1962/63, No. 7 221

HYDROGEN IN IRON AND STEEL

1. SOLUTION AND PRECIPIT,ATION

by J. D. FAST *) and D. J. van OOIJEN *).

When new theories are put forward in a particular field of science, they tend for a while toclaim the undivided attention of research workers in that field, while experiences that do not atonce fit into the new picture are, for the time being, disregarded.

Dislocation theory 'in metallurgy is a case in point, and led to the neglect of a great dealof earliet knowledge of the effects of hydrogen in iron and steel. On the basis of recent ex-periments the authors show that it is only by combining dislocation theory with older conceptsthat a satisfactory picture can be built up, especially in regard to the unexpected ruptures thatcan be caused by hydrogen in steel structures. This subject has latterly attracted considerableattention, since the hydrogen embrittlement of steel is thought to have been one of the causesof air crashes and other accidents.

Introduetion

Of the impurities in iron and steel; hydrogen isone of the greatest sources of danger. In this article,which appears in two parts, we shall review some ,ofthe more important aspects of the effects ofhydrogenin iron and steel. No attempt will be made to dofull justice to the relevant literature, which alreadycomprises over several thousand articles 1). To dealwith the subject at any length would thus amountto compiling a thick and almost unreadable bookcontaining numerous contradictory views. Our aimhere will rather be to present a coherent picturebased on modern insight and on our own experience,and reference will be made to only a fraction of theliterature of interest for our purposes. In Part II weshall deal in particular with the role played byhydrogen in the embrittlement of steel.

To begin with we shall discuss the solubility anddiffusion of hydrogen and its interaction with dis-locations. Attention will be drawn to some effectswhich appear to be bound up with the formation ofgaseous hydrogen at high pressure in lattice defects.It is this high-pressure hydrogen that underliesthe deleterious action mentioned.

Solubility and diffusion of hydrogen in iron

The solubility of hydrogen in iron has been shownby many investigations to be proportional to the

*) Philips Research Laboratories, Eindhoven.1) A large proportion of this literature is referred to in

"Circular 511" oftheNational Bureau of Standard~(U.S.A.),entitled "Hydrogen embrittlement of steel, review of theIiterature", 1951, and in the following reviews: Donald P.Smith, Hydrogen in metals, Chicago Univ. Press 1948; P.Cotterill, The hydrogen embrittlement of metals, Progr.Materials Sci.9, No. 4, 1961; M. Smialowski, Hydrogen insteel, Pergamon Press, London 1962.

546.72:546.11

square root of the pressure of the gaseous hydrogenwith which the metal is in contact 2). It follows fromthis that the solute hydrogen is not present in themetal as molecules H2 but as atoms H. (The sameapplies to hydrogen dissolved in other solid orliquid metals.)

Lattice constant and density measurements haveshown that the hydrogen is dissolved in metalsinterstitially, that is to say, the solute hydrogen atomsdo not replace metal atoms in the crystallattice butoccupy the spaces between the metal atoms, i.e,the interstices.The solubility of hydrogen in a metal increases

with rising temperature, in accordance with well-known thermodynamic laws, if the solution processis endothermic,i.e. one in which, at constant temper-ature, heat is absorbed from the surroundings. Onthe other hand the solubility decreases with risingtemperature if the solution process is exothermic, i.e.one in which, at constant temperature, heat is givenup to the surroundings. As regards the iron-hydrogensystem the first holds true: the solubility increaseswith rising temperature. Fig. 1 shows the solubilityof hydrogen in iron, at 1 atm, as a function of thetemperature, according to the literature alreadycited 2).The fact that the solubility of hydrogen in molten

iron is so much greater than in solid iron, as canclearly be seen from fig. 1, has consequences of theutmost importance to the manufacture .of iron .andsteel. The sudden drop in solubility upon solidifi-cation can cause porosity in castings and welds. An

2) A. Sieverts, G. Zapf and H. Moritz, Z. phys. Chem. A 183,19, 1938/39; M. H. Armbruster, J. Amer. Chem. Soc. 65,1043, 1943;W. GeIlerand Tak-Ho Sun,Arch. Eisenhüttenw.21, 423, 1950.

222

2,5

2,0

l,S

I,D

O,S

500

PHILIPS TECHNH'AL REVIEW VOLUME 24

Fig. 1. The solubility of hydrogen in iron at a H2 pressure of1 atm in percentages by weight, as a function of tempeI'ature in°C. The symbols a, y, Ö and I denote respectively a, y and Öiron and molten iron.

0

IfIII

V~?~( ../.

....

VV--

extreme case is demonstrated in fig. 2, which showsthe pores and cavitics in an iron bar, formed bymelting pure iron in hydrogen at a pressure of1 atm and pouring it in a water-cooled copper chill.

Among the elements that can dissolve inter-stitially in metals (H, C, N, 0), hydrogcn is excep-tional in that its diffusion coefficient is very muchhigher than that of the other elements mentioned;in ferrite (a iron) at 20°C, for example, it is no lessthan 1012 times greater than that of carbon andnitrogen (see Table I). This exceptionally rapiddiffusion can be explained by assuming that thehydrogen moves as a proton from one intersticeto another - the diameter of a proton being a merehundred-thousandth ofthat of an atom or ion ofcarbon or nitrogen. Thisdoes not exclude the possi-bility of the hydrogenbeing present in the inter-stices as an atom; it thenonly "jumps" as a proton.It is probable that a dis-sociation equilibrium IS

involved, of the simpleform

H ~ P + e, (1)

where pand e are respec-tively a proton and anelectron,

/

Table I. Diffusion coefficients D of N, C and H in ferrite.

Temp. D(N) D(C) D(H)°C cln2/s cm2js cm2js

20 8.8 X 10-17 2.0 X 10-17 1.5 X 10-5

100 8.3 X 10-14 3.3 X 10-14 4.4 A 10-5

200 1.7 X 10-11 1.0 X 10-u 1.0 X 10-4

300 5.3 X 10-10 4.3 X 10-10 1.7 X 10-4

400 6.0 X 10-9 5.9 X 10-9 2.5 X 10-4

500 3.6 X 10-8 4.1 X 10-8 3.3 X 10-4

700 4.4 X 10-7 6.1 X 10-7 4.9x 10-4

900 2.3 X 10-6 3.6 X 10-6 6.3 X 10-4

In agreement with the foregoing it is found thatin the metals investigated in this respect -iron 3),palladium 4) and tantalum 5) - solute hydrogen iselectrolytically transporred towards the negativepole when an electric current, produced by a DCpotential, is passed through the metal.

Traps

We have spoken up to now only of the solubilityand cliffusion of hydrogen in an ideal iron crystal.The iron or steel normally used is polycrystalline,and moreover each crystal contains impurities andlattice defects. The impurities occur both in theform of solute foreign atoms and in thc form ofseparate phases (e.g. as carbide or nitride). At these"imperfections" the hydrogen atoms find sites thatare energetically more favourable than the normalinterstitial sites, especially in the dislocations (fig. 3),anel also in the neighbourhood of certain foreign

3) A. Herold, Colloque snr la diffusion à l'état solide (organiséà Saclay, 1958), p. 133, North Holland Pub!. Co., Amster-dam 1959.

4) A. Coehn et al., Z. Physik 62, 1, 1930; ibid. 71,179,1931;ibid. 83, 291,1933.

5) J. Wesolowski, .J. Jarmula and B. Rozenfeld, Bull. Acad.Pol. Sci. chim. 9, 651, 1961 (No. 10).

Fig. 2. Sawn-through bar of pure iron, obtained by melting and casting in hydrogen of1 atm. Owing to the sharp drop in soluhility upon solidification, the metal contains numer-ous gas-filled cavities.

1962/63, No. 7 HYDROGEN IN IRON AND STEEL, I 223

000000000000000000000000000000000000000--Q8HBB-88omS88B--

000000000000000000000000000000000000000t 70345

Fig. 3. Schematic representation of an edge dislocation in asimple cubic lattice of metal atoms. The dislocation can beimagined as produced by the forced introduetion of an extraplane of atoms (arrow) in the lower half of the crystal. (Inreality this situation arises on compression of the lower halfof the crystal.) An interstitial atom finds a site in the middleof the outlined area energetically more favourable than a sitein an interstice of the unperturbed lattice. The dashed lineindicates the slip plane.

atoms, at the grain boundaries and at the ferrite-carbide or ferrite-nitride interfaces. At relativelylow temperatures, then, these lattice imperfectionswill act as "traps" for the H atoms; in other words,the H atoms will spend a very much longer averagetime at these sites than in the normal interstices.Obviously, this implies a lower diffusion coefficientand a higher solubility, In broad lines, however, thepicture is unaltered, since the solubility still de-creases with decreasing temperature, and thediffusion rate can still he called exceptionally high.

Interaction of hydrogen with dislocations

The interaction of the interstitial atoms with thedislocations that act as traps can have a markedinfluence o.n the mechanical properties of a metal.The interaction of carbon and nitrogen atoms withdislocations in iron has been extensively studied, andwe shall briefly discuss this as an introduetion to. thecortesponding interaction with hydrogen, whichpresents more complications and has not been so.thoroughly investigated.

The solute C or N atoms that have diffused to.disloeations are unable to. leave them at roomtemperature, so.that in the lo.ng run, given sufficientatoms, strings of atoms form which extend overthe whole length of each dislocation 6).

The formation of these strings o.f C or N atomsconsiderably affects the plastic properties of themetal, since plastic deformation implies the dis-placement of dislocations, Before they can be dis-

6) In body-centred cubic metals, interstitial atoms cause bothvolume expansionand tetragonaldeformation. Consequentlythey can be taken up in these metals, with an energygain, in screw as well as in edge dislocations.

placed they must be detached from the strings ofatoms, and this requires an extra stress which, oncethe dislocations have broken away, is no longer need-ed. In this way one can understand the occurrenceo.f an upper and a lower yield point in the stress-strain curve of mild steel 7) (fig. 4a). It also. explainswhy immediately after a slight plastic deformation- that is when the dislocations are still free -no distinct yield point is to. be found (fig. 4b). If themetal is then left alone for some considerable time,the C and N atoms again diffuse to. the dislocations,so.that the upper and lower yield points return andthe metal becomes more difficult to. deform (harderand more brittle). This spontaneous pro.cessis kn~wnas strain ageing. It is primarily due to. nitrogenatoms, the solubility of nitrogen in iron being muchgreater than that of carbon,

Further data on the interaction of dislocationswith carbon or nitrogen atoms in iron have beenderived from internal friction measurements. Where-as the C or N atoms in the normal interstices of theiron lattice produce the familiar damping peak, withwhich the name of Snoek 8) is associated, the jumpsof the atoms located in the stress field of the dis-locations give rise to. an "abnormal" damping peakat much higher temperatures.To.make this clearer - although we cannot go.into.

details here and must he content with referring to.the literature 8) - it may be mentioned that the

fc

/

o o-é b 70042a

Fig. 4. Stress-strain curves of mild steel. The tensile stress ais plotted versus the deformation E. a) The steel shows a sharpupper yield point at A. b) After plastic deformation to C andupon renewed loading, the metal shows no distinct yield point.

7) A. H. Cottrell, Dislocations and plastic flow in crystals,Clarendon Press, Oxford 1953.

8) J. L. Snoek, Physica 8,711, 1941 and 9, 862, 1942. See alsoJ. D. Fast and L. J. Dijkstra, Philips tech. Rev. 13, 172,1951/52. .

224 PHILIPS TECHNICAL REVIEW VOLUME 24

position of the maximum. of a damping peak of thiskind, caused by the jumps of interstitial atoms,corresponds to a specific value of the diffusion coeffi-cient of these atoms. If the maximum occurs at ahigher temperature, this means that the value inquestion is reached at a higher temperature, whichin turn implies that the relevant atoms are morestrongly bound. That the stronger binding of theatoms is indeed the consequence of their presence indislocations appears from the fact that the abnormalpeak is found only after iron containing nitrogen orcarbon has been cold-worked, and thus possesses arelatively high dislocation density. The peak is higherthe greater is the degree of plastic deformation,provided the metal contains an excess of C or N.Fig. 5 shows such a peak measured on iron contain-ing nitrogen, and fig. 6 a corresponding peak meas-ured on iron containing carbon.

Conflicting answers have been given to the questionwhether hydrogen is also one of the elements thatcan cause internal friction in iron. Gensamer andeo-workers 9) gave an affirmative answer based ondampingmeasurements ona commercial steel chargedwith hydrogen. Heller 10) also found a dampingpeak in iron containing hydrogen. He charged wiresoffairly pure iron withhydrogen and with deuterium,and at a frequency of 1 cis found a damping peakat 30 oK in the hydrogen-charged wires, and one at

9

IJ\ I1I \V

Vo-...a. V

6

3

oo 15050 100

Fig. 5. Abnormal damping peak measured on iron wire aftercharging with nitrogen in hydrogen containing ammonia at550°C, cold-drawing to 60% reduction of cross-sectional areaand heating for one hour at 350 °C. The quantity 1/Q on theordinate is the logarithmic decrement of the torsional oscilla-tions divided by st, Vibration frequency 0.13 cIs. (After T. S.Kê, Trans. AIME 176, 448, 1948.)

9) L.C. ChangandM,Gensamer, Acta metallurgica 1,483,1953;L.C.Weiner and M.Gensamer, Actametallurgica5,692, 1957.

10) W. R. Heller, Acta metallurgica 9, 600, 1961 (No. 6).

12xlO-J

°

Î KVIf

J__.c'"

[7

f96

3

oo 200 250 300°C-T

Fig. 6. Abnormal damping peak measured on iron wire chargedwith carbon in hydrogen containing heptane, cold-drawn to25% reduction of cross-sectional area and heated for severalhours at 250°C. 1/Q is along the ordinate. Vibration frequency2.2cIs. (After K. Kamber, D. Keefer and C.Wert, Acta metal-lurgica 9, 403, 1961, No. 5.)

50 lOO ISO

35 oK in the deuterium-charged wires. He too attrib-utes the peaks to "Snoek jumps" of solute Hand Datoms, similar to those of C and Natoms.After deformation of their hydrogen-containing

steel, Weiner and Gensamer 9) found a dampingpeak at the much higher temperature of about105 "K. Itseems obvious to assume that the explana-tion for this must resemble that given for the abnor-mal peaks caused by carbon and nitrogen; the peakat 105 "K is thus attributed to hydrogen in thedislocations.The abnormal hydrogen peak, however, IS

associated with some remarkable effects that arenot found in the behaviour of the abnormal nitrogenand carbon peaks. In the first place iron containinghydrogen shows such a peak also without deliber-ate plastic deformation, the peak appearing afterseveral days of ageing at 300 "K, Secondly, the peakdisappears spontaneously if the ageing is continuedlong enough at 300 oK (fig. 7).

Formation of molecular hydrogen in microcavitiesin iron and steel

The disparate behaviour of iron containing hydro-gen as opposed to that containing nitrogen and car-bon can be understood as follows. The interstitiallydissolved H has the specific possibility of precipi-tating in molecular form in lattice imperfections.Although this property has long been known, itsimportance until recently was not sufficiently recog-nized. True, nitrogen and carbon can also precipitatein iron as a separate phase - N as the nitride FesNor Fe4N, and C as the carbide FeaC. However, thisnitride or carbide formation does not reduce thefree energy of the iron to the same extent as when theNand C atoms bind themselves to dislocations. This

1962/63, No. 7 HYDROGEN IN IRON AND STEEL, I 225 .

appears, for example, from the experience thatstrain ageing occurs not only in iron contammgsolute N or C, but also in iron in which nitrogen orcarbon is exclusively present in the form of nitrideor carbide. In this case N or C atoms diffuse fromthe precipitate to the dislocations, implying theentire or partial solution of the -precipitate.As regards hydrogen the opposite is the case:

the free energy of the iron falls more when the hy-drogen precipitates in the form of H2 than when itbinds itself to dislocations. As a result, H2 moleculesform in all microcavities and lattice imperfectionswhere there is space for them. This can give rise tohigh local gas pressures, causing plastic deformationof the surrounding metal. The new dislocationsproduced in this way will absorb part of the hydrogenstill in solution, after which the abnormal H peakcan appear. Since, however, the free energy decreaseseven- more when H2 is formed, the abnormal peakfinally disappears again, as demonstrated in fig. 7.

0.9x10-3

0,8

19 ~ 0 0

~ 0

r _0 0. V16 ~o

n t>0 _/' ~

5 ~

.(t\0/ \/ \I 1 ~o

0

0 _Q_y -0u

~ o IA)"'"~ ~ 0' ~

u

° ~1

C7" ~ ~ V° ~° ° 0..0-

o "" +o

0,8

0,7

0,8

0,7 1,Ox10-3

0,9

0,8

1,0

0,9

0,8 1,2xl0-3

1,1

I,D

0,9

O,9xl0-3

0,7 O,9xl0:"3

0,8

0,9x1O-3

- -0,8

0,7

90 100 110 120~T

1300K

Fig. 7. Internal friction of hydrogen-charged steel, after ageingfor various periods of time (vibration frequency 20 c/s). The.figures in the graph denote the number of days during which thesteel was aged at 300 "K, (After Weiner and Gensamer 9).)After four days of ageing a marked damping peak is observedat about 100 oK, which vanishes again after further ageing.Effects of this kind, which are not found after nitrogen or car-bon charging, can be explained by the formation of hydrogenat high pressure in lattice imperfections,

Mter charging some commercial steels withhydrogen electrolytically, Rogers 11) found that fora certain time they showed no distinct yield point.This remarkable effect can be explained in muchthe same way as the above-mentioned spontaneousappearance of the abnormal H peak after chargingwith hydrogen: many new dislocations producedduring- and after charging, as a result of the plasticdeformation upon the formation of H2' are not yetanchored by N or C atoms and therefore require noextra stress to set them in motion.

Also connected with the precipitation of hydrogen in micro-cavities is the experience that, at low temperatures, the(apparent) solubility values found are much greater than mightbe expected from an extrapolation of the measurements athigh temperatures. The difference is too marked to be ex-plained entirely from the binding of H atoms (or protons) intraps at low temperatures. The difference can only be under-stood from the fact that the equilibrium pressure of the molee-nlar hydrogen steadily increases as the temperature drops(see below), and that the hydrogen, once it has precipitated asH2 in microcavities, is unable to escape at low temperatures.The hydrogen thus occluded is of course not dissolved.In view of the high diffusion rate of hydrogen in iron, this

possibility of hydrogen occlusion is rather unexpected. Attemperatures above about 200 oe the hydrogen willieave theiron rapidly as expected, but at lower temperatures, eventhough the diffusion rate is still very high, there is no longerany question of the hydrogen actually escaping. (If it wereotherwise, hydrogen could not he stored in iron cylinders!)The explanation is that the escape from the iron involves notonly diffusion but also surface reactions, in particular thedissociation of molecular hydrogen into atomic hydrogen (atthe surface of the cavities) and the converse reaction (on theoutside surface). The first reaction in partienlar is extremelyslow at low temperatures 12). Further particulars will be foundui fig-B, -

The precipitation of H2 in lattice imperfectionsis also at the root of the numerous detrimentaleffects caused by hydrogen in steel. It is so importantto the understanding of the action of hydrogen insteel that the remainder of this article will be devotedentirely to H2 formation and its consequences.If a piece of iron is enveloped at e.g. 1100 oe by

H2 at a pressure of 1 atm, it can be seen from fig. 1that 0.0006 g of hydrogen (equivalent to 7 cm3 H2of O°C and 1 atm) will be dissolved per hundredgrammes of iron. When such a piece of iron israpidly cooled to 20 oe, the hydrogen content imme-

11) H. C. Rogers, Acta metallurgica 4, 114, 1956 and Trans.AIME 215, 666, 1959.

12) If one measures at low temperatures the rate at whichhydrogen permeates through an iron wall, or escapes fromiron, and the surface reactions are not taken into account,then the diffusion coefficient calcnlated from the resultswill be much too small, sometimes even 10000 times toosmall. For a detailed discussion of this subj eet, seeJ.D. Fast,Philips tech. Rev. 6, 365, 1941, and 7, 74, 1942.

226 PHILlPS TECHNICAL REVIEW VOLUME 24

Fig. 8. The figure showsschematically that in the iron-hydrogensystem the processes of sorption, desorption and diffusionof the gas have very different temperature coefficients. Thediagram shows versus lIT on a logarithmic scale the diffusionrate DId (solid line), the desorption rate Vu (dot-dash line) andthe sorption rate Vi (dashed line). The position of the latter twolines depends to some extent on the surface state, withouthowever altering the essence of the diagram. It can be seen thatthe desorption and sorption rates decrease faster with fallingtemperature than the diffusion rate. At room temperature thisgives rise to a situation, unexpected at first sight, where on theone hand the permeahility of iron to hydrogen is so low that thegas can safely be stored in iron cylinders, whereas on the otherhand the diffusion rate can still be called exceptionally high.

diately after cooling is unchanged, if not near thesurface then at least in the interior of the metal.The concentration of dissolved hydrogen is thenroughly 104 times higher than it would be in equi-librium at 20 oe and 1 atm hydrogen. The solubilitybeing proportional to the square root of the hydrogenpressure, as we have seen, calculation shows thatthis high concentration at this temperature canonly exist in equilibrium with an H2 pressure of lOSatm. The dissolved hydrogen will therefore attemptto escape from the lattice by diffusion to the out-side and to every internal cavity present in the metal.

This calculation still needs a correction: we havenot taken into account that at pressures as high as108 atm the hydrogen no longer behaves as a perfectgas, so that the above-mentioned proportionality isno longer valid. Ifwe allow for this in the calculation,we find that the relation between the equilibriumpressure and the H content at 20 oe is given by acurve as shown in fig. 9. From this we see that theequilibrium pressure in the case considered is not108 but "only" 104 atm.By exceeding the cohesion of the material, an

equilibrium pressure as high as this can easily giverise to ruptures, but only where imperfections exist.In an ideal single crystal of iron even much higherhydrogen concentrations than those mentioned wouldnot lead to rupture. The harmful effects appear onlyif imperfections are present in the metal (or are

introduced by plastic deformation), in which theprecipitation of H2 up to the dangerous pressure canreally take place.

Most commercial steels have hydrogen contents ofthe order mentioned in the above example, ofteneven higher. The example differs from reality onlyso far as most of the hydrogen in steel does notoriginate from hydrogen gas in the atmosphere. Oneof the principal sources of hydrogen in iron is watervapour, which at high temperatures reacts withboth solid and molten iron as follows:

Fe + H20 ~ FeO + 2[H]. (2)

The water vapour may, for example, come from ruston the scrap used in steel-making, from constituentsin the slag (lime), from the crucible or furnace walland from the gas atmosphere present. In electricare welding with coated electrodes, the coating isthe main source of water vapour.

A second and equally important source ofhydrogenin iron and steel is the hydrogen produced in atomicform at the surface during galvanizing, pickling orelectrolysis processes.

The fact that high H2 pressures can arise insteel containing hydrogen may be demonstratedby simple experiments 13) carried out as long

10-

....l---~

/V

V///

/,//

/

-__.c

Fig. 9.The graph shows the calculated equilibrium pressure pofmolecular hydrogen, taking into account the deviations fromideal behaviour, plotted as a function of the concentration c ofatomic hydrogen in iron for a temperature of 20°C. (After G.Vibrans, Arch. Eisenhüttenw. 32, 667,1961.)

13) C. A. Edwards, J. Iron and Steel Inst. llO, 9, 1924; P.Bardenheuer and G. Thanheiser, Mitt. K. W. I. Eisenforsch.10, 323, 1928.

1962/63, No. 7 HYDROGEN IN IRON AND STEEL, I

ago as 1924. The twofollowing experimentsare typical. 1) Hydrogenis evolved electrolyti-cally on the surface ofa hollow Iron cylinder(of small internal vol-ume), enabling the gas inatomic form to penetratethe metal in relativelylarge quantities. Aftersome time the pressureinside the bore ofthe cylin-der starts to rise, as can beread from apressure gauge.In this way the pressurecan be observed to build

Fig. 10. Blisters formed during pickling of steel. In the pickling process atomic hydrogen isproduced which diffuses inwards and forms molecular hydrogen at inclusions in the steel.The blisters appear where these inclusions are immediately below the surface. Magnifi-cation 10 x. The regular pattern of the blisters indicates a regularity of the inclusions,hrought about during the rolling of the steel.

up to several hundreds ofatmospheres, after whichthe experiment is endedfor reasons of safety. 2) By means of an acid or byelectrolysis, atomic hydrogen is evolved on theinside of an open iron pan which is enamelled on theoutside. After some time the enamel cracks awayfrom the pan - sometimes almost explosively -owing to hydrogen having precipitated at highpressure at the iron-enamel interface.

Difficulties connected with effects of this naturesometimes occur in enamelling processes as a resultof the above-mentioned reaction (2) between the steeland the water vapour. The vapour arises from sub-stances used in enamelling, principally from "frit" 14).After cooling, the metal may then be severelysupersaturated with hydrogen, which may result inserious damage to the enamellayer.

Closely related to the precipitation of hydrogenunder high pressure at the metal-enamel interfaceis its precipitation at non-metallic inclusions insteel. This may occur, for example, as a consequenceof the pickling prior to tin or zinc plating. Part ofthe atomic hydrogen formcd during the picklingprocess diffuses into the steel and forms H2 at thcinclusions. This can lead to surface blistering if theinclusions are so close to the surface that the hydrogenpressure is able to push the supervening metaloutwards by plastic deformation (Jig. 10).

Formation of high-Ilressure CH4

·Where iron and steel containing carbon are in external con-tact with hydrogen gas, methane (CH4) may be formed inter-nally by the reaction:

14) D. G. Moore, M. A. Mason and W. N. Harrison, J. Amer.Ceramic Soc. 35, 33, 1952.

At 300°C the equilibrium constant of this reaction has avalue such tha t a CH. pressure of several thousand atmospherescorresponds to an H2 pressure of 1 atm. At higher temperaturesthe equilibrium pressure of CH. is lower, at lower ternpcratureshigher.

In technology this effect caused many initial difficulties inthe large-scale production of various inorganic and organicchemical compounds, e.g. of ammonia, methanol and petrol.Nowadays it is possible to prevent CH. forming by usingalloying additives in the steel that form highly stable carbideswith the carbon.

It is also worth noting that, as reaction (3) is exothermic, theoccurrence of this reaction may create the impression that hy-drogen in a certain temperature range may dissolve exother-mally in the metal instead of endothermally.

In Part 11 of this article we shall show that theformation of molecular hydrogen under high pressureas discussed here, not only takes place in large im-perfcctions (microcavities) but also in defects ofatomic dimensions. It will be seen that knowledgeof this effect makes it possible to understand manyof the other harmful effects of hydrogen in steel.Among the more important of these are the reductionof ductility and the appearance of brittle fractures.

(3)

Summary. \Vhereas hydrogen dissolves interstitially in iron inthe form of atoms and presumably diffuses in the form ofprotons, molecules (H2) can form in imperfections of the crystallattice. The latter explains such unexpected effects as the tem-porary appearance of an "abnormal" damping peak duringageing, and the temporary absence of a distinct yield pointafter electrolytic charging with hydrogen. The formation ofmolecular hydrogen in microcavities in iron and steel can giverise to extremely high pressures; this is demonstrated by calcu-lations, by experiments and in practice. In Part II of thisarticle an explanation will be given, on this basis, of the moreserious harmful effects of hydrogen in steel, in particularreduced ductility and brittle fracture.

227

228 PHILlPS TECHNICAL REVIEW VOLUME 24

THERMIONIC-CATHODE TESTING

Photo Maurice Broomfield

The emission properties of thermionic cathodes are usually investigated in simple test-valves, containing only an anode besides the cathode. The photograph shows three suchvalves being evacuated on a mercury-diffusion pump. Behind them is the cold trap,which is kept at about - 190°C by liquid nitrogen and condenses the mercury vapourfrom the diffusio n pump. Above the test valves can be seen an ionization gauge 1).In situations like this it is necessary to allow for the fact that the pressure in theion gauge is not the same as in the valves.

These preliminary tests under greatly simplified conditions are followed at a laterstage by tests with the cathodes mounted in the valves for which they are intended.

1) Philips tech. Rev. 20, 153, 1958/59.

1962/63, No. 7 229

ABSTRACTS OF RECENT SCIENTIFIC PUBLICATIONS BY THE STAFF OFN.V. PHILIPS' GLOEILAMPENFABRIEKEN

Reprints of those papers not marked with an asterisk * can be obtained free of chargeupon application to the Philips Research Laboratories, Eindhoven, Netherlands,where a limited number of reprints are available for distribution.

3029: W. L. Wanmaker and C. Bakker: The deter-mination of luminescent properties as amethod for the study of diffusion processesin the solid state (Reactivity of solids, Proc.4th int. Symp. on the reactivity of solids.Amsterdam 1960, editors J. H. de Boeret al., 709-717, Elsevier, Amsterdam 1961).

Luminescence is only produced by a foreignion (activator ion) in a crystal lattice. If a crystalis first made without activator and then heatedin contact with a suitable compound containingthe activator ion, the diffusion of the ion inquestion through the host lattice can be followedby means of the advance of the luminescence.This method has a very wide field of application.A start has been made with the study of the diffu-sion of Sb and Mn ions in calcium halophosphate,3Ca3(PÛ4)2.Ca(F,CI)2' which has the apatite struc-.ture.

This method can also he used to study the dif-fusion of ions of the host crystal. In order to dothis, the luminescent substance is heated in contactwith a suitable compound of the ion in question,e.g. CaCÛ3. Since the intensity of the luminescenceis lower when there is an excess of Ca ions, the dif-fusion can be followed by observing the decrease inluminescence.

3030: H. G. Grimmeiss and H. Koelmans: Analysisof P-N luminescence in Zn-doped GaP(Phys. Rev.123, pp. 1939-1947, 1961, No. 6).

A P-N junction in a semiconductor can emit lightwhen a voltage is applied across it. This articledescribes one phase of an investigation aimed atincreasing the yield of light from this effect. Seealso R 398 and Philips tech. Rev. 22, 360-361,1960/61.

3031: N. V. Franssen: Eigenschaften des natürli-chen Richtungshörens und ihre Anwendungauf die Stereophonie (Proc. 3rd int. congresson acoustics, Stuttgart, 1959, editor L. Cremer,Vol. Il, pp. 788-790, Elsevier, Amsterdam1961). (The properties of natural directionalhearing and their application to stereophony;in German.)

A short survey of material which has been fullydealt with in 2889.

3032: C. M. van der Burgtr Neue keranrische Ultra-schallwandler und deren Kopphmg an dieFlüssigkeit (as 3031, pp. 1219-1221). (Newceramic ultrasonic transducers and theircoupling to liquids; in Ge'rman.)·)

Ni-Cu-Co ferrites can be used for the efficientproduction of ultrasonic vibrations. T4e author.discusses the demands these ferrites must meet whenhigh-power vibrations of a frequency' b~tween 20and 50 kcts are concerned. See also 2902,'

3033: A. Recourt and G. H. F. de V;ries.:An ex-perimental apparatus for corrtact: microradi-ography at 200-500 V (Nature 1?1, 1185-1186,1961, No. 4794):

, --~~', ,~;;"l·.,:;_

: In contact microradiography (see phiÛps tech.Rev. 19, 221-233, 1957/58); the ;.)Ç~~~y.'beam isconsiderably attenuated oil: passage thtough thewindow. This effect is so' strong .ai l~~ :voltages(less than 500 V), whose use is'so;metinjè~'indicated,that normal contact smicroradiography. equipmentsimply cannot be used for this p.~rpose:~The.authorshave built an experimental apparatus i~ which theX-ray source, the sample and the.'fi}m;are all in thesame evacuated space, so that the window is com-pletely dispensed with: , .

3034: J. D. Fast, J. L: Meijering ana M. B. Verrijp:Frottement interne dans les alliages .ferri-tiques Fe-Mn-N (Métaux, Corr., Ind. 36,112-114, 1961, No. 427). (Internal frictionin iron alloys Fe-Mn-N; in French.)

Considerations concerning the relative positions'of the three peaks mentioned in the following pub-lication (3035). It is concluded from experimentson alloys with varying Mn contents that the mainmechanism is based on the presence of Mn-Mnpairs. An N atom is more strongly bound to such apair than to a single Mn atom, and yet is more mo-bile in the former case.

230 PHILIPS TECHNICAL REVIEW VOLUME 24

3035: J. L. lVIeijering: Considérations sur l'effetSnoek dans Ie cas de sites non-équivalentspour les atomes en insertion (Métaux, Corr.,Ind. 36, 107-111, 1961, No. 427). (Considera-tions on the Snoek effect for the case of non-equivalent sites for the foreign atoms; III

French.)

The maximum Snoek damping (see Philips tech.Rev. 13, 172-179, 1951/52) due to interstitial Natoms in iron is at about 25 oe at a frequency ofone els. By substituting some Fe atoms by Mnatoms, the peak in the damping curve is broadenedand shifted to higher temperatures. Three individualpeaks can be distinguished in this broadened peak,the middle one corresponding to damping in theabsence of Mn. An N atom in the neighbourhoodof an Mn atom will alternately occupy two kinds ofoctahedral sites, corresponding to two different freeenergies. -This explains the two additional peaks. Asomewhat simpler case is also treated, where thetwo kinds of positions are the octahedral and tetra-hedral sites in a lattice containing no Mn.

R 426: J.D. Wasscher: Note on four-point resistiv-ity measurements on anisotropic conductors(Philips Res. Repts 16, 301-306, 1961,No.4.).

The resistivity of a homogeneous material can bedetermined with the aid of a slice of the materialto which four point contacts are applied. If a currentis passed through two of these contacts, a potentialdifference proportional to this current is producedbetween the other two contacts. The constant ofproportionality, which is a measure of the resistivity,depends on the geometry of the arrangement. Theauthor shows how the results obtained for iso-tropic material can be applied to anisotropicmaterial, making use of a coordinate transformationproposed by Van der Pauw. The case of four con-tacts in a straight line is discussed, as is the case offour contacts at the corners' of a square; both casesare worked out for two samples, one thin andone thick compared to the distance between thecontacts. The square arrangement proves to be themost sensitive to anisotropy. It is also found thatthe three principal values of the resistivity tensorcan be determined from measurements on one singleplane at right angles to a direction corresponding toone of these principal values. Corrections for the

finite dimensions of the contacts and the sample arediscussed. '

R 427: S. Duinker: Short-wavelength response ofmagnetic reproducing heads with roundedgap edges (Philips Res. Repts 16, 307-322,1961, No. 4).

The calculation of the response of magnetic pick-up heads given by Westmijze for a gap with idealrectangular edges is extended by the author to thegeneral case of edges with a finite radius of curva-ture. An expression for the gap-width loss factor isfound hy confermal mappings of head and potentialfield. The ratio of the radius of curvature of theedge to the gap width occurs as a parameter in thisexpression; it follows that as the radius of curvatureincreases the response of the head decreases and theeffective gap width increases.

R 428: J. J. Scheer and J. van Laar: Photo-electricemission from cadmium telluride (PhilipsRes. Repts 16, 323-328, 1961, No. 4).

The photo-electric effect of CdTe is measured oncrystal surfaces and on evaporated layers. Cleancrystal surfaces are prepared by cleaving a crystalunder high vacuum. Experimental curves of thephoto-current' as a function of photon energyshow a "tail" at long wavelengths; this tail isascrihed to impurities. It follows from the experi-mental results that the most reliahle value of thework function is found from the emission from thesurface of a cleaved single crystal. Finally, thepossibility of a relationship between the photo-electric properties of CdTe and CdS is discussed.

R 429: W. Albers and IC Schol: The P-T-X phasediagram of the system Sn-S (Philips Res.Repts 16, 329-342, 1961, No. 4.).

Two maxima were found in the T-XL :diagramduring a study of the equilibria between the solid,liquid and gaseous phases in the system Sn-S.The first maximum occurs at the composition ofSnS, and at a melting temperature of 88l.5 ± 2 °Cand a partial pressure of sulphur of 0.033 atm.The second maximum occurs at the composition ofSnS2, and at a temperature of 870°C and a pressureof 40 atm. Between 10 and 47 at. % S, and perhapsbetween 70 and 90 at. % S, the different liquidphases are immiscible. In the course of this investi-gation, a compound has been discovered whichprohably has the composition Sn3S4•