[CONTRIBUTION FROM THE CHEMICAL LABORATORY OF STANFORD UNIVERSITY ]

THE ACTION OF BASES ON ORGANIC HALOGEN COMPOUNDS.

I. THE REACTION OF ARYL HALIDES WI TH
POTASSIUM AMIDE
F. W. BERGSTROM, RICHARD E. WRIGHT, CHARLES CHANDLER, AND
W. A. GILKEY
Received February 7, I936
It has long been known that halogen directly attached to the benzene
nucleus is not readily replaced by a hydroxyl, amino, or alkoxy1 group,
unless its activity is enhanced by other groups present in the proper posi-
tions within the molecule. Thus, a nitro group will labilize ortho and
para halogen atoms in a manner perhaps best explained by current elec-
trochemical theories of substitution in the aromatic nucleus.' The effect
of conventional "ortho-para-directing" groups-such as methyl-upon the
mobility of aromatically bound halogen is on the other hand much more
difficult to determine, because, in the majority of cases, the halogen is even
less reactive than if attached to an unsubstituted benzene nucleus. It was
thought that these difficulties could be satisfactorily overcome by carry-
ing the replacement reactions out in a dissociating solvent-liquid ammonia
-using potassium amide, a base of the ammonia system, in place of the
less reactive alkali hydroxides or alcoholates. Since the reactions in
liquid ammonia were found to be extremely rapid, there is available a
method for determining the effects of various groups on the mobility of
aromatically bound halogen. I n the present paper, we have reported at
length only on the reaction between potassium amide and the phenyl
halides in liquid ammonia.
DISCUSSION
Liquid ammonia solutions of potassium amide at -33" react very rapidly
with phenyl chloride, bromide or iodide to give aniline, diphenylamine,
small quantities of p-aminobiphenyl, and occasional traces of triphenyl-
amine. Tarry products are formed at the same time, and the solutions
acquire a red or reddish-brown color. I n harmony with its known inert-
1 INQOLD, Rec. trav. chim., 48,810 (1929); ROBINSON, J . SOC. Dyers and Colourists,
Jubilee Journal, 1934, p. 74.
170
ACTION OF BASES ON ORGANIC HALOGEN COMPOUNDS 171
ness, phenyl fluoride fails to react with potassium amide under the same
conditions, and even at room temperature the reaction is extremely slow.
Sodium amide fails to react with phenyl chloride or bromide in boiling
diethyl ether or in boiling benzene (three hours' refluxing), although the
reaction occurs readily in liquid ammonia.la
Homing*, working in this laboratory has found that lithium diethyl-
amide, (C2Hs)2NLi, reacts with phenyl bromide in ether at room tempera-
ture to give diethylaniline in good yield. Since the formation of by-prod-
ucts has been decreased, in comparison with the potassium amide reactions
described in this article, there is perhaps available a still better reagent
for determining the relative reactivities of aryl halides.
a-Chloronaphthalene, o-, m- and p-chlorotoluene, 0- and p-chlorobi-
phenyl, and o-chloroaniline are all rapidly attacked by potassium amide
solution at - 33" ; a-fluoronaphthalene reacts slowly at this temperature,
but more rapidly at room temperatures, while the fluorotoluenes appear
to be unattacked at -33'. A detailed discussion of the reactions men-
tioned in this paragraph will be reserved for the future.
Calcium amide and barium amide react with phenyl bromide in ammonia
at room temperature to give a mixture of mono- and diphenylamines, with
some tar. Apparently the formation of secondary amine cannot be
avoided by using an amide of low ammonia solubility.
The following equations express the reactions that occur when an aryl
halide is treated with a solution of potassium amide in ammonia.
(1) CoHSX + KNHz = C~HSNH~ + KX
(2) CsH6NH2 + KNHz = C6HsNHK + NH,
(3) COH6NHK + CaH& "'-> (C6Hs)zNH + " 3
(4) (C&Hs)*NH + KNHz = (CsHs)zNK + NH3
( 5) (C&Hs)zNK + CB&X " ' - > (c6H~)aN + K X
( 6) [ 0 - N H - ] K+A [ ">NH] K+NH,? CsHsX
(as X salt)
Halogen is removed almost quantitatively from phenyl chloride, bromide
or iodide by a liquid ammonia solution of an excess of potassium amide.
However, when potassium amide is added to a considerable excess of
phenyl halide in liquid ammonia, potassium halide is formed in yields of
1. Cf. TITHERLEY, J. Chem. SOC. 71, 460 (1897); SACHS, Ber. 39, 3011 (1906).
* Unpublished work.
172 BERGSTROM, WRIGHT, CHANDLER AND GILKEY
fifty to sixty per cent. of the theoretical, as calculated from equation (1)
alone. This can only mean that reaction (2)-in which potassium anilide
is produced-is more rapid than the primary reaction expressed by equa-
tion (1). Since the yield of halide ion (ie., of potassium halide) is slightly
greater than the fifty per cent. calculated from the combined equations
(1) and (2), potassium anilide and potassium diphenylamide must further
react, in the sense of equations (3), ( 5) , and (6), although it is possible that
the aniline has not been completely converted to potassium anilide in
accordance with (2). The amide ion was found to be an essential catalyst
for reactions (3), ( 5) , and (6), which otherwise did not take place under
the conditions of the present experiments. A more complete discussion
is reserved for the second paper of this series.
Competition experiments, in which an ammonia solution of potassium
amide is added to an excess of a mixture of aryl halides in the same solvent,
indicate the following order of decreasing reactivity, as judged by relative
yields of the respective potassium halides.
Br > I > C1 > (>>>F)
The same order is found when potassium amide is added to a mixed aryl
dihalide of the type of p-chlorobromobenzene, although the results are a
little more difficult of strict theoretical interpretation. Bromine appears
in both cases to be more readily removed from the aromatic nucleus than
iodine, although the difference in reactivity is not great.
EXPERI MENTAL
The apparatus used in the present investigation (see Figure 1) consists essentially
of pyrex cells, 1 and 2, the first being cooled by a liquid ammonia bath, the second
being fitted snugly into a large Dewar vessel to hinder evaporation of the solvent.*
Vessel 1 is connected at A to a source of ammonia gas (from a tank of the com-
mercial liquid containing sodium) and to vessel 2 through the siphon B, which allows
one quantitatively to transfer a solution from 1 to 2. With the use of stopcocks C,
D, and E, 1 could be connected to a mercury bubbler, or to the waste. Vessel 8 is
connected to 1 i n the manner described, to the waste ammonia line at E and to an
adjustable mercury bubbler, H, which has at its upper end, G, a three-way stopcock,
so that gases (coming from 2 ) can be diverted to the waste, or to a gas collector, K,
containing water. The contents of vessel 2 are agitated by a motor-driven stirrer,
L, whose shaft passes through a packed joint.
The manipulation will be understood from the following description of the re-
action between a known amount of potassium amide and an excess of chlorobenzene.
Cells 1 and 2 are cleaned, thoroughly dried, and air i n them is displaced by am-
monia. (Air in 1 is swept out when stopcocks A, C, D and E are opened and stop-
cock B is closed; air i n 2 is displaced when stopcocks A, B and E are opened and
* From our experience, it is recommended that this second cell likewise be cooled
by an ammonia bath.
ACTI ON OF BASES ON ORGANIC HALOGEN COMPOUNDS 173
stopcock C is closed.) With ammonia gas passing through about 20 cm. of mercury
in the bubbler (stopcocks A and C open, stopcocks B and D closed) and with liquid
ammonia in the Dewar liquid readily condenses in 1. When this is about two-thirds
full, e)topcocks C and A are closed, B is opened, and about half of the liquid is forced
over into 2, applying ammonia pressure through A' if necessary. (I t is convenient
to have a stopcock sealed to the cap closing A' . ) B is then closed, the ground-glass
joint A' is cautiously removed and a previously broken capsule containing a weighed
amount of potassium is introduced,2 together with some clean iron wire. (A slow
stream of ammonia passes through A and out at M to prevent access of air during
these manipulations. The ammonia must, of course, first bubble through the con-
densed liquid in 1.)
FIG. 1
I n the meanwhile, the stirrer in 2 is started and an excess of
chlorobenzene is introduced (through J ) from a weight pipet. After the disappear-
ance of the blue potassium solution in 1, the potassium amide is slowly forced over
into the liquid ammonia solution of the chlorobenzene in 2 according to the method
described above. An opaque red solution results immediately with evolution of
enough heat to cause ebullition of the ammonia. Twenty or twenty-five cc. of fresh
ammonia is condensed i n 1 and forced into 2, the process being repeated twice to
insure complete transference of the potassium amide. Shortly thereafter, cell 2 is
disconnected and removed from the Dewar, in order to facilitate evaporation of the
* PIECK, J . Am. Chem. soc. , 40, 338 (1918); KRAUS AND CRIU, ibid. , 44, 2001 (1922).
174 BERGSTROM, WRIGHT, CHANDLER AND GI LKEY
ammonia, which generally consumes two to three hours. (Vessel 9 may, if desired,
be dipped i n a large volume of water.) The solvent-free residue is then covered with
benzene, and water is slowly added to effect hydrolysis.
I n the prelimi-
nary runs-where larger and less accurately tveighed quantities of reagents were
used, and where the phenyl halide was often added (from buret F ) to a large excess
of potassium amide, the hydrolysate was acidified and the benzene layer extracted
twice with 1:l hydrochloric acid. The acidic aqueous solution was distilled with
steam, diphenylamine passing over, and aniline hydrochloride remaining behind.
The latter was made basic and steam-distilled to remove aniline, which was identi-
fled by conversion to acetanilide. Diphenylamine was identified by the melting
point of a mixture with a specimen of authentic origin.
In the smaller-scale runs, in which halide ion was to be determined quantita-
tively, the following procedure was adopted.
To the residue left after evaporation of the ammonia was added about 75 cc.
benzene, followed by about 50 cc. water. The benzene layer was extracted three
times with very dilute sodium hydroxide, and halide ion was determined gravi-
metrically as silver halide i n the total aqueous extract. Aniline was extracted from
the benzene with 3N hydrochloric acid, and determined in an aliquot portion by
bromine titration,$ after first filtering off a small precipitate of p-aminobiphenyl
hydrochloride. Diphenylamine was then extracted from the benzene with con-
centrated sulfuric acid (4/5 H2S01 by vol.), either being weighed directly after
precipitation from solution, or being determined by bromine titration. (Extraction
with concentrated hydrochloric acid gave a somewhat better separation, for the sul-
furic acid dissolved a small quantity of triphenylamine and tar. The reported
yields of diphenylamine are nevertheless sufficiently accurate for comparative
purposes.)
Evaporation of the benzene yielded a viscous, sweet-smelling, tarry oil, from
which small quantities of triphenylamine could often be isolated by repeated crystal-
lizations from petroleum ether (identification by mixture melting point).
p-Aminobiphenyl, prepared by addition of sodium hydroxide to its hydrochloride,
melted at 53-54.5 uncorr., and at the same temperature after mixture with authentic
p-aminobiphenyl (prepared by reduction of p-nitrobiphenyl).
The treatment of the hydrolysate varied with the experiment.
The results of these runs are summarized i n Table I.
Competitive Runs
Two phenyl halides were introduced separately from weighing burets into ves-
sel d (see figure), which contained 50-75 cc. ammonia. Potassium amide, i n known
amount, dissolved i n 50-75 cc. of ammonia in vessel 1, was very slowly forced into 8
with mechanical stirring. Vessel 1 was finally washed with two or three 25-cc.
portions of solvent, to transfer all amide to 2. Necessarily, the phenyl halides
must be in excess.
Halide ion in the hydrolysate was determined both gravimetrically and volu-
metrically (Mohr titration) i n aliquot portions of the water extract. The apparent
molecular weight of the precipitated silver halide may be calculated from the data
thus obtained. The mol fraction, F, of the heavier silver halide i n the precipitate
is equal to ( A- B) / ( C- B) , where A is the apparent molecular weight of the silver
~~ ~~
3 KOLTHOFF AND FURMAN, Volumetric Analysis, J ohn Wiley and Sons, New
York, 1928, vol. I, p. 212.
ACTION OF BASES ON ORGANIC HALOGEN COMPOUNDS 175
YIELD OF
ANILINI~
24.5 (43)
25.4 (48)
52.3 (53)
24.2 (43)
22.8 (41)
21.7 (38)
42. 1 (43)
18 (30)
25.8 (46)
21.4 (39)
22.5 (47)
50.7 (52)
22.0 (lo)
(8)
( 9)
halide precipitated, B is the molecular weight of the lighter silver halide, AgX,
and C! is the molecular weight of the heavier halide, AgY.
The ratio, (moles AgX)/(moles AgY) can be regarded as the ratio of the relative
reactivities of X and Y only if the average molar concentration of each of the two
phenyl halides is the same in a given run. Since this is not usually the case, it is
YIELD OF
DIPEENYLAYINE~
12.8 (23)
10.8 (20)
15.5 (15.6)
12.4 (22)
17. 6 (31)
24 (24)
16 (23)
15 (26)
Tr
8.1 (15)
20.4 (21)
TABLE I
REACTION OF INDIVIDUAL ARYL HALIDES WITH POTASSIUM AMIDE
MOLES
1 KNHz
HALIDE
CsHsCl
CaHsBr
caCIs1
0.0160
0.0263
0.0305
0.0261
0.0314
0.0312
0.0383
0.0505
0.0383
0.0235
0.0289
0.0296
0.0424
MOLES
HALIDE
0.0210
0.0256
0.0141
0.15
0.0352
0.0337
0.287
0.0255
0.0470
0.0242
0 .OM8
0.0357
0.0190
0.0240
0.0116
YIELD OF
ALIDE IO1
55.8%
53 .O
99 .oi
56.4
56.1
58.2
56. 8
99.1
60.9
57.0
55.3
54.8
96.2i
58.7
53.9
(a) Potassium amide solution added to a solution of the halide.
(a) Solution of halide added to potassium amide solution. I n the fifth experi-
ment liquid CsHsBr was added to the KNHz solution.
(c) Calculated on the basis of potassium in the equations, CsHsCl + KNHl =
CsHsNHl + KC1 or 2CsHsCl + 2KNHz =(CaH&NH + 2KC1 + NHs. The figures
in parentheses are yields calculated on the basis of the halide ion (= yield, as cal-
culated above X100/yield of halide ion).
( d) 0.018 g. p-aminobiphenyl hydrochloride isolated (0.3% of theory).
(e) 0.05 g. p-aminobiphenyl hydrochloride isolated (1.3% of theory). (See
cf) Diphenylamine not pure.
( 9) p-Toluidine is the chief product formed, other than tar.
(i) Calculated on the basis of the halide; in all other experiments yields are
equation 6.)
calculated on the basis of the potassium amide.
necespary to correct the observed ratio, (moles AgX)/(moles AgY), multiplying i t
by the factor, (avge conc. CsHsY)/(avge conc. C6HsX), the average concentration
being equal to 3 (twice the initial conc. phenyl halide minus the halide consumed).
The halide consumed is the product of the mole fraction of AgX (or AgY) and the
equivalents of halide ion, the latter having been determined by the Mohr titration.
176 BERGSTROM, WRIGHT, CHANDLER AND GILKEY
0.01324
0,01343
0.01375
0.01326
0.01603
0.01580
0.01723
0.01421
0.01419
It has been assumed that within reasonable limits the relative reactivities of the
phenyl halides are independent of the concentration.
As the volumetric determination may be in error by as much as one per cent.,
and as the conditions for such a rapid reaction cannot be perfectly controlled, the
following results give more the orders of magnitude than the absolute values of the
relative reactivities of the phenyl halides toward potassium amide:
Br/Cl =20 (&5)/1; I/C1 =(8.3 & 0.3)/1; I /Br =0.40 (f0.10)/1
TABLE I1
COMPETITIVE REACTIONS OF PAIRS OF PHENYL HALIDES WITH POTASSIUM AMIDE
223.6 2.961
223.1 2.996
200.2 2.773
201.4 2.680
184.4 2.957
184.0 2.905
185.3 3.192
181.3 2 577
182.9 2.596
ATOMS K
0.02306
0.02364
0.02426
0,02401
0,02814
0.02756
0.01593
0.02469
0.02525
MOLES
0 ,0259
0.0267
0.0276
0.0286
0.0312
0.0266
0.0293
0.0250
0.0404
0 ,0261
0.0318
0.0261
0.0156
0.0145
0.0571
0.0270
0.0559
0.0258
MOLE FRACTIONS
AS FOUND
AgCl 0.121
AgT 0.879
AgCl 0.128
AgI 0,872
AgBr 0.737
AgI 0.264
AgBr 0,711
AgI 0.289
AgCl 0.077
-4gBr 0,923
AgCl 0,085
AgBr 0.915
AgCl 0.056
AgBr 0,944
AgCl 0.146
AgBr 0.854
AgCl 0.110
AgBr 0.890
CORRECTED
RATIO
I /Cl
I /Cl
8.6/1 .O
8.0/1.0
I /Br
0.35/1.0
I /Br
0.38/1 .o
Br/Cl
25/1 .O
Br/CI
1811 .O
Br/Cl
2311 . 0
Br/Cl
16/1 .O
Br/C1
23/1 .O
A typical calculation follows (first run, Table 11):
0 0259
( 0.02306
RNHl 0.02306 mole (= atoms of K used)
CsHbCl 0.0259 mole 100 X - =112% of K.))
CaHJ 0.0267 mole (116% of K)
Wt. Ag. halides (XO.l) =0.2961; cc. 0.0498N AgN03, 26.59 (in 0.1 aliquot)
Equivalents halide ion, 0.01324; theoretical equivalents halide ion, calculated from
equation (1) =moles KNHl =0.02306.
ACTIOK OF BASES ON ORGANIC HALOGEN COMPOUNDS 177
HALIDE^
CsHdBrCl
CsH4BrC1
0.01324
loo 0.02306
Yield of halide ion, calcd on basis of equation (l), = 57.4% =
,4vge inol. wt. halide 0.2961/0.001324 =223.6.
Mole yo AgI =(223.6 - 143.3)/(234.8 - 143.3) =87.8
Mole y7 AgCl =100.0 - 87.9 =12.1
Ratio, (moles AgI)/(moles AgC1) =87.9/12.1 =7.3/1.0
Avge oonc. of C B H ~ =4 [(2 X 116) - (57.4 X 0.879)] =90.8% of K.
Avge oonc. of CeHjCl =3[(2 X 112) - (57.4 X 0.121)] =109. 1~o of K.
For coiivenience the concentrations have been expressed i n terms of the potassium
ooncen tration.
Corrected ratio, I-/C1- (= ratio of reactivities of C6H&C6HaC1) =7.2 X
109.1/90.8 =8.6/1.0.
EQUIVALENTS OF
HALIDE ION
Br 0.00303
c1 0.00048
Br 0.00868
C1 0,00133
TABLE I11
REACTION OF PHENYL DIHALIDES WITH POTASSIUM AMIDE
CsHaBrCl
flTOY8 K
Br 0.00708
C1 0.00150
0.00761
0.01684
0.01731
0.01300
0.01901
CeH4BrI
CsHaBrI
Br 0,00442
I 0.00230
Br 0.00648
I 0.00333
I
I
CORRECTED RATIO
REACTIVITIES*
Br/C1
7.6/1
Br/Cl
8.411
BrjCl
5. 7/1
I /Br
0.48/1
I /Br
0.47/1
(a) All dihalides were para compounds, obtained from the Eastman Kodak Go.
(best grade). No. moles used =No. atoms of K.
Equivalents of halide ion (1)
Eauivalents of halide ion (2)
(a) The observed ratios, , were multiplied by the
. ,
inverse ratios of the average concentrations of the two un-ionized halogens during
the run, to obtain the corrected figures i n this paragraph. The calculations were
made a1s in the preceding section. At the beginning of a reaction, the concentrations
of un-ionized halogen were necessarily the same, altering because the rates of re-
moval of halogen were different.
The Action of Potassium Amide on the Phenyl Dihalides
A known amount of potassium amide was added to an excess of a phenyl dihalide
of the type of p-chlorobromobenzene, using the manipulations described i n the pre-
ceding section. The mixed halides in the hydrolysate were determined indirectly
by weighing the mixed silver halides, which were subsequently converted to silver
178 BERGSTROM, WRIGHT, CHANDLER AND GILKEY
chloride by heating in a stream of ~hl ori ne.~ The experimental data are recorded
in Table 111, and, although admittedly incomplete, indicate the following relative
reactivities of the halogens X and Y in compounds of the type, X.CsH4.Y-p.
Br/Cl =7/1; I /Br =0.5/1. Obviously, the results are not as amenable to exact theo-
retical treatment as those previously discussed, since we are concerned also with the
replacement of halogen from the intermediate compounds of the type, NH2.CaH4.Y .
SUMMARY
(1) Chloro-, bromo-, and iodobenzenes react very rapidly with a liquid
ammonia solution of potassium amide at - 33" to give aniline and diphenyl-
amine, with smaller quantities of triphenylamine and p-aminobiphenyl.
The relative proportions of the reaction products depend upon experi-
mental conditions. Under similar conditions, fluorobenzene does not
react with potassium amide. A number of substituted chloro-, bromo-
and iodobenzenes have been shown to react with solutions of the alkali
amides in liquid ammonia. I n boiling ether or benzene, the above reac-
tions do not occur, on else are very slow.
(2) The reaction, CeH5NH2 + KNHz = CaHsNHK + "3, is very
rapid in comparison with the primary replacement of halogen, CaH6X +
KNHz = C6HBNHz + KX. Consequently, the yield of halogen ion
obtained when potassium amide is added to an excess of phenyl halide lies
between 50 and 60% of the theoretical, as calculated from the second
equation. Divergence from the theoretical value of 50% is accounted
for by the occurrence of the reactions that yield di- and triphenylamine,
and perhaps also by the incompleteness of formation of potassium anilide
in accordance with the first equation above.
(3) Competition reactions indicate the following order of ease of replace-
ment of the halogen atoms in the unsubstituted phenyl halides Br > I >
Cl, (F not replaced in NH3 at -33"). The same order is found for replace-
ments involving the phenyl dihalides of the type p-Ce"XY.
4th ed., 1916, vol. 11, p. 334.
4 TREADWELL-HALL, "Analytical Chemistry," J ohn Wiley and Sons, New York.