Level 2 · PractitionerLessonPart 04 · page 4 of 975 minScience

75Minutes

7Chemicals

13Sources

Chemicals on this page7

The Latent Image

Open the camera back after an exposure and the film looks exactly as it did before. Weigh it: no change you could measure. Look at it under any microscope you own: no change you could see. And yet something is different, because a developer can tell which crystals were struck by light and which were not, and it can tell in the dark, minutes or months later. Whatever that difference is, it is small enough to be invisible and durable enough to survive a journey home. This page is about what it actually is.

What a photon does when it arrives

A photon is absorbed only if it carries enough energy to do something. In a semiconducting solid, the something is lifting an electron out of the valence band, where it belongs to a particular ion, into the conduction band, where it belongs to the crystal as a whole and can travel. The energy needed is the band gap, and Hamamatsu’s detector note gives the conversion between an energy and the longest wavelength that supplies it:

λ = 1240 / E

Threshold wavelength from threshold energy

with λ in nanometres and E in electronvolts. A 450 nm blue photon carries 2.76 eV; a 550 nm green photon, 2.25 eV; a 650 nm red photon, 1.91 eV. Below the threshold, nothing happens however many photons arrive — which is the whole reason a red safelight exists.

Absorption creates two particles, not one, and this is the point at which most short accounts go quiet. The electron goes up into the conduction band. What it leaves behind — a missing electron in a full band — behaves as a mobile positive charge and is called a hole. Both of them then have to go somewhere, and the photograph depends on the fates of both.

The band picture, with the two particles a photon makes

Valence band, full — the electrons belong to particular ions and cannot travel
Conduction band, empty — an electron here belongs to the whole crystal
The photon — absorbed only if it carries at least the gap energy
The electron — travels to a trap and stops there
The hole — travels to the surface; its fate decides whether any of this survives
Where a sensitising dye would sit — drawn as a dashed level; owned by the spectral sensitivity page

Energy increases upwards and the axis carries no numbers, deliberately: no band gap for any silver halide has been verified for this course, and a diagram with a fake number on it would be worse than one with none.

Following the hole, which most accounts do not

The hole is the half of the story that decides whether the other half survives. Ware’s summary of the mechanism gives it a step of its own: the positive hole can migrate to the surface of the crystal, where it takes an electron back from a halide ion, leaving a neutral halogen atom.

Br⁻ + hole → Br

What the hole does when it reaches the surface

And then the sentence that matters: that halogen atom can subsequently be removed if some kind of halogen acceptor is in the environment. If not removed, the free halogen may reverse the process by re-oxidising the silver atoms.

Read that twice. The silver atoms you have just made are sitting next to a fresh oxidiser. Photolysis is not a one-way reaction; it is an equilibrium with a strong back-reaction, and the entire art of making a photographic material is the art of tilting it.

What actually takes the bromine

Ware is precise about which acceptor works where, and the distinction is easy to get wrong.

In a development emulsion, gelatin is the important halogen acceptor. It surrounds every crystal, and the amount of halogen a latent image liberates is, in his words, exceedingly minute — a few atoms per crystal. Gelatin has no difficulty absorbing that.

At print-out exposures, gelatin is not an effective scavenger. He says so directly, and the consequence is that a printing-out paper cannot rely on it: there, the acceptors are the excess silver ions and the water around the crystals, which between them consume the halogen and, incidentally, re-form more silver halide.

So the same substance is sufficient for one job and insufficient for the other, and the difference is three or four orders of magnitude in how much halogen has to go somewhere. That is why the experiment two pages back stalls: a test tube of water has almost no acceptor in it at all.

Why a pure crystal never goes black

Pure crystal, no halogen acceptor
With a halogen acceptor present

Show the numbers behind this plot

Two rising curves against exposure. The lower one, labelled pure crystal with no halogen acceptor, rises steeply at first and then flattens completely at an optical density of about 0.02, which is a barely perceptible greying; a horizontal guide line marks that ceiling. Ware gives the corresponding limiting yield as 1.2 times ten to the nineteen silver atoms per square metre and the particle size as not exceeding about 10 nanometres. The upper curve, labelled with a halogen acceptor present, follows the same path at first but does not flatten: it continues climbing past the guide line and off the top of the plotted range, because the liberated halogen is being removed and cannot re-oxidise the silver. A note records that only the 0.02 ceiling is a sourced number and that the shapes of both curves are drawn to teach rather than measured.
Series	Exposure, relative	Optical density of photolytic silver
Pure crystal, no halogen acceptor	0.00	0.00
Pure crystal, no halogen acceptor	0.50	0.01
Pure crystal, no halogen acceptor	1.00	0.01
Pure crystal, no halogen acceptor	2.00	0.02
Pure crystal, no halogen acceptor	3.00	0.02
Pure crystal, no halogen acceptor	4.00	0.02
Pure crystal, no halogen acceptor	6.00	0.02
Pure crystal, no halogen acceptor	8.00	0.02
Pure crystal, no halogen acceptor	10.00	0.02
With a halogen acceptor present	0.00	0.00
With a halogen acceptor present	0.50	0.01
With a halogen acceptor present	1.00	0.02
With a halogen acceptor present	2.00	0.04
With a halogen acceptor present	3.00	0.06
With a halogen acceptor present	4.00	0.07
With a halogen acceptor present	6.00	0.10
With a halogen acceptor present	8.00	0.14
With a halogen acceptor present	10.00	0.17

Only the 0.02 ceiling is a sourced number: Ware gives the limiting yield of photolytic silver in a pure silver halide crystal as 1.2 × 10¹⁹ atoms per square metre, corresponding to an optical density of that order, with the particles growing no larger than about 10 nm. Both curve shapes are drawn to teach and were not measured. The curve is drawn to show the shape, not measured from a real material. Your own materials will differ, and measuring them is what the sensitometry part of the course is for.

Recombination, and a figure this page will not give

The other way the exposure can come to nothing is that the electron and the hole find each other again before either does anything useful. Recombination is real and it is one of the reasons a photographic material is nothing like a perfect converter of photons into image.

The course has read no source that says what fraction of absorbed photons is lost that way, and it therefore quotes none. What it can state, from Ware, is the mechanism it has read: that the halogen, if it is not taken up, re-oxidises the silver atoms, and that irradiating a pure crystal therefore gets you almost nowhere. Whether the dominant loss in a real emulsion is recombination or the back-reaction is a question this page leaves open rather than settles with a plausible sentence.

Where the electron stops

An electron in the conduction band travels until something traps it. Ware’s account names the class: a trap in the lattice, either a defect or an impurity centre, where the electron reduces a silver ion to an atom of the metal.

The specific defects usually named in the literature — kinks in a crystal edge, steps, dislocations — are not named in any source this course has read, so the page names the class rather than the species and moves on to the traps that are documented, which are the ones an emulsion maker puts there on purpose.

What a trap has to do

Nothing below is quoted from a source; it is what the mechanism above requires, and it is set out separately for that reason. A useful trap has to satisfy three conditions at once, and they pull against each other.

It has to hold the electron long enough. The next step is not electronic but ionic: a silver ion has to travel through the lattice and arrive. That takes far longer than an electron takes to cross a crystal. A trap that releases its electron before the ion gets there has done nothing, and the electron goes back to wandering and eventually recombines.

It has to let go afterwards. Once the silver atom has formed, the site must be able to accept the next electron. A trap that permanently buries the charge is a dead end.

It has to be somewhere useful. A developer works from the outside in, so a cluster grown on the crystal’s surface is one the developer can attack and a cluster grown in the interior is not. This is the condition an emulsion maker has most control over, because sensitivity centres are put on the surface deliberately, whereas the accidental lattice defects are wherever the crystal happened to grow them.

Read those three together and Sheppard’s result stops being surprising. A crystal with no deliberate centres is not insensitive because it cannot absorb light; it is insensitive because the electrons it frees have nowhere good to stop, and the clusters that do form are in the wrong places.

Sensitivity centres, in the words of the man who found them

Samuel Sheppard’s Eastman Kodak patent, filed in June 1924 and published in March 1926, is the primary document for sulfur sensitisation, and it is worth reading for its caution as much as its content. It records that the sensitiveness of emulsion grains “has been shown to correspond to the presence in said grains of nuclei of unstated chemical composition” — an admission that the thing had been detected before it was identified.

His own compounds are ones carrying a divalent sulfur-group atom double-bonded to a single metalloid atom, with thiosinamine — allyl thiourea — singled out as especially useful. And he states the mechanism as an attribution rather than a proof: he attributes their sensitising action “to their forming in the emulsion grains small, mostly ultramicroscopic, nuclei” which include silver combined with sulfur, selenium or tellurium, “such as silver sulfid”.

The control that makes the argument is the best part. Take a gelatin, oxidise the sulfur compound that occurs naturally in it, and you get, in the patent’s words, emulsions of impractically low light-sensitiveness. The sensitivity was never in the halide alone; a good part of it was an accidental impurity in an animal product.

And then gold, twenty years later

Ilford’s 1946 patent, by Waller, Collins and Dodd, adds the second sensitiser. Small quantities of a soluble gold salt, added before or during digestion at a pH not greater than about 8, give “a very considerable increase in the light sensitivity of the final emulsions”. Three details from it are worth carrying:

The gold does not work alone. The patent states that all normal gelatin contains small proportions of sulfur compounds such as thiosinamine or allyl isothiocyanate, that their presence is believed essential to the gold effect, and that the inert gelatins benefit from a positive addition of one. Sulfur and gold sensitisation are a partnership, not alternatives.
Adding a little thiocyanate along with the gold improves it further.
The quantity of gold is far too small to change the tone of the developed image — and too much gold loses the speed gain rather than increasing it.

The relative-speed tables in that patent are badly mangled in the scanned text, and this course quotes no number from them.

The Gurney–Mott alternation

Now the pieces fit together. The electron is caught at a trap and reduces a silver ion there. The lattice, because it carries Frenkel defects, has mobile interstitial silver ions available. So the process alternates: an electronic step, then an ionic step, then another electronic step, at the same trap, each one adding an atom.

Ware sets out the mechanism first proposed by Gurney and Mott in four numbered steps, and this is the account the course teaches:

The mechanism, as Ware states it

Absorption of a photon excites an electron from the valence band to the conduction band, and this photoelectron migrates to a trap in the lattice — a defect or an impurity centre — where it reduces a silver ion to an atom of the metal.
Subsequent photoelectrons and mobile interstitial silver ions migrate alternately to the same trap, enhancing the cluster of silver atoms, the “sensitivity speck”.
The positive hole generated by the same electron transfer can migrate to the surface, where it reduces a halide ion to a halogen atom, which can be removed if a halogen acceptor is present — and which, if it is not removed, may reverse the process by re-oxidising the silver atoms.
To form a latent image development centre, only a small cluster of silver atoms — thought to be a minimum of four — is required. That cluster then provides the point of attack for the developing agent, which converts the entire sensitised crystal into metallic silver.

Ag⁺ + e⁻ → Ag

The step that is repeated: the same half-reaction as the stain and the image

Six frames: how a latent-image cluster is built

A photon is absorbed — only if it carries at least the gap energy
Electron up, hole down — both mobile, travelling in opposite directions
Electron trapped; halogen removed — the acceptor step, without which the next frames run backwards
An interstitial silver ion arrives — the ionic half of the alternation; one silver atom now exists
A cluster of about four atoms — this is the latent image; nothing about the crystal looks different
Development — the cluster is the point of attack; the whole crystal goes to silver

Frames 1 to 5 take well under a second and leave nothing visible. Frame 6 happens minutes later in a tray. The threshold of four atoms is Ware's figure, and he gives it as what the minimum is thought to be.

Whose mechanism is that, and how settled is it?

This is the place on the page where it would be easiest to sound certain, so here is exactly what the course knows and how it knows it.

Gurney and Mott, 1938. R. W. Gurney and N. F. Mott, “The Theory of the Photolysis of Silver Bromide and the Photographic Latent Image”, Proceedings of the Royal Society of London Series A 164 (917), 151–167, published 21 January 1938. That is the origin of the alternating electronic-and-ionic model, and every modern account starts from it. This course has not read that paper: the publisher’s site refused access and no open copy was found. The four steps above are Ware’s summary of it, and Ware is what the course has read. The bibliographic details were verified against the paper’s own metadata record.

Mitchell, 1957. The model was not accepted quietly. J. W. Mitchell, its principal critic, published “The nature and formation of the photographic latent image” in the Philosophical Magazine 2 (21), 1149–1170, in September 1957 — with Mott himself as co-author, which tells you something about how live the argument still was nineteen years on. This course has not read that paper either, and it will not summarise an argument it has not read. It records that a differing account exists, by whom, and where.

The cluster size. Ware writes that the minimum is “thought to be” four silver atoms. Note his hedge; the course keeps it. Four is the figure used throughout this part, it is the only figure the course has read in a source it holds, and it is not presented as settled. If you meet three, or five, or “four to six” elsewhere, you have not caught anybody out — you have found the disagreement this paragraph is warning you about.

What is not in dispute is worth saying in the same breath, because a page that only lists uncertainties teaches nothing. That absorption frees an electron; that silver ions move inside the crystal; that a small silver cluster makes the crystal developable; and that removing the halogen is necessary — none of those is contested by anyone. The argument is about the detailed route, not about the destination.

Sub-latent image, and where the cluster sits

If four atoms is the threshold, then three is not. A crystal can hold a cluster too small to develop: real, made by real photons, and useless. That is the sub-latent image, and it is the direct consequence of there being a threshold at all. It matters for two reasons that the next page takes up: a sub-latent cluster can be pushed over the threshold by more light, and it can decay away before that happens.

Where the cluster forms matters as much as how big it is. A cluster on the surface is available to a developer, which works from the solution inwards. A cluster inside the crystal is not, and a crystal carrying only an internal latent image will not develop in an ordinary developer at all, however much light it received.

This is not a theoretical possibility. Kodak’s 1952 patent by Davey and Knott is titled for an emulsion that forms an internal latent image, and opens by stating that such emulsions “are known”, citing Mees’s 1942 text and a 1941 paper by Berg, Marriage and Stevens. Emulsion makers can put the image where they want it, and the same patent is the one that gives 6 per cent as the minimum useful iodide content and 10 to 20 per cent as the preferred range.

Why would anyone want an image where the developer cannot get at it? Because a latent image that an ordinary developer ignores is a latent image you can choose when to reveal, using a developer that does reach it — which is the basis of several reversal and direct-positive processes. The point for this page is narrower and more useful: “exposed” and “developable” are not the same word, and the distance between them is a design variable rather than an accident.

Quantum sensitivity, honestly

How many photons does one crystal need? The question has a name — quantum sensitivity — and a clean definition: the number of photons that have to be absorbed by a grain before it becomes developable.

This course cannot give you a figure for any named modern film. It has read no source that states one, and a number of that kind, invented, would be worse than useless: it is exactly the sort of claim a reader would carry away and repeat. So here is what can be said with the sources in hand.

The floor is set by the threshold. If four silver atoms are needed and every absorbed photon delivered one, the answer would be four. It is not four, because photons are lost — to recombination, to holes that are not scavenged, to electrons trapped where no silver ion arrives, and to clusters that form and then decay.

The gap between a latent image and a visible one is enormous, and that number is sourced. Ware states that to form visible photolytic silver the exposure must be increased by a factor of the order of a million over that needed to form a latent image, because you now have to build a colloidal particle of tens of thousands of atoms rather than a speck of four, and to build many of them.

Two things lower the count, and both follow from the mechanism. A bigger crystal presents a bigger target: it absorbs a larger share of the light falling on the area it occupies, so fewer photons have to be aimed at it for four of them to land. And sensitisation — Sheppard’s sulfide specks, Waller’s gold — gives the electron somewhere to stop before it recombines, so a larger fraction of the photons that are absorbed end up as silver atoms rather than as heat. The Ilford patent’s own claim is the evidence: the same emulsion, the same halide, the same crystals, with a trace of gold added at the right pH, is very considerably faster.

And one thing raises it, which is the subject of the next page: if the light is dim enough that the first silver atom decays before the second electron arrives, a grain can absorb photons all day and never reach four.

How the idea was arrived at

The latent image was discovered before anyone could say what it was, and the gap between the two is almost exactly a century.

1839: Daguerre finds an invisible image can be developed. As Part I sets out, the decisive discovery was not the mercury but the principle: that an exposure far too short to produce anything visible had nonetheless left a real change, and that a second agent could find that change and amplify it. The exposure fell from hours to minutes at a stroke.

1841: Talbot names it. In his account of the calotype to the Royal Society he wrote that the impression “is latent and invisible, and its existence would not be suspected by any one who was not aware of it beforehand”. His developer was gallic acid with silver nitrate; the principle was Daguerre’s.

1885: Abney sees that composition is not the whole story. Silver bromide, he wrote, “may be produced in several molecular states, all of which have different degrees of sensitiveness” — identical in chemical composition, differing physically, the particles “built up of a greater or less number of primary molecules”. He also knew that the halogen had to go somewhere, and used the phrase bromine absorbent for what would later be called a halogen acceptor.

1924–1926: Sheppard finds the impurity. Sensitiveness corresponds to nuclei in the grains; the nuclei are silver sulfide; and the sulfur came, unbidden, from the gelatin. Kodak’s own teaching primer was meanwhile telling readers that speed is made by the temperature and duration of heating in manufacture — which is a description of the same phenomenon from the factory floor, four years before anyone could say what the heating was doing.

1938: Gurney and Mott give the mechanism. Electrons, traps and mobile ions, alternating. The paper is the hinge between a craft that knew what worked and a science that could say why.

1946 and after: the sensitisers are engineered. Gold joins sulfur; the pH and the pAg of the precipitation become the levers; the crystal habit becomes something a patent can specify. Everything after that is refinement of a picture whose outline was fixed in 1938 — and, as the contested block above records, argued over for at least twenty years afterwards.

Recap

A photon above the threshold makes two particles, a conduction electron and a hole, and both have to be accounted for.
The hole’s fate decides everything. It makes a halogen atom at the surface, and unless a halogen acceptor removes it, the halogen re-oxidises the silver you have just made. Gelatin does the job at latent-image exposures and not at print-out exposures.
The electron stops at a trap — a lattice defect, or a silver sulfide or gold speck put there deliberately. Sheppard identified the sulfide nuclei in 1924; oxidise the sulfur out of the gelatin and the emulsion is nearly dead.
The Gurney–Mott mechanism alternates electronic and ionic steps at the same trap, using the mobile interstitial silver ions the Frenkel defects supply.
About four atoms makes the crystal developable, on Ware’s figure and with his hedge. Below that is the sub-latent image; inside the crystal rather than on its surface is the internal latent image, which an ordinary developer cannot reach.
The mechanism is not a settled single account. Gurney and Mott 1938 is the origin; Mitchell, with Mott, published a differing treatment in 1957; this course has read neither original and says so.
No band gap and no quantum sensitivity figure is quoted here, because neither has been verified. The sourced magnitude is Ware’s: a visible print-out image costs of the order of a million times the exposure a latent image costs.

Sources for this page

13 cited · checked 2026-09-04

01Argyronomicon: Silver Photographs on Paper — Chemical History of their Invention, Deterioration, and ConservationMike Ware, 2019§ 21.6 Quantum yields from silver halide photolysis; 23.2 Photolytic silver; 23.3 Significance of halogen acceptors; 23.11 Gurney-Mott model of the latent imagemikeware.co.uk/downloads/Argyronomicon.pdftier 2, specialist2026-09-04
02The Theory of the Photolysis of Silver Bromide and the Photographic Latent Image, Proceedings of the Royal Society of London Series A 164 (917), pages 151-167R. W. Gurney and N. F. Mott, 1938doi.org/10.1098/rspa.1938.0011tier 1, primary2026-09-04
03The nature and formation of the photographic latent image, Philosophical Magazine 2 (21), pages 1149-1170J. W. Mitchell and N. F. Mott, 1957doi.org/10.1080/14786435708242745tier 1, primary2026-09-04
04Photographic light-sensitive material and process of making the same, United States Patent 1,574,944Samuel E. Sheppard, assigned to Eastman Kodak Company, 1926patents.google.com/patent/US1574944A/entier 1, primary2026-09-04
05Production of photographic silver halide emulsions of increased light sensitivity, United States Patent 2,399,083Cecil Waller, Ronald Bernard Collins and Edward Cyril Dodd, assigned to Ilford Limited, 1946patents.google.com/patent/US2399083A/entier 1, primary2026-09-04
06Photographic silver bromide emulsion containing some silver iodide, United States Patent 2,592,250Edward Philip Davey and Edward Bowes Knott, assigned to Eastman Kodak Company, 1952patents.google.com/patent/US2592250A/entier 1, primary2026-09-04
076.17B: Frenkel Defect, in Map: Inorganic Chemistry (Housecroft)Stanley Hsia, University of California, Davis, for LibreTexts; after Housecroft and Sharpe, Inorganic Chemistry, 3rd edition, and Tilley, Understanding Solids§ 6.17B: the interstitial cation and the vacancy it leaveschem.libretexts.org/Bookshelves/Inorganic_Chemistry/Map%3A_Inorganic_Chemistry_(Housecroft)/06%3A_Structures_and_Energetics_of_Metallic_and_Ionic_solids/6.17%3A_Defects_in_Solid_State_Lattices/6.17B%3A_Frenkel_Defecttier 2, specialist2026-09-04
08Si photodiodes, technical note KSPD9001EHamamatsu Photonics K.K., Solid State Division§ Cut-off wavelength: the relation lambda equals 1240 divided by the threshold energy in electronvolts, and the band picture of a semiconductor detectorhamamatsu.com/content/dam/hamamatsu-photonics/sites/documents/99_SALES_LIBRARY/ssd/si_pd_kspd9001e.pdftier 1, primary2026-09-04
09Chemistry 2e, section 7.1: Ionic BondingPaul Flowers, Klaus Theopold, Richard Langley and William R. Robinson, for OpenStax§ 7.1 Ionic Bonding: ionic solids as poor conductors because the ions cannot moveopenstax.org/books/chemistry-2e/pages/7-1-ionic-bondingtier 1, primary2026-09-04
10Photography with Emulsions: A Treatise on the Theory and Practical Working of the Collodion and Gelatine Emulsion Processes, 3rd editionCaptain W. de W. Abney, R.E., F.R.S., 1885§ Chapter I: the molecular states of bromide of silver; the bromine absorbentarchive.org/details/cu31924031278470tier 1, primary2026-09-04
11Elementary Photographic ChemistryEastman Kodak Company, 1928§ Chapter II: sensitiveness obtained by varying the temperature and duration of heating in manufacturearchive.org/details/elementaryphotog00east_0tier 1, primary2026-09-04
12History of PhotographyJosef Maria Eder, translated by Edward Epstean, 1945§ Development with Mercury Vapors; Negatives and positives on paper: Talbot and the calotypearchive.org/details/EderHistoryPhotographytier 1, primary2026-09-04
13An Account of some recent Improvements in Photography, in Abstracts of the Papers Printed in the Philosophical Transactions of the Royal Society of London, volume 4William Henry Fox Talbot, 1841§ The gallo-nitrate of silver and the passage on the latent and invisible impressionarchive.org/download/jstor-110751/110751_djvu.txttier 1, primary2026-09-04

Formulas, hazard statements, historical dates and process descriptions on this page were checked against the sources above on the date shown. Safety data changes: obtain the current safety data sheet for the product you actually buy before you open it.