements in FIB milling stability and SEM scanning speed have expanded the imageable volume by roughly two orders of magnitude while maintaining 4-nanometer voxels, making it possible to image entire cells and even small tissue blocks at extraordinary detail.1Nature. An open-access volume electron microscopy atlas of whole cells and tissues Like SBF-SEM, FIB-SEM is destructive: each milled layer is gone for good.
Array Tomography
Array tomography takes a different approach. Instead of imaging and destroying slices inside the microscope, the sample is first cut into a ribbon of serial ultrathin sections that are collected onto a flat substrate such as a silicon wafer or a glass slide. The ribbon is then imaged by SEM (or sometimes by light microscopy for correlative work).4PubMed Central. A Device for Ribbon Collection for Array Tomography with Scanning Electron Microscopy Because the sections are preserved on the substrate, you can go back and re-image specific slices at higher magnification, stain them with different heavy metals, or even perform immunolabeling. This non-destructive aspect makes array tomography especially attractive when you want to combine ultrastructural imaging with molecular markers. A related variant, automated tape-collecting ultramicrotome SEM (ATUM-SEM), collects sections onto a continuous tape reel rather than a flat wafer, making it easier to handle very long series of sections.
Serial-Section Transmission Electron Microscopy
Serial-section TEM is the oldest form of volume EM, predating the automated SEM-based methods by decades. Thin sections are collected on grids and imaged in a transmission electron microscope, where electrons pass through the section rather than bouncing off its surface. Modern high-throughput implementations use large-format cameras and automated stage movements to tile enormous fields of view. One such pipeline uses a custom camera with a 30.5-centimeter scintillator to capture fields over 20 micrometers wide at 4-nanometer pixel resolution in a single shot.5Nature Communications. A petascale automated imaging pipeline for mapping neuronal circuits with high-throughput transmission electron microscopy TEM-based volume imaging remains a workhorse in large-scale connectomics projects, where the goal is to trace every synapse in a chunk of brain tissue.
Preparing Samples for Volume EM
Getting biological tissue ready for volume EM is arguably harder and more consequential than the imaging itself. The sample must be preserved in a way that halts all biological activity instantly, stained with heavy metals so that membranes produce enough contrast under the electron beam, and embedded in a hard resin that can survive either diamond knife sectioning or ion beam milling. Any misstep here shows up as artifacts in thousands of downstream images.
The most common preparation route is chemical fixation with aldehydes followed by en bloc staining with osmium tetroxide and other heavy metals. A widely used protocol called OTO (osmium-thiocarbohydrazide-osmium) deposits extra osmium onto membranes, increasing both contrast and the electrical conductivity of the sample, which helps reduce charging artifacts during SEM imaging.6Nature Protocols. High-contrast en bloc staining of neuronal tissue for field emission scanning electron microscopy Additional heavy metal stains such as uranyl acetate and lead aspartate are layered in to boost membrane contrast further.7PubMed Central. Double staining method for array tomography using scanning electron microscopy
Chemical fixation, though convenient, can introduce subtle distortions. The alternative is high-pressure freezing, which vitrifies the sample so fast that ice crystals do not have time to form and damage structures. After freezing, the ice is gradually replaced with organic solvents and then resin in a process called freeze substitution. The resulting preservation of fine structures like the nuclear envelope, chromosomes, endoplasmic reticulum, and mitochondria is generally superior to chemical fixation.8PubMed Central. A Workflow for High-pressure Freezing and Freeze Substitution of the Caenorhabditis elegans Embryo for Ultrastructural Analysis by Conventional and Volume Electron Microscopy High-pressure freezing does have limitations: the sample must fit within a carrier only about 200 micrometers deep, so bulky tissues need to be trimmed or dissected first.9PubMed Central. Preparation of cultured cells using high-pressure freezing and freeze substitution for subsequent 2D or 3D visualization in the transmission electron microscope
The Charging Problem and How Researchers Work Around It
One of the most persistent headaches in SBF-SEM and FIB-SEM is specimen charging. When the electron beam hits the sample, electrons accumulate in the resin, which is an insulator. The resulting electric charge deflects the incoming beam, blurring the image and stripping away contrast.2PubMed Central. Serial Block-Face Scanning Electron Microscopy (SBF-SEM) of Biological Tissue Samples Increasing beam voltage or slowing down the scan can improve signal, but both worsen charging and can even physically damage the resin block, distorting subsequent slices.
Researchers have devised several strategies to manage this. One approach injects a small puff of nitrogen gas directly over the block face during imaging. The gas molecules ionize in the electron beam’s vicinity and neutralize the surface charge, effectively draining accumulated electrons while the rest of the chamber stays under high vacuum to maintain image sharpness.10PubMed Central. High-performance serial block-face SEM of non-conductive biological samples enabled by focal gas injection-based charge compensation Another strategy changes the way the beam scans the surface. Instead of sweeping left to right in a conventional raster, a leapfrog scanning pattern distributes the electron dose across non-adjacent lines, giving each strip of surface more time to dissipate charge before the beam returns to its neighbor.11Nature Communications. Reduction of SEM charging artefacts in native cryogenic biological samples These solutions are not exotic add-ons; they are increasingly standard features on instruments designed for volume EM work.
From Raw Images to a Usable 3D Volume
Collecting the images is only half the battle. A volume EM dataset can contain tens of thousands of individual slices, and each slice may have been acquired under slightly different conditions. Shifts of a few nanometers between consecutive images, subtle distortions from charging or drift, and the occasional missing or damaged section all need to be corrected before the stack can be meaningfully reconstructed in 3D.
Alignment pipelines have become increasingly sophisticated to handle this. Modern tools use self-supervised neural networks trained via metric learning to encode image pairs and find the transformation that best aligns them. A technique called vector voting helps the pipeline stay robust even when individual slices are damaged or missing entirely, and the workload is divided across many computational workers to keep processing times manageable.12Nature Communications. Petascale pipeline for precise alignment of images from serial section electron microscopy For FIB-SEM data with fiducial marks, specialized alignment tools can reduce displacement errors from about 0.75 nanometers down to roughly 0.25 nanometers along the z-axis, which matters a great deal when you are trying to trace structures thinner than a mitochondrial membrane.13bioRxiv. A high-performance end-to-end 3D CLEM processing workflow for facilities
Once the images are aligned, the structures of interest need to be identified and outlined, a process called segmentation. Manually tracing organelles through thousands of slices would take years for a single cell. Deep learning has transformed this step. Neural networks trained on expert-annotated datasets can now automatically detect and segment dozens of organelle classes, from endoplasmic reticulum and mitochondria to ribosomes and microtubules, in 4-nanometer voxel FIB-SEM volumes.14Nature. Whole-cell organelle segmentation in volume electron microscopy Other approaches combine organelle detection with image morphology operations and 3D meshing to go from raw data to quantifiable models with less manual intervention.15PubMed. Automated segmentation of cell organelles in volume electron microscopy using deep learning The speed improvement is dramatic, but training these networks still requires substantial expert annotation up front, and segmentation quality varies with sample type and preparation method.
Speeding Things Up with Multiple Beams
A fundamental bottleneck of any single-beam SEM is that the beam can only be in one place at a time. Scanning faster means fewer electrons per pixel and a noisier image. Multi-beam SEMs sidestep this trade-off by firing many electron beams simultaneously through a single column and detecting secondary electrons in parallel. The result is an imaging speed increase of close to two orders of magnitude compared with a conventional single-beam instrument.16PubMed Central. High-resolution, high-throughput imaging with a multibeam scanning electron microscope Multi-beam SEM has been described as a key technology for breaking through the bottleneck between resolution and throughput that has constrained volume EM efforts.17Chinese Physics B. Multi-beam scanning electron microscope (MBSEM): Technological evolution, core breakthroughs, and cross-field applications These instruments are already being used in large connectomics projects where mapping every synapse in a cubic millimeter of brain tissue would otherwise take decades with a single beam.
What Volume EM Has Made Possible
Neuroscience has been the marquee application. The goal of connectomics, mapping every neuron-to-neuron connection in a piece of nervous system, requires tracing fine axons and dendrites across volumes that may span hundreds of micrometers, a task that demands both the nanometer resolution of electron microscopy and the volumetric coverage of serial imaging. Volume EM methods have been assessed and adopted specifically for this purpose, with different techniques suited to different scales of circuit reconstruction.18PubMed. Volume electron microscopy for neuronal circuit reconstruction
Cell biology more broadly has also benefited. One study used ATUM-SEM combined with deep learning to reconstruct all endoplasmic reticulum, mitochondria, lipid droplets, lysosomes, peroxisomes, and nuclei in liver tissue, creating an unprecedented 3D map that allowed systematic analysis of how the ER physically contacts other organelles.19PubMed Central. Three-dimensional ATUM-SEM reconstruction and analysis of hepatic endoplasmic reticulum‒organelle interactions Those contact sites between organelles, often called membrane contact sites, are thought to be critical for lipid transfer and calcium signaling, but their full three-dimensional geometry had never been systematically characterized before volume EM made it feasible.
Correlating Light and Electron Microscopy in 3D
One limitation of electron microscopy is that it sees structure but not molecular identity. You can tell that something is a membrane-bound compartment, but you cannot tell what protein is sitting on it. Correlative light and electron microscopy (CLEM) bridges this gap by imaging the same sample first with fluorescence microscopy, which lights up specific molecules, and then with EM for ultrastructural context. Volume CLEM extends this to three dimensions, and it introduces a serious navigational challenge: after you have identified a fluorescent spot of interest in a light microscopy volume, you need to find the exact same spot in the much higher-resolution electron microscopy volume.
Recent workflows use software tools for multimodal image registration that segment common structures across both modalities and align the datasets in 3D, enabling researchers to target specific regions of interest for detailed EM acquisition.20PubMed Central. A workflow for semi-automated volume correlative light microscopy and transmission electron tomography For cryo-preserved samples, a device called a FinderTOP imprints a fiducial pattern onto the sample surface during high-pressure freezing, providing landmarks that survive all the way through cryogenic confocal imaging and into cryo-FIB-SEM, allowing accurate targeting from millimeter-scale light microscopy down to nanometer-scale EM.21Communications Biology. Precise targeting for 3D cryo-correlative light and electron microscopy volume imaging of tissues using a FinderTOP
The Special Headaches of Plant Tissue
Most volume EM protocols were originally developed for animal tissues, especially brain, which is relatively homogeneous and cooperates reasonably well with standard fixation and staining. Plants are a different story. Large air-filled voids, a rigid cell wall, a waxy cuticle that resists penetration by fixatives, and an enormous central vacuole all conspire to make uniform fixation, staining, and resin infiltration harder.22PubMed. A conventional fixation volume electron microscopy protocol for plants Researchers have adapted chemical fixation protocols originally designed for large brain volumes to address these plant-specific obstacles, but the field remains younger and more challenging than its animal-tissue counterpart.
A newer approach avoids chemical fixation entirely. By isolating plant protoplasts (cells with their cell walls enzymatically removed) and rapidly plunge-freezing them in liquid ethane, researchers have demonstrated cryogenic volume EM of whole vitrified plant cells, capturing their ultrastructure without any chemical fixation, dehydration, resin embedding, or heavy-metal staining.23PubMed. Application of cryo-FIB-SEM for investigating ultrastructure in guard cells of higher plants Cryo-FIB-SEM on plant guard cells from fava bean has shown subcellular detail in what is effectively a near-native, unaltered state, representing a significant step beyond earlier cryo-FIB-SEM work that was limited to simpler organisms like algae.
Volume EM Beyond Biology
Though the life sciences have driven most of the technique’s development, volume EM has found a solid foothold in materials science. Battery research is a telling example. The three-dimensional arrangement of active particles, binder, and pore space inside a lithium-ion battery cathode determines how well the battery performs and how quickly it degrades. FIB-SEM tomography can reconstruct that microstructure in 3D, revealing changes caused by repeated charge-discharge cycling that would be invisible in a 2D cross-section.24Journal of Power Sources. Three-dimensional investigation of cycling-induced microstructural changes in lithium-ion battery cathodes using focused ion beam/scanning electron microscopy FIB-SEM has been described as an indispensable tool for battery research precisely because it enables multiscale analysis from macroscopic structure down to nanoscale features.25PubMed. FIB-SEM: Emerging Multimodal/Multiscale Characterization Techniques for Advanced Battery Development More recent work has integrated machine learning into the reconstruction pipeline to segment battery electrode phases automatically from low-voltage FIB-SEM images, reducing the manual labor involved in turning raw image stacks into quantitative microstructural models.26Advanced Engineering Materials. Microstructure Reconstruction in Battery Electrodes Using Machine Learning Based on Low‐Voltage Focused Ion Beam–Scanning Electron Microscopy Tomography Images
How Volume EM Compares with X-Ray Tomography
X-ray nano-computed tomography (nano-CT) is the most common alternative for non-destructive 3D imaging at high resolution. The two approaches complement each other more than they compete. Nano-CT can survey a volume roughly twelve times larger than FIB-SEM in a single scan, and because it uses X-rays rather than a milling beam, the sample survives intact and can be reimaged or further processed afterward.27Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms. Combining X-ray Nano Tomography with focused ion beam serial section imaging — Application of correlative tomography to integrated circuits The trade-off is resolution and contrast: FIB-SEM typically delivers finer spatial detail and stronger material contrast, and plasma-source FIB-SEM systems have been shown to reconstruct larger volumes than X-ray CT while maintaining superior resolution.28PubMed. Comparison of large-volume 3D reconstruction using plasma FIB-SEM and X-ray CT In practice, some groups use X-ray tomography first to scout a large volume non-destructively, identify regions of interest, and then target those regions with FIB-SEM for high-resolution 3D imaging.
Managing Petabytes of Data
A single volume EM dataset from a modern FIB-SEM or connectomics project can easily run into the terabytes, and multi-instrument efforts are pushing into the petabyte range. Storing, sharing, and streaming datasets this large is a genuine infrastructure challenge. Traditional image file formats were not designed for data at this scale, and downloading an entire dataset just to look at one region is impractical.
The community has converged on the OME-Zarr format as a solution. OME-Zarr splits datasets into small, manageable chunks that can be streamed on demand from cloud storage or a file server, so a researcher on the other side of the world can zoom into a specific organelle in a specific cell without downloading the full volume. The format also bundles essential metadata like multiresolution pyramid descriptions, making it easier to navigate datasets at different zoom levels.29PubMed. Toward scalable reuse of vEM data: OME-Zarr to the rescue Open-access atlases built on this infrastructure are beginning to change the way volume EM data is published and reused, turning raw image stacks into community resources rather than single-lab assets.
Cryo Volume EM and the Push Toward Native-State Imaging
Most volume EM workflows involve chemical fixation and heavy-metal staining, which preserve ultrastructure well but inevitably alter it to some degree. Cryo-FIB-SEM aims to skip those steps entirely by imaging samples that have been rapidly frozen and kept frozen throughout the entire milling and imaging process. Biological tissues preserved this way remain hydrated and close to their living state, avoiding the extraction of lipids and other artifacts of chemical preparation. The technique has been demonstrated on tissue volumes exceeding 10,000 cubic micrometers at resolutions between 5 and 20 nanometers.30PubMed. Cryo-FIB-SEM serial milling and block face imaging: Large volume structural analysis of biological tissues preserved close to their native state
Cryo volume EM is still harder and slower than room-temperature methods. Keeping a sample vitrified while milling and imaging demands specialized cryo-stages, and ice contamination is a constant risk. But for questions where the precise arrangement of membranes, proteins, or lipid droplets in their native configuration matters, the payoff is hard to match with any preparation method that dehydrates or embeds the sample in resin. As cryo-FIB-SEM hardware matures and imaging throughput climbs, this approach is likely to become the reference standard for studies where fidelity to the living state is the primary concern.