Ionizing radiation & safety

An electromagnetic wave delivers its energy in lumps called photons (Chapter 6 tells that story). The energy of one lump depends only on the frequency: double the frequency, double the energy per photon. Brightness, or transmitter power, only changes how many lumps arrive each second.

That matters because the things a photon can do to a molecule come in steps. To ionize an atom you must hand one electron enough energy to leave in a single go. If the photon is short, the electron doesn’t bank the difference and wait for a second photon; it just jiggles and the energy turns into heat.

So there is a threshold, and it lives in photon energy. Hydrogen needs 13.6 electronvolts. Water, the molecule your tissue is mostly made of, needs about 12.6. Oxygen needs 13.6, nitrogen 14.5. Radiation-protection bodies therefore draw the ionizing line somewhere around 10 to 12 eV, which corresponds to wavelengths near 100 to 124 nanometers, deep in the far ultraviolet. The exact number depends on which molecule you care about, so treat it as a band, not a knife-edge.

For scale: one 2.4 GHz Wi-Fi photon carries 0.0000099 eV. The random thermal jostling of molecules in your body at 37 °C is worth about 0.027 eV per molecule, roughly 2,700 times more than a Wi-Fi photon. Radio photons are not just below the ionizing line; they are far below the background shuffling of warm water.

Non-ionizing radiation interacts with the body in one well-established way: heating. The oscillating electric field pushes on charges and twists polar molecules, water above all. Water molecules are lopsided, with a slightly negative oxygen end and slightly positive hydrogen end, so the field tries to rotate them back and forth billions of times a second. They lag behind, bump their neighbors, and the motion ends up as heat. That is called dielectric heating.

The safety number for this is the specific absorption rate, SAR: watts absorbed per kilogram of tissue. In the U.S. the FCC limits phones to 1.6 W/kg averaged over any 1 gram of tissue. Europe and most of the world use the ICNIRP limit of 2 W/kg averaged over 10 grams. Phones are tested at full power pressed against a model head, and the reported SAR is that worst case.

For perspective, your body at rest makes about 80 watts of heat, a little over 1 W/kg across 70 kg. The limits were set from experiments showing that whole-body absorption of about 4 W/kg raises core temperature by roughly 1 °C; the public whole-body limit of 0.08 W/kg is fifty times below that.

The microwave-oven myth. Ovens are often said to work because 2.45 GHz is “the resonant frequency of water.” It isn’t. Liquid water has no sharp resonance there; its absorption is a broad smear whose peak sits near 20 GHz at room temperature (the sharp water line at 22.2 GHz belongs to water vapor). Ovens use 2.45 GHz partly because it was an available ISM band and partly because it is off the peak: food absorbs it weakly enough that it penetrates a centimeter or two before being used up. At 20 GHz, the outside of your lasagne would burn while the middle stayed frozen.

Ultraviolet is the awkward neighbor. UV-A (315–400 nm) photons carry 3.1 to 3.9 eV, UV-B (280–315 nm) 3.9 to 4.4 eV, and UV-C (100–280 nm) 4.4 up to 12.4 eV. Most of that is below the ionizing line, yet UV-B is the main cause of sunburn and skin cancer. The reason is that you don’t need to ionize DNA to damage it; you only need DNA to absorb the photon.

DNA’s bases are ring-shaped molecules that absorb strongly near 260 nm, and the tail of that absorption reaches into UV-B. When two thymine bases sit next to each other on one strand, an absorbed photon can make them bond to each other, forming a thymine dimer (more generally a cyclobutane pyrimidine dimer). It kinks the helix and stalls the machinery that copies and reads DNA. Cells run a repair crew, nucleotide excision repair, that cuts out and rewrites the damaged stretch. Sunburn is largely the inflammatory response when damage overwhelms that crew; the dimers that slip through as mutations are what raise cancer risk.

The ozone layer absorbs essentially all solar UV-C and most UV-B, which is why the sunlight reaching the ground is mostly UV-A. Germicidal lamps put UV-C back on purpose: a low-pressure mercury lamp emits at 253.7 nm, 4.89 eV, close to DNA’s absorption peak, and it kills microbes by the same dimer trick. It is just as hard on your skin and eyes, which is why those lamps run in empty rooms or inside ducts. A newer “far-UVC” type at 222 nm is absorbed in the outer dead layer of skin and the tear film, and is being studied for occupied rooms.

On 8 November 1895, Wilhelm Röntgen noticed a fluorescent screen glowing across his darkened lab while he ran a cathode-ray tube wrapped in black card. Within weeks he had photographed the bones in his wife Anna Bertha’s hand. He called the rays “X” because he didn’t know what they were. They turned out to be light, with photons tens of thousands of times more energetic than visible ones.

A modern X-ray tube works the same way. A heated filament boils off electrons; tens of thousands of volts accelerate them into a tungsten target. As each electron swings past a heavy nucleus it is sharply deflected and decelerated, and an accelerating charge radiates (Chapter 4). That is bremsstrahlung, German for “braking radiation,” and it gives a continuous spread of energies up to a hard maximum: an electron accelerated through 100 kV can produce at most a 100 keV photon. Knocked-out inner electrons in tungsten add sharp “characteristic” lines on top. The process is wasteful: under 1% of the beam energy becomes X-rays, the rest heat, which is why tube anodes spin.

Bones show because of the photoelectric effect. Its likelihood rises steeply with atomic number, roughly as Z³, and falls with photon energy. Bone is rich in calcium (Z = 20) and phosphorus (15); soft tissue is mostly hydrogen, carbon, nitrogen and oxygen (Z ≤ 8). So at diagnostic energies of a few tens of keV, bone stops far more of the beam and casts a pale shadow on the detector.

Gamma rays are the same kind of photon, named for where they come from: a nucleus dropping from an excited state to a lower one, just as an atom’s electrons emit visible light when they drop between levels, only with about a million times the energy. Cesium-137 emits at 662 keV; technetium-99m, the workhorse of nuclear medicine, at 140 keV. PET scans use positron emitters such as fluorine-18: each positron meets an electron, both vanish, and their rest-mass energy reappears as two 511 keV photons flying in opposite directions. Detecting both at once pins down a line through the body.

Absorbed dose is energy deposited per kilogram, measured in grays: 1 Gy = 1 joule per kilogram. A gray is a small amount of energy, enough to warm water by about a quarter of a thousandth of a degree, but delivered as ionization it is a large biological dose.

Not all radiation does equal harm per joule. Alpha particles pack their ionization densely and are weighted 20 times; photons and electrons are weighted 1. Weighting the absorbed dose by radiation type gives equivalent dose, and further weighting by how sensitive each organ is gives effective dose, both in sieverts. For X-rays and gamma rays, 1 Gy to the whole body is 1 Sv. Everyday doses are millisieverts (mSv) and microsieverts (µSv).

Typical figures, all approximate and varying with machine, technique and body size: the U.S. average natural background is about 3 mSv a year, roughly 2.3 of it from radon gas (NCRP Report 160). A two-view chest X-ray is about 0.1 mSv, roughly twelve days of background. A transatlantic flight adds about 0.04–0.08 mSv from cosmic rays. An abdomen-and-pelvis CT scan is about 8–10 mSv, three years’ worth.

Limits: U.S. radiation workers may receive 50 mSv a year (10 CFR 20); the ICRP recommends 20 mSv a year averaged over five years. The public limit from regulated sources is 1 mSv a year on top of background. Acute radiation sickness needs roughly 1,000 mSv delivered at once.

Below about 100 mSv, extra cancer risk is too small to measure directly against the roughly 40% lifetime chance everyone already has. Regulators assume a linear no-threshold model, about 5.5% extra lifetime cancer risk per sievert (ICRP 103), as a cautious planning tool. On that assumption a 10 mSv CT adds about 1 chance in 1,800; whether risk really stays linear at such low doses is debated.

Radiation workers learn three words: time, distance, shielding. Time is obvious: half the time, half the dose. The other two are where physics does the heavy lifting.

Distance. Photons leaving a small source spread over the surface of an ever-larger sphere, whose area grows as the square of the radius. Step from 1 meter to 2 meters and the same photons are spread over four times the area, so the dose rate falls to a quarter. At 3 meters, a ninth. That is the same law that makes radio signals fade (Chapter 8). It only holds for a source small compared with the distance; close to a big, spread-out source, the drop is gentler.

Shielding. Each layer of material removes a fixed fraction of the photons, not a fixed number, so the beam fades exponentially. The handy unit is the half-value layer: the thickness that halves it. For 662 keV gamma rays from cesium-137 that is about 0.55 cm of lead (practical tables often say 0.65 cm, allowing for scatter), or about 4 cm of concrete. Two half-value layers leave a quarter; ten leave about a thousandth. For low-energy X-rays the numbers collapse: a 0.25–0.5 mm lead-equivalent apron stops most of a diagnostic beam scattered off a patient, because lead’s photoelectric absorption is enormous at those energies.

In May 2011 the International Agency for Research on Cancer (IARC) classified radiofrequency electromagnetic fields as Group 2B, “possibly carcinogenic to humans.” That headline gets quoted a lot, usually without what 2B means.

IARC groups rate the strength of evidence that something can cause cancer, not how much cancer it causes. Group 2B means limited evidence in humans and less than sufficient evidence in animals: some studies raised a question that couldn’t be ruled out. The same group contains aloe vera whole-leaf extract and traditional Asian pickled vegetables. The 2011 decision leaned mainly on case-control studies of heavy phone users, where people with brain tumors were asked to recall their phone use years later, a design prone to recall bias.

Since then: the large international Interphone study found no overall increase in brain tumors, with an odd signal only in the heaviest-use group that its authors flagged as possibly biased. Big prospective cohorts, such as Danish subscribers and the UK Million Women Study, found no increase. Brain-tumor incidence in many countries has stayed flat through decades in which phone use went from rare to universal. A systematic review commissioned by the World Health Organization and published in 2024 concluded, with moderate certainty, that mobile phone use is not associated with brain cancer.

On the other side of the ledger: a 2018 U.S. National Toxicology Program study found more heart schwannomas in male rats exposed to 900 MHz radiation nine hours a day for two years, at whole-body SAR of 1.5 to 6 W/kg, far above what people receive. Exposed male rats also outlived the controls, which complicates the reading, and female rats and mice showed no clear effect. The science on very long, very heavy use is still accumulating, and IARC has listed RF fields as a priority for re-evaluation.

Wi-Fi exposures are smaller still. A 100 mW router radiating evenly gives about 0.008 W/m² at one meter while transmitting, roughly a thousandth of the 10 W/m² public limit, and routers transmit only a fraction of the time. Physically, there is no known mechanism by which such photons could break bonds, and heating at these levels is unmeasurably small. The fair summary: no established harm below the limits, a plausible-mechanism problem for anyone claiming otherwise, and continued study because “we haven’t found anything” is never quite the same as “proved impossible.”