In the presence of oxygen at 800-900 âC under reduced pressure, diamond is transformed into black carbon. Surface oxides play a par[ in this transjbrmation.
Surface Oxides of Carbon BY DOZ. DR. H.-P. BOEHM, DR. E. DIEHL, D1PL.-CHEM. W. HECK, AND DIPL.-CHEM. R. SAPPOK ANORGANISCH-CHEMISCHES INSTITUT DER UNIVERSITAT HEIDELBERG (GERMANY)
On oxidation of graphitic carbon, acidic surface oxides (or hydroxides in the presence of water) are formed at the boundaries of the carbon layers. It was found that the reaction of oxygen with microcrystalline carbon at 400-450 “C yields four groups of diflerent acidities. One strongly acidic group and one weakly acidic carboxyl group have been detected, as well us a phenolic hydroxyl group; a carbonyl group is probably also present. With dissolved oxidizing agents, one equiialent of another strongly acidic carboxyl group is formed in addition at room temperature. The possible constitutions o f the acidic surface oxides are discussed. Chemically detectable surface oxides are also formed on the surfarc. of diamond. In the presence of oxygen at 800-900 “C under reduced pressure, diamond is transformed into black carbon. Surface oxides play a par[ in this transjbrmation.
I. Introduction
the absence of oxygen, its surface becomes covered with basic surface oxides on access ol oxygen [5]. The constitution of these oxides is not yet known in detail.
The crystalline forms of elemental carbon are diamond and graphite. All the carbon aroms within the threedimensional diamond lattice or the graphite layer structure are joined by covalent bonds. The free valences at the surface can be saturated by oxygen, for example, surface compounds being formed. Surface oxides affect the properties of carbon surfaces, e.g. their wettability, the adsorptive power of active charcoal, or the dispersibility of ink blacks. The carbon blacks used as reinforcing fillers in rubber also invariably contain surface oxides [ *]. Moreover, surface oxides of carbon play an important part as oxygentransfering agents in atmospheric oxygen electrodes. The acidity and redox behavior of surface oxides can be used for the removal of salt from sea-water [I]. The surface oxides of microcrystalline black carbon have often been described. Active charcoal reacts with oxygen at temperatures as low as 300°C, and part of the oxygen consumed remains bound to the surface, being liberated as CO and COz only on subsequent heating [2,3]. Surface oxides of carbon give a basic or acidic reaction, according to the conditions under which they are prepared [4]. If black elemental carbon is freed from all surface compounds by heating in vacuo followed by cooling in ~~
[*] The surface oxides do not appear to be directly responsible for the reinforcing effect, since other finely divided substances, e . g . Aerosil@(silica which is morphologically very similar to the carbon blacks) have a similar action. However, many of the properties of carbon blacks are affected by surface oxides; for example, channel blacks in aqueous suspension give an acidic reaction, furnace blacks and thermal blacks give a basic reaction. How surface oxides affect the vulcanization and properties of the rubber has not yet been entirely explained. [I] J . W. Blair and G. W. Murphy, Advances Chem. Ser. 27, 206 (1960); B. B. Arnoldand G. W. Murphy, J. physic. Chem. 65, 135 (196 I ). !2] 7: F. E. Read and R. V. Wheeler, J. chem. SOC.(London) 13, 461 (1913). [3] B. R. Puri, D . D. Singh, J . Notli, and L. R. Sharnia, Ind. Engng. Chem. 50, 1071 (1958). [4] H . R. Kruyt and C. S. de Kadt, Kolloid-Z. 47,44 (1929).
Angew. Chem. inlernat. Edit.
Vol. 3 (1964)
No. 10
A c i d i c surface oxides are formed on oxidation with oxygen at high temperatures or with dissolved oxidizing agents, e.g. NaOCl; these have been studied in much greater detail [6,7].The present article is a contribution to the knowledge of acidic surface oxides.
11. Starting Material
Microcrystalline carbon is more suitable than graphite for investigations of the surface oxides of graphitic carbon, since its specific surface area is much larger. Hence it can bind correspondingly larger quantities of oxygen. Black microcrystalline carbon, formerly referred to as “amorphous” carbon, is well known, e . g . in the form of active charcoal, carbon black, and carbon electrodes, and consists of extremely small crystallites built up, like graphite, of layers of six-membered carbon rings. These layers are stacked together in piles, but do not possess the mutual three-dimensional orientation observed in the graphite lattice. The distance between layers is larger than in graphite (3.44-3.65 A and 3.354 A, respectively). Hofmann et al. [8,9] found that typical active charcoal contains layers with a diameter of ca. 30 A ; the average height of the stacks is 10-13 A. In our work we used mainly carbons prepared from s u g a r c h a r c o a l [lo], since this can be obtained in very [5] R. Burstein and A . Frumkin, Z. physik. Chem. A 241, 219 (1929); A. Frumkin, Kolloid-Z. 51, 123 (1930). [6] U . Hofrnann and G . Ohlerich, Angew. Chem. 62, 16 (1950). [7] B. R. Puri, 0. P. Muhajan, and D. D.Singh, J . Indian chem. SOC.38, 135 (1961); B. R. Pwi: Proceedings 5th Conference on Carbon (Pennsylvania State University, 1961). Pergamon Press, London 1962, Vo!. I, p. 165. [8] U . Hofinann and I ) . Wilm, 2. Elektrochem. angew. physik. Chcm. 42, 504 (1936). “)I K . Binsroch and U . Hofinann, Angcw. Chem. 53, 327 (1940). [lo] U . Hofr,iann andF.Sinke!,Z.a~lorg.alIg.Chem.245,85(1940).
669
pure form. To make the pores as large as possible, thus ensuring easy accessibility to the surface, the sugar charcoals were “activated” by oxidation with C02 2t 950 “C. In this way, individual crystallites are burned out of the structure. The specific surface area reaches a maximum when the loss by burning is about 50 %; it is only slightly affected by further oxidation [S, 111. The products were partly coked, i. e. heated at 1100 “C in pure nitrogen. This treatment eliminates the greater part of the oxygen and hydrogen which they contain. The specific surface areas of activated and coked sugar charcoals were found by Brunauer, Emmett, and Teller [12] by nitrogen adsorption to be 900-1100 rnz/g. Technical active charcoal and a few typical carbon blacks were also investigated. The products were c o a t e d with s u r f a c e o x i d e s either by exposing them to air during the cooling stage following activation or by oxidation in oxygen for 5-20 hours at 400-450 “C [13]. Cooling of the activated sugar charcoal in pure COz gave only basic surface oxides. In the “wet oxidation” method, the products were subjected to prolonged stirring with solutions of NaOC1, KMn04, or (NH&Sz08, followed by thorough washing. The MnO2 which was formed in the oxidation with KMn04 was dissolved out by boiling with hydrochloric acid. Crude sugar charcoal heated to a maximum of 450°C was often also studied for comparison, although it is not pure carbon. However, its X-ray diRraction diagram still shows the characteristic reflexes of microcrystalline carbon.
111. Chemical Reactions
1. Neutralization of the Acidic Groups
Direct potentiometric titration of the acidic surface oxides is difficult, since the end value of the potential is often attained only after long periods. We therefore determined the acidity by neutralization with bases of v a r i o u s basicities. The most suitable of these were NaHC03, Na2C03, NaOH, and NaOCzHs. In 0.05 to 0.1 N solution, each of these bases gave characteristic maximum consumptions for neutralization. With higher concentrations of base, only sodium ethoxide gave a slight increase in consumption. Thus, in each case. only those acidic groups which were much more strongly dissociated than the conjugate acids of the bases were completely neutralized. The p K values of the conjugate acids are: 6.37 and 10.25 for the first and second dissociations of carbonic acid, 15.74 for water, and 20.58 for ethanol.
Surprisingly, the quantities of bases neutralized generally bear s i m p l e , w h o l e - n u m b e r r e l a t i o n s h i p s to one another (cf. Table 1). In principle, all types of carbon [ l l ] P. L. Walker, R. J. Forest;, and C. C. Wrig-ht, Ind. Engng. Chem. 45, 1703 (1955). [I21 S. Brunauer, P. H. Emmett, and €.Teller, J. Amer. chem. SOC.60, 309 (1938). [I 31 H. P.Boehm and E. Diehl, Z. Elektrochem., Ber. Bunsengees. physik. Chem. 66, 642 (1962).
670
Table I. Neutralization of acidic surface-oxidcs. The samples were oxidized with 0 2 at 400-450 “C.
Product [*]
I
Oxygen
Consumption rmeq.1 I00 g1
[ % I [**I
NaHCO,
NaOCzHs
42 I1 35 27 15 31 14 57 59
I65 a9 137
9.2 SC 3 coked ox. 7.3 SC 1, act. ox. SC I , act. coked ox. 10.7 18.8 SC I , crude Eponite, ox. C K 3, ox. 5.? CK 3, coked ox. Philblack 0, ox. Spheron 6, ox.
72
so 34 60 27 109 118
102 81 62 89 39
-
106 -
70 233 295
164
I96
1’1 SC: sugar charcoal. The number indicates a specific sample. For brevity, only one example is given for each type in this and in the following tables. Complete tables in: E. Diehr, Ph.D. Thesis, Technische Hochschule Darmstadt, 1964, and in [261. Eponite: active charcoal, Eponit BEN” of Degussa, Hanau (Germany). CK 3: carbon black C K 3@ of Degussa, Hanau (Germany). Philblack 0: carbon black Philblack O@ of Phillips Petroleum Comp.. U.S.A. Spheron 6: Spheron 6“ carbon black of Cabot Corp., U.S.A. act. = activated by partial oxidation with COz at 950 “C. coked = coked by heating at 1100 “C in pure nitrogen.
ox. = coated with surface oxides by treatment with oxygen at 400 to 450 “ C . [
* 1 By difference.
show the same behavior, but the simple numerical relationships are found only with completely oxidized products. Twice as many acidic groups are neutralized by Na2CO3 as by NaHC03. The differences between the quantities of NaOH and Na2C03 used, and between the quantities of NaOC2H5 and NaOH used, are approximately equal to the quantity of NaHCO3 used. If the amounts required for neutralization of various samples did not quite agree, the groups neutralized with NaHCO3 and those neutralized with NaOH alone, as well as the groups neutralized with Na2C03 and those neutralized with NaOC2H5, usually occurred none the less in equivalent quantities. Activated or coked products which had not been oxidized neutralized practically no sodium hydroxide. Carbon blacks required more ethoxide than was expected. Carbon black contains appreciable quantities of disordered, tetrahedrally bonded carbon [14], which readily forms alcoholic or aldehydic groups which react with sodium ethoxide.
The agreement between these whole-number ratios of the quantities of base required for neutralization do not appear to be fortuitous, but rather indicates the presence of e q u i v a l e n t quantities of (at least) four d i f f e r e n t types of acidic groups of definite acidities. Thus, all four functions are probably constituents of a larger grouping. We shall refer to the groups as I to IV in order of increasing acidity, e.g. group I1 is neutralized only by Na2CO3 (or stronger bases), but not by NaHC03. The simple numerical ratios observed in the neutralization experiments were also found in many other reactions. Thus, for example, the various carbon preparations were treated with a solution of KI and K103. This solution reacts with H’ ions with liberation of equivalent quantities of iodine: the pH of the solution remains at 7.5. The quantity of iodine -
1141 U. Hofmann and E. Groll, Ber. dtsch. chem. Ges. 65, 1257 (1932); L. E. Alexander and E. C. Sommer, J. physic. Chem. 60, 1646 (1956); C. W. Snow, D. R. Wallace, L. L. Lyon, and G . R .
Crockerc Proceedings 3rd Conference on Carbon (Bufallo, N.Y., 1957). Pergamon Press, London 1958, p. 279.
Angew. Chem. infernat. Edit.
/
Vul. 3 (1964)
/ Nu.
I0
liberated by the acidic surface oxides was always equivalent to the quantity of N a H C 0 3 required for neutralization.
The ratios of the neutralization values after “wet” oxidation at low temperatures were somewhat different (Table 2). Here two groups which could be neutralized with NaHC03 were formed for every one group which could be neutralized with Na2C03, NaOH, or NaOCZH5. Table 2. Neutralization of the acidic surface oxides on wet-oxidized products. Consumption [meq./100 g]
Product
NaHCO,
SC 12, act. coked, ox. KMn04 Eponite, coked, ox. KMnOl Eponite, coked, ox. NaOCl Eponite, coked, ox. (NH&S208 CK 3, coked, ox. (NH)&08
61 79 107 145 21
1
’
Na2C03
89 1 I6 163 203 30
1 NaOH 1 125 160 214 269 36
NaOCzHs lp7
220 262 339 57
Thus, the quantities of base required for neutralization are in ratios of 2:3:4:5. On subsequent heating, one of the two more strongly acidic groups is destroyed with liberation of the equivalent quantity of C02; this process starts at about 200 “C. The titrant ratios are then again found to be 1:2:3:4. The other groups are successively broken down at somewhat higher temperatures. Again, the simple whole-number ratios were observed only after repeated, complete oxidation. The oxidation was complete as soon as small quantities OF a brown colloid separated from the solid phase in alkaline media. Group IV and one of the two groups I are formed first in the oxidation, and group I11 and the second of the more strongly acidic groups I last (with simultaneous increase in the quantities of the other groups).
the bases chosen. All the acidic groups which could be neutralized with NaOH, i.e. groups 1-111, were methylated in dry ether. On the other hand, the methoxy values found after methylation in aqueous ether corresponded to the quantity of ethoxide required for neutralization. As expected, the methoxy values obtained with crude sugar charcoals in aqueous ether corresponded to the sum of the NaOH and NaHCO3 consumptions. The methoxyl groups which remained after hydrolysis with hot dilute hydrochloric acid corresponded, with only one exception, to group 111, i. e. to the group which is neutralized by NaOH but not with NazCO3. This behavior implies the presence of phenols. All the hydrolysable methoxy groups in the products which had been methylated in dry ether must have been esters of carboxylic acids or groups of similar acidity. The extra methoxy groups formed when moist ether is used in place of dry ether were not stable to hot dilute hydrochloric acid, and are not, therefore, methyl ethers of an alcohol. Diazomethane generally forms readily hydrolysable methyl esters with carboxylic acids, but non-hydrolysable phenolic ethers with phenols. Diazomethane reacts with alcohols only in the presence of catalysts such as BF3, ZnCl2 [17], or H20 [18]: lactones of the fluorescein type are rearranged to their quinonoid form with methylation of the carboxyl groups [19].
After hydrolysis, the neutralization behavior towards NaHCO3 and Na2C03 remained unchanged, but the NaOH consumption fell to the Na2C03 value (Table 3). This behavior was the same, regardless of whether the methylation had ‘been carried out in dry or aqueous ether. Table 3. Methylation of acidic surface oxides with diazomethane. Methoxy groups [meq./100 gl [*I
Product
SC 3, coked, ox. SC 1 act. ox. Sc 1 , act. coked ox. SC 1, crude Eponite, ox. CK 3, ox.
I
1
Consumption of base [meq./100 g ] after hydrolysis
Total
nonhydrolyshydrolysable NaHC03 able
124 72 (92) (140) 82 (109) (107) 89
43 30 (30) (41) 32 (30) (61) 33
81 42 (62) (99) 50 (79) ( 146) 56
Nap203
NaOH
84 41
40 21 36 27
-
49
81 42 69 54
-
-
-
32
57
55
[*I The values in brackets are for solutions in ether saturated with water,
2. Methylation of the Acidic Groups
a) With Diazomethane The reaction with diazomethane has often been used f o r assaying acidic surface oxides [6,15,16]. Some of the resulting methoxyl groups can be readily hydrolysed with hot dilute hydrochloric acid. The methylation was carried out using diazomethane, dissolved in some cases in dry ether and in others in ether saturated with water. As may be seen from comparison of Tables 1 and 3, the individual neutralization values are not accidentally governed by the strengths of
the others for solutions in pure ether.
Dimethyl sulfate gave the same results as dry diazomethane solutions when dilute sodium hydroxide was added slowly, so that the solution always remained slightly alkaline. If the methylation was carried out i n the usual fashion in strongly alkaline medium, the methoxy content corresponded to the group 111 content.
b) With Methanol A few products were esterified with methanol plus hydrogen chloride. The quantities of methyl esters formed were equivalent to the group I or group I1 contents (Table 4).
~~
[I51 M . L. Studebnker, K. W . D . Huffrnann, A. C. Wulfr,and L . G. Nnbors, Ind. Engng. Chem. 48, 162 (1956). [161 V. A . Garten, D . E. Weiss, and J . R. Willis, Austr. J. Chem. 10, 295 (1957). Aitgew.
Chem. internat. Edit. / Vol. 3 (1964) No. 10
1171 E. Miller and W. Rundel, Angew. Chem. 70, 105 (1958). 1181 H. Biltz, Ber. dtsch. chem. Ges. 55, 1069 (1922). 1191 0.Fischer and E. H e p p , Ber. dtsch. chem. Ges. 46, 1951 (I 9 I 3).
67 1
Table 4. Esterification of acidic surface orides with methanol and hydrogen chloride.
1
Content [iiieq./100 g1
Product
Gronp I [hl
SC 3, coked, ox. S C 1, act. ox. SC I , act. coked. ox. C K 3, ox. Eponite, coked, ox. NaOCl Eponite, coked, ox. K M 0 4 [a]
42 21 41 31 107 14
1 1
Group I I [cl
Very weakly acidic phenolic hydroxy groups, such as that i n salicylic acid (pK 13.4), also give n o reaction, but can be methylated with dimethyl sulfate [20].
The close agreement of the phenolic hydroxy contents determined by four characteristic phenol reactions can be seen from Table 5 .
Methoxyl
40 22 53 29 56 15
4. Detection of Carboxyl Groups
[a] Incompletely oxidized. [ b ] The group-I content was obtained by neutralization with NaHCO3.
[ c ] T h e group-I1 content was determined from the difference between the N a 2 C 0 3 and N a H C O J consumptions.
The facts that the more strongly acidic groups can be neutralized by NaHC03 and that they form hydrolysable methoxy groups suggest that these are carboxyl groups. This assumption was confirmed by other reactions.
[d] These products had been esterified by passing a mixture of methanol vapor and nitrogen over the material at 12OOC.
Which of the two groups had been esterified was determined by the neutralization behavior of a n esterified product (SC 3, coked, ox.): the NaHCO3 and Na2C03 consumptions were equal (44 and 40 meq./100 g, respectitely); thus: group I1 had been esterified. At the same time, the sodium pthoxide comumption fell to 82 meq./100 g. The esterification of group 11 apparently also led to inhibition of the reactivity of group IV towards ethoxide, so that only groups IT and 111 remained neutralizable.
a) Reuction with Thionyl C h h i d e In the reaction with thionyl chloride, part of the chlorine is bound to the carbon in a readily hydrolysable form, whereas the remainder is stable even to hot sodium hydroxide solution [ 131. The readily hydrolysable chloride must be present as the chloride of a carboxylic acid. Hydrolysis occurs only very slowly with pure water, and more rapidly with alkali, especially on warming. This behavior corresponds to that of aromatic acyl chlorides. Table 6. Reaction of acidic surface oxides with thionyl chloride.
1 Content [meq./100 g ] I
1
Product
Group11
1
Chloride
I
15 SC 1, coked, ox. 46 SC 3, act. ox. 28 SC 1, act. coked, ox. 35 SC 3 , act. ox. red. LiAIH4 SC 2, crude (degassed at 100 “ C ) 61 9 SC 2, crude (degassed a t 3 3 0 “C) Eponite, coked, ox. ( N H ~ S Z O S 30
3. Detection of Phenolic Hydroxyl Groups
The quantitath e agreement between the amount of nonhydrolysable inethoxy groups and the difference between the NaOH and Na2CO3 titers indicates the xesence of phenols. This assumption was confirmea by reaction with two typical phenol reagents, 2,4-dinitrofluorobenzene and 11-nitrobenzoyl chloride. Dinitrofluorobenzene does not react with carboxylic dcids, and reacts only extremely slowly or not at all with alcohols. Table 5 . Detection of phenolic hydroxy groups in acidic surface oxides.
Content [meq./100 gl
Conversion [meq./100gl with
42 29 39 31
42 30 37 28 l3
Product benzene [b], chloride S C 3 , coked, ox. SC 1, act. ox. S C 4, act. coked, ox. SC I , crude C K 3, ox.
I 32
43 30 -
32 33
I
[a1 T h e group-I11 content was found from the difference between the N a O H and NazCO, titers. [bl Reaction accordins t o [21], followed by nitrogen determination.
672
I Change in consumption of base [meq./100 g1
1
NaHCO3 -
16 48 36 35 61 11
+- 47 + 35 +- 70
31
f- 30
1
NazCOl
+ 20 +- 45 1 35
+ 61 + 32
1
NaOH
1
+ 18 f 41 + 33 + 35 +-62
NaOC2Hj 0
- 34 - 31 -
+ 30
t S
The chloride titrated was equivalent to the quantity expected for group I1 (Table 6). This is clearly shown by the products “SO 3, act. ox.”, “SC 3, act. ox., red. LiAIH4”, and a partially thermally decomposed sugar charcoal “SC 2”; most other preparations contained equal quantities of group 1 a n d It acids. The products M hich had been treated with thionyl chloride were titrated with the four bases. Owing to neutralization of the hydrogen chloride formed, t h e titer corresponded to t h e sum of the original neutralization figure and the hydrolysable acyl chloride. However, the products described as “activated” required Iess ethoxide than originally (Table 6 ) . ‘The decrease corresponded t o half of the N a H C 0 3 consumption of the products. The low NaOC2HS consumption can be explained by an
assumption which was later confirmed experimentally. The more strongly acidic carboxyl g r o u p s occur in pairs on directly adjacent sites and react with thionyl chloride to form a carboxylic anhydride. Acyl chlorides and anhydrides each consume only 1 mole of sodium ethoxide: -CO-CI -CO-0-CO-
+
N a O C z H j + --CO-OCzHj NaOCzHs + -CO-OCIHj
-1- NaCI,
t- -COONa.
Thus, acyl chlorides consume the same quantity of ethoxide as the free carboxylic acids, whereas anhydrides consume only half as much. The anhydrides bind two moles of NaOH. [20] F. Sachs and V. HeroM, Ber.d(sch.chem. G e s . 40,2714(1907). [21] H. Zahn and A . Wiirz, Z. analyt. Chem. I 3 4 , 183 (1951). .41igew. Cliem. iriternnt. Edit. { Vol. 3 (1964)
Nv. 10
Nothing is yet known regarding the bonding of the hydrolysis-resistant chlorine which is present in a quantity corresponding to twice thc N a H C 0 3 titer.
The carboxyl groups were detected directly by Sclzr.oc~t w ’ s variant [22] of the Schmidt degradation. After
b) Reaction with Arnnioniu The p K values of two adjacent carboxyl groups generally differ by 2-3 units. Consequently, the ammonium salt of the more weakly acidic carboxyl groups has a strong tendency tou ards hydrolysis, and decomposes readily o n heating. The behavior of the carbon preparations towards ammonia can give information as to the position of their more strongly acidic groups.
“Coked” and “crude” sugar charcoals which had been treated with 0.05 N ammonia and degassed at 20 “C contain a quantity of ammonia equivalent to the NaHC03 consumption. The same results were obtained when the products were allowed to react with gaseous or concentrated aqueous ammonia and degassed at 100 “C. In the products which had formed anhydrides on treatment with thionyl chloride, half of the group-1 ammonium salts were decomposed. Thus, the more strongly acidic carboxyl groups (I) in the activated sugar charcoals occur in pairs occupying adjacent sites, so that anhydrides or imides can be formed. When products which had been treated with concentrated aqueous or with gaseous ammonia were degassed at only 20”C, more NH3 was retained, but the quantity again corresponded t o the NaHC03 or Na2C03 consumption. The ammonia contents of the products in these experiments were determined by expelling the NH3 with hot 2 N NaOH. Since amides or imides both decompose under these conditions, the true ammonium salts were again determined with the weak base Mg(0H)Z. The consumption of 0.05 N HCI (taking into account the basic surface oxides already present) and the quantity of NH4+ ions formed were also measured. The quantity of ammonium ions almost always agreed with the NH3 content determined by expulsion with NaOH. Thus the carboxyl groups I had not formed amides or imides even at I00”C.
c) Friedel-Crafts Reactions Several products, e . g . “Eponite coked, ox. (NH&S208’’, were converted into acyl chlorides by treatment with thionyl chloride (cf. Table 6), and were then treated with a solution of dimethylaniline and AIC13 in nitrobenzene. The nitrogen content of the reaction product corresponded to the chloride content, i.e. half of the bicarbonate consumption, and was therefore equivalent to group 1 or group 11. The NaHC03 consumption decreased by half, and the neutralization values for the four bases were in the ratios of 1:2:3:4. Thus, in this wet-oxidized product, one of the two groups I (neutrdizable with NaHC03) had formed a carbonyl chloride on reaction with thionyl chloride, as shown by the smooth progress of its subsequent Friedel-Crafts reaction. The other group I was thermally stable at 200 “C. Group I also reacted in “coked” sugar charcoals which had been oxidized with oxygen; neutralization after the FriedelCrafts reaction gave ratios of the acidic groups of 0: I : 2: 3 . With an “activated’ sugar charcoal which had been oxidized with oxygen, both the anhydride formed by reaction with SOC12 and the acyl chloride of group I1 reacted under the same conditions. Alzgew. Chem. internat. Edit.
Vol. 3 (1964) 1 No. 10
treatment with thionyl chloride, the product “Eponite coked, ox. (NH&S20s” was treated with sodium azide in ethanol; the product contained a quantity of nitrogen corresponding to the replacement of one carboxyl group by a urethane group. The neutralization values decreased by half of the NaHC03 neutralization value.
5. Detection of Carbonyl Groups
The acidic groups detected so far account lor only part of the oxygen content of the surface oxides. In the literature, the surface oxides are often formulated as carbonyl groups bonded to the periphery of the carbon layers. The consumption of ethoxide in the neutralization experiments may be due to carbonyl groups which react in the presence of a sufficient excess of alkali to form the salt of a hemiacetal. Secondary reactions, such as aldol condensations or Cannizzaro reactions, which take place readily under these conditions, cannot occur with carbonyl groups fixed at the peripheries of carbon layers. This assumption is plausible, since many aldehydes with adjacent negative substituents (hydroxy groups) contain appreciable quantities of acetal in aqueous solution [23]. In the presence of sodium ethoxide, the intermediate (hemiacetal) is fixed as the salt. The consumptions of characteristic carbonyl reagents such as hydrazine, semicarbazide, or hydrogen cyanide were in the region of the quantity of NaHC03 or Na2C03 required for neutralization. However, the consumptions of the various reagents were unequal. After reaction with HCN, followed by hydrolysi?, the NaHC03 consumption of a sample of activated and oxidized sugar charcoal increased to twice the initial value; thus an equivalent quantity of new carboxyl groups had been formed. Almost all the carbonyl reagents used are also strong reducing agents, and furthermore, they can form salt-like compounds. Unambiguous interpretation of the results I S therefore difficult.
6. Acetylation Several products were treated with acetyl chloride. The acetyl contents of the purified products always corresponded to the quantity of NaOH required for neutralization. Pairs of adjacent carboxyl groups react with acetyl chloride to give cyclic anhydrides. Other carboxylic acids, phenols, and even alcohols bind acetyl groups. Thus, the subsequent reaction with sodium ethoxide is not a simple neutralization of an acidic hydroxyl group, since otherwise the acetyl content (of the “coked” products at least) should correspond to the NaOCzHs coiisum;%ion. One product (SC I , act. ox.) \vas boiled under reflux with acetic anhydride and a little sodium acetate. After the usual [22] G . Schroeter, Ber. dtsch. chem. Ges. 41, 3356 (1909). [ 2 3 ] N . C . Melchior, J. Amer. chem. SOC. 71, 3651 (1949).
673
purification, 182 meq. of NaOH/100 g of charcoal were required for neutralization. (The acetyl groups were not determined directly). If, as in the reaction with acetyl chloride, all the groups which could be neutralized with NaOH (70 meq./100 g) had been acelylated, the NaOH consumption should have doubled i n alkaline hydrolysis. However, an additional 42 meq. of acetyl groups per 100 g of charcoal had apparently been bound (the formation of new acidic groups during acetylation is unlikely). These additional acetyl groups presumably did not react with hydroxyl groups, since acetyl chloride introduced 70 meq. of acetyl groups/100 g.
reviewed briefly below. In comparing the work of difl‘erent authors, it should be noted that the products used were rarely even similar.
Oxidation of single crystals of graphite with absolutely dry oxygen leads to an “arm chair” (1 12 1) boundary (1) of the carbon layers, while a “zig-zag” (IOT 1) boundary ( 2 ) is formed in the presence of water [25].
It is possible that a geminal diacetate -CH(OAc)z had been formed from an aldehyde [24]. We presume that the group 1V (20 niey./100 g in the strating material) is an aldehyde or a ketone (see below).
7.Reductions The acidic surface oxides behaved differently toward reducing agents (Table 7). On Clernrnensen reduction with amalgamated zinc and hydrochloric acid, all four Table 7. Change in the quantity of acidic groups after reduction of the acidic surface oxides.
Reduced with
Froduct
Change in groups [bl
I SC 3, coked, ox. [a1
Zn/HCl(atZO”C)
SC 3. act. ox.
-
Zn/HCI (at 20 “C) Zn/HCI (at 1OO’C) LiALH4 NaBH4
1
11
1 111
IV
(1) -1
(20)
(61) -16 -I0 0
(47) -13 -18
0
(21) -19 (46) 0 -15 -11
0
-21
-3
-1 1
(20) +I8 (60) 0 -11
-19
-I4
[a] This product was obtained by rapid quenchins of oxidized sugar charcoal. Practically no groups 111 were formed during this treatment.
[hl The values in brackets give the content [meq./100 gl in the starting material; the figures without brackets are the changes after reduction.
groups are usually, and group IV is always, attacked. The contents of groups I1 and 111 frequently decrease in the same proportion; the two groups are probably closely related. In two products (“SC 3, coked, ox.” reduced by the method of Clemmensen, and “SC 1, act. ox.” reduced with LiAlH4), the decrease in group I1 content was accompanied by an equivalent increase in group 1. Thus, group I1 had become more strongly acidic as a result of the reduction. Group I1 is probably, like group I, a carboxyl group. However, in other cases, e.g. in “SC 3, act. ox.”, group I1 was reduced with LiAlH4 or NaBH4 without formation of a more strongly acidic carboxyl group. This suggests that group I1 is normally present as a lactone. This result agrees with a structure similar to that of fluorescein [16], but other structures are possible also (see below).
IV. Possible Structures of Carbon Surface Oxides
Our starting materials always contained hydrogen, and consequently, oxidation probably gave form (2). It is not known which boundary is formed on oxidation with C02, i. e. on activation. We observed differences in the behavior of the surface oxides on coked and on activated sugar charcoals. It seems possible that the original boundary is largely retained on coating with surface oxides. The acidic surface oxides are bound at the edges of the carbon layers. Carbon black graphitized at 2700 “C neutralizes no sodium hydroxide after oxidation at 450 “C with oxygen [26]. The surface of graphitized carbon black consists only of the basal planes of graphite [27] and contains only bery few edge carbon atoms which can form acidic ( e . g . phenolic) groups. We detected f o u r acidic groups of various strengths, which were generally present in equivalent quantities : a more strongly acidic carboxyl group (I), a more weakly acidic carboxyl group (II), a phenolic hydroxyl group (IIi), and a fourth group (IV) which is probably a carbonyl group. Groups I1 and IV appear to be closely related. One explanation for this would be that the more weakly acidic carboxyl group I1 is present as a hydroxy lactone which resembles phenolphthalein or fluorescein. A plasuible arrangement of lactone and hydroxy groups is shown in formula (3a). This would afford a satisfactory expIanation of the lower acidity of the carboxyl group 11, since phenolphthalein is known to form its colored salt, analogous to the form (Sb), with sodium carbonate, but not with sodium bicarbonate. This as-
m
0
HO H 0 - C O
N ~ ~ C ONaOH ,,
(30)
H
COO-Na’
-0% (36)
.~
Tho data available from the hterature and from our own investigations are insufficient to afford complete and unambiguous information about the constitution of the acidic surface oxides. However, the most important results are ~
~-
[24] E. Knoevenagel, Liebigs Ann. Chem. 402, 127 (1914).
674
[ 2 5 ] G . R . Hennig, Proceedings 5th Conference on Carbon (Pennsilvania State University, 1961). Pergamon Press, London 1962, Vol. I, p. 143; Z. Elektrochem., Ber. Bunsenges. physik. Chem. 66, 629 (1962). [26] H . P . Boehm, E. Diehl, and W. Heck, Rev. gen. Caoutchouc 41, 461 (1964). [27] H . P. Boehm, 2. anorg. allg. Chem. 297, 315 (1958). Angew. Chem. internat.
Edit.1 Vol. 3 (1964) 1 No. 10
sumption is confirmed by an absorption band in the infrared spectrum of the strongly oxidized color black “Carbolac which was attributed to the carbonyl group of a?{-lactone [16]. @”,
The H - C bond in (3n) or (3b) could be the result of decarboxylation. Carbon oxidized at room temperature loses a carboxyl group o n heating; o n oxidation of the carbon a t 40OoC, this carboxyl group is lost immediately after its formation.
Sugar charcoal activated with C02 always contains pairs of carboxyl groups I which are so close together that they can forin an anhydride (cf. the reaction with SOC12). A structural model of the surface oxides, which also takes these observations into account, is discussed in detail elsewhere [26]. Q u i n on o i d groups have also been detected on the surfaces of carbon blacks [28]. During polarographic reduction, a wave occurred at a voltage corresponding to the reduction of quinones. The polarographic reduction wave disappeared after the carbon black had been treated with LiAIH4. Hydroquinones have been detected in the same manner by polarographic oxidation. The carbon black could no longer be oxidized after methylation with diazomethane. The infrared spectrum of the color black “Carbolac” contains an absorption band at 1580 cm-1 due to a hydrogen-bonded carbonyl group [28]. This absorption was displaced on methylation to ca. 1690 cm-1, corresponding to a normal carbonyl group. Free radicals can be detected in oxidized carbon blacks [29,30]. The radicals are probably closely connected with the quinonoid groups [30]. Another explanation for the coupled occurrence of the carboxyl group I1 and the carbonyl group IV, which is also compatible with experimental findings, would be the presence of a lactone of a formylcarboxylic acid or of a ketocarboxylic acid.The model of a zig-zag edge containing all the groups detected, as well as such a lactol, is shown in (4). (A further carboxyl group is formed on oxidation in acidic solutions.) The lactol is very similar to the fluorescein-like lactone. 0 O II
H
O
M
11
(4)
OH
HOOC
I
I
CO 0
I1
IV
0’
I11
The presence of such a lactol is supported by the following data: Oxidation of pyrene (5) with ozone in glacial acetic acid [31] or with H202 and Os04 [32] gives 41281 J. V . Hallum and H . L. Drushel, J. physic. Chem. 62, 110, 1502 (1958). [291 J. B. Donnet and A . Henrich, Bull. SOC.chim. France 1960, 1609. I301 J. B. Donne!, G. Henrich, and G. Riess, Rev. gkn. Caoutchouc 38, 1803 (1961). [3 I ] H . Vollmann, H. Becker, M . Corell, and H . Streeck, Liebigs Ann. Chem. 531, 66 (1937). [32] F. G. Oberender and J. A . Dixon, J. o r g . Chemistry 24, 1226 (1959).
Angew. Chem. internat. Edit. J Vol. 3 (1964)
I No. I0
formylphenanthrene-5-carboxylic acid (7) which normally exists as the lactol (6) [33,34]. Compound (6) or (7) reacts with diazomethane to form a hydrolysable methylcarboxylate [34]; on the other hand, alcohol plus mineral acid gives the pseudoester of the lactol form
[33,34]. Both compounds are reduced to the diol (8) by LiAlH4 [35]. On reduction of ( 7 ) with amalgamated zinc and hydrochloric acid, both the aldehyde group and the carboxyl group are attacked, with formation of 1,&dihydropyrene [33]. Ketocarboxylic acids whose functional groups are at the peri positions of a condensed ring system behave in a similar manner to formylcarboxylic acids [34J.
The behavior of the acidic surface oxides on reduction can be explained on the assumption that a lactol is present. On reduction with amalgamated zinc, the carboxyl groups I1 and IV are attacked to an equal extent. The reaction with sodium ethoxide, when the latter is added in large excess, may be regarded as the formation of the salt of a hemiacetal of the aldehyde group. Secondary reactions, such as aldol condensation or the Cannizzaro reaction, cannot occur. I t is very difficult to decide whether it is preferable to formulate the surface oxides as a lactone (3a) or as a lactol ( 4 ) . The principles of organic chemistry derived from work on relatively small molecules cannot be applied directly to very large oxidized ring systems. It is also conceivable that in each case only two acidic groups, e. 6. I and 111 or I1 and IV are adjacent. The equal frequency of occurrence of all four groups could then be due to the fact that the two combinations are bonded to carbon edges with different structures occurring with equal probability.
The functional groups detected explain, at the most, 50 of the analytically determined oxygen content. This has been confirmed by other authors [I 5,361. When graphite was extremely finely ground and then oxidized, it evolved elemental oxygen as well as carbon dioxide on degassing in vaczm between 100 and 250 “C [36]. The bonding of this oxygen has not yet been explained. It is perhaps chemisorbed on sites of high rc-electron density associated with defects in the layer planes. [33] M. S. Newman and H. S. Whitehouse, J. Amer. chem. SOC. 71, 3664 (1949). [34] G. M . Badger, J . E . Campbell, J. W. Cook, R . A . Raphael, and A . I. Scott, J. chem. SOC.(London) 1950, 2326. [351 M. S. Newman and C . W. Muth, J. Amer. chem. SOC.73, 4627 (1961). [36] Y. and A. Sar$, W. F. Kisselev, N . N . Leshnev, I . S. Novikova, and G . G . Fedorov, Akad. Nauk S.S.S.R. 143, 1358 (1962).
675
V. Surface Oxides on Diamond
Table 9. Chemical reections of the surface oxides of diamond. I
Pretrestment
The diamond which is found in the "blue ground" of volcanic pipes of South Africa is hydrophobic, as is any freshly comminuted diamond. This permits its easy separation from the hydrophilic matrix rock, e.g. by flotation or, as formerly, with tarred or greased beds, to which the diamonds adhere, while the rock is washed away by water. On the other hand, diamonds which occur in sediments are hydrophilic. K. A . Hofmann observed that finely powdered diamond forms stable suspensions in dilute ammonia after treatment with calcium hypochlorite solution [37]. It was obvious to assume that this was due to the presence of surface oxides. Investigation of the surface oxides on diamond seemed particularly attractive, since diamond can be regarded as the parent substance of aliphatic compounds, whereas aromatic compounds are derived from graphite. The surface oxides of diamond should therefore be expected to be different: e.g. phenolic hydroxyl groups cannot be formed. Only relatively small specific surface-areas were available for the investigations on diamond, and correspondingly small quantities of surface oxides were therefore formed. Owing to the high price and limited availability of sufficiently finely-powdered diamond, the sample weights used could not be increased as desired. We managed to obtain diamond powder with a specific surface area of about 17 mz/g, which is just high enough for investigations of surface oxides using modified analytical methods. Diamond purified with aqua regia and hydrofluoric acid was oxidized with oxygen at 400 O C , or with solutions of sodium hypochlorite or ammonium persulfate at about 20 " C . The products were thoroughly washed, if necessary, and dried by degassing in vucuo at 100°C.
The oxidation with oxygen was studied with the aid of a thernioniicrobalance. Surface oxides start to form at about 260 "C. Perceptible liberation of oxides of carbon gradually begins at 350 "C and becomes very pronounced at 380°C. Potentionietric titrations of the acidic groups gave a curve which rose relatively slowly and showed no distinct inflexions. No simple ratios of the quantities of base consumed were observed on neutralization by shaking with bases of various basicities (Table 8). The results Table 8. Neutralization of oxidized diamond powder. Consumption [meq./100 g1
PiOdllCt
NaHCOJ Diamond ox. 0 2 Diamond, ox. NaOCl Diamond, ox. (NN&S>Os
1
I
NazC09
I NaOH I NaOCzHs
0
1.2
2.7
3.4
4.8
of some other reactions are listed in Table 9. As can be seen from the agreement between the results of direct titration (to pH 7), neutralization with NaHC03, and reaction with thionyl chloride, the diamond contained about 2.5 meq. of more strongly acidic carboxyl groups per 100 g of sample. These were probably bound to the corners and edges of the diamond crystals. [37] K. A. Hofinam: Anorganische Chemie. 1st Edit., Vieweg, Braunschweig 1918, p 287.
676
titration to pH 7 NaHCOJ
Oh. 0 2
ox.
I
I
Reaction
0 2
0s. 0 2
SOCl2
ox. NaOCl ox. NaOCl ox. 0 2 , degassed at 100 "C OY. OZ, degassed at 500 "C degassed ia variio at 800 "C
diazoniethane acetyl chloride potassium potassium Clz (at 100 "C) Clz (at 500 " C ) potassium
degassed at 800 "C, then Hr at 800 O C
Conversion [nieq./100 g ] 2 [a1 2.5 2.5 7.4 [a] 3.7 [a] 20 11 20 17 4.1
[a] lhese results were obtained with starting materials other than those used Tor the other values.
We assumed that the octahedral faces of the diamond crystals are saturated with tertiary hydroxyl groups. The active hydrogen content can be determined by salt formation with metallic potassium. This determination was carried out ~y distilling potassium in I'IICUO onto oxidized diamond; the excess potassium was then distilled o f f at 300°C. On degassing between 25 and 400"C, about 20 meq. of potassium per 100 g of dianiond always remained fixed ; this could be determined acidimetrically after hydrolysis (cf. Table 9). The number of active groups decreased above 400 "C; this agrees with the result of thermogravimetric analysis. At 800 OC, almost all the surface oxides were decomposed, and the consumptions of NaHC03 and NaOH required for neutralization fell respectively to zero and 0.5 meq./lOO g. We believed that f r e e r a d i c a l s were formed during the thermal decomposition of the surface coating. The quantity of these should be determinable by direct reaction of diamond degassed at 800°C with elemental chlorine, without prior contact with air. The quantity of chlorine bound agreed very well with the result of the reaction with potassium, if this reaction and the subsequent degassing were carried out between 100 and 400 "C (conversion about 20 meq./100 g). The bound chlorine was resistant to hydrolysis, even with hot sodium hydroxide. The diamond powder was treated with hydrogen in a similar manner at 800 "C: it was then able to bind only very much less potassium (cf. Table 9). The products which had been degassed at high temperatures were hydrophobic, in contrast to the hydrophilic oxidized diamond. The heat of wetting [*I of an oxidized diamond preparation with water was 0.7 cal/g; the value for the same product after degassing at 800°C was 0.5 cal/g, and 0.35 cal/g after treatment with hydrogen at 800°C. These reactions prove the existence of surface oxides on diamond. Some surface oxides are formed even on storage in air. This can cause appreciable interference in the technical separation of diamonds by flotation [38]. The coefficient of friction of diamond that has been heated in vacuo decreases significantly on admission of air [39]. _____ [*I We wish to thank Dip[.-Chem. W. Kabich for conducting the measurements. [38] I. N . Plaksin, V . S . Alekseyev, Chem. Abstr. 59, 2194 (1963). I391 F. P . Bowden and A . E. Hanwell, Nature (London) 201, 1279 (1964). Angew. Chem. internat. Edit.1 Vol. 3 (1964)
/ No. 10
The same phenomenon is known to occur with graphite [40] [**I. The surface oxides formed at room temperature in air are not acidic.
and black microcrystalline carbon [42]. The catalytic efficiency is a measure of the specific surface area [43].
25
VI. Transformation of Diamond into Black Carbon
j
-"yo
\
20 15
During the thermal decomposition of surface oxides we made an interesting observation: at oxygen pressures of about 10-2 to 10-1 mm, diamond becomes dark in color. This blackeniny, which had been described earlier [41], is due to the transformation of diamond on the surface into the black, microcrystalline form of carbon, which is thermodynamically favored. The transformation is cataiysed by traces of oxygen. The dark coating on the diamond was rapidly oxidized to oxides of carbon at 400 "C and an oxygen pressure of 1 atm. I n large dia mond crystals, th e (1 11) a n d ( 1 10) faces become covered with velvet-black car b o n a t a n oxygen pressure of 0.4 m m between 650 a n d 850°C; above 85O"C, th e (100) surfaces become covered t o o [41]. Diam o n d remains unchanged in vacua even a t 1350 "C. In th e presence of a little oxygen, the transformation proceeds more rapidly th an the oxidation of the black carbon produced; th e latter oxidizes more rapidly only at higher oxygen pressures. Surface oxides must play a p ar t in this catalysis by oxygen. In the stationary state of the black layer, th e rate of transformation is governed by th e diffusion of oxygen through t he layer. Thus, oxygen is continuously being consumed; surface oxides a r e presumably formed, a n d these decompose t o oxides of carbon a n d black carbon. We observed no transformation with pure CO at 900°C, a n d only a slight darkening with pure C02. A Boudouard equilibrium such as
can therefore be ruled o u t.
We have proved that the darkening was actually due to the catalytic activity of the black microcrystalline carbon in the synthesis of hydrogen bromide. The equilibrium HZ I- Brr
+ 2 HBr
lies almost completely to the right at 15Q"C, but no HBr is formed in the homogeneous gas phase in the absence of cataIyst. The reaction is catalysed by graphite .
~~~
[40] R. H. Savage, J. appl. Physics 19, I (1948); F. P. Bowden and .I. E. Young, Proc. Roy. Soc.(London), Ser. A 208,444 (1951). [**I One result of this is that diamond bearings and graphite-
lubricated bearings cannot be used in spacecraft, since they wear out very rapidly. Monomolecular oxide layers initially present would be very rapidly removed from such bearings. 1411 T. Evans and C. Phaal, Proceedings 5th Conference on Carbon (Pennsiivania State University 196 I). Pergamon Press, London 1962, Vol. I, p. 147.
Angew. Chem. internat. Edit. 1 Vol. 3 (1964)
1 No. 10
m
0
10
05
m
1
3
2
1
Fix. 1. Change of catalytic efficiency with time in the synthesis of HBr = 67 mm).
(T = 15O.OOC; Hz pressure Q 1 atm: Brz pressure I: Diamond coated with surface oxides
I T : The same product as in J, after degassing in 111: The same product as in
VUC~O at
TI. after darkening at low
800°C 0 2
pressure
at 800°C. Ordinate: Catalytic efficiency [mniole of HBr per hour per p of carbon]; Abscissa: Time [h].
Figure I shows the catalytic efficiency of the same sample of diamond after various pretreatments. The diamond was used in three forms: after oxidation with NaOCl, after subsequent degassing in vacuo (< 10-5 mm) at SO0 "C, and finally after covering with the black coating by heating at SO0 "C in oxygen at ca. 10.-1 mm. With the third sample, the catalytic efficiency fell off gradually in a peculiar manner dtiring the measurements, and became almost zero. The diamond powder was again light in color at the end of this experiment, i. e. the dark coating had volatilized off under the conditions of the HBr catalysis. It must be concluded from this that the coating of black carbon was extremely finely-divided and reactive. This is supported by an earlier observation [44] that diamond which had been heated in an evacuated Crookes' tube by a glow discharge and had become coated with carbon became lighter in color after storing for some years in air. We wish to thank the Deutsche Forschungs~emeinschaft and Union Carbide European Research Associates, Bruss e k , for their support in these investigations. Messrs. biamanten- Winter, Hamburg, supplied us with finely powdered dianiond at a special rate. We are deep1.v indebted fo Prof: U. Hofnnnn Jbr his help and valuable encoiiragetnent. Received, April 3rd, 1964 [A 388/180 IE] German version: Angew. Chem. 76. 742 (1964) Translated by Express Translation Service, London
[42] U. Hofmann and W. Lemcke, 2. anorg. allg. Chem. 208, 194 (1932). [43] A. C!auss, H. P. Boehni, and U. Hofinann, 2. anorg. allg. Chem. 290, 35 (1957). [44] W. Crookes, Chem. NeRs 74, 39 (1896).
677
