Getting a sample into an SEM that doesn't ruin your day
Scanning Electron Microscopy Sample Preparation
The whole point of sample prep is making something that can sit in a vacuum, handle an electron beam, and not charge up or melt while you're trying to get an image. That sounds simple until you're looking at a biological tissue section or a porous polymer and realizing it's doing neither of those things well. Your sample needs to be dry. Water creates problems at every stage - outgassing in the column, ice crystals forming under the beam, and charging when you finally get to imaging. For anything biological, chemical, or hydrated, dehydration is step one. You can do this through a graded ethanol series starting at 30% and going up to 100%, typically spending 10 to 15 minutes per step depending on the sample size. Some labs use critical point drying with CO2 to avoid surface tension damage during the drying process. If you skip critical point drying for delicate structures, you will watch your features collapse as the liquid evaporates. It is painful to witness. Mounting is where most people waste time. Your sample has to be level on the stub, and the stub has to make solid electrical contact with the stage. Carbon tabs work for quick mounts and light samples. For anything heavier or requiring stable imaging over time, epoxy and a copper stub are better. I used to use double-sided carbon tape exclusively and kept getting intermittent charging issues at high magnification. Switching to silver-loaded epoxy for the mount completely eliminated that problem. The conductive path from sample through epoxy to stub matters more than people realize.
Conductivity depends entirely on what you're imaging. Metals and graphite don't need coating. Semiconductors often don't either if you're working at low kV. Everything else needs a thin conductive layer. Sputter coating is the standard approach - gold, gold/palladium, platinum, or carbon are the usual choices. The thickness matters more than the material. A 5 nanometer gold coating on a high-resolution scan of a semiconductor cross-section will round off your finest features. I learned this after ruining three FIB-prepared cross-sections by sputtering at too high a rate. Using a fine-coat carbon sputter at 0.5 nanometers per second instead fixed the resolution without adding visible coating artifacts. For non-conductive samples that need thicker coatings, say 10 to 20 nanometers, carbon coating is preferable to metal. Metal coatings create crystalline artifacts under the beam that show up as spotty contrast. Carbon is amorphous and gives a uniform conductive layer. The downside is lower conductivity, so you may need to lower your accelerating voltage or work at a slightly higher beam current to compensate. Cross-section preparation deserves its own consideration. If you're doing failure analysis or looking at layer structures, cleaving is free and fast but rarely gives you a clean surface on anything hard or composite. Mechanical polishing introduces scratches and debris. Focused ion beam milling gives you the best surface but damages the near-surface layer with gallium implantation unless you use a low-energy final pass. I usually mill at 30 kV down to about 2 microns and finish with a 5 kV cleaning pass that removes roughly 200 nanometers. This leaves a region clean enough for imaging at 5 kV or below without significant beam damage artifacts.
One thing nobody tells you about sample prep: your environment matters more than you think. A lab with poor HVAC control will produce inconsistent results on hydrophilic samples because the absorption of ambient moisture before coating varies day to day. I once spent three days troubleshooting charging on polymer samples only to realize the humidity in the coating room had dropped from 45% to 18% between Monday and Wednesday. The drier samples absorbed less moisture but actually charged worse because the surface resistivity increased. Keeping humidity at 40 to 50% in both the preparation area and coating chamber makes a measurable difference for sensitive materials. There are cases where sample prep simply cannot save you. Highly heterogeneous composites with phases that have vastly different sputtering rates will develop topographic artifacts during coating that mimic real structure. Soft polymers that deform under the beam even at low kV need cryo preparation or specialized resin embedding to hold their shape. Biological samples larger than about 1 millimeter after fixation rarely survive the dehydration process intact - they crack and distort. In those cases, serial block face imaging or thin sectioning through resin is the only option, and that adds considerable time to the workflow. The bottom line is that good SEM imaging starts before the sample ever goes into the microscope. Budget realistic time for preparation - a proper metallurgical cross-section with mounting, polishing, and coating takes about 90 minutes if you know what you're doing. A biological sample through fixation, dehydration, critical point drying, and coating might take half a day. Rushing prep to save time guarantees you will spend more time troubleshooting charging artifacts, beam damage, or poor contrast on the microscope itself.
Get the Full Details
