Feb.

, 1927 THE PREPARATION O F ALIZARIN 473

[CONTRIBUTION FROM THE COLOR LABORATORY OF THE BUREAUOF CHEMISTRY,

DEPARTMENT O F AGRICULTURE,
NO. 1211
THE PREPARATION OF ALIZARIN FROM PHTHALIC
ANHYDRIDE AND ORTHO-DICHLOROBENZENE
BY MAXPHILLIPS
RECEIVED
SEPTEMBER
25, 1926 FEBRUARY
PUBLISHED 5,1927
The synthesis of anthraquinone from phthalic anhydride and benzene,
by the application of the Friedel and Crafts reaction, has of late become
commercially important. This synthesis was made commercially feasible
by the development in recent years of improved processes for the manu-
facture of phthalic anhydride and aluminum chloride. The applications
of this synthesis are quite general, and a number of derivatives of anthra-
quinone are thus prepared commercially.
I n this paper, experiments are described on the preparation of 2,3-
dichloro-anthraquinone from phthalic anhydride and o-dichlorobenzene
and the conversion of the dichloro-anthraquinone into alizarin.
o-Dichlorobenzene is obtained as a by-product in the chlorination of
benzene, particularly in connection with the preparation of p-dichloro-
benzene. A large quantity of o-dichlorobenzene is annually produced in
this country and is now utilized only t o a limited extent. It was with the
view to possible development of a commercial utilization of this product
that the investigation described was undertaken.
Sprent and Dodd' claim that when phthalic anhydride and o-dichloro-
benzene are condensed with anhydrous aluminum chloride a t 115', chloro-
benzoylbenzoic acid is formed. This they condensed with sulfuric acid
into 2-chloro-anthraquinone (P-chloro-anthraquinone) . Their claim could
not be confirmed. Several experiments were conducted following the di-
rections given in the Sprent and Dodd patent and the only reaction
product obtained was 2,3-dichloro-anthraquinone. The o-dichlorobenzene
which Sprent and Dodd used may have contained monochlorobenzene as
an impurity, which would account for the result they obtained.2
When o-dichlorobenzene is condensed with phthalic anhydride, two
isomeric dichlorobenzoylbenzoic acids may be formed (I and 11).
The 3 ,4f- and 2f,3f-dichloro-2-benzoylbenzoic acids, upon condensation
with concd. sulfuric acid, yield 2,3- and 1,2-dichloro-anthraquinone,
respectively, as indicated below.

'OH
acid
3 ',4'-Dichloro-2-benzoylbenzoic 2,3-Dichloro-anthraquinone
I
Sprent and Dodd, Brit. pat. 204,528(1923).
* Phillips, Ind. Eng. Chem., 17, 721 (1925).
474 MAX PHILLIPS VOl. 49

2',3 '-Dichloro-2-benzoylbenzoicacid 1,2-Dichloro-anthraquinone

I1
The dichlorobenzoylbenzoic acid obtained by the method described
in this paper (m. p., 191', corr.), when heated with sulfuric acid (d., 1.84),
was converted into 2,3-dichloro-anthraquinone.It is thus reasonably
assured that the dichlorobenzoylbenzoic acid obtained has the structure
represented by Formula I. There was no indication of the formation
of the other isomeric acid (Formula 11).
I n the alkali fusion of 2,3-dichloro-anthraquinoneone would expect
t o obtain the 2,3-dihydroxy-anthraquinone (hystazarin). However,
Kircher3 found that dichloro-anthraquinone (m. p., 261') obtained by the
oxidation of dichloro-anthracene, when fused with alkali, was converted
into alizarin and from that concluded that the dichloro-anthraquinone
a t hand must have been the 1,2 derivative. It is now known, however,
that 1,2-dichloro-anthraquinonemelts a t 208' (coIT.),~so that Kircher's
dichloro-anthraquinone was, without doubt, impure 2,3-dichloro-anthra-
quinone. Furthermore, Grandmougin6 has recently shown that 2,3-di-
bromo-anthraquinone, on fusion with alkali, is converted into alizarin and
has also proved that the dibromo-anthraquinone used by Graebe and Lieber-
mann6 in their now classic synthesis of alizarin was the 2,3 compound.
That 2,3-dichloro-anthraquinone should yield alizarin instead of hystazarin
is, therefore, not a t all surprising.
Experimental Part
Materials Used.-The phthalic anhydride used in these experiments
was the C. P., sublimed product.
The anhydrous aluminum chloride used was the commercial article.
The o-dichlorobenzene was a commercial product obtained from The
Niagara Alkali Company, Niagara Falls, New York.
When subjected t o distillation, 89% distilled a t 175-180', 3% below 175'
and 8% above 180'. It was used directly, without previous distillation.
Method.-In a preliminary experiment 37 g. (0.26 mole) of phthalic anhydride,
184 g. of o-dichlorobenzene (1.25moles) and 73 g. of anhydrous aluminum chloride
(0.27 mole) were placed in a flask provided with a reflux condenser a t the top of which
was inserted a calcium chloride tube. The mixture was heated in an oil-bath first at
3 Kircher, Ann., 238, 348 (1887).
4 Ullmann and Billig, Ann., 381, 27 (1911).
5 Grandmougin, Compt. rend., 173, 717 (1921).
Graebe and Liebermann, Ann. Suppl., 7,289 (1870).
Feb., 1927 THE PREPARATION OF ALIZARIN 475
120" and subsequently heated to the boiling point of the mixture (temperature of the
oil-bath, 175"). After two hours' heating, the reaction mixture, which was black, was
allowed to cool, and was poured into ice water; 100 cc. of concd. hydrochloric acid was
added to this mixture and the o-dichlorobenzene removed by distillation in a current
of steam. The black, tarry residue was digested on the steam-bath with a 20% sodium
carbonate solution. The greater part of the tarry matter remained undissolved. The
filtrate from this was boiled with animal charcoal, filtered and acidified with sulfuric acid.
The dichlorobenzoylbenzoic acid was filtered off and dried a t 120". The yield, before
crystallization, amounted to 15 g., or 20.3%. It was crystallized twice from hot benzol
and obtained as colorless, prismatic crystals; m. p., 191.2' (corr.). It is readily soluble
in chloroform, ethyl alcohol and glacial acetic acid; it is difficultly soluble in carbon
tetrachloride and practically insoluble in petroleum ether.
Anal. Subs., 0.1971,0.1934: COZ, 0.4128,0.4037; HzO, 0.0558,0.0505. Subs.,
0.1837,0.1692: AgC1, 0.1800,0.1661. Calcd. for ClaH803C12: C, 56.96; H, 2.76;C1,
24.03. Found: C,57.11,56.92;H,3.16,2.92;C1, 24.24,24.28.
Conversion of 3', 4'-Dichloro-2-benzoylbenzoic Acid into 2,3-Dichloro-anthra-
quinone.-Fourteen g. of the acid obtained as described above was treated with 20 times
its weight of sulfuric acid (d,, 1.84)and heated a t 150" for four hourd. The reaction
mixture, after being cooled to room temperature, was poured into ice water, filtered and
washed with water, then with sodium carbonate solution, and again with water. The
dichloro-anthraquinone precipitate was dried a t 120";yield, 13 g. It was crystallized
from glacial acetic acid, separating out as light yellow, needle-like crystals; m. p., 267"
(corr.). The yield of the crystallized product amounted to 9 g. (69.2y0of the theoretical,
calculated on the dichlorobenzoylbenzoic acid taken). This product is identical with
the 2,3-dichloro-anthraquinone described by Ullmann and Billig.?
Inasmuch as in the preliminary experiment just described, the yield of
the 3',4'-dichloro-2-benzoylbenzoic acid was far from satisfactory, owing
to the formation of a tarry matter, experiments were conducted for the
purpose of increasing this yield, if possible. Several preliminary experi-
ments showed that if the condensation were conducted a t looo, the forma-
tion of tarry matter would be entirely avoided. It was also indicated
that the proportion of aluminum chloride used was an important factor in
determining the yield. Accordingly, a series of experiments was con-
ducted a t the reaction temperature of looo, in which the ratio of aluminum
chloride to phthalic anhydride was the only variable. I n all of the experi-
ments an excess of o-dichlorobenzene was used ( 5 moles of o-dichloro-
benzene t o 1 mole of phthalic anhydride). The procedure was, briefly,
as follows.
Twenty-nine and six-tenths g. (0.2 mole) of finely-ground phthalic anhydride and
147 g. (1mole) of o-dichlorobenzene were placed in an Erlenmeyer flask provided with an
inverted condenser, at the end of which was attached a calcium chloride tube; the requi-
site amount of powdered aluminum chloride was added and the mixture heated on the
steam-bath until there was no further evolution of hydrogen chloride (about 14 hours).
To the reaction mixture, after cooling to room temperature, water and 100 cc. of concd.
hydrochloric acid were added and the mixture was distilled in a current of steam to re-
move the excess of o-dichlorobenzene. The residue was filtered off, washed with hot
water and successively digested on the steam-bath with sodium carbonate solution
Ref. 4,p. 11.
476 MAX PHILLIPS VOl. 49
until free from 3’,4‘-dichloro-2-benzoylbenzoic acid. This sodium carbonate solution
was boiled with animal charcoal, filtered and made acid with dil. sulfuric acid, where-
upon the 3‘,4‘-dichloro-2-benzoylbenzoicacid separated out as a grayish, crystalline
precipitate, which was filtered off, dried at 110” and weighed.
The result thus obtained is recorded in Table I, under the heading
“Yield of 3f,4f-dichloro-2-benzoylbenzoic acid (crude).” This material
was crystallized once from boiling benzene, giving a colorless, crystalline
product which was dried a t 110’ and weighed. This result is recorded
in Table I, under the heading “Yield of 3’,4f-dichloro-2-benzoylbenzoic
acid (pure) .” The purity of the 3’,4’-dichloro-2-benzoylbenzoicacid
was determined by titrating with 0.1 N sodium hydroxide solution. The
yield of this pure 3’,4f-dichloro-2-benzoylhenzoicacid was calculated on
the basis of the phthalic anhydride taken.
TABLE
I
EFFECTOF QUANTITY
OF ALUMINUMCHLORIDE ON THE YIELD OF 3‘,4‘-DICHLORO-2-
BENZOYLBENZOIC ACID
Temperature, 100”; molal ratio of phthalic anhydride to o-dichlorobenzene, 1: 5
(29.6 g. of phthalic anhydride and 147 g. of o-dichlorobenzene).
Calcd. yield
Molal ratio o-Dichloro- of 3’,4 -di-
of phthalic benzene Yield of 3’,4’-dichloro- chloro-2-
anhydride recovered, 2-benzoylbenzoic acid, 6 . Purity benzoylbenzoic
Expt. to AlClr g. Crude Pure 7% acid, %
1 1:0.5 137 0 0 0 0
2 1:l 107 34.9 30.5 100 51.7
3 1:l.l 106 50.6 36.6 98.7 62.0
4 1:1.2 104 46.1 36.6 100 62.0
5 1:1.5 99 51.4 41.3 100 70.0
6 1:2 91 60.4 43.1 98.8 73.0

It will be observed from Table I that practically the maximum yield

of dichlorobenzoylbenzoic acid was obtained when the molal ratio of
phthalic anhydride to aluminum chloride was 1:1.5.
Determination of the Quantity of Sulfuric Acid to Be Used for the
Conversion of 3’,4’-Dichloro-2-BenzoylbenzoicAcid into 2,3-Dichloro-
anthraquinone
I n the preliminary experiment previously described, the 3‘,4’-dichloro
2-benzoylbenzoic acid was condensed into 2,3-dichloro-anthraquinone,
usiiig a rather large excess of concd. sulfuric acid. The reaction was com-
plete after heating for four hours a t 150’. I n order to determine the
minimum quantity of sulfuric acid necessary in this condensation, experi-
ments were conducted in which the proportion of sulfuric acid (d., 1.84)
to 3’,4’-dichloro-2-benzoylbenzoicacid was the only variable. The pro-
cedure was briefly as follows.
Eight and eighty-five hundredths (8.85) g. (0.03 mole) of 3’,4’-dichloro-2-benzoyl-
benzoic acid and the requisite amount of sulfuric acid (d., 1.84) were heated together at
150”for four hours. The reaction mixture was subsequently cooled to room temperature
Feb., 1927 THE PREPARATION OF ALIZARIN 477
and then poured into several times its volume of cold water. The precipitated 2,3-
dichloro-anthraquinone was filtered off, washed with dil. sodium carbonate solution,
then with water, and finally dried a t 110”. The product thus obtained was pure, as is
indicated by the chlorine determination recorded in Table 11.
TABLE
I1
EFFECTO F VARYING THE QUANTITY
OF SULFURIC ACID
Temperature, 150 O ; 0.03 mole of 3’,4’-dichloro-2-benzoylbenzoicacid (8.85 g.) :
reaction period, four hours.
R a t i o of 3’,4‘-
dichloro-2-benzoyl- Yield of 2,3- Calcd. yield of
benzoic acid to dichloro-anthra- 2,3-dichloro-anthra-
Expt. sulfuric acid quinone, g . CI,C % quinone, %
1 1:2 7.94 25.67 95.6
2 1:4 8.15 25.61 98.0
3 1:6 8.17 25.58 98.3
4 1:8 8.34 25.20 98.4
5 1:lO 8.27 25.50 98.6
a Calcd. for CI4H6O1Cl2:C1, 25.60.

The results in Table I1 show that the conversion of 3’,4’-dichloro-2-

benzoylbenzoic acid into 2,3-dichloro-anthsaquinoneis practically quanti-
tative, using four parts by weight of concd. sulfuric acid to one part of
3’,4’-dichloro-2-benzoylbenzoicacid.
Condensation of Phthalic Anhydride and o-Dichlorobenzene According
to the Method of Sprent and Dodd
Inasmuch as the results obtained were a t variance with those claimed
in the patent of Sprent and Dodd,l an experiment was carried out following
the description of their method as given in the patent. The reaction
product consisted entirely of 2,3-dichloro-anthraquinone(m. p., 2 6 i 0 ,
corr.). I t was not possible to find any 2-chloro-anthraquinone which,
according to Sprent and Dodd, is supposed to he the chief product of this
reaction.
Alkali Fusion of 2,3-Dichloro-anthraquinoneinto Alizarin
In a typical experiment, 20 g. of 2,3-dichloro-anthraquinone,80 g. of potassium
hydroxide and 20 cc. of water were heated in a nickel crucible which was inserted in an
oil-bath. At 225 (thermometer in the reaction mixture) a reaction took place and the
fusion mass became dark blue. This temperature was maintained for 20 minutes and
the reaction product then allowed t o cool. The cold melt was dissolved in water,
heated to boiling and filtered. No insoluble residue was left on the filter paper. The
filtrate was acidified with dil. sulfuric acid, and the orange precipitate was filtered off,
washed and dried; yield, 5.4g., or 31.2%. The alizarin was purified by sublimation and
identified through its melting point and also spectrophotometrically.8
Summary
(1) A synthesis of alizarin from phthalic anhydride and o-dichloro-
benzene is brought about by the condensation of phthalic anhydride,
8 The spectrophotometric measurements were made by Mr. W. C. Holmes of the
Color Laboratory.
478 0. F. HEDENBURG AND I,. H. CRETCHER VOI. 49

o-dichlorobenzene and anhydrous aluminum chloride into 3',4'-dichloro-

2-benzoylbenzoic acid, and the conversion of this acid, by means of sulfuric
acid, into 2,3-dichloro-anthraquinone. Upon fusion with alkali the di-
chloro-anthraquinone is converted into alizarin.
( 2 ) The claim of Sprent and Dodd' that 2-chloro-anthraquinone is
obtained by the condensation of phthalic anhydride and o-dichlorobenzene
could not be confirmed.
D. C.
WASHINGTON,

[CONTRIBUTION FROM THE DEPARTMENT

O F RESEARCHIN PURE CHEMISTRY, MELLON
RESEARCH,
INSTITUTE O F INDUSTRIAL UNIVERSITY OF PITTSBURGH]

T H E PREPARATION OF CRYSTALLINE d-TALONIC ACID

AND LEONARD
BY OSCARF. HEDENBURG H. CRETCHER
RECEIVED
SEPTEMBER
27, 1926 PUBLISHED
FEBRUARY
5, 1927

Introduction
I n 1885 Kiliani' prepared crystalline d-galactonic acid. So far as the
authors are aware, this is the only crystalline hexonic acid described in
the literature. Later, Nef and one of us2 made a study of the substance
and prepared some of its derivatives.
It is known that when a freshly prepared, aqueous solution of d-gluconic
acid, or its epimer d-mannonic acid, is rapidly concentrated a t low tem-
perature and pressure, the resulting gum yields on trituration with absolute
alcohol the so-called /3-lactone of the corresponding acid.3
When, however, a freshly prepared solution of d-galactonic acid is treated
in the same manner, crystalline d-galactonic acid is ~ b t a i n e d . ~
This work was undertaken to determine whether the epimer of d-galac-
tonic acid, that is, d-talonic acid, on treatment as described above, would
yield a /%lactone or an acid. Our results have shown that talonic acid
solutions, similarly to solutions of galactonic acid, yield a crystalline acid
on evaporation. From an examination of the space formulas of the hex-
ose sugars it is evident that they may be classified in two series on the basis
of the configuration of the groups attached to Carbon Atoms 3 and 4.
One of these series, characterized by the fact that the hydroxyl groups,
in the conventional representation of the sugar molecule, are found on
opposite sides of the carbon atoms, may be called the glucose series. It
includes also mannose, idose and gulose. The other series, characterized
by hydroxyl groups on the same side of the carbon atoms, may be termed
the galactose series and embraces also talose, allose and altrose. The
monobasic acids corresponding to glucose and mannose-gluconic and
1 Kiliani, Ber., 18, 1551 (1885).
(a) Nef, Ann., 403,277 (1914). (b) Hedenburg, THISJOURNAL, 37,364 (1915).
Ref. 2 a, pp. 310, 323; 2 b, pp. 347, 355.
Ref. 2 b, p. 364,