The Color of the Ruby - The Journal of Physical Chemistry (ACS

---

THE

C'OLOR O F THE Rt-Bl_____

BY CHARLES W. STILLWELL

In 1 8 3 ; Gaudin' found that by atlding a minll anioiint of green cllromjc oxide t o aluminuni oxide and fusing the mixture in an ox?-hydrogen flame, the color of the natural ruby could be reprotluced. Since that time, m t l tloubtless long before, it has heen cssunietl thet the ruby owes its lirillia'nt color t o a tract of chromium. This is justifieblr Iiecawe the natural ruby is known to contain a trace of this nietxl. On the other hand. chromium is also held responsjhle for the green color of the The rraction of the average man of int,elligence who is confrontetl n-ith these facts is one of wonder, if he be of n tolerant nature: antl one of ridicule ant1 tlenial if he lie intolerant. The attitncle of investigators antl authors in the field of precious stones has been open-minded t o the point of neglect. Some3leave the matter coiiipletelj1111 in the air as does Gootlchilcl,-L"Traces of ferric oxide, silica antl oxide of chromium are found in even the purer varie:ies often antl it is supposed (though it is by no means certninj that these substances. slightly vnrying in anioiint ant1 proportion, give to the ruby, the sa,ppliire m t l the other precious forms of corunduni their beaut~ifulcolors on which the value so hrgcly depends." Others; state the fects;.ai; though firmly esta1)lishecl. n-ith the mntterof-factness of Kraiis and "This green color [of the emerald] is t'hought to be due t o a trace of chromium. -4s the rctl color of the ruby is also ascribed t o the same element, it ohviourly mixt exist in different states of combination in t'he tn-o minerals. ;Iccortlinglj- in miner compounds chromium may produce either a green or a ret1 t o violet color." The authors who have heen quoted represent fairly the attituck of the literature in geneid on this question. K e have ::tlvancetl t o the point wkere we are willing t o say definitely that the ret1 color i s 1:rctluced 11y chroniium. For the most part', however. we assert, it glihly ~s thciigh i t viere a n everyday occurrence to color a substance retl by aclrling a green coloring matter to i t ; and lit'tle progress has been iiiatle toward an es1:l:in::tion for this ~:heccnienon. The issue has bctheretl those of the ceramics industry in the form cf the chrome-tin pink. They have been com~elleclthrough nece.esity t o investigate some of the con:litions under which chromic oxide causes a retl or pink color in stannic oxide. The data which haT-e been collectetl are ~-oluniinoiisant1 full of cont'radictory evidence. The results obtained are largely empirical G n u d i n : C o m p t . rend., 4, 999 ( I IT-. R.Cnttelle: "Plec~ioiisStonc Gem lIaterinls," page 124 i 1 9 2 ji . . K u n z : "Gcms :incl Prccious Stone

pagc~39 ( 1892 I : IV, I?, ('ntt:~ll(~:"Pwciou-;

c i o w Stone.;," pngc' 18: i 1gc;8). l l u s c ~ u mBullctin 118, I . " Kmus nnd Holdm: "Gems :ind Geni lI:iterial$." p g c 124 1925).

j

c;. P.lIerri11: I-.7 . S n t i o n d

1342

CHARLES TY. STILLlYELL

a.nd bring u s little neerer a n explnnaticn of the retl color. -1s simmarized by Bancroft' they d o indicate xveral general conclusions. The red color is ohta'ined n-hen a comparatively low lmcentege of chromic oxide is added to stannic oxic!e C Y other white oxides or mixtures of v-l-ite osic!es: and as the chromic oxide content is increafed the color chacges t o green: usually going through an intermediate 1)ron.n. The roint at xhich the color change occurs, in reference to chromic osicle content, is influenced by the particular oxide which is present. Finally. in a niajcrity of cares a t least, the formation of the red or violet color takes p1:tce in an oxidizing atmosphere while a retlucing a tmosphcre produces the green. The present' investigation was untlertalien for the purpose of determining the conditions untler which clironiic oxide i i n p r t s a ruhy color to corundum as distinguished from those conditions causing a green c8oloration : ancl secondly t o ascertain in what form or state of coiiihinntion the chromium exists in corunduin t'o produce the red color. Possible explanations of this red or riihy color caused 11)- the presence of chromiiini ha\-e been offered. It has l w n suggested2 that, it W V R S due t o colloitlal chromic oxi(!e present in the aluniiniini oxide, although no experimental evidence either for or against this x-iew ~ n furnished s wit8hthe suggestion. I t is at least a plausihle theory. for we know that a change in the size of the Iiarticles of many metals is ::cct;mpanietl by a change in color, p ~ has lleen prepared which usually \-cry s triking.:' Finely tli\.-itletl c o ~ i oxide is blue.4 Scheet'zjfound that ferric oxide was \-ellon jn t'lie presence of alumina. Since the 5:niounts and nature of the oxitlw to TT-hich the chromic oxide is added npparcnt'ly afTect t h e concentration :It8Tvhich the mixture changes from red t o green, we might slippow, therefore, that these other oxidcs in some way stabilize colloidal chromic ositle. In order t o test this hypothesis. finely tlivirled chromic oxide was pxparecl tiy electrical disinicgration accortiing t o thc method of Svctllierg.fi -1hightension owillating arc n-:is passetl, untler w.tcr, lietween tu-o chromic oxide electrodes TI-hich hat1 hern preprec! by fusing the r:ow-c!er in a con;-cnient shape. The clectroc!e clisintcgrntetl noticeably hut tcntlctl ton-art1 a, yellowish green color nntl at no point in the experiment n thcre any suggestion of retl in the color of the ljarticles formed. This ycllon-ish color TKW found to lie clue t o chrciiiic acid. Finel)- tlivitlctl chromic oxitlc vias eventually prepared using tke method tlcscrilitd 1)). AIott.? ( ':irlmn clcctrodes were set vertically and the metnllic ozi,!c plncetl in the lon-cr. ciip-~lialxdelectrode. 1 5 0 volt'. 2 0 nni!:ere direct current furnishes the erc, ca~isiiigtl:c oxide t o volatilize and , Chcin.. 2 3 . 6.3 1,1919). i'nni. Ain. C -Chc111.. s. 21 , i i O ( I g I i " . 5 d r c t l i ~ e r g :"Die 1Icthoileii zur Hcrstelliinp kolloitlcr T,iisungen :morgnnischer Ytoffe," page 424 !19c91. ' T r a m . -1m. ~lcc.troc.2iemic~~~l So?., 34, 2 , 5 ~'1918). - r

T H E COLOR O F THE RUBY

1443

deposit on the uppei electrode. Mott found that the size of the particles deposited decreases as the distance from the arc increases. The results which he obtained with ferric oxide were first' duplicated, a greclual change in color from red to yellow occurring as the distance from the arc increased. This agrees with the Itnonn fact that finely divided ferric oxide is yellow. K h e n the experiment ~ ~ 2repeated , s with chromic oxide, however, there was no color change evident except a paling of the green toward white, n-hich is expected t o accompany a decrease in the size of the particles antl is noticeable n-hen the green is ground in a mortar. It is evident from theFe results that finely divided chromic oxide is not red antl that 1%-e must look for another explanation of the red color. The results of some investigaticns of the chrome-tin pink, to which reference has already been made. suggest that the red color mE8ybe caused by the oxidation of chromic oxide to some higher oxide. This view has enjoyed considerable popularity in the past. Kat'ts' states thet n-ith mixtures of chromic oxide antl hydrous aluminum oxide ground together antl calcined in an oxidizing flame. he obtained a pink colcr n-hen there was 9.47 percent. or less, 3f chromic oxide in the find mixture. If the same mixture were calcined in a reducing flame a hriglit chrome green resulted. Seger? says that with mixtures of chromic oxide, stannic oxide antl calcium oxide the red is destroyed 11)reducing conditions and reappears when oxidized. Khead3 notes a mixture colored b y chromic oxide which gave a green in a fritt kiln antl a red in a muffle kiln. ;In attempt has been made t o duplicate the results obtained Iiy K a t t s . Hydrous aluminum oxide was precipitated from a solution of alum. dried a t 140' c'. and ground together with chromic oxide in the proper amounts to form a final mixture conhining 90.6 1:ercent of aluminum oxide antl 9.4 percent' of chromic oxide. This mixture was ciilcinetl a t a tcnirerature of approximately 1490' C'. in a n .Ijax induction furnace. T o ensure a strong- reducing atmosphere, the mixture was placed in m iiliintl~.ini crucible which was in turn fitted into a piece of graphite. -4s a result. the furnace TWS filled with carlion monoxide throughout the run. Several trials were niatle 11 ith .this mixture varj-ing the temperature and the diiretion of the calcinntion. ant1 in ever!. case the resulting pon-der n-as of a distinct Iaventler gr retltlish lavender color: antl there TI-as not, the least tendency ton.artl a grcen. The,:e results ir-ere independent of the method of cooling the pow!er. I t could he removed when red hot antl cooled in the air or it coulcl tie cooled to room temperature in the reducing atniosphere without' a n y effect on t h e final color. There is a minimiin3 temperature lielow which no rezction takes place, leaving the powder the original green color. Tlik may I:e msjly tlistinguishetl from z green which niight be a product of the reaction. however. since a microscopic examination of the powder shon-9 a mixture of green :mtl white particles,

* J. I m . Cclnmic

Soc., 13. j o r :1911I . Tiam. Ceramic doc.. 4,230 (1902'1. J. -1m. Ceramic i5oc., 13, 3 2 1 (1911).

I444

CHARLES I T . STILLKELL

different in appearance from the blended color resulting when the temperature is high enough t o cause the reaction t o take place. I t is interesting t o note that this diffusion of the lavender color into the nhite aluminum oxide takes place far below the melting point of either constituent. The same results are obtained as regards color in a reducing atmosphere if the mixture is completely fused. K e have three semyles of alpha aluminum oxide furnished by the Xorton Company. containing roughly 0 . 5 > 5.0 and 15.0 percent of chromic oxide respectively, with traces of other impurities. These were prepared by fusing the oxideq in an electric arc, in which case the atmosphere is doubtless reducing. The samples are all of a bluish red or lilac shade, increasing only in saturation as the chromic oxide content increaces. Theee results show that we do not obtain a green color n i t h a reducing atmosphere necessarily: anil it follou-s, of course. that a n omlizing atmosphere is not necessary t o proJuce the red. It is equally true that v i t h a reducing atmosphere n-e get a bluichred color rather than a good ruby red and a possible explanation of thi. i c offered later.' I t has been shown? that with solutions of chromic chloride and man! other chroniium salts, the color depends on the thicknesc through which light is transmitted or on the concentration of the solution. A chromic chloride solution goes from green t o violet by transmitted light as the thicknesq of the solution transmitting the light is increaced. If the thickness he kept conPtant and the concentration of the chromic chloride increwetl, the color of the solution changes from preen t o Yiolet. The suggestion has been mer'e th2.t the red or lavender color caused by chromic oxide in aluminum oxide is Pimilady a concentration effect. T h i c is iniyorsible, lionever, since we find an increase in concentration of chromic o ~ i d ein cluniiniim oxide rausing the color to change from bluish-red or red t o green, nhile in the czse of the solution it is just the reverce. I t may aleo be seen from (lata given on page 1447 that mixtures of chromic oxide and aluminum oxide of identical composition may be either red or green depending upon the conditions under which they vere prepared, Recently the S o r t o n Company has found that sinal1 amounts af chromic oxide color d F h a alumina red or bluish-red, while the same yercentage of chromic oxide always colors beta alumina green. This suggests the hypothesis that there may be two modifications of chromic oxide, one red or bluish-red and the other the usual green color; provided, of course, that the alpha and beta alumina have different crystalline structures. The existence of beta alumina w-es first discovered a t the Geophysical 1,aboratory and ha. been described by Rankin and Mervin.3 The beta alumina used in our work is prepared by the Sorton Company by adding m sulfate were clissolved in water in the proper proportions to yield the desired ositle misture, Frecipitated with ammonia and the h\-drous oxide precipitete \\--ashed. filtered, dried and calcined to remove the water. The powder thus formed 112s the advantage of being as nearly homogeneous as possible. es this is a nece:.ar>- factor in General Electric Review. S p p t . 119221. "Structure of Crystals." page 149 (1924'1. '.Inn. Chim. Ph>-a..(8)3 , 20 i1904).

I446

CHARLES Ti. STILLX-ELL

1-erneuil's process. I t as found t o he very fluffy and easily blown away the force of the blast, hcwever: and so the mixtures finally used were made up by mixing aluminum oxide and chromic oxide directly. The matter of homogeneity is not so important in this work beceuce the ent8iremass beccnies fused and the entire saniple is used, and hence no discrepancy in composition can occur. An oxy-hydrogen hlast l a m p has been used in all the experiments recorc!ed here to effect the fusion of the mixtures. Aismall pile of the powdered mixture is placed on an alunduni slab and a very gentle oxy-hydrogen flame directed upon it. If this first step is done carefully. little p o ~ d e ris h!onn away ljy the force of the flame. h consitlerable portion of the pile will soon sinter together somen-hat' and the force of the flame may then be increased until c jmplete fusion occurs, when the melt will become mobile and assiinie a spherical shape. Unless the heat' becomes so intense that this molten sphere is flattened against the alunduin by the force of the blast'. no difficulty is encountered in removing the sphere from the slab when it cools antl the surface of cont,act between the two is so sninll that, th.ere can be no appreciable contamination of the product hy the slab. In general it was found convenient t o prepare globules Qf the I x d u c t of not more than two or three niillimeters in diameter. The main objection t o preparing larger samples by this method lies in the fact' that if the mixtures lie heated w r y much above the fusion tem1:erature the chromic ositle begins t o volatilize, and if a larger sample is taken, boiling occurs on the suface hcfore the inner part has hecome fused. This takes place even with comparatively gr:ttlual heating. Since in all this work the composition of the product is assumed to be that of the mixture liefore fusion it is essent'ial that any boiling during fusion be prevenletl. The following mixtures were prepcred antl fused : Percent .41?03 00

I ' < ~ l ~ ~Cr l- percent chromic oxide in :in oxidizing atmosphere ~ ~ - o u lproduce tl sciiie of the higher retl t h a consequentm tendency ton-artls yellon- in .the color of t h e niixturc. Therc is no noticenlik difference in t h e shac!c of the green protlLicts. lioiwver. It is evit!cnt. then. that the tlifference in color cannot be tlue to different tlpgrees of oxitlation of the chromiuni and it naturally follows that 110th t h e red nntl preen colors arc tlue t o the presence of chronic oxide. In other words we 1ia.i-e t h e same I h y l e i g h : S a t u r e , 3 , 234 i 1871).

1338

CHARLES W-.STILLTI-ELL

chemical compound differing in at least one physical property, that of color: and this is strong evidence of the existence of chromic oxide in two different modifications, one red and the other preen. There is an intermediate range fr3m 3 0 to 4.;percent of chromic oxide a t least n-here either a red or a preen color may be formed. A t any definite concentration of chroniic oxide nithin this range. therefore, either one of the modifications me;\ be ohtained but one must he stable and the other metastable, In order to Jetermine the effect of either color or chromic mide content on the crystal structure of the pro lucts. X-ray diffraction patterns of Eeveral were made by the ponder inethocl. A casual inspcction of the films macle it irnmedia tely obvious that the a x i d ratios in the crystals were affected noticeably by the chramic oxide content and also hy either the color or the conditions under which the fusion takes place-or both. Davey’ has measured the axial ratio of alpha alumina as 2 . 7 2 6 and that of preen chromic oxide a s 2 . 7 5 4 , both substances being of the same c q s t a l form. The possibility of measuring quantitatively axial ratios IJ ing lietneen these two values and having the figures bear any significance was immediately abandoned and all the following statements concerning axial ratios are base:l on qualita ive comparisons nhich TTere obtained very definitely hy taking the X-ray patterns in suitably arranged pairs af indicated. The figures refer t o the amount of chromic oxide in mixtures containing only chromic oxide and alpha alumina. I.

o 5 percent red

2.

1;

0

*‘

3.

15 0

”

4.

30

0

’*

5. 6.

30

0

”

30 o

”

7.

35

0

1

red red red green green green

1 5 o percent red 1 5 . 0 ” bluish-red 30 0 ” red 30 o *’ green 60 o ” green 4.i 0 ‘‘ red 3j o ” red

Each film reccmied two setq of lines, one for the mixture t o the left, the other for the mixture to the right. The lines on both sets were arranged according t o the same general pattern, indicating that the crystal forms 3f the two arc the same. If the size of the crystals and their axial ratios were identical, each line nould he a t exactly the same distance from the zero point on the film as the corresponding line of the comparison pattern and we would see each line as the continuation of its mate on the other pattern. If the axial ratios of the two mixtures were different, however, some of the lines on the pattern of the crystal with the larger axial ratio R-ould be shifted slightly toward the zero point. This latter condition was found t o exist in certain of the films examined and by inspection it was possible t o arrange these mixed crystals in the order of magnitude of their axial ratios. I n the f3llowing table Phys. Rev.,

(2)

21, 716 I 1923).

THE COLOR OF THE R C 3 T

I449

the composition of the crystal 711th the largest axial ratio, a value nearest to that of green chromic oxide, is placed at the top and the others in regular rlecreasing order. I. 6 0 0 percent chromic o d e green 2. 45 0 green gieen 3. 30 0 ‘ ‘ red 1. 4 i 0 ;. jo 0 ‘’ retl 6. I : o ” * ’ bluish-1 et1 7 . 1.: 0 ‘ ‘ retl 8. 0 : ” retl The difference betneen six and seven 1s veiy ~ n i a l lcompwed to the magnitude of the difference betneen t n q antl four, or three and five, for eumiple.

FIG.I Schematic Dinprnm ;1study gf these axial ratio relations offers a possible clue as t o the conditions under which the red modification of chromic oxide will be formed in a given mixture and the conditions under which the green modification may occur in the same mixture. I n Fig. I we have plot.ted axial ratios against composition. The curves, of course. do not represent ahsolute values of axial ratio but are constructed from the following qualitative deductions drawn from the tahles above. I. I n the red crystals the axial ratio increases with chromic oxide content. I n the green crystals the axial ratio ilso increases with the chroniic 2. oxide content. j. Since we are dealing niLh alpha alumina in both cases the curves must obviously meet at IOO percent alumina. 4. The quantitative values of pure alumina antl pure green chromic oxide are fixed. j. The approximate amount of divergence shown is justified by three of the qualitative observations : namely, the axial ratio of 30 percent green is greater than 3 0 percent retl 4.; percent green is greater than 4 s percent retl 3 0 percent green is greater than 45 percent retl.

1450

CHARLES IT. STILLTYELL

X considerat'ion of the color change of samples of the various products

as a whole and a microscopic examination of these same products finely powdered. make it evident that' there is never a gradual change from retl t o green or from green t o red. If both red and green existed in the s x n e product antl were intimately mixed, the color of the product shoultl go through an intermediate bron-n or yellon- color as the change from red t o green. or vice-versa, occurred. I t does not do this. Furthermore. the microstq:e does not reveal any 1xon.n or yellow particles in the poi\-deret1 1;rotliict. There is always a distinct. sharp change from red t o green. ar green to rctl. indicating that the retl antl green modifications d o not occur in thc Came mixture zt the same time. This is best explained hy the assumption that in n mixture of a given composition the retl form may occiir normally at a lon-cr axial ratio than the green and that there is a definite break in the asial ratio when one form changes to the other. This fact has been experimentally shon-n for mixtures containing hetn-een 3 0 a,nd 4%; percent of chromic oxide a t least, nntl iq also brought out 117the curves in Fig. I . \\-e may a,ssume. thefi. that. a. chromic oxitle is atltled t o a81umina. the red modification is normally formed 111-1 to thirty I'ercent chromic oxide, and the axial ratios of the mixctl c tals follow the red curve. At thirty percent, if the atmosphere in nhich the fusion is affected be reducing: we have a sharp change t o the green modification ncconipanietl by a lireali in the axial ratio of the crystal. A%lpparentlythe ret1 form is stabilized by the alumina antl. having an axial ratio more nearly equul .io that of alpha aluiiiina than ha. the green form, it occurs most rcaJily :tt Ion. concentrations of chromic oxidc ani1 n-ould tend t o forin at any concentration. -1s the chromic oxidk contcnt is increased; h o m v e r , there conies a point n-here there is not a sufficient anicunt of alumina to st'abilize the red modification ant1 so it jumps over into the green form. This change may lie retarcled t o 35 percent chromic oxicle instead 3f 3 0 percent, for reasons not tlefinitely I;no~m.if the fusion be done in a,n oxidizing atmosphere. Above 45 ~icrcent,of chromic oxide there is not a large enough proportion of aliiriiina present t o hold the chromic oxide in the red motlification, and as it goes over into the green form the axial ratio relation jumps oyer on t o the corresponding curve in the tlirpmn. The portims of the curves filled in with solid lines h3,z-e been quditatively tletermined in these exrerjnients. The conditions under which points on the dotted portion could he realized are riot known as yet. The red modification is of the same crystal form a s nlpha a,luniina. If it were not, new lines ~ o u l dappear in the diffraction pattern of a mixture ccntaining as high as thirtJ- percent chromic oxide which did not helong t o the pattern of alpha alumina. Theoretically. then, retl chromic oxide is isomerphous n-ith alpha alumina and forms a continuous series of sdid solutions with it. Green chromic oxide is also isomorphous with alpha alumina and forms a second continuous series of solid solutions Tvith it. I n practice, up t o the present neither of ,these complete series hns been realized and Tve cannot consider green chromic oxide as being isomorphous with alpha alumina in the usual sense. I t forms solid solutions up t o i: certain point antl then shift,sover

THE COLOR O F THE RYBT

14.71

to the ret1 nicdification: so we really have a portion of eacti of the continuous ,series of solid solutions comhininp t o form a cliscontiniions series. K e heve in this discontinuity definite evic!ence of the existence of a difference in I: second physical property of the tn-o niodificaations: namely, t h e axial ratios of the crystals. This c:erw u p the apparently unreamnab!e hehavior of chromic osi;'e ton-3rd alrha and 1;eta alumina. I t TV:,~ found that alpha aliiriiina had ta.1 structure as preen chromic oxide. n-hi!? 1:etz nliimina hat1 a ,ely different: antl it n-as tlificult to see n-lip the normall;\- green oxide should not dissolve green in the crystal n.hich it most close]!- resemliletl. The answer is obvious n.hen we linon that there is a red modification of t h e caliromic oxide n-hich has a structure 111orc nearly identical with t h a t of d p h a alumina than has the green. The reason why chromic oxide s h o d d tlissolve in licta alumina as the green riiodification at low concentrations is not l m o m . -4 study of the c h m g e in crystid structure antl color with 7-arying n m o u n t i of cl~rcmicoxide might thron- linht on this point. Alpp:?reiitl;\-at the aliiiiiina entl of the ciirvc the effect of :I reducing :!tmospliere is not strong enough to pull the red niotlification over into the gree11. There i q one possible instance in nature. lion-ever. n-here this map have heen ('one. N r , Sosinan. of the C;col;t:ysicnl Laboratory. rq;orts that the O 4 m t a l Fmeraltl. thc green form of cori:ntliim. is alpha nlunijna ant1 not l w t n . This h i n g the cme, nntl if the col of the Oriental I-ontIn-hich it is 1:resent in too small a n anioiint t o prevent the change into the green form. It goes one step fnrther hon-ever, in suFgesting n-hy, Trithin certain limits, the aliiminn may or may not lie able to stnhilize the retl modification. tlepencling on the condition of the atmosphere during the fusion. Page 1463.

It has long been known that the ruby. either natural or synthetic, turns green a t a comparatively low temperature n hen heated and remains green as the temperature increases until it becoineq red hot. On cooling, it goes through this same preen stage and does not acsume its characteristic red color until nearly co1d.l S o satisfactory explanation of this phenomenon has been offered and it remains as much of a mystery as ever. It cannot be a temperature reaction from a body transparent t o red and absorbing green, because the cdor change occurs a t too 1 0 i ~a temperature. Of course several possible eyplanations at once conie t o niind in view c ~ fthe reqults presented in this paper. K e might assume that the chromic oxide is changed from the red t o the green modification by an increase in the axial ratio of the alpha iliiinina as heat is applied. If this were the case, hon-ever, one would suppose that hy quenching. the preen modification ~voiildbc retained. Quenching does not accoiiipliqh this. Furthermore X-ray powder data indicate that the only change which takes place when alpha durnina is heated, aside from a regular increace in intensity of all lines. is an abnormal decrease in the intensity of the line for the ( I I I', I plane.? Theee facts also eliminate the possibility sf lieta alumina being formed at the higher temperature n hich is instable at ordinary temperatures. I n a further attempt t o s d v e the riddle. the effect of cathode rays on the following samples was observed as they m r e heated. Alpha alumina 0.5 percent chromic oxide (red) Beta alumina 0.5 percent chromic oxide (green) 3 . Xlpha alumina (nhite) 4, Beta alumina (white) I.

2.

S u m h e r one shoned the typical ruhy fliioreqcence qpectrum3 nhich persisted aq [he powder was heated t o a dull red heat. although the lines gradually faded until only the brightest red line was faintly visible. At the same time the point of ma\iniuni brightness of the band shifted w l l into the green. Suniber two did not show any ruby lines h i t instead a red band (about iOop/.L600 pp) and a green hand (about j~o,u/.L-joop/.L~approximatcly equal in intensity and extent. These hands gradually merged n i t h a rise in temperature. Sumber three shoned a green band (j8opp-500pp) nhich graduall> faded on heating. S u m b e r four also showed a green band, narrower and more intense than that of the alpha form. Seither shxvetl any lines or bands in the red. Samples one and t n o were examined spectroscopically by Dr. Papish who reports the impurities as follons: Alpha-Si, Beta-Sa,

large trace: Alg, Fe. S a , N n , traces. large quantity: Mn, Fe. Si, traces.

The only difference is that the red alpha sample contains a trace of magnesium which the beta does not contain. The significant difference in the spectra of Bnuei : "Edelbteinkunde," page 334. Wjckoff: "Structure of Cry-tnls,' pngr 412 (1924). 3Crookes Proc. Ro:. Soc , 42, z j (1887)

THE COLOR O F THE RUBY

1457

the two is that the alpha form gives the typical red lines of the ruhy spectrum. nhile the beta form gives only a red band and no red lines. Hence there is nothing in the analyses t o account for the absence 3f the retl lines in the lieta, form spectrim and the presence of the red liantl. The difference in thc spectra must therefore be tlue t o a difference in the chromic oxide antl adds weigh-t to the evidence that it exists in two modifications. The results do not give any help on the problem of the change of color f r a n retl to green with an i ncrease in temperature, however. =Ilthough the reason for the red color in synthetic mixtures of ~liuniinuni oxide and chromic oxide is clear, there are facts in the literature concerning the natural ruby which seem to create the necessity for another explanation of its color. I t has aln-ays been takrn for granted that the color of the n:-tural ruby was due to a minute trace of chromiuni-or some other oxicle’: ant1 no definite figures for the emount of chromium are given even in (1et:iiletl analyses.? On the other hand, 1-erneuil uses 2 . 3 percent of chromic oxide in iliaking artificial rubies to reproduce the color of the natural stone.3 -1cilia11 trace of rhromiuni does not color :duminum oxide when the tn-o x c xtificially f w e d together. Bordas4 lielieves that the color of the natural rtone has heen intensified hy the :ction of radium. He fused nluminiim oxitlc containing :I trncc of chromium and found it to he colorlcss. h i t cn exposure to ratliuni lromitle it turned roFe colored. It i p well known that the coloring power of‘ a colloid is relatively high and it might lie that the color of the natural rnhy ]vas tliie to colloidal chromiiim. The action of radiinn is generally counter:?etet1 hy heat, and if re,tliiini were respcnsilile for the color we should expect the ruby to bleach when heated. Furthermore. heat should lileach a stone colored liy colloidal chromium. since it nould tcntl t o agglomerate the particles. -1natural ruby ~ m heated s up t o fusion without any fading of the color. When it actually began t o fuse. fading occurred locally at the part, of the scone in the hottest part of the flame. ;1niicroscopic examinetion of this area showed it to lie pitted. indicating that boiling had occurred on the surface and that the fading n-as doubtless tlue to the volatilization of chromic oxide. -1 synthetic ruby containing between two antl three percent of chromic oxide n-as prepared t o duplicate as nearly as possible the intensity of color of the natural stone. antl this was heated in the same m y . The same amount of fading occurred in the same relative position of the stone arid was apparently tlue t o the rolatilization of chromic oxide. natural riiliy crystal exaniinetl with the ultramicroscope showed no evidence of a colloidal suspension. I t appears, then, that, the cauce of the color must be the same in both the synthetic and the natural rubies, and it follow that the coloring agent, must he G. F. K u n z : ,‘Gems anti Precious Sto 11.. Goodchild: “Prccious Stones.“ p. 18, L. Smit,h: l h e s (4) 18, 289 (1850): 1

’’

Compt. rend., 145, 800 (1907).

present in equal anioiiiits in the two to protluce the same intensity of color. of :I, natural ruhy was made n-ith the Accordingly, a sr,ect ro5cq:ic anal) following result : Keight of s:cmp!c 6.0 mg. -4ctual n-cight ( i t chrcmium be,n-een O . O I - O . of chromium 0.04 mg.

I

mgs.

IT-P fintl that the partjciilar natura! ruby il-liich as anclyzet! certainly mntninetl lietween 0.5 percent and 5.0 percent of chromic ositle, nntl the wtinictec! content is 2 . . ; percent. This is far from the slight ant1 sii1q:osedly untletectah!e trace of chromium nhich the ruby was thought to contiin m t l s h o w IT-ithout a tlxiht that the natural ruhy is identicd in this respect. as in all ot,hers, ni-th the artificial stone. Incitlentally, the spectroscopic annlysis shon-e(! on1y n trace of iron ant1 the puhiishec! analyers of rliljy report varying amount:: of ferric oxide up to 1.5 1:ercent. This leads one t o ?uspert that the c h r x i i i u m was in Pome naj- niistalien Cor iron in the oI(1er analyxs. =Inother ~eric::of solitl solutions n.hich hehaves in a manner similar to the solitl solutions of :ilphn a1 ciiiiinn m d chromic oxides has recently come t o our n.ttention. I n inaliing photographic cmu!sions the practice is t o atltl ;. percent of silver iodide t o 05 1:crcent of silver hroniitle. It has been noticed that the color of the resiilting mixtiire is n tlnrlqer )-ellon- shade than eit'her of the pure constituents. :lit1 until the present this has not lieen explained. Trivelli' made a crystallographic stiitly of thew solid scluiions, however. and found that as si!rer iodide is acldcd t o silver liromide, the resulting inixetl crystal is in the cubic system, which is the form of pure silver liromlde. The addition of silver iodide shrinks the cubic lattice in proportion t o the crnount added. For a considerable intermediate range the results are uncertain. Silver iodide is hexagonal in the sti.lile form and the mixed crystals containing seventy percent of the iodide, or more, ai-lsume this structure. I t is known further that a t allout I ~ O O C ' . the ye!low, hcxagonsl silver iodide changes into an orange-red, cuhic form.' (lhviouslj- the mixture containing 5 1;ercent of silver iodide is of a deeper shade than the pure yellow iodide because the iodide is present in this mixed crystal in the cubic form. The silver iodide, then, is a substance which exists in two modification?; the form stable at room temperature? is hexagonal, the form stable ahove 140' c'. but metastable at room lemperatures, is cubic. Thia metastable, cubic silver iodide in this case is stabilized at ordinnrg temperature hy adding t o it a sufficient amount of another c u h c substance, the silver bromide, whose crystal structure it more closely resembles than does the hexagonal, stable form. This case offers a beautiful analogy t o that of the two chromic oxides and alumina. There v e have a ret1 form, metastable a t ordinary temperatures but stabilized by the presence of a sufficient eniount of alpha alumina, whose crystal form it re-

Rcc. Tiar. chim., 42, 7 1 1 (1923). J. TI'. llellor: ''-1 Comprehpnsiw Treatise o n Inorganic and Theoretical Chemistry." 3. page 12;.

THE COLOR O F T H E RL-BT

14'9

semhles more closely than does the green chromic oxide. K e have noticed that n.hen the silver iodide changes over into t h e cuhic form there is some chznge in color from yellon- to L: tleerer yellon-. Then, as the temperature rises. the color continues to change gradually t o a (lee!, orange. mise3 crystal having the composition 8-c!lon- cn1ir.o.t tw niaintninctl by qucnchinp. The green co!er of the riilij-, ohtainetl 11)- heating i t , a l ~ ocnnnot he preserved 1))-qi1.enchicg ant1 i t is very possi1;le that \Then mineone fin(!$ tlic expla.nation for the liehavier of any one of these three suhstences ton-art1 heat they n-ill h::w the ansn-er for all three. -4t a n y retc. the 1:ehavior of the ruliy in this res1:ect is not unique. Still another case of a series of solid solutions analogow to the aluminum oxide-chromic oxide niirtures is that of the ~li~continiious series of solutions of niercnric iodide antl mercuric lironiide. This has lie stuclied by ,J. S. table at temperavan Sest' nntl others.? Red, tetragonal inercuric iotlicle tures helow I 28'C.. while yellon-. rhonihic mercuric iotlitle is stalile :thore that temperature. I\Ierciiric lxomide is white. antl rhoinhic in crystal form. T7an S e s t obtained a n unbroken series of yellon- inixecl crystals of mercuric io-liile ant1 mercuric hromide varying in composition from zero u p to 97'; of mercuric iodide. That is. the yellow form. stnhle at high temperatures, is Lstatdizetla t ordinmy temperatures li>- the presence of a, sufficient amount of rhonihic mercuric liromide n-hose crystal form it niorc closely resernhle;: than does :he red. tetragonal forni. K i t h less than 3 ' ; of the tiroinide. its influence is not sufficient to stabilize the high temperature form and only red crystals occiir. I t is evident that the analogy lietn-een this case antl the nlpha aluniinachromic oxide series is even more complete than the case of silver liromicle and silver iodide t o n-h,ich reference hxs already lieen rnatle. I n the case of the silver halides we have merely a change in shade of the yellow color n-hen the series of ~olitlso!iition? is liroken. K i t h the niercuric halides. h o m v e r , a marked change in color from recl t o yellow gives evidence of thc lire& in t h p series. In the case of the aluniinuni oxide antl the tn-o forms of chromic oxide n-e have n change in color no less distinct from ret1 to green. I\Iy nttention n-as dran-n t o this case 1 ) ~ -Dr. D. T. Kilhar. Summary The results of this section are as follows: I. The red color i n i p r t e d t o alpha d u m i n n 11:- chromic oxick is not due to cdloidal chromic oxide. "Hritriige ziir I\cnntniss cler Qu~clisi1t)~rhaloarnitIe" i 1909 I . 3Irllor: "Treatise on Inorganic antl Theori.tica1 Chemietr!~." 4. 904.

1460

CHARLES W. STILLTVELL

I t is not due t o a higher or loner oxide of chromium. I t is not d i i ~to a change in the concentration of the chromic oxide as is the case with solutions of chromic salts. 4. The red color is tlue t o a second modification qf chromic oxide, of the came crystal struetiire as the green modification and alpha alumina. whose axial ratio is nearer t o that of alpha alumina than is the axial ratio of the preen modification. Therefore, the red modification tends t o forni when chromic oxic'e is added t o alpha alumina and does form under ordinary conditions up to a certain point. lieyond which the effect of the alumina ih not strong enough to stabilize it. ;. The occurrence of the red or green modification depends on the value of the axial ratio of the mixed crystal. There is a discontinuity in the change of the axial ratio Tvith change in chromic oside content when one modification changes t o the other. There is alsq a marked difference in the axial ra+io of a mixed crystal containing the red forni of chromic oxide and a cryital containing the qaine amount of chromic o d e in the green form. This change in axial ratio may lie affected in t n o different n-ajs. a. By (.hanging the chromic osi(!e content. b. By a variation of the atmoyihere in which the fusion iq carried o u t , for mixtureu containing between 30 anti + j percent of chromic oxide. 6. I t has been indicated that the tivo modifications differ in at least three properties; namely color, axial ratio of the crystals, behavior n hen exposed t o cathode rays. ;. The bluish-red color of uome varieties of iuby is tlue t o rhe reduction of a limited amount d' the red chromic oxide to hlue chr3mous oxide. 8. The chiomic oxide content of the natural ruby has been determined RS approximately two percent and the cause of the color of the natural ruby has thereby been shonn to lie the same as that of the artificial ruby. 9. The temporary change of color of the ruby to green on heating cannot 1:e explained by any of the above results sntl must still he accounted for. IO. \\-hen there is no stahilization of the red modification, as in the emerald, a green color is obtained n i t h very lon- concentrations of c h r m i c oxide. 2.

3.

THE SA4PPHIRE There is little agreement among inveatigators as to the cause of the blue color of the sapphire. Goodchild' assumes i t t o he due t o a trace of chromium. as does Doelter;L although the latter finds that the behavior cf capphire> nhen heated varieq n i t h their source, and therefore suggests that there may be more than one substance vhich causes the color. C"attelle3states that the cdor may be cauqed by the presence of iron; while T'erneui14 and Nerrillj " P r t c i o u i Stones." page 18; ( 1 9 ~ 8 ) . 1Ionatshcft. 29: 1115 (19cXj. 3 "Precious Stones," pagc I O I (I9C3!. Compt. rend.. 151, 1c63 (1910). j L-. S.S a t i o n a l 1 1 u e c u m l?ull.. 118, I . 2

THE COLOR O F THE R E 3 Y

1461

assert that, it is the result of a trace of titanium. Iiraus antl Holden1 give an analysis of both natural and artificial (TTerneuil)sapphires. the former containing. in addition t o aluminum oxide, 0.92 t o 1.95 percent 3f ferric oxide. depending on the source. antl less than 1.0 percent of silica. ,-lccording to this analysis the natural stone contains no titanium. T7erneuil has made analyses of natural sapphires from three source? and finds that they all contain from 0.03 t o 0.06 percent of titanium tiiosjde. Bailer? s:i>-s the cause of the color is unknown, hut suggests3 that it is prohahly due to either iron or c hroiiii urn. The first note in reference to the production of the b!ue color of the sa,pphire artificially seems to be that of Sninte C'lsire Del-ille' who IT-as working on the preparation of artificisl ruhies hy heating a mixtiire of aluminuni fluoride. chromiiim fluoride and hoi ic oxide in a plntinum crucihle, He remarks that a small amount of chromium gives the typical ruby retl, while a slightlj- larger amount produces the sapphire hlue. The Mue is att'ributeti liy hiin t o this change in chromic oxide content or "perhaps t o the state of oxidation." K i t h a larger amount of chromic oxide he gets green. He admit?, h o w v e r , that the difference in chromic oxide content hetween the red and hlne products c3uld not lie detected 11)- the ordinary analytical means. E l ~ e l n i e no1ita.ined ~ t v o kinds of crystals in preparing artificial rubies-the tals and n larger quantity of bluish needles. The chromium may have heen the cause of this colnr. Fremy6 notes the occurrence of some red ruby crystals, artificially prepared, n-hick are hlue in the middle: and others that are r d on one side and blue on the other. Ih1nz7 describes several specinients of natural coriinduni which are both red antl blue. The prolilem came up in this inwstigation u-hen it was noticed that many of the alumina-chromic oxide mistures n-hich n-ere fused contained an appreciahle number of vivid sapphire blue particles when examined under the microscope. The conditions favoring the formation of the blue could not lie definitely established. except for the fact that the higher amount of Hue appeared most frequently in the mistures containing from 1 5 t o 30 percen-t of chromic oside. I t is as likely t o appear in a product which ha,s heen fused in an oxidizing flame as in one which has been fused in a reducing flame. It' seemed unique in that it n-as always in isolated fragments! never forming part of a retl. or bluish-red or green fragment. Even when ohserved through the microscope in a large piece of the substance before it had been pon-clered, the })!lie ~ ~ o uappear ld as a distinct' antl separate unit embedded in the lump which served as a matrix. The total amount of hlue, estimated by counting the number of particles in the field of the microscope in a given sample, was "Gem l\Iaterinls." p u p I G O I 19251. ,'Etlelstcinl;untlc." page 356. 3 Bnucr: Z. angen-. Chem.. 2 2 , '177 f 1 9 0 9 ~ . ' Conipt. r e n d , 46. 76+ (1858). :.inn. Chim. Phys., (3' 3 3 , 34 (18g1). fi F w m y and 1 - e r n r i d : Cornpt. rend.. 111. 667 (1890). .,Gems and Precious Stones." page 46 (1892). 1

2

1462

CHARLES W. STILL11 ELL

probably never greater than one t o two percent. The color and degree of its saturation were always the same- an evenly distributed cornflower blueregardless of the relative amounts of aluminum oxide and chroniic oxide in the original sample. I t might be assumed that this blue color is due to a trace of titanium dioxide in the materials which does not diffuse throughout the mixture, hut xhose particles simply dissolve in the alumina immediately adjacent t o them, To test this, the following mixtures were made up and fused. The aluminum oxide used is the alpha alumina employed in the mixtures with chromic oxide in which the blue particles occur. Percent Ti02 in -ll$.la.

Color

gray with faint bluish tinge. bluish gray grayish blue, smoky grayish blue, smoky

trace 0.5 I

.o

;.0

A\ microscopic examination of the pc8rticles of e,ll these mix.tures rew:ils a mixture of colorless. fairly transpnrenLparticles and pzsticles varyicg in color as indicated in the table. There is no suggesiion of the cornfloJT-er blue color and the tit:inium dioxide apparentlF diff ueea inore or less evenly throughout the mass, since there arc co iFolated blue ci ta,ls in the product. The isolated flashes of blue ohserveti in the ruby are therefore not dzt t o titnniiirii dioxide. Moreover: the mere fact that artificial sapphires can lie made n-ith titaniuni dioxidc does not mean that this osic’e is the coloring mecdium in the natural stone. Fremyl made nrtifir.i::l sepphires hy coloring aluminiuiii oxide wit’h cohalt, but nobody has suggested that cobalt colors the natural sapphire. The fact that mixtures of aluiiiinum oxide ant! chroniic oxide of lo\\- chromic oxide content, are red IT-hen fwetl in an oxidizing atmosphere and bluish-red when fused in a reducing atmosphere. ofers the suggestion that the blue may he the result of a reduction. X x t u r e s of aliiiiiinum oxide containing varying amounts of chroniic oxide u p to thirty percent were fused in a reducing flame and held at the fusion temperature for varying lengths of time and cooled a t different rates. Kone of there factors seemed to have any consistent effect en the number of isolated cornfiower blue particles in the product. Cooling t o room temperature in an a,tniosphere of hydrogen did not increare their number. It was found. as has been noted before. t h i t the samp!e containing ten percent of chromic osidk which \vas heated for thirty minutes in the Xjax induction furnace in a strong reducing atmosphere 11-as decidedly more bluish than the sthers, although it contained no more of the isolated blue particles. Duboin2 gives a receipe for making a blue glass colored by chromium. He states specifically that the fusion must be carried on in a reducing atmosphere. This work of Duboin’s was repeated anti a greenish-blue product obtained.

?

Fremy and Fril: Compt. rend.. 8 5 , Duboin: Ber., 31, 1977 (1898).

1029

(1877’1.

THE COLOR O F THE RVBT

I463

The blue color was next discovered in a mixture of lierylliiini oxide, aluminum oxide a,nd silicon dioxide in the proportion in which they occur in the emerald. and containing one percent of chrcmic oxide. I t was found, moreover, that the production of this ldue color required a reducing atmossphere. An oxidizing atmosphere produced green and no blue. Furthermore. upon heating in the oxy-hydrogen flame, the globule \vas seen t o boil and the amount of the blue formed depecded on the duration of this lioiling. By increasing the proportion of oxygen in the flame the boiling could lie stopped, shon-ing definitely that the apparent lioiling TI-as an evolution of oxygen or steam resulting from a reduction in the gloliule. -1single globule was changed from green to blue to green perera1 times over liy alternating the conditions of fusion lietn-een a reducing antl a oxitlizing atmosphere. The change in color cnnnot be due to a change in composition: since. if it w r e . the colors could not lie reproduced at will after the heating rntl poesilile volatilization had continued for some time. The effect of varying the nnioiint of chromic oxide was investigated and it was foiintl that the blue color. though more saturated, n-as ohtained in mixtures containing up t o fifteen percent of chromic oxide. K c tiid riot go beyond this point. A curious result of the method of heating n-as observetl in a11 of these capes. The hlue c3lcr np1;ears to rise ton-art1 the fiaine-or tow-artl the top of the globule. That is. the upper portion of any liexi after fusion is blue antl comparatively trans1:arent and compared to the Ion-er portion. n-hich is green a n d opaque antl granular earmcc. It is natural that the part ton-art1 the flnnic shoulcl I:c nffectcd first : I n i t if the head he turned o.i-er 11-it11the green side up n-e ~ ~ o i i~l dx p e c tto have the entire liead blue Zfter fusing. This is not the case. The 1,liie an:l green ap1:arentl.v change places antl 11-e have t h e neJv upper sitle hlue and the loner sitlc green. a result, it is practically impo?sible t o get a uiiiiorni lilue head. It can he done liy heating the second ride only monientaril!-. dthoiiph in this case the liead does not have the glossy. transparent appearnncc. I t might be done hy heating siinulteneoiisly from hoth side?. Thcsc i w u l t s shon- heyond ::doubt that chroiiiiiim can inipart a t>-I:icalsal; t o certain oxide mixtures at any rate, ant1 that this blue colo m o m oxide. Furthermore. X-ray tliflracticn patterns of t h e green ::nd Iilue beryllium aluminum silicates shon- 1:oth t o lie ::morphoLis. -1s supgcsted ehen-here in this paper.’ the fact that the mistmiireis not crystallice i11a>*explain n h y it is so easy t o reduce the green t o The hlue in t,hi,s case. T o return to the occurrence of this Pepphire liluc color in the artifickl riitq-, we ha\-e the effect of the blue brought to notice in tn-o distinctly tiiWerent n-ays. First as the isolated. vivid blue partic!es occurring in either a n oxidizing 3r a reducing atmosphere. At present, no adequate explanation of the forniation of these blue crystals can be brought fori-ard. Several possil;!e reasons have been considered and found wanting. For example, the alpha alumina is known t o contain a m a l l amount of beta alumina. and the chrcmoiis oxide might cliesolve in that. This is not likelj- for two reasons. Beta dumirw when l

Page 1455.

I464

CHARLES TI-. STILLWELL

fused changes t o the alpha form hecause the soda, which apparently stabilizes the beta form, is driven off; and secondly. the samples of beta alumina containing chromic oxide obtained from the Sorton Company are green. These are macle in a rcducing atnioqphere and hence should have been a t least a bluish-green if the above assumption were true. I t might also be argued that the chromous oxide formed is soluhle in the chromic oxide-aluminum oxide melt t o only a limited extent and therefore separates ?ut as a second phase. If this were so. however, we should expect t o find these blue flashes only in those samples n hich were already lduish-red in color, indicating that they were saturated with reqpect t o chromous oxide. X s has been noted, the blue particles found just as frequently in mixtures which have been heated in an osidizing flame and have a pure red hotly color. Furthermore, if it were possible for the chromous oxide to separate out in a second phase, there is no reason why the change should not go further. A11 n e can say definitely in regard t o the-e blue particles, then, is t o point out that chromous oxide is known to be exactly that shade of blue. at least in beryllium aluminum silicate mixtureq containing from 1.0to 15.0percent of chromous oxide; and that titanium dioxide does not color alumina in that manner or with the shade of blue described. I n the second place, we have the effect of hlue coming in when the color of the complete mass of the globule is changed from red t o bluish-red by a reducing atmosphere. This is a perfectly consistent occurrence and may be duplicated a t nill. Xe has already been itatcd, this change in color is doubtless due t o the formation of a limited amount of chromouq oxide. So far, it has been inipossihle t o cause a complete change of the red ciiromic oxide t o blue chromous oxide. The reduction may be effected in the case of the emerald where we have an amorphous beryllium aluminum silicate containing the chromic oxide as the green modification. This tendency for the red chr imic oxide t o turn blue a t least makes it legitimate for us t o assume that the natural sapphire is colored hy chromous oxidc under conditions which we have not as yet been able t o duplicate. TTe also have here a t least a plausible explanation to account for the color of purple corundum. the Oriental Amethyst. I t may well he a partially reduced product, an intermediate step between the red ntby and the blue sapphire. I t s color has been attributed by Ilerrill' t o the presence of traces of chromium, giving the red, and titanium giving the blue componcnt. This is on the assumption that the amethyst contains titanium-which is not knon n-and that titanium does cause the blue color of the natural sapphire. Keither of these fact. is very probable. d possible duplication of the conditions used by nature t o form these various cdored corundums has been sought from time t o time in the use of X-rays and radium emanation. Bordas? and Escard3 claim that a ruby subjected t o radium emanation turns from red to violet t o blue to green t o yellow.

THE COLOR O F T H E Rt-BI-

1465

Doelter' in a detailed clescription of many individual stones exposed t o radium, records two rubies ivhich went from red t o a purplish-red and one which hecame more re+. The ex1:erirnental results. therefore are not consistenr enough to assume that the a c t i m of radium on the ruhp has produced the natural sapphire. The discrepancy in results may lie due in part t o the fact that ratljuni emanation of different intensities was used in different cases. Unfortunately the strength is not stated in every case. Lind and Rardwll' insiet that the ruby is not changed noticeably when exposed to 2 0 0 1nilligr:mis of radium for a considerahle period of time. Rlost of th0.e n-ho have investigated the matter agree that, under the influence of radium, the sapphire changes from blue t o green t o a final yellow color. Of these invest'igators. Doelter alone has noticed a difference in stcncs in this respect as well as in their behavior ton-ard heat. He states that several sniall Ceylon sapphires exposed t o radium fcr as much as fifteen clay!: untlermnt no change in colw antl ir-ere not hleachrtl hy heat. \-e may conclude .that the,se sapphires which are stable toward heat antl radium are coloretl by some stable metallic oxide, while those which are riot stable may he colored jn some other ~ a y . Several small Ceylon sapphires vere procured antl their properties investigated. It, as found that they did not lose their color when hcatecl-at least' not permanently. They did seem to fade somewhat n-hen heated, but were just as blue after they were cooled again. K h e n heated u p to the melting point in an oxidizing atmosphere the sa,pphireturned to a retltlisli liron-n color. This is an interesting hut slightly ambiguous result. If the hlue color were diic t o chromous oxide, the oxidation product should he red chromic oxide : whereas if the hlue color n-ere caused 1iy a hlue modification of ferro-ferric oxide, the oxidation product should lie the reddish-honn ferric oxide. -4 spectroscopic anal)-sis of another of these same sa,pphiresrevealed a surprising lack of impurities of any kind in the>alumina, there lieing present onlja small amount of iron. possibly a trace of titanium. and no chromium. A casual inspection of these sapphires inakes i t plain, even t'o the amateur, that there is not as much coloring matter in them as in the ruhp of a corresponding size. One gets the impression that the ~ ~ i is h ywell saturated with the red color: while the hlue of the sapphire, although t l x k in shade, gives the impression that the colxing matter is spread out too thinly to he w r y efficient, antl is therefore present in smaller amounts than the coloring matter of the ruby. -411 this discussion makes it olivious that the real cause of the color of the natural sapphire is as much in the dark as ever-in fact, much more so than one would guess after reading the statements corninon in t,he literature. For sapphires not bleached lip heat the color may lie caupetl by chromous oxide, although it \vas not found in the particular one we analyzed. I t mamylie tlue t o a hlue ferro-ferric oside, or possibly to titanium. I t is certain that those sapphires of a deep hlue or cornflower shade could not he colored tiy titanium, >Ionatsheit, 29. 1 1 4 j (1908). J. Franklin Inst., 196,3 7 j (1923)

1466

CHARLES W . STILLX-ELL

but the smoky blue stones may be. As for those sapphires which bleach when heated, all we can say is that it is very improbable that their color is due t o any of the causes just mentioned. Summary The results of this section are: I. Xlpha alumina colored b j red chromic oxide turns hluish-red when fused in a reducing atmosphere due to the formation of some hlue c h r ~ m o u soxide. It remainc t o determine under what conditions thiq reduction could be carried to completion, giving n sapphire blue coruntlirni. a h observed in preparations of artificial ruby may be colored by chromoiiq oxide because a. It has been shonn that chromous o d e IC. exactly that shade of blue, a t least when dissolved in beryllium aluminum silicate. b. There is no other impurity present in the :iii\tiire u q d which would produce that shade of blue. 3. The cause of the hlue color of the n a t u r d sapphire iq not knon-n. I t is verv probable that the color of all sapphires iq not caused by the same impurity. Acknowledgment It is discouraging t o one who en,ioy$ a diversity of interests t o hear from all sicks that only through a narrowing of attention niay one hope t o become accoiiiplishetl in his chosen field. I t lins been the writer's privilege. however, t o work under the guidance of one who insists-and whose reputation most certainly proves-that in general an ever broadening viewpoint will produce better results than a narrowing one. Professor Kiltler D. Bancroft, hy his sense of humor. his ehility to see the things of life with true sense of proportion and his evident belief that one need not take himself or his work too seriously at the expense of everthing else in order to be a good scientist, has maintained and increased the n-riter's interest in chemistry and in science, and has given t o him a guide on which t o base his future effcrt9. The writer also wishes to thank Professor c'. c'. Murdock and A h , H. K. lltiissell for t'heir aid and a,dvice in procuring the X-ray spectrograms: and Profesor J. Papish for making the several spectrographic analyses. Coriiell C T ~ ~ i o e r s i t g .