For a hundred years, food safety had one answer: heat. Cook the meat, kill the pathogens, done. Raw feeding removes that answer on purpose, because heat also degrades taurine and denatures the proteins a raw diet exists to preserve. That leaves a real engineering question. How do you kill Salmonella inside a product whose whole identity is that nothing was cooked? The most precise answer in use today is an electron beam.
An electron beam is electricity, not a radioactive source
Start with what the machine is. An electron accelerator takes ordinary electrons, the same particles that flow through a wall socket, and speeds them to near light speed inside a vacuum tube. Aimed at a sealed pack of food, the stream passes through and collides with whatever it meets. There is no radioactive material anywhere in the building. Switch the machine off and it is as inert as the X-ray unit at a dental clinic.
This is the first separation from the technology most people picture when they hear the word irradiation. Gamma sterilisation uses Cobalt-60, an isotope produced in a nuclear reactor, which emits radiation continuously and cannot be switched off. The two share a word. They do not share a power source, a dose rate, or a chemistry.
What the beam does inside the pack
Bacteria die when their DNA breaks faster than they can repair it. Accelerated electrons break DNA. The physics favours the food: a Salmonella genome is an enormous target compared with a vitamin molecule, so the beam wrecks bacterial DNA at doses far below the level that damages nutrition. Treatment takes seconds, not the hours a gamma cycle needs. And e-beam processing is not an oxidising process. The flood of incoming electrons pushes the chemistry toward reduction, which is why the fat rancidity documented under gamma treatment does not carry over.
The numbers
Dose is measured in kilograys, written kGy. The US FDA regulation for irradiating animal diets sets the working arithmetic: 1 kGy removes one log, 90 percent, of a Salmonella population. Peer-reviewed work on ground chicken and turkey found 2 kGy reduced Salmonella and Campylobacter, and 3 kGy pushed total bacteria below the detection limit.
A raw pet food kill step runs at roughly 5 to 10 kGy, stacking a wide safety margin on top of those figures. For scale: the WHO assessed food treated up to 10 kGy as wholesome decades ago, and a joint WHO, FAO and IAEA study group later concluded that even doses above 10 kGy are safe and nutritionally adequate under good manufacturing practice. Hospitals feed high dose treated food to immunosuppressed patients. Astronauts eat it in orbit.
Sealed first. Treated second
Order of operations decides recontamination risk. A cooked product is made safe by heat, then cooled, handled and packed, and every step after the heat is a chance for bacteria to return. An e-beam product is sealed in its final vacuum pack, frozen, and then treated. The beam passes through the plastic. The pack a customer opens at home is the exact pack the electrons swept, with nothing touching the food in between. Zero handling after the kill step. That single fact carries most of the argument.
Proof, not description
Any process claim should survive a laboratory. Batch testing at an ISO/IEC 17025 accredited laboratory in Bangkok recently compared a Thai raw cat recipe before and after e-beam treatment below 10 kGy. Taurine measured 222 mg per 100 g before and 232 after. Vitamin A read 121 micrograms RAE before and 124 after. Peroxide value, the standard marker of fat oxidation, moved from 0.12 to 0.11 meq per kg. Thiamine dropped from 0.70 to 0.58 mg per 100 g, a loss of about 17 percent, which is exactly why serious raw formulas carry thiamine at several times the minimum before treatment. The US regulation says the same thing in drier words: treated feeds should be formulated to account for nutritional loss.
The cat expert team at Pau & Sun reads certificates before recommending anything. This is the kind of certificate worth asking every brand to produce.
Raw does not have to mean untreated. The kill step exists, the physics is public, and the results are checkable. Ask for them.
