In my book, Camera Lenses: From Box Camera to Digital, Chapter Two is entitled Films and Emulsions. What follows here is an expanded and augmented version of Section 2.5, How a Photographic Emulsion Works.

Many photographers do not clearly understand what happens inside a photographic emulsion when it is exposed to light. They do not know about latent images and related photographic effects. So here I give first a discussion of the formation of latent and developed images. This is followed by an explanation of reciprocity failure and some of its consequences.

This discussion is recommended to anyone who is curious about emulsions and wants to know how they operate. Although what I say is generally applicable to ordinary photography, much can be understood best by examining the effects that astronomers encounter when recording faint images of stars and nebulae at the telescope.

The discussion applies only to classical photochemical photography. It does not include any effects encountered with the new photoelectric image detectors such as CCD and CMOS sensors.

Latent Image Formation

The light-sensitive part of a photographic emulsion consists of a myriad of tiny (about a micron) crystals of silver halide (mostly bromide). These crystals are suspended in a medium consisting mainly of a very pure form of gelatin (the same stuff that is in Jell-O). The resulting emulsion is thinly spread on a supporting substrate such as a glass plate or plastic film. Actual photographic products have additional constituents and features, but this is the essential form.

The atoms in a silver halide crystal exist as ions. Each bromine is a negative ion with an extra electron. Each silver is a positive ion with one electron missing. The combination, of course, has no net electrical charge.

During an exposure to light, a photon event occurs when an incoming photon knocks off the extra electron from one of the bromine ions. This photoelectron is now free and mobile, and it wanders about in its crystal until typically it finds a dislocation or flaw in the crystal lattice, where it becomes trapped.

A few of the silver ions are not locked in the crystal lattice, and these interstitial ions are also mobile. The negative electric field set up by the trapped photoelectron draws one of these positive silver ions over to it. When the two meet, the ion and the electron combine to form a neutral silver atom. This silver atom is the beginning of a latent image.

As the exposure continues, the same process is repeated several times until, in at least one place in the crystal, a compact group or aggregate of about four to six neutral silver atoms is eventually created. When this happens, a threshold is crossed. This clump of silver atoms is now large enough that it has become a fully formed latent image.

A silver halide crystal containing a fully formed latent image has become developable. Developable means that when an exposed plate or film is placed in a developing solution, the latent image aggregate acts as a development center. Starting around the clump of silver atoms, the chemical action of the developer progressively converts the whole crystal into a grain of metallic silver. The developer can be thought of as a chemical amplifier that multiplies the size of the original clump of silver atoms formed by light.

If the recommended development procedure is followed, and if the crystal contains no latent image aggregate of sufficiently large size, then the developer has no effect on that crystal. For each individual crystal, it is either a go or a no-go situation. Those silver halide crystals that are not converted into silver metal during development are removed from the emulsion during the fixing process that follows.

The final overall developed photographic image is thus a vast array of tiny silver grains (which are largely opaque) in the gelatin layer (which is otherwise transparent). Areas that received heavier exposure had a greater fraction of its crystals made developable and later converted into metallic silver.

The invisible latent image has thus been converted into a visible developed image. When the developed plate or film is viewed by transmitted light, more silver grains absorb more light. Thus heavily exposed areas appear darker. The result is the familiar photographic negative.

The Law of Reciprocity

Ideally, a given amount of light should produce the same photographic effect regardless of whether it is bright and arrives over a short period of time, or is faint and arrives over a long period of time. This yields the "law" of reciprocity, which states that the photographic effect of an exposure should be proportional to the product of the light intensity (image irradiance in photons/mm²/second) multiplied times the exposure duration (in seconds). In particular, the photographic effect should be independent of the individual values of intensity and exposure duration.

In normal photography, this law holds reasonably well. Photographers know that 1/250 second at f/4 yields the same exposure as 1/60 second at f/8. But very high and very low light intensities requiring very short or very long exposure times are a different matter. At these extreme intensities, there are significant departures from this law. These departures are referred to as failure of the law of reciprocity, or reciprocity failure.

It is interesting that the most widely accepted explanation of photographic reciprocity failure is completely wrong. This misconception asserts that during a long exposure (such as in astronomy), the sensitivity of an emulsion decreases, that is, the emulsion speed slows down over time. Somehow the ISO/ASA rating supposedly goes from, say, 1000 at the beginning of an exposure to something like, say, 200 or 100 or 50 after an hour or so.

This description of what happens is so widespread and repeated so often that few question it. However, the theory of latent image formation refutes it. Something else is happening. The correct explanation is actually quite different.

Low-Intensity Reciprocity Failure

The formation of a developable latent image, as described above, actually occurs in two distinct stages. In the early part of the first stage, the clump of neutral silver atoms contains only one or two atoms. At this time, the process forming the latent image is reversible. The silver aggregate is unstable because thermal agitation (Brownian motion) can reionize one of these atoms. If this happens, then the reionized atom is lost from the latent image.

Not until the clump contains three or four atoms is it finally stable against thermal disintegration. When this number is reached, this stable, but not yet developable, latent image is called a latent sub-image.

During the second stage, the stable latent sub-image is converted into developable latent image. To do this, more exposure is necessary. The additional photoelectrons add one or two more silver atoms to the growing silver clump to finally give it the required threshold number of four to six atoms.

Thermal instability during the first stage of latent image formation has consequences for photography of faint objects. Because the photons arrive slowly during the long exposure, there is plenty of time for the reionization process to work. The potential latent sub-image can therefore fall apart almost as fast as it is being formed.

Not only are there fewer photons/mm²/second at low light levels, but the emulsion's net response to these photons is less. In other words, the efficiency of the process is reduced. This loss of efficiency (and thus sensitivity) at low intensities is the cause of low-intensity reciprocity failure.

Note carefully that this process is unchanged throughout the length of an exposure. In fact, the properties of an emulsion remain (nearly) the same from the time of its manufacture until it is placed in the developer. The emulsion has no built-in clock that decreases its responsiveness with time.

The controlling variable is the intensity (brightness, irradiance) of the incident light. Reciprocity failure is a light intensity effect, not an exposure duration effect.

It is of further interest that reciprocity failure can be occurring differently at different locations across the emulsion at the same time. Under the image of a bright star, emulsion efficiency is relatively high. But simultaneously in a nearby region under the image of a fainter star, emulsion efficiency is lower.

These differences in efficiency yield a rule of thumb: if it is desired to photograph stars one magnitude (2.512 times) fainter than those that can be barely recorded with a given exposure, then the exposure length must be increased about three times (neglecting the sky fog limit).