JOURNAL OF THE OPTICAL SOCIETY OF AMERICA VOLUME 40, NUMBER 9 SEPTEMBER, 1950

Growth of Silver Filaments and Dendrites* CHESTER R. BERRY

Kodak Research Laboratories, Eastman Kodak Company, Rochester, New York (Received May 24, 1950)

Filaments and dendrites of silver have been formed by transfer of silver through a layer of silver halide which has been put in contact with a silver block at about 400°C. According to the scheme proposed for explanation of the observations, silver dissolves in the silver halide and migrates as interstitial ions, transfer occurring because of the existence of a concentration gradient. At the same time vacant anion sites are formed in the reaction at the interface of the block and the silver halide which permit the migration of halide ions toward the silver block. When the rate of silver migration is relatively slow, as with silver bromide or silver chloride, dendrites usually form by migration of silver on the silver already formed. When the silver transfer is fast, as with silver iodide, the migration process does not remove the silver quickly enough and the silver is extruded as long filaments.

INTRODUCTION

THE growth of silver filaments and dendrites has been the subject of considerable study. As a re­

sult, it is well known that such deposits may be pro­duced by many processes, a few examples of which are: thermal decomposition of certain silver compounds, condensation of the vapor, electrolytic deposition, and reduction of silver halide grains in hydrogen or by photographic developers.

The mechanism of photographic development and the formation of hair-silver in the process are par­ticularly interesting and have been discussed by Mott and Gurney1 and re-examined by Mott.2 According to the mechanism which they describe, silver is pushed out bodily from the grain because of pressure produced within the grain by the rapid migration of interstitial ions under the influence of the electric field of electrons given to the latent-image speck by the developer.

On the other hand, James3 has proposed that the filaments are produced by migration of silver from the site of formation (at the triple interface of silver, silver halide, and developer) along the surface of silver pre­viously formed.

The purpose of the present experiments is to attempt to gain further information on the mechanism of fila­ment production from silver halides by a somewhat

* Communication No. 1349 from the Kodak Research Labora­tories.

1 N. F. Mott and R. W. Gurney, Electronic Processes in Ionic Crystals (Clarendon Press, Oxford, 1940), Chapter VII.

2 N. F. Mott, Phot. J. 88B, 119 (1948). 3 T. H. James, J. Chem. Phys. 11, 338 (1943).

different experimental procedure from those indicated above.

DESCRIPTION OF THE BASIC EXPERIMENTS The experiments require a thin, compact layer of

silver halide in intimate contact with a silver block. This layer may be obtained by heating the silver support and melting the silver halide powder which has been placed on it. Then, if the specimen is maintained at a temperature of about 400°C, transfer of the silver from the block through the solid silver halide layer is observed.

Attempts to produce specimens which would give reproducible quantitative measurements of the rate of silver transfer were not successful. Since it was ob­served that there was reaction only at very few points in the interface, it was supposed that the reaction may be accelerated by impurities. Although only qualitative results were obtained, the rate of silver transfer using a silver iodide layer was obviously much faster than that observed with a silver bromide or silver chloride layer. With a silver iodide layer about 0.2-mm thick, silver was often detected in about ten minutes. On the other hand, several hours were required before the first traces of silver could be seen emerging from a similar layer of silver bromide or silver chloride.

After a large amount of silver had been transferred through the silver iodide layer, bundles of long fila­ments were found, as in Fig. 1. The length of the fila­ments shown in Fig. 1 was about one inch. When the transfer was through silver bromide or silver chloride, the silver formed some filaments but more often it

G R O W T H OF S I L V E R F I L A M E N T S AND D E N D R I T E S 615

616 C H E S T E R R. B E R R Y

flowed out on the surface of the halide, forming den­drites, as shown in Fig. 2.

While there was a flow of silver away from the block, there was a flow of the halide into the block. This was determined by noting, after a considerable transfer of silver, the halide lying at the bottom of deep holes which had been hollowed out in the silver block.

THE TRANSFER PROCESS The flow of silver through the halide layers must be

assumed to depend upon the existence of some gradient, such as electrical, thermal, or concentration. Since the flow was observed in the absence of applied electrical or thermal fields and counter to the gravitational field, the flow in these experiments has been attributed to concentration gradients in the halide layers.

Although the flow of silver might conceivably be either through the interior or on the surface of the silver halide crystallites, the interior flow seems to give a better explanation of the observed phenomena. If this is the case, then there is little question that the silver is dissociated into interstitial ions and electrons.4

The process of silver solution and flow in the silver halide is apparently the same as that observed in the oxidation of a metal such as copper.5 Evidence of the solution of silver in silver bromide is the observation that silver bromide specimens which had been in con­tact with silver at high temperature colored much more rapidly, owing to photolysis, than specimens similarly treated but in contact with glass. It is well known that such darkening is greatly accelerated when excess silver atoms are present. Warfield6 suggests that, in his experiments on conductivity in silver chloride in­duced by electron bombardment, the greater trap density near the surface of the silver chloride may be

explained by the solution in the silver chloride of the surface silver film used as an electrode.

In addition to the excess of silver ions and electrons which is established in the halide at the interface, a concentration gradient of vacant anion sites must be produced, giving a flow of halide into the block. The large diameter of the halide ion prevents its movement in interstitial positions.

To explain the growth of long filaments or dendrites it is assumed that the concentration gradients persist even though there is silver on both sides of the silver halide layer. The coverage of the surface from which the silver emerges is not nearly as complete as at the interface with the silver block. For this reason, the quantity of silver redissolving after filament or dendrite formation cannot be as large as that dissolving from the silver block. Another factor which is probably important in slowing the counterreaction is the lack of impurities in the silver threads of abundance greater than about one part in 50,000.

The greater rate of silver transfer in silver iodide is probably related to the higher silver-ion mobility in silver iodide than in silver bromide or silver chloride and suggests that the rate-limiting factor in the transfer process is this ionic mobility. In silver iodide very high ionic mobility exists because of the peculiar crystal structure above 146°C, in which the silver ions are dis­tributed among many possible sites separated by rather small potential barriers.7 The similarity of structures of silver iodide and silver sulfide8 at high temperatures suggests that silver filaments might also be grown rapidly using silver sulfide as the conducting layer. It was found, indeed, that silver was transported through silver sulfide much more rapidly than through silver bromide or silver chloride, and filament formation resulted.†

FIG. 1. Bundles of silver filaments about an inch long, grown through a silver iodide layer.

4 J. H. Simpson, Proc. Roy. Soc. 197, 269 (1949). 6 Reference 1, Chapter V11L 6 G. Warfield, "A study of electron bombardment induced conductivity in silver chloride single crystals," Thesis, Cornell University (1950), p. 55

DESCRIPTION OF SILVER FORMATIONS The growth of relatively long straight filaments

through the iodide and sulfide layers afforded speci­mens which could be conveniently studied using the x-ray diffraction technique. Diffraction patterns were taken in such a way that preferred orientation along the filament axis would be easily recognized, but no preferred orientation of any kind was observed. The diffraction rings were very spotty, and an estimate of 25μ for the crystal size was obtained. This would indi­cate that a single filament may contain roughly 100 crystallites in its diameter. This is consistent with the observation by microscope that what appears to be a single thread in the relatively low magnification of Fig. 1 is actually composed of many smaller filaments.

Bundles of filaments up to one inch in length were 7 L. W. Strock, Zeits. f. physik. Chemie 25B, 441 (1934). 8 P. Rahlfs, Zeits. f. physik. Chemie 31B, 157 (1936). † A description of silver filament production from silver sulfide

in contact with silver at high temperature has been given by W. Ostwald, Kolloid Zeits. 102, 35 (1943).

G R O W T H O F S I L V E R F I L A M E N T S A N D D E N D R I T E S 617

grown from silver iodide. In general, the diameter of a filament was nearly uniform along its entire length. In some cases, however, filaments were found which had narrowed and finally stopped growing completely.

There seems to be no question that the long silver filaments are formed by squeezing out, owing to pres­sure within the silver halide crystals. On the other hand, the dendritic silver formations usually observed on silver bromide or silver chloride must be due to flow of silver over the material already deposited. Whether silver is formed by extrusion or surface migra­tion is apparently dependent upon the rate of silver transport.‡ If the transport is sufficiently slow, the mobility of the silver will permit its removal from the site of formation. On the other hand, transport which is so rapid that the silver cannot remove itself from the path by migration causes extrusion to occur.

In the case of very rapid transport observed for a silver iodide layer, the rate of filament growth was about 1 cm in 24 hours. This amounts to about 1/10 micron per second and is not very different from the value Mott2 calculated as the rate of filament extrusion which may occur in development of a silver bromide grain, according to the Gurney-Mott1 mechanism. Mott found a value of 1/100 micron per second as a lower limit of the rate at which silver is pushed out, and the implication is that this is sufficient for the formation of extruded filaments. If, in the development process,

‡ In a private communication, G. I. P. Levenson has stated similar conclusions on the basis of observations of the develop­ment of silver bromide grains at different rates.

FIG. 2. Silver dendrites spreading over silver bromide surface about 4 millimeters from the point of origin.

silver does migrate through the grain at the rate given by Mott, the present experiments indicate that ex­trusion of threads will occur.

The correlation of the silver transport and growth phenomena in the present experiments with the phe­nomena occurring in the development process cannot be stated with certainty. It does seem likely, however, that there is considerable similarity and it may be that the only essential difference is the nature of the force which causes the silver ions to move: in the present case, a thermally activated migration giving a net flow because of a concentration gradient; and in develop­ment, an ionic drift in the electrical field produced by electrons of the developer.