вход по аккаунту


Functional Groups on the Surfaces of Solids.

код для вставкиСкачать
JUNE 1966
PAGES 5 3 3 - 6 2 2
Functional Groups on the Surfaces of Solids
Atoms or groups foreign to the structure of a solid are often bonded to its surface. On
diamond, graphite, and even silicon dioxide, the foreign atoms are bonded covalently,
whereas the bonding of the structural groups to titanium oxide and alumina is predominantly
ionic. Oxides are normally covered with a monomolecular hydroxide layer. Changes in
the valence of the metal atoms lead to changes in the acidity of the surface; for example,
reduction of surface Ti4f ions gives the surface of Ti02 an acidic nature.
1. Introduction
The three-dimensional periodic arrangement of atoms in
a crystalline solid is interrupted at the surface. The surface may therefore be regarded as an extreme lattice
defect. Surfaces of molecular lattices contain entire
molecules; the only forces acting at the surface are those
that hold the molecules together in the crystal. These
van der Waals forces can form only reIatively loose
bonds with molecules or atoms foreign to the crystal,
i.e. only physical adsorption can occur on such surfaces.
In ionic lattices and lattices consisting of covalently
bonded three-dimensional networks, on the other hand,
the surface ions or atoms are coordinatively unsaturated, i.e. they possess “free valences”. Thus foreign atoms,
groups, or ions can be relatively firmly bound (chemisorption).
In most compounds the bonding is intermediate between
ionic and covalent. Accurate measurement of the electrondensity distribution has shown that even in LiF, there are
indications of the presence of “bonds”, i.e. increased electron
density, between the cations and anions.
The situation in amorphous solids is similar to that in
crystals, since the former also contain small regions of
quasi-crystalline short-range order due to the geometrical arrangement of the valence directions. For example,
in amorphous silica, such as silica gel, the sp3 hybridization of silicon gives rise to SiO4 tetrahedra.
[I] J. Krug, B. Wagner, H. Wittr, and E. Wo(fe, Naturwissenschaften 40, 599 (1953).
Angew. Chem. internat. Edit. / VoI. 5 (1966) No. 6
The strength of the bonds formed with the chemisorbed
atoms can vary widely. A surface compound is formed
when the bond corresponds in nature and strength to
the bonds in the body of the solid or in ordinary chemical
Really “clean” surfaces are rare, since reactive substances, particularly oxygen and water, are present in
the atmosphere. Under normal conditions, therefore,
many substances are covered with surface oxides or
hydroxides. This covering has a decisive influence on the
character of the surface.
It is important for many technical applications to know how
these surface coverings behave. Surface compounds play an
important part, not only in heterogeneous catalysis, but also
in friction and lubrication processes, in the reinforcement of
rubber with fillers or of synthetic resins with glass fibers, in
the behavior of pigments in lacquers and inks, in textile processing agents, in the adhesion of phosphors to television
screens, and in many other processes. Nevertheless, very
little work has been done with a unified approach to this
field. This may be partly because only in substances with
large specific surface areas does the number of foreign atoms
bonded to the surface become analytically important. The
usual chemical methods of analysis can only be used for
substances with specific surface areas of several hundred
square meters per gram. For smaller specific surface areas,
down to values of the order of 1 mZ/g, speciallyadapted micromethods are required, while for still smaller specific surface
areas, such as are found in coarse-grained material, the foreign atoms bound to the surface can be determined only by
physical methods, if at all.
The structures and reactions of some typical surface
compounds are described below. Chemisorption on
metals, which is important in many catalytical processes,
cannot be treated here.
2. Surface Compounds on
Diamond-Like Structures
Diamond is a particularly suitable substance for the
study of surface compounds. In the diamond lattice,
which is also the lattice structure of silicon and germanium, the atoms are linked tetrahedrally because of their
sp3 hybridization. Diamond is the prime example of a
crystal with purely covalent bonds. The most frequent
cleavage plane is (1 1l), i.e. the octahedral face, in which
one bond is broken per C atom. It was observed as
early as 1918 that diamond powder, which is iiormally
hydrophobic, becomes hydrophilic on treatment with
oxidizing agents 121. Natural diamonds from volcanic
pipes (e.g. at Kimberley) are hydrophobic; on the
other hand diamonds from secondary deposits (placers)
are hydrophilic.
This behavior is due to the formation of surface oxidesL31. Diamond powder with surface oxides has a
higher heat of wetting in water than diamond powder
without surface oxides. When diamond powders with
surface oxides are treated with water, carboxyl groups are
formed (Fig. la). Determination of the active hydrogen
by the method of Zerevitinof or by deuterium exchange
shows that nearly all the hydrogen atoms present are
in the form of carboxyl groups 141. However, most of the
oxygen is not bound in the carboxyl groups, but in a
form that is not yet definitely known. No confirmation
has been found for the conjecture that tertiary hydroxyl
groups are bonded to the octahedral face (Fig. lc).
Other groups containing oxygen that could conceivably
be attached to the surface of diamond are indicated in
Figures lb, Id, and le.
1151711 Id1
Fig. 1 .
Functional groups o n the surface of diamond.
Carboxyl groups on an edge (detected);
Carbonyl groups o n a (100) face (presumed);
Tertiary hydroxyl groups on a (111) face (not detectable);
Ether grouping o n a ( I 1 1 ) face (presumed);
Ether grouping o n a (1 10) face (presumed).
A surface oxide structure with unpaired electrons on the
oxygen has also been ruled out. Electron-spin resonance
measurements on a sample with an oxygen content of
320 pg-atornlg of diamond indicated only 7 pequiv of
unpaired electrons/g of diamond 151. Similar results were
obtained when diamond powder was heated to remove
surface oxides, and then treated with fluorine, chlorine,
or hydrogen at elevated temperatures, whereby 200
[2] K . A . Hofmann: Anorganische Chemie. Verlag Vieweg,
Braunschweig 1918, p. 287.
[3] H.-P. Boehm, E. Diehl, W. Heck, and R. Sappok, Angew.
Chem. 76,742 (1964); Angew. Chem. internat. Edit. 3, 669(1964).
[4] R. Sappok and H.-P. Boehm, unpublished work.
[5] We thank Prof. U.Hausser for these measurements,
5 34
pequiv of halogen per gram and 230 pequiv of hydrogen
per gram were bound to the surfacer31. The infrared
spectra of the substances contained corresponding absorption bands. Though the ESR signal showed only a
small decrease in intensity during these reactions, its
shape and width changed appreciably. It follows that
the unpaired electrons are situated close to the surface.
Even diamond that had been freed from surface oxides
in a high vacuum (10-5 to 10-6 mm Hg) at 800 “C did
not exhibit a much greater ESR absorption (11 pequiv
of free radicals/g of diamond). A possible explanation
is the rapid contamination of the surface by impurities
such as vapor from the vacuum grease. Even at this low
pressure, however, surface oxides are probably reformed on cooling. The coefficient of friction of diamond increases markedly when the surface oxides are
removed in an ultrahigh vacuum (5 x 10-10 mm Hg).
If the surface of this diamond is exposed to the residual
gas pressure of 5 x 10-10 mm Hg, the frictional coefficient decreases within 20 min to the value normally
found in air [61. However, chemical analysis revealed
practically no bound oxygen on diamond that had been
heated at 950 “C in a high vacuum and allowed to cool
therein [41. Confirmation of the formation of surface
oxides under these conditions has been obtained in
diffraction studies with low-energy electrons [71.
Low-energy electron diffraction [81 is particularly suitable for
the investigation of the structures of surface compounds, as
well as of the “bare” surface. The surface of a crystal with a
diameter of a few millimeters is irradiated with an electron
beam, the energy of which is varied between 20 and 200 eV.
The corresponding wavelength (0.6 8, at 20 ev) is of the same
order as the interatomic distances in the lattice. Since the
penetration depth is small, amounting to only a few layers of
atoms, the diffraction pattern represents the structure of the
surface. Owing to the low energy of the electrons, the measurements must be carried out at 10-4 to 10-10 mm Hg, i.e.
under conditions in which some surface compounds are not
A pure, fault-free surface gives the diffraction pattern of the
basic structure. If the foreign atoms or the surface atoms of
the crystal have a periodic arrangement with unit cells larger
than those of the crystal lattice, the reflections due to the
latter are accompanied by others having fractional indices
with respect to the “normal” reflections. A structure analysis
of the surface layer is thus possible, but the difficulties involved are considerably greater than in normal X-ray analysis.
The principal results obtained by diffraction of slow
electrons can be summarized as follows. Whereas
“atomically pure” metal surfaces, e.g. the (100) face of
nickel [91 or chromium [lo], have the ideal face-centered
and body-centered cubic lattices, respectively, “clean”
silicon or germanium surfaces (diamond lattice) give
reflections with fractional indices 111-131. In a fresh (1 11)
[6] F. P. Bowden and A . E. Hanwell, Nature (London) 201, 1279
[7] J. B. Marsh and H . E. Farnsworth, Surface Sci. I , 3 (1964).
[8] The results obtained by this method are reviewed by J. J.
Lander, Surface Sci. I , 125 (1964).
[9] A . U. MacRae, Surface Sci. I , 319 (1964).
[lo] C. A . Haque and H. E. Far-nsworth, Surface Sci. I , 378
[l I] R. E. Schlier and H. E. Farnsowrth, J. chem. Physics 30,917
[12] J. J . Lander and J. Morrison, J. chem. Physics 37, 729(1962).
1131 J . J. Lander and J. Morrison, J. appl. Physics 34, 1403(1963).
Angew. Chem. internat. Edit.
Vol. 5 (1966)
1 No. 6
cleavage surface, the Si tetrahedra are distorted to such
an extent that the distance between pairs of surface Si
atoms is only 2.6 8, compared to 3.84A within the
crystal. Naturally, the Si atoms in the next two layers
are also displaced from their normal positions. This
structure is still unstable. When the crystal is annealed
at 700 “C (300 “C for Ge), about 25 % of the surface Si
atoms diffuse to other positions. The result is a surface
structure containing puckered six-membered rings [121.
Calculation of the Si-Si distance in the rings from the
atomic coordinates gives a value of 2.2 A, whereas the
normal bond length is 2.35 A. This indicates some
double-bond character; Lander [*$131refers to the rings
as “warped benzene rings”. A structure consisting of Si
chains with conjugated double bonds of length 2.2 8,
has also been suggested 1141.
In both of these structures, there is a regular pattern of
vacancies in the topmost layer. However, it has been
observed that surface structures with ideal periodicity
are formed when oxygen [111 or iodine 1131 is chemisorbed
on these surfaces at low temperatures 1151. Moreover,
compounds of silicon or germanium containing double
bonds are unknown.
The free valences in the surface exhibit a marked tendency to saturate one another, and the energy liberated
in this way is sufficient to cause severe distortion of the
lattice. These results explain the observation that diamond does not exhibit strong electron spin resonance.
It is apparently impossible t 3 obtain a clean diamond
surface with two-dimensional order. Diffraction patterns
are observed only after treatments that lead to the formation of surface compounds, e.g. after treatment with
aqua regia or hydrogen (at 700”C)[71. The normally
very effective process in which clean surfaces are obtained by bombardment with Arf ions leads to an “amorphous surface” with no periodic order [*I.
After treatment of diamond powder with H2 at 700 OC,
the diffraction pattern of the surface contains strong
reflections with integral number indices, and only weak
reflections of half-integral order. Presumably this treatment generates C-H groups, as is also indicated by the
low heat of wetting in water [31.
Significant effects were also observed on treatment of
the diamond surface with oxygen at 2 x 10-6 mm Hg 173.
The diffraction pattern changes only slowly at room
temperature, and more rapidly at higher temperatures.
Whereas chemisorption of oxygen on silicon leads to
extinction of the reflections of half-integral order C111,
these reflections occur frequently in surface oxides of
diamond [71. No structure analysis has been carried out,
but it may be assumed that the free valences of the C
atoms are saturated by oxygen bridges, again with
distortion of the diamond lattice (Fig. Id and le).
[14] R. Seiwatz, Surface Sci. 2, 473 (1964).
[15] Cf. N . R . Hanserz and D . Haneman, Surface Sci. 2, 566(1964).
[*] Note added in proof: J. J . Lander and J. Morrison, Surface Sci. 4, 241 (1966), report a structure with puckered (2.5
rings for a clean (111) face of diamond, similar to that of Si.
After treatment with H2 a t 900-10OO0C, the idea1 periodicity
was observed, whereas 0 2 did not produce an ordered surface
Aitgew. Chem. internnt. Edit.
1 Vol. 5 (1966) / No. 6
One atom of iodine or bromine per atom of Si (or Ge)
is chemisorbed on the (111) face of silicon (or germanium). Contrary to expectations, the halogen atoms are
not situated directly above the surface Si atoms, but
above the centroid of a triangle formed by three Si
The bonding presumably involves the dorbitals of the halogens.
Further peculiar surface structures are formed when
phosphorus 1121 or aluminum 1161 is chemisorbed on the
(1 11) surface of silicon.
The bonding of chlorine on diamond is different from
that of oxygen or hydrogen. The infrared spectrum of
diamond treated with hydrogen contains an absorption
band at 2850 cm-1[41 (C-H bond). In diamond treated
with chlorine, on the other hand, a weak absorption
occurs at 1410 cm-1, and not between 600 and 800 cm-1
as expected. Diamond treated with oxygen absorbs at
1760 cm-1, indicating the presence of carbonyl groups.
These could easily be formed on the (100) surface of
diamond (Fig. 1); on the (111) surface, however, their
formation requires the cleavage of another C-C bond.
Carbonyl groups on the (100) surface form part of a
warped, rigid c6 ring if the C atoms in the second and
third layers are also considered. This would agree with
the relatively high frequency of the absorption. A very
broad absorption at 1000-1300cm-1 could be due to
ether bridges on the (111) and (110) surfaces (Fig. Id
and le).
Microgravimetric and gas-analytical studies of the oxidation showed that diamond powder is not continuously
oxidized below 360 “C [31. The weight increases to a
terminal value. Nevertheless C02 is formed, presumably
by oxidation of exposed C atoms. The surface oxides are
almost completely decomposed in a vacuum at 950 OC,
with further formation of C02 in addition to CO. If it is
assumed that one 0 atom is bonded to each surface C
atom, isolated C atoms must remain on the surface after
decomposition; when the surface is oxidized again,
these atoms are removed as C02 before a new oxide
layer is formed.
3. Surface Compounds on Black Carbon
Elementary carbon occurs as diamond, graphite, and
graphite-like microcrystalline carbon. Whereas aliphatic
carbon compounds are derived from the diamond
lattice, the graphite layer lattice forms the basis of aromatic compounds. Each atom in these layers is linked
by o-bonds to three neighbors (sp2 hybridization). The
fourth valence electron is bound as a x-electron, which
can easily migrate from one atom to another within
the 1ayer.The layers are relatively loosely held together by
van der Waals forces (distance 3.35 8,).Microcrystalline
carbon consists of packets of layers, in which some 3 to
30 aromatic layers about 10 to 1000 A in diameter are
stacked in an unordered manner (turbostratic order) 1171.
The surface compounds correspond to the end groups
[I61 J. J. Lander and J. Morrison, Surface Sci. 2, 553 (1964).
[I71 U. Hofmann and D . W i h , Z . Elektrochem. angew. physik.
Chem. 42, 504 (1936).
of polymers. It follows from the highly-condensed aromatic layer structure that surface compounds on graphite or microcrystalline carbon can be expected only at
the layer edges (prism faces of graphite). Foreign atoms
or molecules can be only weakly adsorbed on the basal
faces by means of the x-electrons, except where they are
bound at lattice defects.
When microcrystalline carbon is treated with oxygen
just below its ignition temperature, surface oxides with
acidic properties are formed. If, on the other hand, the
carbon is heated to remove surface compounds, and
cooled to room temperature in vacuum or in an inert
gas, it can bind acids with siniultaneous uptake of
oxygen and formation of basic surface oxides. Thus
depending on its pretreatment, the same sample may
carry predominantly acidic groups or only basic
groups [18J. The quantity of the basic surface oxides
formed is always smaller than the quantity of acidic surface oxides ; however, all charcoals with acidic surface
oxides also carry basic surface oxides. Typical neutralization values are about 80-250 mequiv of NaOH/100g
and 10-15 mequiv of HCljlOO g.
The first attempts to elucidate the constitution of the
acidic surface oxides by the methods of chemical analysis were made by Villars [I91 and by Hojmann and Ohlerich [201.
The acidic surface oxides formed by the action of oxygen
on microcrystalline carbon at about 400 "C contain four
types of acidic groups. Carboxyl, phenolic hydroxyl,
and carbonyl groups have been detected. One carboxyl
and one carbonyl group probably combine to form a
lactone ( I ) [211 or (2).
In fully oxidized samples, the four groups are present in
equivalent amounts. When the carbon is treated with
aqueous solutions of strong oxidizing agents (e.g. ammonium persulfate) at room temperature, two carboxyl
groups are found for every one of the other groups, one
of the two carboxyl groups decomposing at about
200°C with liberation of C02. Fig. 2 shows a model
containing all the functional groups that have been
The acidic surface groups are most easily detected by neutralization with bases. Free carboxyl groups react with hydrogen
carbonate ions, the lactone rings are opened by sodium
carbonate, and the phenolic groups require sodium hydroxide for neutralization. The carbonyl group adds sodium
ethoxide to form a salt of a hemiacetal, C(OC2HJO~--.
[IS] H . R . Kruyt and G. S. de Kudt, Kollold-Z. 47, 44 (1929).
[19] D . S . Villars, J. Amer. chern. SOC.69, 214 (1947); 70, 3655
(1. Hofmann and G. Ohlerich, Angew. Chem. 62, 16 (1950).
[21] V. A . Gurten, D . E. Weiss, and J. B. Willis,Austral. J . Chern.
10, 295 (1957).
Open form
Lactone form
Fig. 2. Model containing all the functional groups detected in acidic
surface oxides of carbon. The arrangement of the groups in a closely
spaced row has not been proved. Two carboxyl groups or t w o carbonyl
groups often appear t o be close together. The carboxyl groups can then
form anhydrides [31.
I a . Carboxyl group which is removed at about 200 "C (occurs only in
products that have been oxidized at 20-I5O0C):
I b. Carboxyl group, removable only above 325 "C:
11. Carboxyl group that occurs as a lactone;
111. Phenolic hydroxyl groups:
1V. Carbonyl group that reacts m,ith the carboxyl group 11 to form the
lactone (or lactol).
Na+[221. Moreover, unpaired electrons have also been detected in oxidized carbon [*31.
The acidic surface groups are usually determined by neutralization with dilute sodium hydroxide solution.The samples are
shaken with standard NaOH, and aliquots of the solution are
back-titrated. It is important that shaking is continued t o
complete neutralization 13,229 241. The time required for complete neutralization is 96 hours with 0.01N NaOH, or 12
hours with O.05N NaOH [24a1. When concentrated sodium
hydroxide solutions are used, more base is consumed, corresponding to the sodium ethoxide value; this value is reached after 96 hours with 0.25 N NaOH, and after 36 hours with
1 N NaOH. The use of barium hydroxidec241 has two disadvantages: 1. Porous charcoals nearly always contain adsorbed C02, which forms insoluble BaCO3 (Na2CO3, on the
other hand, is included in a back-titration, and so does not
affect the result); 2. exchange of acidic hydrogen for bivalent
cations may occur either between equivalent quantities or
between equimolar quantities (Fig. 3). Equivalent ion exchange occurs when the charged surface groups are situated
close together, e.g. in synthetic ion exchange resins. In equimolar ion exchange, a bivalent (or multivalent) cation, together with additional anions to preserve electrical neutrality,
is bound instead of a monovalent cation: this occurs when
the charges are relatively far apart. The possibility of equimolar ion exchange with multivalent ions must be borne in
mind in all reactions used for the determination of surface
groups. Charge compensation is better in this case than in
Fig. 3. Reaction schemes for equivalent (a) and equimolar (b) ion exchange. Example: neutralization of surface carboxyl groups.
[22] H . - P . Boehm, E. Diehl, and W . Herk, Ind. Carbon and
Graphite, in press.
[23] J. B. Donnet and G. Henrich, Bull. SOC.chim. France 1960,
1609; J.B. Donnet, G. Henrich, and G. Riess, Rev.gen. Caoutchouc
Plastiques 38, 1803 (1961).
[24] B. R . Puri and R . C. Bunsal, Carbon 1, 457 (1964).
[24a] H . P . Boehrn and H . J . Kuhn, unpublished work.
Angew. Chem. internat. Edit. 1 VoI. 5 (1966)
1 No. 6
equivalent exchange. Pauling 1251 has pointed out that the
positive ions in a crystal lattice occupy the positions at which
the negative potential is greatest, and vice versa. The same is
true on the surface[261. The state adopted is that with the
minimum potential and the smallest possible distance between opposite charges. The phenomenon of equimolar ion
exchange was first observed in kaolinitc by Weiss[261, and
was subsequently also observed on the surface of Si02[27].
However, the more strongly acidic groups of the surface
oxides of carbon appear to lie sufficiently close together to
ensure equivalent exchange [3,241.
It has been shown[281 that oxygen is almost entirely
chemisorbed on the prism faces of graphite crystals.
Single crystals 2 mm iri diameter and about 0.1 rnm high
were treated with 0 2 at 700 "C. The oxides ofcarbor. given
off in vacuum at 900 "C were determined by mass spectrometry. The experiment was then repeated, but the
crystals were first cleaved five times parallel to the layer
planes, so that while the area of the prism faces remained the same, the area of the basal faces was increased by
a factor of six. The quantity of gas given off was not
affected by this treatment.
Carbon blacks that have been graphitized at 2400 to 3000 "C
have a very homogeneous surface [291. The surface of the isometric particles consists almost entirely of the basal surfaces
of pyramidal graphite crystals, the apices of which are situated in the center of the particle [301. Consequently, graphitized
carbon blacks form practically no acidic surface oxides 1311.
The functional groups on the carbon surface undergo
the usual reactions of organic chemistry. The carboxyl
groups react with diazomethane [3,20,22,31-331, alcohols [31, acetyl chloride [31, or SOC12 [3,22,311. FriedelCrafts reactions and Curtius rearrangements can be
carried out [3,221. The phenols react with p-nitrobenzoyl
chloride or with 2,4-dinitrofluorobenzene 13,221. The
carbonyl groups add ethoxidc, and react with hydroxylamine to form oxiines 1221.
Although the basic surface oxides have been known
longer [341 than the acidic ones, nothing definite is
known as yet about the constitution of the former.
Carbon binds mineral acids only in the presence of
oxygen [351. The extent of adsorption depends on the
partial pressure of oxygen, at least below 20 mm Hg [36,
[25]L. f'auling, J. Amer. chem. SOC.51, 1010 (1929).
1261 A . Weiss, Kolloid-Z. 158, 22 (1958);Z.anorg. allg. Chem.
299, 92 (1959).
[27] H.-P. Boehm and M . Schneider, Z. anorg. allg. Chem. 30;.
326 (1959).
[28] G.R. Hennigin: Proc. 5th Conf. on Carbon, 1961.Pergamon
Press, London-New York 1962, Vol. I, p. 143.
[29] R . A . Beebe and D . M . Young, J. physic. Chem. 58,93 (1954).
[30] E. A . Kmrtko in: Proc. 1st and 2nd Conf. on Carbon, University of Buffalo, N.Y.(1956), p. 21 ; H.-P. Boehm, Z. anorg.
allg. Chem. 297, 315 (1958).
[311 H.-P. Boehtn, E. Diehl, and W. Heck, Rev. gCn. Caoutch3uc
Plastiques 41,461 (1964).
[321 M . L. Studebaker, E . W. D . Hrrfmnn, A . C . Wove, and L . ti.
N a b o r ~ Ind.
Engng. Chem. 48, 162 (1956).
[33] H . - P . Boehni and E. Diehl, Z . Elektrochem. Ber. Bunsenges.
physik. Chern. 66, 642 (1962).
[341 F. E. Bartell and E. J . Miller, J. Arner. chem. SOC.34, 1866
(1922);45, 1106 (1923); E. J . Miller, ibid. 46, 1150 (1924);47,
1270 (1925).
[35] R. Burstein and A. Frumkin, Z. physik. Chem. A 141, 219
( 1 929).
[36] N . Schilow, G . Schn~unowskaja,and K . Tsrl7mutoU, 2. physik.
Chem. A 149, 211 (1930).
Angew. Chem. internat. Edit.
I/ Vol. 5 (1966) No. 6
Part of the bound acid is undoubtedly physically
adsorbed, particularly at higher concentrations, since it
can be displaced by solvents such as tolueneI3*.391.This
result indicates that a relatively weak interaction occurs
with the x-electrons of the layers; perhaps proton complexes of the aromatic layers are formed 138,401.
According to Burstein and Frumkin [411, the reaction of basic
surface oxides with acids yields H202:
+ 2 H + + 2 X-
+ Cx?'(X-12
However, this product has never been detected in the expected
quantities, possibly because of carbon-catalysed decomposition. A substance that oxidizes KI has been repeatedly observed [421. More recent studies [431 have shown that the acid is
adsorbed relatively slowly, and that the concentration of
Hz02 in the surrounding solution passes through a maximum.
The salts of the basic surface oxides decompose in vacuum
above 60 "C 1431. Garten and Weiss [391 believe that the oxygen
is present in chromene-like structures ( 3 ) , in which the positive charge is delocalized after the addition of acid as in benzopyrylium salts.
Free radicals such as ( 4 ) could also be responsible for the
reaction [391.
On the other hand, carbonium and oxonium bases are normally very weak, and washing would be expected to lead to
complete hydrolysis; the charcoals, however, cannot be completely freed from bound acids by washing with water [44,45].
The anion must be ionically bound, since it is exchangeable
[38,451. Cationic sites have been detected in carbon blacks
by hydride-transfer reactions [451; thus isopropanol in 50
HzSO4 has been oxidized to acetone:
RtX--I- CH3 CH(0H) CH3
The original carbon black, after oxidation with triphenylmethyl perchlorate, can also oxidize isopropanol:
R--H+ (C6H5)3C-CID;
+ R-CIO;
+ (ChH5)3CH
The corresponding formate decomposes with evolution of
COz 1451
> R--H-{-CO2
Since acids are bound only in the presence of oxygen,
which can be detected in the reaction product 1431, it is
justifiable to speak of basic surface oxides. These prob[37] 0 . Bretschneider, Z. physik. Chem. A 159,436 (1932).
[38]B. Sternberg, Dissertation, University of Uppsala 1944.
[39] V. A . Garten and D. E. Weiss, Austral. J. Chem. 10, 309
[40] See, e.g., H. H . Perkampus and E. Baumgarten, Angew.
Chem. 76,965(1964);Angew. Chem. internat. Edit. 3,776(1964).
[41] R . Bursteiu and A . Frumkirt, Doklady Akad. Nauk SSSR 32,
327 (1941).
[42] A . B. Lamb and L. W . Elder, J . Amer. chem. SOC.53, 137
(1931);J. M . Kolthof, ibid. 54, 4473 (1932); A . King, J. chem.
SOC.(London) 1934, 22.
[43] H.-P. Boehm and M . Voll, unpublished work.
[44] E. J. Miller., J. physic. Chem. 36, 2967 (1932).
[45]D . Rivin: Proc. 5th Conf. on Carbon, 1961. Pergamon
Press, London-New York 1963, Vol. 11, p. 199.
ably play an important part in the transfer of oxygen by
carbon in fuel cells using atmospheric oxygen or in the
use of carbon as a depolarizer in other cells.
Hydrogen and halogens, as well as ather elements, are
bound on the surface of graphite and of microcrystalline
carbon. In carbons (e.g. carbon black) obtained from
hydrocarbons or other organic materials, most of the
free valences at the edges of the carbon layers are saturated by hydrogen.
4. Surface Compounds on Silica
Silica occurs in many crystalline forms and in an amorphous form. Apart from stishovite, which has the rutile
structure, and the very unsiable fibrous silica, which has
the SiS2 structure [4'51, all the modifications have the
well-known three-dimensional network of Si04 tetrahedra with common corners. The Si-0 bond is partially
ionic[47,481. The bond angle at the oxygen atom in the
siloxane bond Si-0-Si is close to 140 " 1491. This value
suggests that the electron pairs of oxygen that are not
used in the a-bonds are also involved in the bonding
The surface chemistry of silica appears at first sight to
be much simpler than that of carbon. Under normal
conditions, only two types of "end groups" are conceivable on the surface, namely siloxane groups (5) and
silanol groups (6).
products obtained by condensation of low molecularweight silicic acids also contain silanol groups in the
interior of the particles, since some of the silanol groups
fail to find partners during the condensation. These
groups are mostly, but not entirely, decomposed at
1000 "C. Small amounts of O H groups have even been
detected in vitreous silica [52,531 and in rock crystal [541
by their infrared absorption. Water molecules may also
be very strongly adsorbed, and they may fail to escape
on drying at 150 to 200 "C[551. This adsorbed water can
be titrated by the Karl Fischer method. If a suitable
correction is made to the loss on ignition, the resulting
silanol contents agree satisfactorily [561 with the values
obtained e.g. by reaction with thionyl chloride "271.
Calculation from the loss on ignition often gives a very
high value for the packing density of the silanol groups
on the surface. The maximum possible packing density
can be estimated from the structure of the crystalline
modifications of Si02. In this way de Boer 1571 obtained
a value of 4.55-4.85 OHjl00 A2 for cristobalite and
tridymite. The calculated values [581 for various crystal
faces are listed in Table 1. If the lower density of amorphous Si02 is taken into account, averaging over all the
crystal faces gives a packing density of about 5 OH/
100 A2 for amorphous Si02. However, the values found
by measurement are nearly always lower [591.
Table I . Packing density of silanol groups on various crystal faces of
silicon dioxide.
(d = 2.655)
(d = 2.32)
By far the most investigations have been carried out with
amorphous silica with a large specific surface area, mainly
because this material has many industrial uses. Porous silica
gel, for example, is used as a n adsorbent and drying agent,
while other types with spherical particles are used as fillers in
rubbers and plastics. SiOz is also used as a powder base, as a
toothpaste additive, and in the textile industry.
The very pure and non-porous "Aerosil" [*I and Cabosil [*I,
which are produced by flame hydrolysis of SiC14, have proved
to be particularly suitable for many measurements.
U.Hofmann et al. [SO] were the first to recognize that the free
valences at the edges of the silicate layers in the clay minerals
must be saturated by silanol groups. Curman 1511 came to the
same conclusion for colloidal amorphous SiOz.
The number of surface silanol groups was at first determined from the loss on ignition at 1000-1 100 "C. However, this method raises certain difficulties. Many
[46] A . Weiss and A . Weiss, Z. anorg. allg. Chem. 276, 95 (1954).
[47] W . Noll, Angew. Chem. 75, 123 (1963); Angew. Chem.
internat. Edit. 2, 73 (1963).
[48] H. Schmidbaur, Angew. Chem. 77, 206 (1965); Angew.
Chem. internat. Edit. 4, 201 (1965).
[49] F. Liebau, Z. Naturforsch. 15b, 468 (1960); Acta crystallogr.
14, 1103 (1961).
[*I Aerosil is manufactured by Degussa, Hanau (Germany), and
Cabosil by Cabot Corp., Cambridge, Mass. (U.S.A.).
[SO] U.Hofmann, K. Endell, and D. Wilm, Angew. Chem. 47,539
[51 J
P. C. Carman, Trans. Faraday SOC. 36, 964 (1940).
(d = 2.26)
Packing density
(OH/100 A2)
[a] for crystalline SiOz.
[b] calculated value for amorphous SiOz, based on density d
Hydroxyl groups on the SiOp surface have also been detected
by infrared spectroscopy. Aerosil that had been degassed at
30 "C showed absorptions at 3750 cm-1 (isolated silanol
groups), 3660 cm-1 (hydroxyl groups with weak hydrogen
bonds), and 3520 cm-1 (strong hydrogen bonds) [601. When
the sample was heated, the absorption at 3520 cm-1 receded
first; outgassing at 940°C led t o the disappearance of the
peak at 3660cm-1, and the intensity of the absorption at
3750 cm-1 increased at the same time. It can be concluded,
therefore, that more than two hydroxyl groups must be
present in the complexes that absorb at 3660 cm-l, since no
OH groups would otherwise remain after the removal of
[52] A . J. Moulson and J. P . Roberts, Nature (London) 182, 200
[53] H. Schafer and K . Etrel, Z. anorg. allg. Chem. 301, 137
[54] G. 0.Brunner, H . Wondratschek, and F . Laves, Z. Elektrochem. angew. physik. Chem. 65, 735 (1961).
[55] W . Stober, Kolloid-Z., 145, 17 (1956).
[56] W . Noll, K . Damm, and R. Fauss, Kolloid-Z. 169, 18 (1960).
[57] J. H . de Boer and J. M . Vleeskens, Proc. Kon. nederl. Akad.
Wetensch., Ser. B 61, 2 (1958).
[58] M . Schneidzr, Dissertation, Universitat Heidelberg 1962.
[59] H.-P. Boehm and M . Schneider, Kolloid-Z. 187, 128 (1963).
[60] R. S. McDonald, J . physic. Chem. 62, 1168 (1958).
Angew. Chem. internat. Edit.
Vol. 5 (1966)
/ No. 6
water. The observation that the OH groups with the stronger
hydrogen bonds are the first to be split off has been repeatedly
confirmed [61,621. Different reaction rates were also observed
in the OHjOD exchange on the silanol groups[621. The combination and overtone vibrations in the region between 1 and
3 p 163-651 are particularly suitable for differentiation bttween
silanol groups and water.
A silica particle may be regarded as very highly polymeric polysilicic acid. The silanol groups on the Si02
surface enter into many chemical reactions. They are
weakly acidic, and can be titrated in concentrated NaCl
solution with NaOHC661. In our experience, all the
silanol groups are neutralized at p H 9.0; further addition of NaOH leads to cleavage of siloxane bonds. On
neutralization with Ca(OH)2 an equimolar reaction
occurs [271.
The reaction with thionyl chloride
Si-OH+ SOC12 + sSi-CI
+ SO2-t HCI
can also be used for the determination of silanol groups
if the substance does not contain narrow pores [27,67,681.
The reagent and the reaction products are held so firmly
in narrow pores that they cannot be completely removed
by degassing at 200 "C.
However, the content of active hydrogen found by Zerevitinoff determinations was higher than that corresponding to
the chloride uptake from SOCI21691, but this may be due to
very strongly adsorbed water (see below). The OH absorption
in the infrared spectrum of porous Vycor glass (which is 96 %
Si02) disappeared when the glass was treated with SOC12, and
after subsequent treatment of the glass with NH3, the spectrum indicated the presence of Si-NH2 groups "01.
Esterification with alcohols can be used for the analytical
determination of silanol groups via the methoxy or carbon
content. The silica is heated with the alcohol in a n autoclave
or, for higher alcohols, in a vessel fitted with a rectifying
column, the water formed being distilled off as a n azeotrope 171-731. The silanol groups can also be methylated with
diazomethane 1279 741. Esterification makes the surface hydrophobic, and leads to a finely divided,silica which can be readily
incorporated into oils or waxes[711. .
Of the many other reactions ofthese surface hydroxides,
we shall mention only those with Si(CH3)3Cl[75,761, BCl3
[61] G. I. Young, J. Colloid Sci. 13, 67 (1958); A . V. Kiselev and
V . I. Lygin, Kolloidnyi Zh. 21, 581 (1959).
[62] J. J. Fripiat, M . C. Gastuchc, and R . Brichard, J. physic.
Chem. 66, 805 (1962).
[63] G. Wirzing, Naturwissenschaften 50, 466 (1963).
[64] G. Wirzing, Naturwissenschaften 51, 21 1 (1964).
[65] J. H. Anderson Jr. and K . A . Wickersheim, Surface Sci. 2,252
[66] G. W . Sears Jr., Analytic. Chem. 28, 1981 (1956); W . M .
Heston Jr., R . K. ller, and G . W . Sears Jr., J. physic. Chem.46, 147
[67] H.-P. Boehm, M . Schneider, and F. Arendt, Z. anorg. allg.
Chem. 320, 43 (1963).
[68] J. Wartmann and H. Deuel, Chimia 12, 82 (1958).
[69] J. Uytterhoeven and H. Naveau, Bull. SOC. chim. France
1962, 27.
[701 M . Folman, Trans. Faraday Soc. 57, 2000 (1961).
[711 R. K . Iler, U.S.-Pat. 2657149 (1953j, DuPont.
1721 W . Stober, G. Bauer, and K . Thomas, Liebigs Ann. Chem.
604, 104 (1957).
[731 C . C. Ballard, E. C . Broge, R . K . ller, D . S. St. John, and
J . R . McWhgrter, J. physic. Chem. 65, 20 (1961).
[741 G. Berger, Chem. Weekblad 38, 42 (1941); K . H . Ehert, Mh.
Chem. 88, 275 (1957).
[751 W . Stuhtr, Kolloid-Z. 149, 39 (1956).
Angew. C h e m . internat. Edit. / Vol. 5 (1966)
No. 6
or AIC13 [671, with trialkylaluminum 1771, and with
B2H6 178,791, in which the hydrogen of the silanol groups
reacts to form HCI, alkane, and H2 respectively. Coating as in (7) with trimethylsilyl groups leads to a marked
decrease in the rate of dissolution of finely divided
silica in water [751. Although some of the silanol groups
remain unchanged, the surface is protected by the trimethylsilyl "umbrellas". Coating with a monoatomic
layer of AP+ practically prevents the removal of Si(OH)4
from the surface of Aerosi1[67.771. These ions are adsorbed hydrolytically on the Si02 surface from basic
aluminum chloride solutions containing 1 to 2 hydroxide ions per Al3f ion [sol; one Al3+ ion is firmly bound
per silanol group 1811. During hydrolytic adsorption, the
polynuclear complexes [Alx(OH),(H20),](3x-r,f in the
basic aluminum chloride solution give up aluminum
hydroxide to the SiOz surface.
The solubility of orthosilicic acid in water is about
1.67 mmole/l. In a solution of NaCI/NaHCO3 buffered to pH
8.2, 123 pg of Si02 dissolves per ml ofsolution at 20 "C. When
the solution is shaken with Aerosil (specific surface area about
150m2/g), this value is reached in about 24 hours. If the
solvent is agitated under the same conditions with Aerosil
coated with AI3+ ions, only 6 pg/ml dissolves in three weeks
[671. The aluminum layer can be readily removed with mineral
Many other ions are hydrolytically adsorbed o n the silanol
groups of the Si02 surface[82,831. This behavior can be used
for the analytical separation of ions on silica gel columns 1843
Table 2 summarizes results of many reactions on Aerosil. The
packing density of the surface silanol groups can be calculated
fromthe specific surface area and the conversion.Table2 shows
remarkable agreement between the results obtained by various methods, particularly in the experiments carried out
with the same Aerosil sample of specific surface area 145 m2/g.
[76] H . W . Kohlschiitter, P . Best, and G. Wirzing, Z. anorg. allg.
Chem. 285, 236 (1956); G. Wirzing and H. W. Kohlschiitter, Z.
analyt. Chem. 198, 270 (1963).
[77] M. Lieflander and W. Stober, Z . Naturfcrsch. I%, 41 l(1960).
[78] I. Shapiro and H . G. Weiss, J. physic. Chem. 57, 219 (1953);
H . G. Weiss, J. A . Knight, and I. Shapiro, J. Amer. chem. SOC.81,
1823 (1959).
[79] C. Naccache, J. Frangois-Rosetti, and B. Imelik, Bull. SOC.
chim. France 1959, 404; C. Naccache and B. Imelik, C.R. hebd.
Seances Acad. Sci. 250,2019 (1960); Bull. SOC.chim. Francel961,
[80] H . W. Kohlschiifter, S . Miedtank, and H. Gefrost, Z. anorg.
allg. Chem. 308, 190 (1961).
[81] H.-P. Boehm and M . Schneider, Z. anorg. allg. Chem. 316,
128 (1962).
[82] J. H . Stanton and R. W . Maatman, J. Colloid Sci. 18, 132
(1963); D . L. Dugger, J . H . Stanton, B. N . Irby, B. L. McConndl,
W. W . Cummings, and R . W . Maatman, J. physic. Chem. 68,157
[83] S. Ahrland, I. Grenthe, and B. NorPn, Acta chem. scand. 14,
1059 (1960).
[84] G . - M . Schwab and K . Jockers, Angew. Chem. 50,546 (1937).
[85] H . W . Kohlschiitter, H . Getrost, G. Hofmann, and H. H .
Stamm, Z. analyt. Chem. 166,262 (1959); H . W. Kohlschiitter and
H . Getrost, ibid. 167, 264 (1959); H . W . Kohlschiitter, S. Miedtank, and H. Getrost, ihid. 192, 381 (1963).
Table 2. Amount and packing density of silanol groups on the surface
of Aerosil.
Weight loss at 1000 "C
- free HzO ( K . Fischer
Reaction with SOClz
Zerevitinoff method with
CH3MgI or CH3Li
Reaction with SOCI?
Reaction with SOCIz
Titration with NaOH [661
Reaction with BCIJ
Sorption of AlJ+f r o m
AI(0H)zCI solution
Sorption of uo:+ from
at pH 5.4 [82]
Methylation with CH30H
at 200-250 "C
DzO exchange after
outgassing at 200OC
Zerevitinoff method with
CH3MgI after
outgassing at
120-200 "C
Infrared spectra
Silanol groups
100 A2)
Values much smaller than 5 OH groups per 100A2 were
found by all methods except the Zerevitinoff active hydrogen
determination and the D20 exchange method. The value
found by the active hydrogen determinations was exactly
twice as high. Stober 1551 attributed this result to the fact that
for every two silanol groups, one molecule of water is adsorbed so strongly that it is not released even in a high vacuum at
100-200 "C. This argument is supported by the results of the
reaction with BC13 or AlC13 [671. However, the results of the
Zerevitinoff determinations do not fit this explanation, since
only one H atom of water reacts with CH3MgI. The infrared
spectrum of Aerosil treated with SOC12 still contains OH
absorptions [891. Aerosil deuterated with D 2 O was treated with
SOC12; it then contained half as much exchangeable deuterium as before. Thus the OH groups that are stable towards
SOClz are situated at the surface.
It is difficult to see why only half of the silanol groups
react with the other reagents, since it is very unlikely that
the surface contains only silanediol groups =Si(OH)2.
These conflicting results require further experimental study.
The packing density of silanol groups is influenced by the
particle size of the SiOz, and appears to decrease with
increasingly fine division 1901.
The existence of siloxane groups on the surface of silica
has been inferred only from the observation that the
number of silanol groups detected is not sufficient t o
cover the surface completely. Silica that has been heated
t o a high temperature contains practically n o residual
silanol groups. The dehydration is reversible if carried
out below 450°C. If, however, Si02 is heated to 800 t o
1000 "C, no rehydroxylation occurs a t room temperature o n exposure to water vapor below saturation
H.-P. Boehm, unpublished work.
H.-P. Boehm and D . Brand, unpublished work.
C . GokGek, Diploma Thesis, Universitat Heidelberg 1963.
H.-P. Boehm and H. Wistubrr, unpublished work.
M . M . Egorov, V. F. Kiselev, and K. G. Krasil'nikov, Zh.
fiz. Khim. 35, 2234 (1961); M . M . Yegorov and V. F. Kiselev, ihid.
36, 318 (1962).
pressure [61,91,921, and the silanol groups are regenerated slowly over a period of several months in aqueous
suspension [93 ,941. This difference in behavior is ascribed
t o the stability and inertness of the siloxane grouping.
Siloxane bridges formed at low temperatures are still
strongly "strained". The strain can lead t o distortion of
the bond angles o r bond lengths and to strongly p d a r
bonds. Only above 450°C is the thermal energy sufficient for rearrangement of the tetrahedra to such anextent
as t o remove the strain.
The inertness of the siloxane linkage is due t o the d,-p,
component; it is the only reason for the resistance of
silica vessels to many reagents at relatively low teniperatures. The Lewis basicity of oxygen is greatly reduced
in the siloxane group, and there is practically n o tendency for the oxygen to form hydrogen bonds [951. Consequently, the surface of silica that has been heated t o
high temperatures becomes hydrophobic. Water-vapor
adsorption isotherms are of type 111[611 in Brunauer's
classification [961, or type V for narrow-pored substances [9*,92,97J(Fig. 4).
Fig. 4.
Adsorption isotherms of water vapor on silica gel at 18 'C,
after vacuum treatment at (a) l O O " C , (b) 1000°C. The desorption
branches are not shown (after [971).
pipo = equilibrium pressurelsaturation pressure at 18 'C; xlin - quantity of water adsorbed in mm3/g of silica gel.
Siloxane linkages are cleaved by nucleophilic attack on the
silicon. The best-known example is the dissolution of SiO2 in
alkalis. Similarly, Aerosil is attacked by organolithium compounds 1981. The reaction with phenyl-lithium gives tetraphenylsilane, triphenylsilanol, diphenylsilanediol, and polymeric siloxanes. The SiOz surface contains phenyl groups
bound to Si.
[91] J . H . de Boer, M . E. .4. Hermnns, and J . M . Vleeskms, I'roc.
Kon. nederl. Akad. Wetensch., Ser. B 60, 45 (1957).
f921 M . M. Egorov, T. S. Egorova, V. F. Kiselev, and K. G .
Kfasil'nikov, Doklady Akad. Nauk. S.S.S.R. 114, 579 (1957); A .
V . Kiselev and G . G . Muttik, Kolloidnyi Zh. 19, 562 (1957).
[93] S. P. S h d a n o ~ , Doklady Akad. Nauk S.S.S.R. 123, 716
[94] B. V. I / j i n , V . I;: Kiselev, and K . G . Krasil'nikov, Vestnik
Moskovskogo Univ., Ser. Mat., Mechan., Astron., Fiz. Khim. 12,
35 (1957), Chern. Abstr. 53, 807 (1959); A . A . Issirikyan and A. V .
Kiselev, Doklady Akad. Nauk S . S . S . R . 115, 343 (1957).
[95] C. M . Huggins, J . physic. Chem. 65, 1881 (1961).
1961 S . Brunauer, L. S . Deming, W. E. Deming, and E. Telkr,
J. Amer. chern. SOC.62, 1723 (1940).
[97] H . W. Kohlschiitter and G . Kiimpf, Z. anorg. allg. Chern. 292,
298 (1957).
[98] H.-P. Boehm, M . Schneider, and H. Wistubn, Angew. Chern.
77, 622 (1965); Angew. Chem. internat. Edit. 4 , 600 (1965).
A n g e w . Chern. internat. Edit.
1 Vol. 5 (1966) No. 6
Silanol groups have been detected on the surface of
crystalline quartz 1553. Chemical detection is difficult in
this case owing to the small specific surface area. The
Nab uptake from sodium hydroxide solution labelled
with 22Na gives a value of 4.25 OH1100 & for the packing density of the silanol groups[991. OH groups have
also been detected in the high-pressure modifications
coesite and stishoviteflml. Stishovitc, in which the silicon atoms have the coordination number 6, dissolves
more rapidiy in water than the other modifications of
Si02, but iscompletelyinsolublein hydrofluoricacid E1003.
Xts surface chemistry should resemble that of its structural analogue, rutile, i.e. it should point to the presence
of serongly polar Si-0 bonds.
5, Surface Compounds on Titanium Dioxide
Titanium dioxide has three naturally occurring modifications, rutile, anatase, and brookite, the first two o€
which are produced industrially as important white pigments. In all three crystal lattices, the Til+ ions are
surrounded by slightly distorted octahedra of six 0 2 ions, and the coordination number of the oxygen with
respect to titanium is 3. The bonding may be regarded as
essentially ionic. This leads to differences in behavior
between the surfaces of titanium dioxide and of silica.
Hydroxyl groups have been detected on the surface of
titanium dioxide by their infrared absorption [101,1021.
Whereas anatase still contained water molecules (bending vibration at 1605cm-') after vacuum drying at
150"C, two types of hydroxyl groups, which absorb at
3715 and 3675 em-1, remain at 350 'C. Under the same
conditions, rutile gives only one band at 3680 cm-1.
Water vapor adsorption and the heats of adsorption f103,1041,
and the heat of wetting in water [lo51 a190 indicate that OH
groups are still present on the surface of Ti02 after evacuation
at high temperatures. NMR measurements showed the
presence of very firmly bound water in rutile after outgassing
at 350°C[1Q61.
How can the presence of hydroxyl groups be explained?
Figure 5 shows a section through an anatase crystal in
the (100) plane, Xf the crystal is imagined to be cleaved
along the (001) plane (broken line), it has the surface
structure shown in Figure 5a: each TPt ion lacks one
0 2 - ion in its coordination sphere, and each 0 2 - ion has
only two Ti4+ neighbors. It may be assumed that in the
1991 A. M . Gaudin, H. R. Speddcii, and P. A. 14nseiz,Trans. Instn.
Mining Engr. 4, 693 (1952); Chcm. Abstr. 46, 7485 (1952)
[I001 W. Stb'bcr, Reitr. Silikose-Forsch.,Special Vol. 6, 35 {1964}.
[roll D. J. C. €'ares, J. physic. Chern. 65, 746 (1961).
(1021 f. T. Smith, Nature (London) 201, 67 (1964).
11031 L. G. Ganiclteitkn and V. F. Kiselev, Doklady Akad. Nauk
S.S.S.R. 138, 608 (1961); L. 6. Gnnicjieti~u,V. F. Kisdev, and
W. Mr. Mrtrinn, Kinetika i. Katalir, 2, 877 (1961).
[iO4]C. M. r ~ o l / a b a i and
J . J . Cliessick, J . physic. Chem. 65,
I09 (1961).
[lOS] W'. H. Wmie and N. Hackennomt, J. physic. Chcm. 6.7, 1681
[I061 J. iM. Moys and G. W. Brad.v, 5. chem. Physics 23, 583
[I071 L. P d i n z g : The Nature of the Chemical Bond. Cornclt
Press, 3rd Edit., Ithaca, N.Y. 1960, p. 548.
Aitgerv, Cliem, infertiat. Edit.
/ Vool. 5 (1966) No. 6
Fig. 5. Addition of water and formation of hydroxide ions on &I fresh
cleavage plane of anitasc ( .:Ti; 0:O).
(a) Fresh cleavage plane along (031); (b) completion of coorrlination of
the Ti'& ions by addition of water: (c) formation OF hydroxide ions by
proton transfer t o the 0 1 - ions on thc surface.
presence of water vapor, the coordination gaps on the
Ti4+ ions are immediately filled by H20 molecules
(Fig. 5b). The equilibrium vapor pressure of the water
adsorbed in this way can be expected to be very low.
However, even this state is not stable. Proton transfer
from H20 to 0 2 - (Fig. 5c) gives a surface covered with
hydroxide ions, in which the charge distribution is much
more favorable.
According to Pardirtg's electrostatic valence rule [1071, the
negative charge on any anion in a stable ionic lattice is equat
or nearly eqitat to the sum of the "strengths" of the electrostatic bonds joining the anion to the surrounding cations
(and vice versa). The strength of a bond is defined as the
quotient of the charge, 2, and the coordination number, v, of
the cation. Thus the charge on the anion, <, is given by
Thus in anatase, the charge on the anion situated above the
Ti4+ ion should be -213, while the charge on the oxygen ion
bound to two cations should be -4i3. This requirement is
better satisfied by the charge -1 on the hydroxide ion than
by the arrangement of Fig. 5b, in which the charge on the
water i s zero and that on the oxygen is -2, In other words,
the stable state is that with the lowest potential, i.e. with the
lowest free energy, and with the lowest possible charges and
the minimum separatioa of opposite charges. Since the bonding forces on the surface OH groups act in only one direction,
the distance from these groups t o the Ti4+ ions should be
slightly shorter. A very similar situation is found on the other
cleavage planes of anatase[(lll) is the most common] and on
those of rutile. This is also true of many other metal oxides.
In experiments with titanium dioxide, its sensitivity to reduction is important. When finely divided titanium dioxide is
heated in vacuum, it turns gray to blue-gray, whereas it
remains white when heated in air. This color change can be
attributed to adsorbed organic impurities which are nearly
always present [I042 1081. On decomposition these form carbon,
which reduces Ti4* to Ti3*.
Organic contamination of the surface is a very common
phenomenon, and the contaminants can be determined analytically in substances with sufficiently large specific surface
areas. Finely divided anatase (Ti02 P 25[*I, specific surface
area 56 d i g ) was found to have a carbon content of 0.11 >$,
and Aerosil (SiOz, specific surface area 145 m*:g) also to contain 0.11 y: of carbon 11091,The organic impurities are presumably derived from volatife constituents (e.g. antioxidants) of
I1081 J . Gebhnrdf and K. Hwriizgtoir, J. physic. Chem. 62, 120
( I 958).
[*] Ti02 P 25 is a very pure, lineiy divided anatase, produced by
Drgussa, Hanau (Germany), by flame hydrolysis of TiC14.
[to91 M. Herrmanrr, Dissertation, Universitlt Heidclbcrg 1965.
the plastic containers, from solvents in the laboratory atmosphere, from vacuum greases, and from respiration products
of plants [11OJ. The carbon content of titanium dioxide could
be reduced to 0.06% by extraction with petroleum ether
followed by vacuum outgassing or by treatment with ozone
at 100 "C[log]. The carbon remaining after this treatment was
traced to C 0 2 , which is adsorbed on anatase even at low
partial pressures. At 25 O C and a COz pressure of 7 mm Hg
(corresponding to a relative pressure pipo = 1.4 x lO-4), the
product Ti02 P 25 adsorbs 2 mg of COz per g of TiOz. The
adsorbed C 0 2 can be detected in the infrared spectrum [loll.
The oxidizing power of Ti02 is greatly increased by
ultraviolet light [111J. Under these conditions, triphenylmethane derivatives can be oxidized to the leuco bases
of the corresponding dyes [1121. Owing t o this oxidizing
power on irradiation, paints containing untreated Ti02
rapidly deteriorate in light. To avoid this "chalking",
the pigment particles are coated with thin layers of other
oxides, such as alumina or silica.
The hydroxide ions o n the surface of Ti02 can be
detected by their reactions, e.g. by esterification with
alcohols [77,113,1141. Reactions with trimethylsilyl chloride and with triisopropylaluminum have also been
described [77,1141 (see also Table 3).
Table 3. Reactions of OH groups on the surface of anatase (Ti02 P 25),
specific surface area 56.1 m2/g [I091
Deuterium exchange with DzO after
outgassing at 100 OC
Sorption of F- from unbuffered N a F
Sorption of F- from NaF solution at
pH 4.65
F- content after washing with H2O
Sorption of Na+ from NaOH
Sorption of Ba*+ from Ba(OH)2
Sorption of A!3+ from basic AICI, solutions
Sorption of H2P04- from Na2HPO4 solution
Chlorination with SOCl2 vapor,
outgassing at 250 'C
Sorption of NH3 at 20 "C,
NH3 determined after
outgassing at 20 "C
Sorption of NO2 at 20 "C, N determined after
outgassing at 20 "C
Sorption of SO, from solution in C2H2C12,
SO, determined after outgassing at 200 "C
Methylation with diazomethane,
OCH, determined
Benzylation with phenyldiazomethane,
C determined
Reaction with the (CHa)zC'-CN radical,
N determined
Same as above, N determined after
extraction with ethanol.
Reaction with the (CsHs)N radical,
N determined
Reaction with the 2,4-dichlorophenyl radical
C! determined
(mmole/100 g
of Ti0,)
Fig. 6. Electrokinetic potontial < o f anatase (Ti02 P 25) as a function of
the pH, calculated from the electrophoretic mobility. Isoclzctric paint:
pH 6.6.
The hydroxylated surfaces of anatase and of rutile are
amphoteric [109,1151. Figure 6 shows the relationship
between the electrokinetic potential and p H for T i 0 2
P 25. The isoelectric point is a t p H 6.6. Consequently,
Ti02 can exchange hydrogen ions for cations in alkaline
solutions, and the OH groups can be neutralized with
alkali hydroxides [1091. Anions, and particularly fluoride,
sulfate, and phosphate ions, are bound in acidic media.
Industrial Ti02 nearly always contains phosphate, most
of which can be washed off with NaOH, and which must
therefore be attached t o the surface. The exchange of
fluoride proceeds t o completion even a t low equilibrium
concentrations (cf. Fig. 7). Unbuffered solutions are
weakly alkaline.
c [mole/ll-
Fig. 7. Adsorption isotherms on anatase (Ti02 P 25) at 23 "C.
and Ba(OH)2, respectively.
F- from N a F at pH 4.65.
PO:- from Na2HP04.
- - - - - - - Al3+ from A!(OH)zCI.
Ordinate: m = millimoles of bound ions/100 g of anatase.
Abscissa: c = equilibrium concentration (molejl).
~Na+ and Ba*+ from NaOH
[I101 I. J . Bear and R. G. Thomas, Nature (London) 201, 993
[ I l l ] C. Renz, Helv. chim. Acta 4, 961 (1921); A. E. Jacobsen,
Ind. Engng. Chem. 41, 523 (1949); J. Petit and R. Poisson, C.R.
hebd. Seances Acad. Sci. 240, 312 (1955).
[I121 W . A. Weyl and T. Forland, Ind. Engng. Chem. 42, 257
[I 131 A. A. Isirikyan, A. V . Kiselev, and E. V. Ushakova, Kolloidnyi Zh. 25, 125 (1963).
[ I l ? ] W. Srobrr, M. Lieflunder, and E. Bohn, Beitr. SilikoseForsch., Special Vol. 4, p. 1 1 1 (1960).
An interesting observation was made in the adsorption of Fions from NaF [1091: whereas finely divided Ti02 (anatase)
forms very stable suspensions in aqueous media, which
separate only slowly even in a high-speed centrifuge, rapid
flocculation occurs when fluoride is added. This is explained
by the replacement of OH- ions on the surface by F- ions,
so that dissociation to form H+ is no longer possible. The
surface charge of the Ti02 particles in neutral or weakly
alkaline media is then insufficient to stabilize the suspension.
[I 151 P. G. Johanstn and A. S. Buchnnan, Austral. J. Chem. 10,
392 (1957).
Arrgew. Chem. internnt.
Edit. , Vol. 5 (1966) i No. 6
The results of a number of reactions on an anatase
sample are summarized in Table 3. Either 16-20 or
about 40 milliequivalents of OH groups react per 100 g
of Ti02. in agreement with the presence of two infrared
absorptions (3715 and 3675 cm-1) 11011. 40 mequiv/lOOg
corresponds to a packing density of 4.3 OH/100 A2. The
average packing density estimated from the structure of
the various crystal surfaces is 14 OH/100 A?, and 7 OH/
1OOA2 in the outermost layer. As in Si02, fewer OH
groups are found than expected. From the water loss on
ignition of anatase, a value of 12pmole of H20/m2,
corresponding to 14.4 OH/100 A2, was obtained [1031.
When the sample was heated to high temperatures and
rehydroxylated, the content of chemisorbed water decreased to 4.1 pmole/rn2, corresponding to 4.9 OH/
100 A2.
Fig. 7 shows a number of adsorption isotherms for the
same sample. In the presence of the alkalis, half of the
hydroxyl groups are neutralized even at very low equilibrium concentrations (0.002 M). Further sorption involves less energy, as shown by the small slope of the
isotherm, perhaps because half of the OH groups are
much less acidic, or because additional OH- ions are
adsorbed on the surface. In the reaction with F- ions,
20mmole of OH- ions/lOOg again react at very low
equilibrium concentrations. The iwtherms rise much
more slowly in the sorption of phosphate ions (Na+
ions are adsorbed at the same time) and in the hydrolytic adsorption of Al3+.
The dye diniethylaminoazobenzene (butter yellow) is
not adsorbed from benzene solutions on anatase, whereas methyl red is adsorbed as the red cation. Butter
yellow (pK, = 3.3) changes color at a lower p H than
methyl red (pK, = 4.9). When anatase is exposed to
sunlight or ultraviolet radiation, however, butter yellow
is also adsorbed to give a red color (1 8 mmole of butter
yellow/100g of anatase P 25). Evidently the surface
becomes more strongly acidic on reduction, as is also
shown by the increase in catalytic activity in the elimination of water from tert.-butyl alcohol [1091. Partially
reduced Ti02 also catalyses the Fischer-Tropsch synthesis [1161. Magnetic susceptibility and electrical conductivity measurements [1161 show that the Ti3+ ions
formed on reduction are situated at the surface. It may
therefore be assumed that protons are added to equalize
the charge, and that these protons are responsible for the
more strongly acidic reaction. The acidity of the reduced
surfaces probably also contributes to the breakdown of
paint binders, particularly in the case of esters such as
alkyd resins.
6. Surface Compounds o n Alumina a n d o n
The surface of aluminum oxide is fundamentally similar
to that of TiO2. The number of crystalline modifications
of A1203 is large. Apart from x-AI2O3 (corundum),
there are at least eight other oxides, which are generally
[ I 161 Y. L. Sandler, J. physic. Chem. 58, 54 (1954); T.J. Gray,
C. C. McCain, and N . G . Masse, J. physic. Chem. 63, 472 (1959).
A n g e w . C h e m . internat. Edit.
/ Vol. 5 (1966) / No. 6
grouped together under the name y-A1203[1171. Some of
these forms are probably non-stoichiometric oxidehydroxides. Hydroxyl groups are present even in samples
that have been heated to high temperatures, and can be
detected by infrared spectroscopy [118,1191 and by proton
magnetic resonance measurements [1201. The presence of
hydroxyl groups on the surface was demonstrated by
deuterium exchange [1211 and by esterificaiion with alcohols [114,1221. The surface esters are very readily hydrolysed.
Interest in the surface chemistry of alumina and of the amorphous mixed oxides of silicon and aluminum is concentrated
mainly on the strongly acidic groups formed when the samples
are heated to 500-600 "C. Owing to these acidic properties,
the oxides are used on a large scale as catalysts in petroleum
chemistry. The question of whether the acidity is due to
readily dissociable protons (Br~rnstedacids) or to electron
acceptors in the surface (Lewis acids) is a subject of lively
A description of the many experimental arguments for and
against the wo views would lead beyond the scope of the present paper. A dehydrated alumina catalyst is readily poisoned
by water, whereas the catalytic activity of Si02-AI203 mixed
oxides is impaired, but not inhibited completely. The spectrum of pyridine adsorbed o n y-AI203 dehydrated at 450 "C
indicates that only Lewis acids are present "231. On the other
hand, both Lewis acids and Brmsted acids have been detected
in SiOz-Al203 mixed oxides [I239 1241. Accurate measurement
of the wavelength of the X-ray fluorescence of the aluminum
shows that up to relatively high A1203 contents, all the Al3
ions are tetrahedrally coordinated, i . P . they presumably
replace Si4+ in the three-dimensional SiO2 lattice [1251. Electroneutrality is probably achieved by protons incorporated in
the network. Chemical reactions have so far revealed only
silanol groups, and n o AI-OH groups c871.
Aromatic hydrocarbons such as perylene are oxidized to
radical cations on adsorption on the surface of such a
catalyst, as is shown by absorption spectra and electron
spin resonance measurements [126-1291. This oxidation
occurs even in the absence of oxygen"z'1. Triphenylamine [I301 and iodide anions [I311 are also oxidized. The
number of oxidation equivalents of a mixed-oxide
catalyst of this type is much lower than the number of
Lewis or Brarnsted acid equivalents. For this reason,
suggestions that these acidic centers are connected with
[117] See, e . g . , S. H . Ginsberg, W . Hiittig, and G. Strunk-Lirhter:berg, Z. anorg. allg. Chem. 293, 33, 204 (1957).
[I181 0. Glemser and C. Rieck, Angew. Chem. 63, 182 (1956);
Z . anorg. allg. Chem. 297, 175 (1958).
[I191 J . B. Per; and R . B. Hnnnnn, J. physic. Chem. 64, 1526
[I201 0. Glemser, Angew. Chem. 73, 785 (1961).
11211 J. K . Lee and S. W. Weller, Analytic. Chem. 30, 1057
(1958); W. K . Hall and F. E. Lutinski, J . Catalysis 2, 51 8 (1963).
[I221 M . M . Egorov. L . A . Ignntjevn, V . F. Kiselev, K . G. Krasilnikov, and K . V . Topchieva, Zh. fiz. Khim. 36, 1882 (1962).
[123] E. P . Parry, J . Catalysis 2, 371 (1963).
[I241 M . R . Basila, T. R . Kantner, and K. H . Rhee, J. physic.
Chem. 68, 3197 (1964).
11251 A . Leonard, S . Suzuki, J . J. Fripiat, and C . de Kinipe,
J. physic. Chem. 68, 2608 (1964).
[I261 J . J. Rooney and R . C. Pink, Trans. Fardday SOC.58, 1632
[I271 D. M . Brouwer, J. Catalysis I , 372 (1962).
[I281 A . E. Hirschler and J. 0. Hudson, J. Catalysis 3, 239 (1964).
[I291 R. L. Hodgson and J . H . Rnlej, J . Catalysis 4 , 6 (1965).
[I301 F. R . Dollish and W. K. Hall, physic. Chem. 69, 2127(1965).
[I311 S. D. M d l o r , J . J. Roomy, and P . B. Wells, J. Catalysis 4 ,
632 (1965).
the oxidizing power [m+1291 appear unlikely. The oxidizing power and the catalytic activity may be due to
defects in the structure, i.e. in the coordination of the
metal ions in the oxides 11321. Different types of reactions
are evidently catalysed by different active sites; thus the
addition of alkali, which suppresses the catalysis of one
reaction, often has relatively little influence on another [87,1331. Cracking processes may proceed via carboni11321 J . 8.Peri, J . physic. Chcm. 69, 220 (1965).
[I331 P . Stright and J. D . Donforth, J. physic. Chem. 57, 448
(1953); W . K . Hall, F. E . Lutinski, and H. R. Gerberich, J. Catalysis 3, 512 (1964).
um ions formed on acidic centers, as well as via free
radicals formed at oxidizing centers [134J. The nature of
the products depends on which of these routes is followed.
The author acknowledges valuable support from the Deutsche Forschungsgemeinschaft and the Fonds der Cheinischen Indusrrie for his work described in this paper,
Received: May 25th, 1965
[A 517 IE1
German version: Angew. Chem. 7S, 617 (1966)
Translated by Express Translation Service, London
[I341 S. E. Tung and E. Mcirrinclr, J. Catalysis 4 , 586 (1965).
The Structure of Glasses
A melt can readily solidifv to a glass with a three-dimensionalor two-dimensional network,
or a chain structure, provided that an irregular bonding system can be formed by virtue
of free rotation about the bonds between a central atom and the ligands which function as
bridging atoms. Such an irregular structure can arise when the system contains a sufficient
amount of bridging atoms such as 0, F, and S, or bridging groups such as CH2, with bond
angles less than 180 O. When the network is formed predominantly by trivalent and tetravalent elements, such as As and Ge, the glasses though they cannot be prepared by cooling
of melts - can be obtained by other processes, e.g. by condensing the vaporized substances
onto a surface (glasses in the wider sense of the word). As a result of extensive network
formation, the bonding system and, therefore, the short-range order of the atomic arrangement in the melt difSer from those in the glass or the crystal. A liquid mixture of substances
having unlike molecular size and shape also can form a glass on solidiJication. Moreover,
glasses can be formed even when the system contains only one component opposing regular
packing into a crystal lattice.
I. Introduction
In 1945, the American Society for Testing Materials
(ASTM) defined glass as “an inorganic product of
fusion which has cooled to a rigid condition without
crystallizing”. This definition is unsatisfactory, since it
excludes transparent plastics and glasses prepared by a
process other than fusion. Thus, vitreous selenium can
also be obtained by the condensation of selenium vapor
on a slightly warm surface. Furthermore, p- or y-arsenic
exhibiting a conchoidal fracture typical of glasses, and
amorphous to X-rays like vitreous selenium, can be
prepared only by condensation of arsenic vapor on a
surface heated to about 100 “C121, since it crystallizes at
270 “C (far below the softening point). Using the term
in the wider sense, we can, therefore, define glass as a
solid substance characterized by dense packing of the
atoms and a lack of long-range order in the network.
According to this definition, amorphous Al(OH)3 and
Cr(OH)3,for example, are not regarded as glasses; they
[ I ] J . D . Mackenzie: Modern Aspects o f the Vitrous State. Butterworth, London. a) Vol. I, 1960; b) Vol. 11, 1962; c) Vol. 111,
[2] H.Stohr, Z. anorg. allg. Chem. 242, 138 (1939).
are loose powders containing empty spaces ranging in
size down to atomic dimensions.
Tamrnann[31was the first to carry out systematic investigations concerning the vitreous state by means of
modern scientific techniques. Goldschmidt [41 correlated
structural parameters with glass formation, and his
pupil, Zachariasen [51, developed these concepts further
and expressed them in the simple form which has since
become well known.
11. Causes of Glass Formation
1. Glasses Having a Three-Dimensional Network
Zachariasen has attributed glass formation to the
possibility of the formation of a three-dimensional
network differing from a crystalline lattice by the lack
[3] G . Tummann: Aggregatzustiinde. Voss, Leipzig I923 (transl.
by R. F. Mehl: The States o f Aggregation. Van Nostrand, New
York 1925); Der Glaszustand. Voss, Leipzig 1933.
[4] V. M . Goldschmidt Geochemische Verteilungsgesetze der
Elemente, VIII. Vid. Akad. Ser. Oslo, N o . 8, 137 (1926).
[5] W. H. Zachariasen, J. Amer. chem. SOC. 54, 3841 (1932).
Chem. internnr. Edit.
Vol. 5 (1966) / No. 6
Без категории
Размер файла
1 414 Кб
solids, group, surface, function
Пожаловаться на содержимое документа