Field-emission Studies of the Adsorption of Chlorine and theDissociation of Carbon-Chlorine Compounds on TungstenBY M. 3. DUELL, B. J. DAVIS AND R. L. MossWarren Spring Laboratory, Stevenage, Herts.Received 17th January, 1966The adsorption/desorption of chlorine and the dissociation of carbon tetrachloride and chloro-form on tungsten were studied in a field-emission microscope observing the pattern variations andchanges in work function derived from the Fowler-Nordheim equation. Chlorine from the gas-phase and from dissociated carbon-chlorine compounds gave rise to positive surface potentials atlow coverages (oxygen was examined for comparison), accompanied by a reduction in emittingarea. This was ascribed to chlorine penetration below the metal surface ; final surface potentialswere about - 1 eV.The adsorption of methane and its attack on a chlorine-covered surface werealso examined. Under a continuously applied field tungsten whiskers were formed in carbontetrachloride or chloroform vapour, believed to arise from the migration of surface complexesinvolving both carbon and chlorine.Previous field-emission studies of the adsorption of halogen gases on tungstelr haveshown that while iodine 1, 2 is sensitive to surface topography and produces a smallincrease in work function, chlorine 3 and bromine 2 increase it by 1 eV or more,at constant coverage. At low bromine coverages, the work function decreases toa minimum accompanied by a reduction in emitting area, which was explained interms of the partial burial of the adatom in the metal surface as a result of the migra-tion of tungsten atoms.A corresponding study of chlorine adsorption is reportedhere, with emphasis on observations at low coverages together with some resultson oxygen, in order to compare their respective corrosive effect on the tungsten sur-face. Also, as background to work on carbon-chlorine compounds, methane wasexamined ; carbon has little effect on the work function 4 compared with chlorine,although carbon patterns 5 on heating are particularly characteristic. The field-emission technique could then be applied to study the dissociation of carbon-chlorine compounds, i.e., carbon tetrachloride and chloroform, on a clean tungstensurface with respect to both the adsorbed species formed and their action on themetal surface itself.The reaction between adsorbed chlorine and gas-phasemethane was also examined.EXPERIMENTALThe techniques used in field-emission microscopy have been described 6.7 and alsoexperimental details particular to the present work, including an indication of the pre-cautions necessary to obtain a sufficiently pure supply of the gas being studied.2 In experi-ments where the field-emission tip was flashed before admitting the gas ((ii), below), specialcare was taken to flush the metal valve, through which gases were admitted to the field-emission tube, to avoid desorbing unwanted gases at the start of the experiment. Thesame electrical arrangements 2 were used to measure the applied voltages and field-emissioncurrents from which Fowler-Nordheim plots were derived.Plots corresponding to theearly stages of adsorption moved rapidly in the direction of increasing voltage for constantemission current in some cases (e.g., fig. 3 for chlorine). Since individual voltage-currentcharacteristics required at least 30sec to record, work functions derived from the slopes444 FIELD-EMISSION OF CI2 A N D C-CI COMPOUNDSof such Fowler-Nordheim plots would be in error. Accordingly the actual time was re-corded at which each single voltage-current observation was made, adjusting the currentin turn to one value in a set of chosen values. From these results, sets of interpolationplots were constructed of applied voltage against time at constant emission current, andhence Fowler-Nordheim plots obtained from current-voltage values referred to the sameelapse of time from the start of the experiment.RESULTSThe basic observations, i.e., field-emission patterns and voltage-current data,were made in various ways : either (i) the gaseous compound under study was passedcontinuously through the field-emission tube before starting the experiment byflashing the tungsten tip at 2500°K; or (ii) since the early stages of interaction withthe clean tip, particularly of interest, might be affected by adsorption during theshort time required to cool to room temperature, the tip was flashed in ultra-highvacuum, before admitting the gas.Also the field was applied: either (iii) inter-mittently, but nevertheless for a substantial fraction of the total time to derivevoltage-current data, as described above ; separate experiments were made to checkfor the effect of applied field; or (iv) continuously in the presence of chloroform orcarbon tetrachloride vapour which produced unusual field-emission patterns andvoltage-current data.The Fowler-Nordheim equation 6 describes the relationship between the currentdensity j (total current Ilemitting area a) in A/cm2 and the applied field F thus :1.54 x 10-6P2 4%- 4t'CYI F j = exp -6.83 x 107--v[y],where the functions t[vJ and o l y ] have values given in ref.(6). The field F is givenapproximately bywhere Y is the applied voltage, r is the tip radius and K-0-2.7 A plot of log (J/V2)against l / V will be approximately linear with a slope M given bywhere s b ] lies between 0.833 and 1.000.The value of (r/K) for clean tungsten wasobtained from the Fowler-Nordheirn plot, taking the accepted work function,4 = 4.52 eV, and making a series of successive approximations in eqn. (3), startingwith s[y] = 1. Assuming ( r / K ) to remain constant, the value of the work function,after adsorbing a particular gas for a known time was also calculated, Usingeqn. (I) the quantity log A given byF x KVfr, (2)M = - 2.96 x 1074t(,lK)~[y], (3)1 * 5 4 ~ 1 0 - ~ Klog A = log [ ($1 4t"YI(4)was found from the magnitude of log (I/V2) at the mid-value of (1/V) in the Fowler-Nordheim plot.In the following sections variations in field-emission pattern (at constant emissioncurrent) and changes in work function and log A are reported for chlorine, carbon-chlorine compounds, and, for purposes of comparison as discussed later, oxygenand methane.CHLORINEFig.1 (a)-(c) shows some patterns observed, starting with clean tungsten whenchlorine was adsorbed, initially at 5 x 10-9 torr ; the pressure was then reduced anM. J. DWELL, B. 3. DAVIS AND R. L. MOSS 45the chlorine desorbed, (d)-(f), by heating the tip at a series of temperatures. Thetypical granular appearance of the adsorption patterns was evident after 3 min,and after 9 min the pattern continued unchanged up to the longest time for whichchlorine was adsorbed (36 min). The patterns were photographed at constant emis-sion current and the applied voltage had to be increased to produce this currentthroughout the adsorption period.However, since the dark (1 12) planes (cf. fig. 2)remain a constant size, the decrease in the dark area of the central (011) and corres-ponding " corner " planes in fig. l(c) is probably a real effect. The work functionwas approximately constant between 20 and 36 min at 56-57 eV in a final pressureof 2 x 10-9 torr and gave an indication (4 = 5-26 eV) of desorption on heating for1 rnin at 1170"K, fig. l(d). The work function decreased on further heating (at2100°K for 30 sec, q5 = 443 eV ; fig. l(e)) to a minimum at 2200°K (4 = 3-86 eV;fig. l(f)). Flashing at 2400°K produced an apparently clean tungsten surface but4 at 4.36 eV was still below the expected value, 4.52 eV.FIG.Z.-Disposition of the p h e s exposed on tungsten tip with (011) plane central, based onstandard stereographic projection.Fig. 3 shows Fowler-Nordheim plots observed for chlorine adsorption at1.5 x 10-9 torr, derived from interpolation plots as described above for times up tot = 250 sec, after which the normal method of taking a voltage-current character-istic could be used. Although the plots change position rapidly in the early stagesof adsorption, nevertheless satisfactorily linear plots were obtained. Values oflog A are recorded in table 1 and work function values calculated from these plotsare shown in fig. 4 (open circles). A " spot " value was obtained, without using theinterpolation procedure, by admitting chlorine at 1.5 x 10-9, then closing the gasinlet valve immediately (half-filled circle). At t = 30 min, the work function wasTABLE 1 .-Loglo A VALUES FOR CHLORINE ADSORPTIONtime (min) 0 0.8 1.7 2-5 3.3 4.2time (min) 7 10 15 20 25 30h 9 0 A -7.98 -8.33 -8.65 -8.98 -9.00 -8.51log10 A -88.27 -8.22 -8.13 -8.12 -8.00 -77.946 FIELD-EMISSION OF Cl;! AND c - c l COMPOUNDS5.44eV; on raising the pressure tow5 x 10-8, holding at 2 x 10-8 and finally re-ducing to 1 x 10-9 ton, 4 = 5.40 eV was found.In another adsorption experi-ment the field was only applied briefly to photograph the pattern to confirm thatoperating according to (iii) above had not increased field effects above the essentialminimum.Ih 2 G a b -&- ,W14.0-I-I2-5 3.0 3.5 4.01041 vshown.FIG. 3.-Fowler-Nordheim plots for chlorine adsorption at 1.5 x 10-9 torr after elapse of timesi? L5 .0 0 rI i4.75 c iI~ _ ~ ~ L10 15time (min)FIG. 4.Variation of work function with time (lower scale) during chlorine adsorption at roomtemperature : 0, calculated from plots in fig. 3 ; (3, chlorine " dose ", cf. text ; a, effect of heatat low coverage ; 0, removal of adsorbed chlorine with methane, upper time scale.A characteristic feature of the change in work function during chlorine adsorptionwas the initial decrease at low coverages, and this was examined further in thefollowing experiments. Chlorine at 1 x 10-9 torr was admitted to the microscopM. J. DUELL, B. J. DAVIS AND R. L. MOSS 47and the tip Sashed, the inlet valve was then fully closed and also the chlorine wasremoved from behind the valve.In the falling chlorine pressure, the work functionwent slowly through the minimum (fig. 4, filled circles) to 4-64 eV. The temper-ature of the tip was then increased at intervals of -70" and at 1320"K, 4 fell backto 4.39 eV, again less than the clean tungsten value. Chlorine was also adsorbedaccording to procedure (ii) above, i.e., the tip was flashed in - 10-10 torr background,allowed to cool for a short time, voltage-current data recorded and then 1 x 10-9 torrchlorine admitted. Work function values observed in the early stages were 4-52,4.50, 4-54, 4-59, 4.62, 4.62 eV for t = 0, 50, 100, 150, 200,250 sec respectively.The activity of hydrogen and methane towards the chlorine-covered tip was alsoexamined.As a result of experiments described above, it was thought that anychanges observed at 920°K or less could reasonably be ascribed to the hydrogenor methane rather than desorption with temperature. Chlorine was adsorbed for60 min (# = 5.79 eV), and 2 x 10-8 torr hydrogen admitted to the field-emissiontube and the temperature raised in stages to 920°K with little effect. A more positiveresult was obtained using methane at 6 x 10-8 torr at room temperature whichsteadily reduced the work function of a chlorine-covered tip from 5-46 to 4-86 eVover 50 min (fig. 4). Field-emission patterns observed during the removal of ad-sorbed chlorine with inethane are shown in fig. 1 (g)-(j).METHANE AND OXYGENShort studies were made of methane and oxygen for comparison with the resultsreported above for chlorine and carbon-chlorine compounds.With methane, itwas of interest to examine how far the patterns and work function changes ob-served with chloroform and carbon tetrachloride might be attributed to carbonresidues. The occurrence of a minimum in work function and other evidence forthe corrosion processes which are believed to occur when chlorine is present, weresought using oxygen; such effects were expected to be absent during methaneadsorption.Fig. 5 shows that a small increase in work function occurred, without a minimumat low coverages, when methane was adsorbed at 2 x 10-8 torr ; at t = 20 and 50 minthe patterns shown in fig. I@)-(Z) were observed. At a pressure of 5 x 10-9 torrmethane, # was 4.52, 4-57, 4.60, 4-63 and 4.67 eV after 0, 95, 260, 470 and 650 sec.In contrast, with oxygen at 4 x 10-9 torr, there was an initial minimum, followedby a sharp rise in work function (fig.5) ; in a lower oxygen pressure, 1 x 10-9 torr,the observed minimum was 3.97 eV. Fowler-Nordheim plots are not reproducedbut values of log R are summarized in table 2.TABLE 2.-LOglo A VALUES FOR METHANE AND OXYGEN ADSORPTIONtime (min) 0 0.8 1.7 2.5 5 10 15 20 25oxygen -8-30 -8-91 -9.41 -7-99 -8.62 -8.79 -8.84 -8.74 -8.82methane - 8-61 -8.24 -8.18 -8.21 -8.36 -8.53 -8.55CARBON TETRACHLORIDE AND CHLOROFORMCarbon tetrachloride at a pressure of 3 x 10-9 torr was passed continuouslythrough the field-emission tube, the tip flashed and the field applied intermittentlyto obtain the patterns shown in fig.6(a)-(d) (and also to record voltage-current data).At 2 min, the (1 11) planes appeared relatively bright and there was also emissionfrom the (001) and (010) positions (cf. fig. 2). The bright granular band whichdeveloped around the central (011) plane after 5 min, linked to diamond-shape48 FIELD-EMISSION OF Clz AND C-CI COMPOUNDSbright areas centred on the (001) planes, fig. 6(c), is associated with adsorbed chlorine.This pattern then evolved gradually into fig. 6(d) as the diamond-shaped areasdarkened ; at 8.5 min the circular dark areas centred on (001) and (010) thus producedhad a reasonably clear outline. After 30 min the pattern was essentially similarbut more granular in appearance.0 5 10 15 20 25time (min)during methane adsorption at 2 x 10-8 torr, both at room temperature.FIG. 5.-Variation of work function with time: 0, during oxygen adsorption at 4 X 10-9; 0,The tip was then heated after removing carbon tetrachloride vapour from thefield-emission tube.With increasing temperature up to 1050"K, the circular darkareas became brighter and heating at 1170°K for I min (fig. 6(e), q5 = 5.35 eV),restored the bright diamond-shaped areas observed during the adsorption sequenceat 5 min. Further changes in pattern occurred at 1370 and 1465°K with a reduc-tion in the work function to 4-65 and 4.58 eV respectively, but the tip was stillpartly covered with adsorbed residues (fig. 6 (f)-(g)), apparently completely removedby heating for 1 min at 1860°K.The work function values corresponding to the above adsorption sequence areshown in fig.7 (open upright triangles); the interpolation procedure was not usedand the exact depth of the minimum, but not the later values, is uncertain. A furtherexperiment is also shown (open inverted triangles), where the inlet valve was closedafter admitting 1.5 x 10-9 torr carbon tetrachloride vapour to the continuouslyevacuated field-emission tube and the tip flashed. The slower adsorption shouldenhance the accuracy of the data in the region where the minimum occurs, but alsoincreases the possibility of contamination by residual gases. Fig. 8 shows Fowler-Nordheim plots obtained with a reasonably large carbon tetrachloride pressure(final pressure 5 x 10-9 torr), admitted to a cold tip, but the interpolation procedurewas used to increase the accuracy.The occurrence of the minimum is apparentfrom the slope of the plot for t = 50 sec; the work function values derived fromthese plots are also shown (filled triangles) in fig. 7 and log A values in table 3.TABLE 3.-Loglo A VALUES FOR CARBON TETRACHLORIDE ADSORPTIONtime (min) 0 0.8 1.7 2.5 3.3 4.2 5 6.7log10 A -8.68 -10.67 -9.56 -8.99 -9.00 -8.68 -99.00 -8.8FIG. 1 .-Field-emission patterns for chlorine adsorption/desorption : (a) clean tungsten ; (b) ad-sorption in 5 x 10-9 torr chlorine for 3 rnin ; (c) after 9 rnin ; ( d ) chlorine desorbed by heating for1 min at 1170°K ; (e) for 30 sec at 2100°K ; (f) for 30 sec at 2200°K.Reaction of adsorbed chlorinewith methane at room temperature : (9) chlorine covered tip ; (h) after 7.5 rnin in methane ; (i)after 24 min ; (j) after 40 min. Methane adsorption at 2 x 10-8 torr on clean tungsten : (k) after[To face page 48.20 rnin ; ( I ) after 50 minFIG. 6.-Field-emission patterns for carbon tetrachloride adsorption/desorption : (a) clean tungsten ;(b) adsorption at 3 x 10-9 torr for 2 min ; (c) after 5 rnin ; ( d ) after 8.5 min ; ( e ) after heating at1170°K for 1 min in vacuum ; (f) at 1370°K for 1 rnin ; (9) at 1465°K for 1 min ; (h) chloroformat 2 x 10-9 torr adsorbed for 38 min on clean tungsten ; (i) after heating in chloroform vapour at1170°K for 1 min; ( j ) at 1260°K for 1 min; (k), (I) at 1370°K for 1 min, after which the field wasapplied continuouslyM.J. DUELL, B. J. DAVIS AND R. L. MOSS 49The work function at these low pressures of carbon tetrachloride apparently increasesto a steady value of 5.7 eV.Similar adsorption experiments were performed with chloroform, producingfield-emission patterns which closely resembled those from carbon tetrachloride,6 . 0 j.- i A,/< A / . - -' +'* /' /------- 5.5 c-I01 1 I 110 20 30 40time (min)FIG. 7.-Variation of work function with time during carbon tetrachloride and chloroform ad-sorption at room temperature : A, carbon tetrachloride corresponding to series of patterns shownpartly in fig. 6(a)-(d) ; v, gas " dose ", cf. text ; A, gas admitted after flashing tip, calculated fromFowler-Nordheim plots in fig.8 : 0, chloroform adsorption at 2 x torr.I I2 . 3 2.5 3'0 3.5104117FIG. 8.-Fowler-Nordheim plots for carbon tetrachloride adsorption at 5 x 10-9 torr using inter-polation procedure, cf. text.e.g., fig. 6(h) is the " steady " pattern, 4-53 eV (cf. fig. 7). However, experimentswere also carried out where chloroform vapour was reacted over the heated tip.Increasing temperature did not produce the gradual brightening of the circulardark areas around (001) and (010) positions observed when the tip was heated inthe absence of vapour. Instead at 1120"K, the circular dark areas were sharper in50 FIELD-EMISSION OF (212 AND c-c1 COMPOUNDSoutline and at 1170°K showed fine structure, fig. 6(i). The presence of carbonindicated by these patterns was shown clearly by heating to 1260"K, fig.6(j), whichresembles closely the carbon pattern,5 with its characteristic (334) planes.Operating the microscope with the field applied continuously, procedure (iv)above, in the presence of carbon tetrachloride or chloroform vapour can give riseto patterns such as fig. 6(k) where intensely emitting " spots " are " superimposed "on the expected pattern. In this case, the tip had been heated at 1370°K in chloro-form vapour, continuing the experiment described above. The " spots '' also showedfine structure which changed rapidly between various simple forms, doublets, quadru-lying pattern was suppressed, fig. 6(2). It is believed that these intensely emittingspots are due to the occurrence of projections or whiskers of tungsten on the originaltip.They are not formed by chlorine or carbon alone, nor indeed by oxygen, andcan be formed on the unheated tip after adsorption at room temperature of thecmbon-chlmine compunds.I#&%, p,tr,. ; C%n-t>PJJ.$3 tJxy d4Lwi'TPJbtd tb2" ewi%%kYA &a.wJb!i**iG% Q& that& thfz WXkT-DISCUSSIONSurface potential values for the adsorption of nitrogen on individual tungster iplanes vary not only in magnitude but also in sign,8-10 and average values for all[planes exposed on the emitter tip must be treated with caution. However, thc:observation of a positive surface potential, i.e., the initial work function decreascto a minimum shown in fig. 4, with a halogen, in this case, chlorine, merits furthei-consideration.Observations were started by flashing the tip to produce a clean surface, after.switching off the ionization gauge used to regulate the chlorine supply through the field-.emission tube.This avoids any unusual species produced by the gauge 11 and minim-.izes contamination of the tip by foreign gases desorbed in the early stages of the:experiment. Under these conditions, adsorption started on a tungsten surface:which had been subjected to thermal disordering and was still cooling to roomtemperature. When chlorine was admitted after allowing the tip time to cool,,the minimum was only just apparent. Heating the tip at low chlorine coverage:produced an unambiguous, if small, positive surface potential (cf. fig.4), and a larger.one, +0.66 V during desorption from a well-covered tip.Carbon tetrachloride, and also chloroform, are believed to dissociate on tungsten,,yielding adsorbed chlorine, as discussed later ; again positive surface potentialswere observed initially (fig. 7), but of greater magnitude. The minimum was present.even when carbon tetrachloride was admitted after the tip had been allowed to cooldown. Results with these compounds and also with chlorine suggest that reductionin pressure delays the eventual increase in work function after the minimum andprobably also increases the positive surface potential observed.There are many references 12 to enhanced photoelectric emission or to positive.surface potentials when a small quantity of oxygen is admitted to clean surfaces.The effect of halogens on the work function of metals is less well explored but positivesurface potentials have been observed at low coverages of chlorine on nickel13and titanium 14 films and in field-emission studies of bromine on tungsten? Thesize of the halogen is also important, thus, fluorine (and also oxygen) on iron at20°C produce positive surface potentials, unlike chlorine or iodine.15It now seems generally agreed that electronegative gases can penetrate below themetal surface producing a positive surface potential, providing 16 a certain criticaltemperature is reached.Direct evidence of the penetration of oxygen below M. J. DUELL, B. J. DAVIS AND R. L. MOSS 51tungsten surface is available from field-ion micrographs observed after the removalof successive metal layers by field evaporation.17 Penetration was found two layersdeep at 20°K within a few minutes and as deep as five layers if the tungsten surfacewas exposed to oxygen at room temperature.Attack began at lattice steps butpreferred oxidation at lattice imperfections could not be discovered. The detectionof positive surface potentials by field-emission using the procedures described, wasconfirmed by adsorbing oxygen which showed an initial work function minimum,and by adsorbing methane which as expected, did not (fig. 5). Again, the observedpositive surface potential was greater at a lower oxygen pressure.Therefore, the work function values observed when gas-phase chlorine wasadsorbed on tungsten or when carbon tetrachloride or chloroform dissociated togive adsorbed chlorine are believed to arise from two processes.(i) Penetration ofchlorine below the tungsten surface, as discussed later, causing a reduction in workfunction. The process is enhanced when the temperature of the tip is raised, eitherduring cooling after flashing in the gas, or by deliberate resistive heating. (ii) Ad-sorption of more chlorine above that which has penetrated the surface, i.e., essentially" chloride " formation, or adsorption on areas which it cannot penetrate. Sinceemission is primarily from low work function areas, further adsorption over theareas initially penetrated would appear to be responsible for the rise in work functionafter the minimum.Reducing the pressure would be expected to slow down thisadsorption process and hence the work function minimum due to (i) becomes morereadily observable. Further, the lower work function at the minimum observedwith the compounds compared with chlorine can be understood in terms of therelatively slow release of chlorine from, say, carbon tetrachloride. In effect, thisis the equivalent of a very low chlorine pressure.As a preliminary to examining process (i) further, the basis of the work functioncalculation from the Fowler-Nordheim plots, i.e., the constancy of ( r / K ) during anexperiment, might be considered briefly. Certainly, before the start of each ad-sorption sequence and after flashing clean, (r/K) is reasonably constant for a giventip.This indicates that the macroscopic dimensions of the tip were little changedby the adsorption process, but protrusions could have been formed leading to fieldenhancement and an apparent reduction in work function. However, the reductionwas observed at low coverages and disappeared on further adsorption which wouldbe expected to follow the surface contour.Penetration of the surface by chlorine, observable in the initial stages, appearsto be accompanied by a reduction in emitting area, derived from further analysisof the Fowler-Nordheim plots (fig. 3 and 8). For a partly covered surface, theFowler-Nordheim equation should be modified for polarization of the adsorbateby the field.where N is the number of adsorbed species per cm2 and a is the polarizability of theadatom.Hence the ratio of the pre-exponential terms for clean tungsten, Ao,and after a given period of adsorption, A , is given by 2, 10Polarization produces a work function change A# given by7A+ = 4nNaF ( 5 )=--- '' A , a0 (4+A+) t"Y1'"yo' exp (--$4;nNor#* x 6.83 x lO'v[y]).A a -At low coverages, say 8 = 0.1 ( N = 5 x 1013 atoms/cm2), a = 3.6 x 10-24 cm3,184 = 4.45 eV (i.e., the minimum for chlorine) and u[y] = 0 6 , the exponential termis evaluated as - 1-3. Since A+ -0.1 eV, and also t[yo] and tb] are sensibly unity,the ratio (A/&) is a reasonable guide to the variation in emitting area. For example,352 FIELD-EMISSION OF C12 AND C-C1 COMPOUNDSat the observed minima for chlorine, oxygen and carbon tetrachloride, the emittingareas have decreased by approximately 10, 10 and 100 times (tables 1-3).Examination of fig.6(b), corresponding to the minimum work function withcarbon tetrachloride, shows clearly that substantial areas, compared with the cleanpattern, were still brightly emitting. Hence it would appear that electrons wereemitted from small areas distributed over the surface where chlorine has penetratedand decreased the work function. The relative brightening of the (111) planes,the areas around the (001) planes and the ring around the central (011) plane, mayindicate preferential attack on these highly stepped regions ; initial attack at stepswas noted above for the corrosion of tungsten by oxygen.17 A further point ofinterest is the subsequent increase in log A, when the surface was covered, to ap-proximately the value observed at the start of chlorine or carbon tetrachloride ad-sorption. This indicates that emission from the covered tip was occurring gener-ally, whereas as low coverages it occurred through “windows” of lower workfunction where chlorine had penetrated the surface.DISSOCIATION OF CARBON-CHLORINE COMPOUNDSIt has been assumed in the discussion above that carbon tetrachloride and chloro-form dissociate on clean tungsten at room temperature producing adsorbed chlorineand field-emission observations on this point are considered first.(a) In additionto the work function minimum observed with both chlorine and the carbon-chlorinecompounds, the final work function values were similar, i.e., 5-6-57 eV for chlorine,5.4 eV at lower pressure, compared with 5.7 and 5.5 eV for carbon tetrachlorideand chloroform respectively. Further, carbon has little effect on the work function 4and as shown in the present work, adsorbed methane has only a small effect.(b) Patterns observed early in carbon tetrachloride adsorption and after heating acovered tip (fig.6(c) and (e)) and also in chlorine adsorption (fig. l(c)) are closelysimilar. The pattern shows a bright granular band around the central (011) plane,linked to diamond-shaped bright areas centred on the (001) planes and is associatedwith adsorbed chlorine. (c) There was an indication of chlorine desorption at1170°K (heated for 1 min, Q) = 5-26 eV) and with carbon tetrachloride, the patterndescribed above was restored (4 = 5.35 eV).Thus the presence of adsorbed chlorine from the dissociation of carbon tetra-chloride and chloroform shows up reasonably well on the patterns and in the workfunction measurements.However, the continued dissociation of these compoundson tungsten at room temperature eventually produces patterns with one readilyobservable difference with respect to adsorbed chlorine. This difference is thesubsequent darkening of the bright diamond-shaped areas producing large circulardark regions also centred on the (001) planes (cf. fig. l(c) and 6(d) or (h)). Thesedark areas were not observed when a chlorine covered surface was attacked by gas-phase methane (fig. l(g)-(j)) which clearly removed the chlorine, i.e., the character-istic chlorine pattern disappeared, to be replaced by the methane pattern (cf.fig.1 (I)). Also the work function decreased substantially, levelling out towards thevalue for adsorbed methane, i.e., 4.8 eV (fig. 4). Since hydrogen had little effecton adsorbed chlorine, it would appear that chlorine was removed as a compoundwith carbon.There is evidence, however, that at room temperature, carbon tetrachloride andchloroform yield in addition to chlorine, and possibly carbon, fragments of thetype CCl,. Under a continuously applied field, intensely emitting spots believedto be due to the formation of tungsten whiskers are produced. Since whiskerM. J. DUELL, B. J. DAVIS AND R. L. MOSS 53cannot be formed by carbon or chlorine alone, it would seem that a complex involv-ing both of these and surface tungsten atoms is required. This complex migratesto the end of the whisker where it dissociates and deposits tungsten.1 Moss and Kemball, Trans. Faraday Soc., 1960,56, 1487.2 Duel1 and MOSS, Trans. Faraduy Soc., 1965, 61, 2262.3 Silver and Witte, J. Chem. Physics, 1963, 38, 872.4 Klein, J. Chem. Physics, 1953, 21, 1177.5 Muller, in Physical Methods in Chemical Analysis, ed. Berl, (Academic Press, New York,. 1956),6 Good and Muller, in Handbuch der Physik (Springer-Verlag, Berlin, 1956), 21, 176.7 Gomer, Field Emission and Field Ionization (Oxford University Press, London, 1961).8 Oguri, J. Physic. SOC. Japan, 1964, 19, 83.9 Holscher, J. Chem. Physics, 1964, 41, 579.10 Delchar and Ehrlich, J. Chem. Physics, 1965, 42, 2686.11 Archer and Gobeli, J. Physics Chem. Solids, 1965, 26, 343.12 Klemperer, J. Appl. Physics, 1962, 33, 1532.13 Anderson, J. Physics Chem. Solids, 1960, 16, 291.14 Anderson and Gani, J. Physics Chem. Solids, 1962, 23, 1087.15 Burshtein and Shurmovskaya, Surface Sci., 1964, 2, 210.16 Riviere, Brit. J. Appl. Physics, 1965, 16, 1507.17 Muller, in Structure and Properties of Thin Films, ed. Neugebauer, Newkirk and Vermilyea18 Syrkin and Dyatkina, Structure of Molecules and the Chemical Bond (Butterworths, London,3, 135.(Wiley, New York, 1959), p. 476.1950)