]\gt ; On the Structure of .
Part VI .
Titanic Oxide , its Isomorphs .
By W. J. LLAS , Sc. D. , F.R.S. , Professor of in the University of Oxford .
( Received January 3 , \mdash ; Read January 23 , 1908 .
) key to crystalline structure is furnished by a study of molecular volumes , and one of the best criteria of its powers will be found in its application to polymorphous compounds .
If it can explain consistently with our knowledge oi crystalline symmetry the change in volume which accompanies the passage of a substance from one crystalline system to another , this fact alone would seem to offer presumptive evidence in its favour .
There are inherent difficulties in the subject which render progress laborious and slow , so we cannot at present offer an exhaustive account .
of all cases of polymorphism , and on this occasion we collfine our attention to the single but remarkable instance of titanic oxide , which presents itself in the three -kncwn forms of anatase , brookite , and rutile .
The chemical formula of the oxide , generally taken as , is bosed on analogy with zircon , , which is isomorphous with rutile .
It may be represented graphically as ' and the simplest lyptic renderin of this gives at once the general uration of the crystal molecu ] , Haiiy 's " " molecule integrante out of which anatase , and , with slight modifications , brookite , and rutile also , are built up ( see , p. To obtain the molecule in its true relative nensions , we must first determine the relative sizes of its constituent atoms .
The ross atomic volnme of metallic titanium is .
This follows from the relation the molecular weight ( mw ) being , and the density ( d ) .
On the assumption that the atoms in titanium are in open cubic order , this gives as the diameter of the atom , or radius The gross atomic volume of oxygen cannot be obtained from the free element , and it is difficult to arrive at a completely satisfactory result from its compounds ; in commencing this investigation , I selected the number as representing an of considerable probability ; this gives for the diameter , or radius .
In an earlier part of Prof W. J. Sollas .
[ Jan. 3 , this paper*I made use of a somewhat higher value ( 189 ) , but it is possible the dimensions of the oxygen atom are not absolutely constant throughout all its combinations .
In any case , the number is that which harmonises with all the facts presented by the minerals we are about to consider .
Anatase.\mdash ; The fundamental molecule of anatase consists of four atoms of arran ged in contact , with their centres situated at the corners of a square : one atom of titanium rests upon the four oxygen atoms , its centre the centre of the square ; the other atom is situated in a position below .
On joining the ceutres of the titanium with those of the oxygen atoms , we obtain a pyramid of the tetragonal an resemblance to the primitive pyramid , which is the characteristic form of anatase .
These molecular pyramids may be built up into a crystalline edifice by arranging them so that the oxygen atoms lie in square order in one sheet , taken as horizonfftl ; the titanium atoms form corresponding sheets Plan .
The larger circles in all the figures represent titanium , the smaller , oxygen atoms .
above and below .
On this layer of molecules a second may be superposed in such a manner that the titanium atoms of its lowest sheet rest upon those sets of four oxygen atoms which are formed by the four corner atoms of four squares in contact below ( fig. 2 ) .
The structure thus obtained is tetragonal and holohedral ; it may be regarded as a tetragonal lattice with the centres of the molecules situated on the nodes .
The packing is fairly close : to determine the volume occupied by one molecule of titanic oxide ( taken for convenience as ) , we must now partition the space occupied by the configuration homogeneously .
In plan 'Boy .
Soc. Proc , 59 , p. 294 , 1902 .
1908 .
] On the Structure of ( fig. 1 ) the structure may be completely divided into equal square areas , such as a , ench including one atom of titanium and four -atoms of ; the of the side is , and this squared gives 8 , or .\mdash ; Elevation on a Plane passing through A of fig. 1 .
Each square again is the base of a prism , which containls one atom of titanium and quarter-atolns of oxygen , in all one molecule of ; its height is shown by the line in fig. 2 , and is measured by , or ; the volume of the prism is .
The best determinations of the specific gravity of anatase vary from to , and its molecular volume lies consequently between and , with a mean of .
The accordance between the observed and calculated volumes is thus remarkably close .
FIG. 3.\mdash ; Elev ation on a Plane passing through a , fig. 1 , but with the Oxygen Atoms separated as in fig. 4 .
It is evident that the configuration we have described affords data for calculating the axial ratios or parameters of the crystal .
For our horizontal axes we may select either or ; we take the latter and make the parameter equal to 2 the corresponding parameter on the vertical axis is given by the length of the elementary prism from which the volume was calculated ; we have , therefore , for the horizontal parameter a length of , and for the vertical of , and this gives a ratio of 1 : , while Prof W. J. Sollas .
[ Jan. 3 , the ratio obtained by direct measurement of the mineral is 1 : .
Ths agreement in this case is only roughly approximate .
We may make it a little closer , however , by slightly modifying our original conception of the molecular configuration ; we have supposed the oxygen atoms to lie in contact with each other , let us now separate them by a short interval ; this ] allow the titanium atoms to approach one another , and the total effect will be to lengthen the horizontal parameters and shorten the vertical parameter , at the same time the volume will be increased ( figs. 4 and 5 ) .
FIGS .
4 and ] to figs. 1 and 2 , but readjusted into a nearer approximation with the actual axial ratios of Anatase .
In determining the amount of the change which can be produced without unduly the molecular volume , we shall make use of topic axes .
horizontal axes we have already chosen do not correspond with the sides of the square which forms the base of our elementary prism ; we must compensate for this by the axial ratios from 1 : to or 1 : the volume we take the mean , Then , since , we have , and .
Thus the parameters on the topic axes are and ; and from we obtain corresponding parameter on by dividing by , which gives ; but , and thus the interval between adjacent oxygen atoms amounts to With the data at our disposal , we can now calculate the length of the vertical pal.ameter as given by our tion when modified by the separation of the oxygen ; it is found to be , while the topical axes show it to be actually , a difference of 0.084 , or about per cent. * According to Schrauf 's measurements the axial ratio of anatase is 1 : ; this gives for , and our value deduced from the configuration becomes , or a difference of 2 per cent. 1908 .
] On the ucture of Our results , however , are based on the improbable assumption that all atoms possess a spherical form , and are thus first approximations only .
It is extremely likely that the atoms of those elements which crystallise in other systems than the cubic are not true spheres .
The of oxygen with sulphur and of titanium with tin gests that this may be the case with the atoms under consideration .
Anatase cleaves readily along the planes of ( 111 ) and ( 001 ) .
But it may be here that the form ( 111 ) does not correspond to the form of the inte rant molecule : the symbol for this , according to the have chosen , must be ( 101 ) .
primitive pyramid of the rapher only rises when the molecules are built up into the edifice ; its basal section is given in the square of fig. 1 .
The connection between and structure is thus not immediately obvious , but on referring to the ( fig. 1 ) it will be seen that alternate squares of four oxygen atoms exist over which titanium atoms are absent , both above and below , and these umoccupied spaces run in files parallel to the horizontal of the pyramid ( 111 ) , all the other squares covered with titanium atoms .
It would seem probable that the structure should part along these lines in preference to any other .
The basal cleavage ( 001 ) follows readily from the structure .
The optical sign of is negative , , and ; hence the axis of greatest elasticity responds -with the axis of the crystal , or the titanium axis of the molecule .
The coefficient of expansion is greatest the principal axis , least along the axes normal to it .
According to Fizeau , it is parallel to , and in any direction normal to this ; according to Sehrauf , it is in the first direction , and in the second .
To this relation we shall recur later .
Butile.\mdash ; Although rutile crystallises in the same system as anatase , it presents very diHel.ent forms , and cannot be eferred to the same axes .
Its mdamental molecule has the same constitution as that of anatase , there is no polymerism ; but it differs in two respects : first the titaninm atoms have approached each other the vertical axis so far as to come into contact ; and next , as a consequence , the atoms haave been pushed outwards as far as possible , so that they lie comparatively far ough still at the corners of a square , its plane at angles to the vertical axis ; which passes its centre and the centres of the titoninnl atoms above and below ( fig. 10 ) .
The arrangement of these molecules differs completely from which obtains in the case of anatase .
The first layer is formed by the molecules with the titanium axes horizontal and end to end , so that the VOL. LXXX.\mdash ; A. '1 ' Prof. W. J. Sollas .
[ Jan. 3 , titanium atoms form continuous horizontal files ; on these together side by side they will be found to fit together , the molecules of adjacent files with each other .
The titanium atoms are thus marshalled in square order in a single sheet , and the vgen atoms lie one over the centre of each square of four titanium atoms , forming two sheets , one above and one below ( fig. 6 ) .
If the bonding of the atoms be disregarded , this configuration will be ; but when the bonds are taken into account it becomes rhombic ; consequently .
to preserve the tetragonal character , the next layer of molecules , which is built up like the first , must be turned in a horizontal plane through an angle of , and then superposed so that the two titanium FIG. 6.\mdash ; Rutile\mdash ; Plan .
atoms of one molecule lie immediately over the two oxygen atoms of the layer beneath , the centres of the two molecules on the same vertical axis ( figs. 7 and 8 ) .
The structure so produced is and holohedral ; the .
7.\mdash ; Plan of Two Molecules stlperposed .
FIG. 8.\mdash ; Elevation on 1908 .
] the Intimate Structure of gyrohedry , since it amounts to a erht atYle , no raphic effect .
* We may partition the structure eneously into square prisms , having as a base the square of the figure with an area of , or ; the height is shown by the line ff ' , and nearly equals or .
The volnme is , therefore , .
The specific gravity of rutile lies between and , and its molecular volume between and 19 , but when heated in the speciiic ravity rises to , or the volnme diminishes to .
The reenlent between theory and observation is thus as close as we can expect .
If we take the height of the prism just determined ) as the vertical parameter of the crystal , then the horizontal parameter will be or , and the ratio 5 0831 : .
The actual ratio found by angular measurement of the crystal is 1 : The most perfect of rutile is parallel to the prism faces ( 110 ) , fact consistent with the ration ; there is a parallel to the faces of the second prism ( 100 ) also , but this is less perfect than preceding , though there in the ruration which would have enabled us to predict this fact .
Rutile exhibits a characteristic tendency to form reatly eated prisms , and this is quite in harmony with the tion , which a closer texture over the prism faces than at to them , that is to , the surface density is greater parallel to the axis than in any other direction .
The optical of rutile is positiye , and consequently the clirection of reatest elasticity corresponds with the horizontal axes of the crystal , ) it is along these axes that the titanium of the molecules are disposed , so that we have in this case precisely the same relation as that which found to obtain in anatase .
In agreement with the optical characters , the constant is at a parallel to the axis , its value in this dil'ection being 173 ; at to it only 89 , or about one-half .
If the truth of our configuration should be confirnled , this ratio should afford some insi , ] into the difficult question of the connection between the optical ) erties i the molecular structure of a crystal .
The thermal conductivity is at a maximum parallel to the principal axis .
The thermal expansion , as determined ) Fizeau , ives a coefficient for the * Schrauf , however , has observed gyrohedry in some crystals of rutile : it would appear , therefore , that the rotation not always complete .
The error oduced by this formula only affects the fifth place of decimals .
Prof W. J. Sollas .
[ Jan. 3 , vertical axis of the crystal of , and for the horizontal axis of ; if the volume of the oxygen atom expahds to a greater extent than that of the titanium atom for the same rise of temperature , this relation between the coefficients is such as the configuration lead us to expect .
It not infrequently happens that prisms of rutile occur " " attached to plates of haematite in such a position that each prism lies with its face ( 100 ) upon the basal plane of the haematite , and with its prism edge perpendicular to the edge of the combination ( 111 ) : ( 100 ) of the haematite , fig. 220.\ldquo ; * Some seems to be thrown on this interesting phenomenon by the structure of the two minerals ; an attempt to formulate the structure of haematite shows that the atoms on the basal plane lie on the nodes of a hexagonal network ; and when the structure we have yned to rutile is observed as it presents itself on the side of the prism ( 100 ) , it also is found to exhibit a hexagonal arrangement , which , although not quite regular , may perhaps be sufficiently so to allow of the co-adjustment of the two crystalline structures ( fig. 9 ) .
FIG. 9.\mdash ; Elevation on of the Prism ( 100 ) .
Brookitc.\mdash ; This mineral crystallises in the rhombic system .
Its molecule has the same chenlical formula as anatase and rutile , and there is no polymerism .
The titanium atoms have approached each other and come nearly not quite into contact , and the atoms , though pushed out , are not equally separated , but lie at the corners of a rectangle having two sides longer than the others .
Thus the molecule itself possesses a rhombic 'Mineralogy , ' by Professor Miers , 1902 , p. 86 .
1908 .
] On the lntimate of form , which is therefore inherent in the crystal from first ; in the same way the tetragonal form is already determined in the molecule in both anatase and rutile .
The difference between the three molecules will be seen by inspection of fig. 10 , which represents them in plan .
FIG. 10.\mdash ; Molecules of Anatase ( 1 ) , Brookite ( 2 ) , and Rutile 3 ) , shown in Plan .
The first layer of molecules is constituted very much as in rutile , but with this difference that the adjacent files of titanium atoms are not in contact with each other , but separated by oxygen ; they a rectangular but not a square lattice .
FIG. 11.\mdash ; Plan of First Sheet of Molecules .
The second layer of molecules has the same constitution as the first , and is superposed with the same orientation , but with a horizontal displacement parallel with the axis , so as to bring the atoms of the upper Prof. W. J. Sollas .
[ Jan. 3 , layer immediately over the interspaces between the most closely approximated titanium atoms of the layer below ( fig. 12 ) .
The third layer is similarly superposed , but with a compensatory displacement in the opposite direction .
FIG. 12.\mdash ; Plan of Two Sheets of Molecules .
The structure so produced fulfils all the conditions of rhombic symmetry .
The simplest means of ' this hypothetical configuration is furnished by the so-called topic axes .
specific gravity of brookite ranges from to , but the most probable value lies near This ives for the molecular volume .
The accepted ratios of the parameters are .
The topic ratios are , therefore , as follows:\mdash ; In our construction ( fig. 11 ) we make the sides and of the rectangle equal to and , but we now proceed to deternline the vertical parameter means of diameters of our atoms , and we obtain as a result the number , while the observed value is , or a difference of , i.e. , about per cent. This is the sole difference which exists between the results obtained by hypothesis on the one hand and observation on the other .
The volume is only slightly affected , still remaining well within the limits of its observed values .
* Miers , 'Mineralogy , ' p. 365 .
1908 .
] On the of Crystals .
The refractive indices for soditun The acute bisectrix coincides with the axis and is ; thus the direction of maximum elasticity coincides with the axis , or with the titanium axis of the molecule , recisely as it does in both anatase and rutile .
The thermal expansion has been studied by Schrauf , who obtained the following coefficients at .
: ) We may here return to the question of thermal expansion in connection with all the members of this polymorphous series .
There are at least three ways in which changes of temperature may affect the relative dimensions of a crystal .
our attention to a rise of temperature , this may act : ( 1 ) by the space dominated by one set of atoms , oxygell , for instance , to extent than that of the others .
This willaccount for the relative iu the length of the axes in the case of rutile , where the question is not complicated by changes of uration ; but it is inapplicable to anatixse , the coefficients of which are in the iuverse order of magnitude to that which this change would pdce ; ( 2 ) In the next place , a of temperature cause the atonls of one kind , say the oxygen , to become more remote each other , the other titanium atoms to make a closer approach .
This is what , by hypothesis , must happen in the case of the lnolecules of natase could but in this case also the change in the dimensions of the axes , at least in the as of anatase , is in the relation to that by observation .
3 ) There remains then the third factor : the rise in tenlperattlre m initiate those chan , in the configuration ) which ulti1nately one passes into another .
The well be ) ) utthel.e would seem to be no good eason w ) not be set up long before the final change .
Looked at from this point of view , the relative length of the topic in the several polymorphs be expected to throw the matter ; they give the efl'ect of the various changes definite directions .
In the table the topic axes are , the axis of rutile homologised with the axis of tase : ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2 .
ookih . . .
2380 Prof W. J. Sollas .
[ Jan. 3 , The corresponding thermal coefficients are as follows:\mdash ; It will be seen that the axis consistently diminishes throughout the series , this is in harmony with Schrauf 's observation of a negative coefficient , in the case of anatase .
But the second axis of anatase ( in the table ) should a positive coefficient .
It is barely possible that it does so , and in this might be found reconciliation between the apparently discordant obserservations of Schrauf and Fizeau .
On the other hand , the complicated change which occurs in passing from anatase to brookite would seem to render it necessary to consider the axes and of brookite together , and to compare them with the second axis and the axis of anatase taken together .
We then have for anatasc , and for brookite , a very ficant expansion .
Turning next to relations between brookite and rutile , the topic axes gest a contraction along the axis , and an expansion along the axes and , but to a greater amount along than .
In correspondence with this , we find that the expansion ( instead of contraction ) which takes place along , is represented by the minimum coefficient , and that the expansion along is greater than along Considering how many factors enter into the problem , this amount of agreement is as great as we can expect .
It does not perhaps lend much confirmation to our hypothetical configuration , but relieves it from serious objections .
Using constant dimensions for our atoms , we have been able to build them up according to the strict laws of crystalline symmetry into the three forms characteristic of the different polymorphs of titanium oxide , we have obtained for our hypothetical configurations similar volumes and similar parametral ratios to those determined by observation , and we have shown how the properties of these configurations are in harmony with what is known of the thermal and optical properties of the minerals they represent .
If , however , our speculations possess any basis of truth , we should be able not only to explain the polymorphs , but also the isomorphs of titanium oxide , such as tinstone and zircon , which crystallise in similar forms to rutile .
I have not as yet sufficiently studied the volumes of silicon and zirconium , and must therefore confine myself for the present to tinstone .
1908 .
] On the Intimate ucture of ystals .
The metal tin crystallises in the tetragonal system , and consequently its utoms are probably not true spheres , or , what possibly comes to the same bhing , their poles are not of equal value .
In the absence of information to the parametral ratio of the crystals , it is impossible to estimate what rmount of deviation from the spherical form exists ; but we may expect find some anomalies consequent upon it in studyiug the metal in its ompounds .
We have already had occasion to refer to one of these compounds* when we took for the diameter of the atom of tin the numbel ) , which we will again adopt .
The specific gravity of tinstone lies veen 6 and 71 , nd the molecular weightis the volume ranges , therefore , from We the crystal to be built up in precisely the same way as utile .
The area of the base of our elementary prism is consequently eqnal or , i.e. , to ; the height is given approximately by the ormula , which is correct to the fourth place of decimals ; it amounts , and this multiplied into gives as the volume found the configuration : thus very nearly equal to the mean of the observed alues .
The parametral ratio found from the configuration is 1 : by bservation it amounts to 1 : , a difference ) .
This is more ccordant than might have been expected .
It may be pointed out that if molecular volumes assist us to mderstand rystalline structure , the collverse is also true that crystalline structure throws ome on the anomalies presented by molecular volumcs .
Taken in the ross the volumes of the three forms of titanium oxide can only wake istrust in the rnificance of atomic volumes ; thus , if we deduct the ooross , tomic volume of titanium from the volumes of the dioxide in its ] orms we have the results:\mdash ; Anatase volume of one atom Brookite , , Rutile , , , , , , by this method , the volume of one atom of oxygen is found to be inconstant and unduly low , but a study of the crystalline hows that the irregularity may be merely apparent , and restores us the xygen with a constant and normal volume .
Before writing this account I was unaware that the subject already treated by previous writers ; a valuable literature , however , exists .
'Ro .
Proc vol. 03 , On th , Intimate Structure of Crystals .
Schrauf has published some yhly speculative views , attributes rutile the structure we have assigned to anatase .
Prior has made estive remarks on the of titanium oxide , to which I hop to recul when treating further of the isomorphs of this substance .
work which most nearly touches my own is by BaumhauerJ who , his attelllion to anatase , arrives at a si1nilar molecular and crystallin structure to that adyocated here , and he supports his conclusions by a variet of tant aeruments based on the crystalline form , and etch ures i in relation to a Bravais net ; but the atoms of the molecule are indicated their centres only , and no attempt is made to discuss the question of molecula volumes .
echt Schrauf , " " Ueber die Trimorphie .
die Ausdehnungscoefficienten vo Titandioxyd ' Zeits . .
Kryst .
, vol. 9 , p. 433 . .
T. Prior , " " Molecular Volume and Chemical ition .
Mag 190 vol. 1.3 , } ) . .
Baumhauer , " " Die uctur des Anatas Zeits .
1895 , vol. 2 p. 555 .

