Some examples -
(1) Compendium Methods for Microbiological examination of Foods, 4th Ed.
Frozen bulk foods may be sampled with sterile corers, augers and other sharp sampling instruments. A presterilised auger or hollow tube may be used to obtain enough material for analysis.
Frozen samples should be kept frozen until arrival at the laboratory. Thawing and refreezing of samples must be avoided.
For larger solid food samples – frozen or unfrozen, test units should betaken aseptically from several areas using sterile knives and forceps. These portions should be mixed as a composite to provide a sample representative of the food to be evaluated.
Use aseptic technique when handling product. Before handling or analysis of sample, clean immediate and surrounding work areas. In addition, swab immediate work area with commercial germicidal agent. Preferably, do not thaw frozen samples before analysis. If necessary to temper a frozen sample to obtain an analytical portion, thaw it in the original container or in the container in which it was received in the laboratory. Whenever possible, avoid transferring the sample to a second container for thawing. Normally, a sample can be thawed at 2-5°C within 18 h [!!]. If rapid thawing is desired, thaw the sample at less than 45°C for not more than 15 min. When thawing a sample at elevated temperatures, agitate the sample continuously in thermostatically controlled water bath.
When sampling block frozen products (e.g., frozen blocks of fruit), try taking samples before the company freezes the product (this is the easiest way to sample and it allows the company to top up the container before they freeze it). If this is not possible, try taking samples when the company intends to use the product (i.e., take the sample from the product being thawed by the company).
Otherwise, use a sterile sharp utensil (e.g., a knife of chisel) to break pieces off the block to obtain the samples. You may need to let the product thaw just a little in order to be able to take the samples. Thawing could potentially harm the product by increasing the microbiological load if not done properly.
My own experience with "difficult" samples is similar to (3) but primarily via scalpels. [IMO (1) looks ideal if you have the equipment.]