Chemical mechanism of the radical feedback loop in the classical BZ reaction. Malonyl bromite and oxalic acid as flow-through intermediates

Article abstract of DOI:10.1039/b004903o

High-pressure liquid chromatography (HPLC) and measurements of the CO₂ produced were performed in the induction period of the classical Belousov-Zhabotinsky (BZ) reaction (malonic acid-bromate-cerium catalyst in sulfuric acid medium). It was found that oxalic acid is a flow-through intermediate of the reaction. This was confirmed with an independent qualitative test with thiobarbituric acid. The concentration of oxalic acid grows in the induction period together with that of bromomalonic acid and dibromomalonic acid intermediates. It is known that there are two negative feedback loops in the BZ reaction: one is via bromide and the other via organic free radicals. Oxalic acid and also CO₂ are products of this second loop where organic radicals react with BrO₂ radicals. The induction period was chosen for the present experimental studies because the above radical-radical reactions are most intense during that time. Based on the experimental results mechanistic proposals are made for the radical feedback loop. A method to accumulate multivalent organic acids present in very low concentrations in the BZ reaction was also developed. Applying this and a thermal decomposition method ethenetetracarboxylic acid (EETA) was identified as an oxidation product of ethanetetracarboxylic acid (ETA).

Full text of DOI:10.1039/b004903o

View Article Online / Journal Homepage / Table of Contents for this issue
Chemical mechanism of the radical feedback loop in the classical BZ
reaction. Malonyl bromite and oxalic acid as Ñow-through
intermediates
Laszlo Hegedus,a Horst-Dieter Forsterling,a Eniko Kokai,b Krisztina Pelle,b Gabriella Taba,b
Maria Wittmannb and Zoltan Noszticzius*b
a Fachbereich Chemie, PhilippsÈUniversitat Marburg, D-35032 Marburg/L ahn, Germany.
E-mail: foersthd=mailer.uni-marburg.de
b Center for Complex and Nonlinear Systems and the Department of Chemical Physics,
Budapest University of T echnology and Economics, H-1521 Budapest, Hungary.
E-mail: noszti=phy.bme.hu
Received 19th June 2000, Accepted 11th July 2000
Published on the Web 16th August 2000
High-pressure liquid chromatography (HPLC) and measurements of the CO produced were performed in the
2
induction period of the classical BelousovÈZhabotinsky (BZ) reaction (malonic acidÈbromateÈcerium catalyst
in sulfuric acid medium). It was found that oxalic acid is a Ñow-through intermediate of the reaction. This was
conÐrmed with an independent qualitative test with thiobarbituric acid. The concentration of oxalic acid
grows in the induction period together with that of bromomalonic acid and dibromomalonic acid
intermediates. It is known that there are two negative feedback loops in the BZ reaction: one is via bromide
and the other via organic free radicals. Oxalic acid and also CO are products of this second loop where
2
organic radicals react with BrO radicals. The induction period was chosen for the present experimental
2
studies because the above radicalÈradical reactions are most intense during that time. Based on the
experimental results mechanistic proposals are made for the radical feedback loop. A method to accumulate
multivalent organic acids present in very low concentrations in the BZ reaction was also developed. Applying
this and a thermal decomposition method ethenetetracarboxylic acid (EETA) was identiÐed as an oxidation
product of ethanetetracarboxylic acid (ETA).
course of the oscillations, especially in the periods when the
Introduction
autocatalytic reaction is ““switched o† ÏÏ. In this state the
Ce4`Èorganic substrate reactions proceed undisturbed. On
the other hand, a very new pathway is opened whenever the
autocatalytic reaction is ““switched onÏÏ. This is because in the
The classical BZ reaction, the Ce4` catalysed oxidation of
malonic acid by bromate in a sulfuric acid medium, is by far
the most studied chemical oscillator.1h7 However, many
important details in the organic chemistry of the reaction are
still unknown. This lack of information is due to the fact that
appropriate analytical techniques (like NMR and HPLC)
were missing or not generally available at the time when the
reaction was discovered. Thus in the last few years we have
started a program applying mostly HPLC to identify various
organic products and intermediates of the BZ reaction.8h12
In the Ðrst stage of this research, however, we dealt not with
the whole BZ system but only its so-called organic subset,
more precisely with the Ce4`Èmalonic acid and the Ce4`È
bromomalonic acid reactions. Obviously, the whole BZ
system is more complex than the simple sum of its organic
and inorganic subsets as the intermediates of the two sets can
cross-react with each other. Nevertheless this Ðrst stage of the
research was unavoidable because a good understanding of
the organic subsystem was a prerequisite for understanding
the whole BZ system. These organic subsystem studies have
shown that the primary organic radicals produced in the
Ce4`Èorganic substrate reactions do not disproportionate, as
was previously assumed,13 but recombine. The recombination
products are ethanetetracarboxylic acid (ETA)8 and malonyl
malonate (MAMA)9 in the case of the Ce4`Èmalonic acid
reaction and bromoethenetricarboxylic acid (BrEETRA)10
when the organic substrate is bromomalonic acid. The same
reactions obviously take place in the whole BZ reaction in the
time of the autocatalytic reaction BrO radicals appear which
2
react with malonyl (and most probably also with other
organic) radicals in a fast di†usion-controlled way.14 Thus as
long as BrO radicals are present malonyl radicals have little
2
chance to react with each other to form ETA and MAMA.
The situation is qualitatively similar to the case when the
Ce4`Èmalonic acid reaction takes place in the presence of
oxygen.12 In that case oxygen reacts with the malonyl radicals
rapidly and as a result the reaction pathway is diverted from
the formation of ETA and MAMA toward the production of
oxalic acid (OA).
The present work is the Ðrst HPLC study on a whole BZ
system. Our main aim here is to identify the products of the
malonyl radicalÈBrO radical reaction. It has been known for
2
10 years14 that the two radicals react with each other and this
is the basis of a second negative feedback loop and the so-
called radical controlled oscillations,15,16 nevertheless no
intermediates or products of the reaction have been identiÐed
until now. It is reasonable to assume that as a Ðrst step the
two radicals recombine with each other and form malonyl
bromite as an intermediate:
DOI: 10.1039/b004903o
Phys. Chem. Chem. Phys., 2000, 2, 4023È4028
This journal is ( The Owner Societies 2000
4023

View Article Online
The real question is the subsequent decomposition of this
intermediate. One possibility is a rapid hydrolysis of malonyl
bromite which can be regarded as an ester of bromous acid
with the alcoholic hydroxyl group of tartronic acid (TA):
tion (in 1 M sulfuric acid). Thus the initial concentrations
were: 0.5 M malonic acid, 0.1 M bromate, 5 ] 10~3 M Ce4`
and 1 M sulfuric acid. For the CO measurements the total
2
volume of the reaction mixture was 2 mL only but the same
composition was applied. The constant nitrogen stream
created anaerobic conditions and also mixed the solution. To
analyse the intermediates of the induction period with HPLC
the reaction was quenched abruptly.
(DR1)
Such a process would be analogous to the hydrolysis of
MAMA to tartronic and malonic acids.9 If reaction (DR1)
were the only decomposition route of malonyl bromite then
Quenching methods
Dilution. A hundredfold dilution was suddenly made by
injecting 0.25 mL of the reaction mixture to approximately 24
mL distilled water in a 25 mL volumetric Ñask, mixing and
adjusting it exactly to 25 mL.
the malonyl radicalÈBrO radical reaction could only delay
2
the autocatalytic production of HBrO but would not be able
2
to inhibit it permanently.
Another decomposition route was suggested by Sirimung-
kala et al.17 just recently (reaction (Z15) in their scheme):
Quenching with NH OH and precipitation with Ce3‘. 1 mL
4
of the reaction mixture was added to a centrifuge tube con-
(DR2a)
taining 0.3 mL 25% NH OH solution. Then 1 mL 0.1 M
4
Ce (SO ) solution was added to precipitate insoluble Ce3`
2
4 3
salts of certain organic acids together with a large amount of
Blocking of the autocatalytic process can be easily understood
Ce(OH) . The precipitate was settled and washed with 0.005
this way as no reappearance of HBrO (the autocatalytic
3
2
M Ce (SO ) solution three times. Finally, the bromate-free
intermediate) is possible here. It is known12 that mesoxalic
acid (MOA) is oxidized by Ce4` to oxalic acid very rapidly
thus we can assume that reaction (DR2a) is followed by reac-
tion (DR2b) nearly instantaneously:
2
4 3
precipitate was dissolved in 2.5 mL 0.1 M sulfuric acid in a
volumetric Ñask and diluted with water to 25 mL.
Analytical methods
HPLC. Experiments were performed with Shimadzu
equipment using an ion exchange column at 45 ¡C and a UV
detector working at 220 nm. The eluent was 0.01 M H SO ,
Ñow rate 0.40 mL min~1. Identical HPLC conditions have
already been applied successfully in a series of previous inves-
tigations to separate various organic acid intermediates of the
BZ reaction. For further details see ref. 9È11.
The eluent was Ðltered with a membrane Ðlter (Poraphil,
pore size 0.2 lm) to remove the solid particles. All the mea-
sured solutions were also Ðltered (Sartorius Minisart NML,
pore size 0.2 lm).
(DR2b)
2
4
The subject of the present paper is a systematic search,
mainly by HPLC, for the above-mentioned hypothetical inter-
mediates, namely for oxalic and mesoxalic acids. These acids,
however, should be only Ñow-through intermediates,
occurring in very low concentrations in the reaction mixture
as Ce4` oxidizes both of them rapidly to CO . Consequently
if reactions (DR2a) and (DR2b) are important processes in the
BZ reaction then the start of the autocatalytic reaction should
2
be followed by a sharp increase in CO production. To check
this supposition, besides HPLC analysis measurements of the
CO measurements. CO was purged from the reactor with
2
2
2
a nitrogen stream, it was then mixed with hydrogen. Next
CO produced were also performed in this work, focusing on
CO was converted to methane and measured with a Ñame
2
2
the induction period of the BZ reaction. This preoscillatory
ionization detector (FID). The details of the method and the
period was chosen for our studies because the autocatalytic
reaction is very active in this time, thus higher than normal
concentrations of malonyl bromite and the other Ñow-through
intermediates can be expected here. Elevated cerium ion
(5 ] 10~3 M) and malonic acid (0.5 M) concentrations were
chosen for similar reasons.
apparatus are given in ref. 7. The only di†erence here was a
splitting of the nitrogen stream: only 12.3% of the nitrogen
carrier gas was introduced to the detector to avoid a too-high
methane concentration resulting in a nonlinear response of
the FID.
Spot test for oxalic acid. 2 mL of the reaction mixture was
added to a centrifuge tube containing 1 mL 25% NH OH
Experimental
4
solution, then 1 mL 0.02 M Ce (SO ) solution was added to
2
4 3
precipitate Ce3` salts. The precipitate was settled and washed
Chemicals
with 0.005 M Ce (SO ) solution three times. Next D0.1 g
2 4 3
precipitate was heated with 0.1 mL 1 M H PO solution over
Malonic acid (Fluka, puriss.), Ce(SO ) É 4H O (Merck, pro
analysi), oxalic acid (Merck, pro analysi), Ce (SO ) (Fluka,
purum), 2-thiobarbituric acid (Fluka, purum), ortho-
phosphoric acid 85% (Merck, pro analysi), NH OH 25% (J.T.
4 2
2
3
4
a free Ñame. After cooling it was evaporated to dryness with
2
4 3
0.2 mL 25% NH OH solution. At that point 2-thiobarbituric
4
acid was added in great excess and carefully heated over a free
4
Baker) and H SO 97% (J.T. Baker) were used as received. All
Ñame. If oxalic acid is present then the condensation product
gives a brick-red color. Compared to the HPLC sample prep-
aration a smaller amount of Ce3` was applied. In this way the
characteristic brick-red color of the test was not weakened by
the white Ce3` salts.
2
4
solutions were prepared with doubly distilled water.
Reactions
The BZ reaction was examined under anaerobic conditions at
room temperature (22 ^ 2 ¡C) in a batch reactor. 5 mL 1 M
malonic acid solution (in 1 M sulfuric acid), 1 mL 1 M
Results
NaBrO solution (in water), 0.9 mL 5 M sulfuric acid and 2.6
mL (doubly distilled) water were mixed in a 30 mL reaction
CO measurements
3
2
vessel and bubbled with nitrogen for 2 min. The reaction was
First we studied the rate of CO evolution in the induction
period of a BZ system with elevated cerium and malonic acid
2
started by injecting the catalyst, 0.5 mL 0.1 M Ce(SO ) solu-
4 2
4024
Phys. Chem. Chem. Phys., 2000, 2, 4023È4028

View Article Online
Fig. 1 CO evolution rate (FID signal) vs. time diagram registered in
2
the BZ reaction. Initial concentrations: [MA] \ 0.5 M, [NaBrO ] \
3
0.1 M, [Ce4`] \ 5 ] 10~3 M, [H SO ] \ 1 M. Total volume: 2 mL.
2
4
The reaction was started by injecting Ce4` together with acidic
bromate. Registration of the FID signal started before initiation of the
reaction to show the baseline. t : Time of the Ðrst maximum, start of
0
the preinduction period (delay time18 of the system). t : Start of the
1
autocatalytic reaction (the ““realÏÏ induction period). t : End of the
2
real induction period, start of the oscillatory regime.
concentrations. The reaction was started by injecting the
cerium catalyst, in the form of Ce4`, together with acidic
bromate into a mixture of malonic and sulfuric acids. The
recorded evolution rate vs. time diagram is shown in Fig. 1
together with the experimental parameters.
A characteristic feature of the diagram is that before the
real induction period18 a single maximum appears imme-
diately after the addition of the Ce4` catalyst to the malonic
acid. (In a di†erent experiment similar behavior was observed
by Forsterling et al.19) This maximum is produced by the
Ce4`Èmalonic acid reaction which generates CO at a rate
2
proportional to the Ce4` concentration. The rate of this reac-
tion is maximal right after the start but decreases gradually
with decreasing Ce4` concentration. It is important to note
that the autocatalytic reaction cannot start for a while, partly
because there is not enough Ce3` initially but mainly because
of the inhibitory e†ect of the malonyl radicals produced by
the Ce4`Èmalonic acid reaction. When the Ce4`, and conse-
quently the inhibitory malonyl radical concentration, falls
below a critical value the autocatalytic reaction starts. This
point is marked in Fig. 1 by a sharp increase in the CO pro-
duction rate to values well above the initial maximum. It is
Fig. 2 Chromatograms of BZ samples quenched by 100 times dilu-
tion with water. See Table 1 for assignment of the various peaks.
Quenching time (a) 5 s, (b) 1 min, (c) 2 min after starting the BZ
reaction.
2
obvious that the Ce4`Èmalonic acid reaction alone cannot be
responsible for this high CO output. It is logical to assume
2
that in this stage of the reaction the decomposition of malonyl
bromite and further oxidation of its decomposition products
(oxalic and mesoxalic acids) by Ce4` are responsible for the
extra CO . A further contribution to the extra CO might
The chromatograms were recorded with high sensitivity to
show minor components of the mixture. Abbreviations of the
intermediates belonging to the separate peaks are given in the
Ðgures. Full names and retention times for the intermediates
can be found in Table 1.
2
2
come from the Ce4`Èbromomalonic acid reaction which gen-
erates more CO that the Ce4`Èmalonic acid one. This contri-
2
bution can be signiÐcant if bromomalonic acid is produced
rapidly in the induction period.
The high bromate content of the sample results in a long
tailing after the Ðrst peak which does not disturb identiÐcation
of the species but hinders quantitative analysis.
HPLC of BZ samples collected after quenching the reaction by
dilution
To obtain direct evidence about the supposed intermediates
we performed HPLC measurements on reaction mixtures
which were quenched somewhat after the start of the autocat-
alytic reaction (D1 min reaction time) and close to the end of
the induction period (D2 min reaction time).20 For compari-
son practically unreacted mixtures (samples quenched after 5 s
reaction and mixing time) were also analyzed with HPLC
(Fig. 2a).
The reaction was quenched simply by diluting the reaction
mixture with water 100 times. This was necessary not only to
stop the BZ reaction but also to get a solution whose sulfuric
acid concentration was equal to the eluent concentration (0.01
M). The HPLC curves are shown in Fig. 2.
Our main interest is focused on MOA and OA, the
expected decomposition products of malonyl bromite. While
no trace of mesoxalic acid (Table 1) can be discovered in the
chromatogram there is a peak at 600 s which is the retention
time of oxalic acid. Unfortunately ETA has the same retention
time thus interference is possible, especially because some
ETA should be produced in the short ““preinduction periodÏÏ
when only the Ce4`Èmalonic acid reaction is active and the
autocatalytic reaction has not yet started. The presence of
ETA is also supported by the appearance of MAMA, the
other recombination product. However, in these solutions the
ETA : MAMA peak ratio is di†erent from the usual one
(1.6 : 1, see Fig. 1 in ref. 9), the ““ETA peakÏÏ in Fig. 2b and c
Phys. Chem. Chem. Phys., 2000, 2, 4023È4028
4025

View Article Online
Table 1 Assignments of the various HPLC peaks found in our
experiments. Small peaks in the Ðgures (e.g. the peak of acetic acid or
formic acid in Fig. 2) can be seen as well deÐned peaks high above the
baseline at higher sensitivities
Retention
time/s
Symbol
Component name
480
480
480
500
500
570
600
Inj.
Injection
Bromide
Bromate
Dibromomalonic acid
1,1,2,2-Ethenetetracarboxylic acid
Monomalonyl malonate
Oxalic acid
Br~
BrO ~
3
Br MA
2
EETA
MAMA
OA
600
638
700
723
ETA
1,1,2,2-Ethanetetracarboxylic acid
Mesoxalic acid
1,1,2-Ethenetricarboxylic acid
Bromomalonic acid
Tartronic acid
MOA
EETRA
BrMA
TA
800
840
863
965
ETRA
MaleA
GOA
MA
1,1,2-Ethanetricarboxylic acid
Maleic acid
Glioxylic acid
Malonic acid
Dibromoacetic acid
Succinic acid
UnidentiÐed intermediate
Bromosuccinic acid
Formic acid
Acetic acid
Fumaric acid
1049
1123
1242
1295
1380
1386
1511
1568
1810
Br AcA
2
SA
u. i.
BrSA
FA
AcA
FuA
BrAcA
Bromoacetic acid
are signiÐcantly greater in this case compared to the small
MAMA shoulders. This means that besides ETA another
component with the same retention time, most probably
oxalic acid, should also be present.
Comparing Fig. 2b and c it can be seen that the OAÈETA
peak is larger at the end of the induction period than it is at
the beginning. Also the tartronic acid and dibromoacetic acid
peaks have grown during this period. Most probably tartronic
acid is a hydrolysis product of malonyl bromite (see reaction
(DR1)) while dibromoacetic acid can appear via decarboxyl-
ation of dibromomalonic acid. Because of the high sensitivity
applied here bromo- and dibromomalonic acid peaks are
reduced but their amount also grows during the induction
period.
Fig. 3 Chromatogram of the same samples as in Fig. 2 but after
keeping them at 90 ¡C for 30 min. (a) 5 s, (b) 1 min, (c) 2 min reaction
time.
While inability to resolve ETA and OA peaks was certainly
a drawback of the HPLC column used in our experiments this
column was able to resolve practically all other components.
Thus it would not be reasonable to apply a new column with
all the consequent experimental difficulties just to achieve a
resolution of ETA and OA peaks, especially because a reten-
tion time alone is not enough for positive identiÐcation.
end as the decrease of the ETAÈOA peak due to the heat
treatment is larger in Fig. 3b than in Fig. 3c. Other changes
caused by the heat treatment are the following: Because of the
decarboxylation of BrMA its peak decreases while the decom-
position product, BrAcA peak grows. In a similar manner, the
amount of Br AcA, which is a decarboxylation product of
2
Br MA, also increases. Some decarboxylation of MA also
2
takes place, indicated by the growing AcA peak. Decarboxyl-
Thermal decomposition of ETA
ation of ETA leads to ETRA and SA.
Next we compare Fig. 3b and 3c. In both Ðgures the peak
at 600 s retention time is due entirely to oxalic acid without
any interference from ETA. Regarding data in Fig. 3.
([OA] \ 7.7 ] 10~4 M in Fig. 3b and [OA] \ 3.2 ] 10~3 M
in Fig. 3c) it can be seen that the oxalic acid concentration
grew considerably during the induction period. Furthermore
A di†erence between the thermal decomposition properties of
ETA and OA was applied to di†erentiate between these two
components. It was found12 that after keeping an aqueous
solution of ETA at 90 ¡C for 30 min practically all ETA
decomposes while no measurable thermal decomposition of
OA can be observed under similar conditions. Thus after
quenching the reaction by dilution the cited heat treatment
was applied. The resulting chromatograms of samples for 1
and 2 min reaction times are shown in Fig. 3b and c, respec-
tively. Comparing Fig. 2 and 3 it can be observed that while
the peaks at 600 s decreased more (Fig. 3b) or less (Fig. 3c) the
MAMA shoulder disappeared completely in both cases. It can
be seen that the relative amount of ETA in the ETAÈOA peak
is higher at the beginning of the induction period than at the
the amounts of BrMA and Br MA have also grown in this
2
period as peaks due to their decomposition products, BrAcA
and Br AcA are much higher in Fig. 3c than in Fig. 3b.
2
HPLC of BZ samples enriched in oxalic acid
Regarding the oxalic acid peaks of Fig. 3, they are character-
istic but not entirely convincing because here OA is only a
4026
Phys. Chem. Chem. Phys., 2000, 2, 4023È4028

View Article Online
relatively minor component. Moreover, they are strongly per-
turbed by the tailing caused by the high bromate content of
the samples. Thus it was desirable to prepare a sample
without bromate interference and with a relatively enriched
oxalic acid content. To this end we precipitated the oxalic acid
content of a BZ mixture after 2 min reaction time in the form
of Ce3` oxalate and removed bromate and a part of the other
more soluble components by washing the precipitate (see
Experimental for details). After redissolving the precipitate in
0.01 M H SO a bromate-free HPLC sample was produced
that EETA is an oxidation product of ETA according to the
following mechanism:
2
4
whose oxalic acid concentration was higher than in the orig-
inal sample. The chromatogram of such an enriched sample is
shown in Fig. 4.
The precipitation method changes the relative amounts of
the various components. Comparing Fig. 2c and 4 it can be
seen that the OA-ETA peak is much higher in the latter. No
bromate is present and the peaks of the brominated acids are
missing or considerably decreased. Obviously the Ce3` salts
of these acids are more soluble. On the other hand, peaks due
to acids with more carboxylic groups are enhanced, thus we
can expect more interference from ETA in this case also. To
avoid this the heat treatment method was applied again. A
chromatogram taken after this is shown in Fig. 5.
In Fig. 5 oxalic acid gives a major peak and also no inter-
ference from bromate can be observed, thus we achieved our
aim. Comparing Fig. 4 and 5 we can make an additional
observation. The peak in Fig. 4 at 500 s retention time is cer-
tainly due to EETA. This assignment of the peak is supported
by Fig. 5 where all its characteristic decarboxylation pro-
ducts: EETRA, MaleA and FuA can be observed. On the
EETA is a minor component but due to the small solubility
of its Ce3` salt it was enriched in the sample. Also it has a
high UV absorption, this explains the high EETA peak in Fig.
4. Finally we remark that the peak in Fig. 4 close to the place
where tartronic acid (TA) appears is not due to TA since TA is
stable under our heat treatment but this component is not
(compare Fig. 4 and 5). Most probably it is a multivalent acid
present in low concentrations in the BZ mixture but accumu-
lated in this sample by our Ce3` precipitation method.
Thiobarbituric acid test on enriched BZ samples
To obtain a Ðnal proof for the presence of oxalic acid we per-
formed a thiobarbituric acid test. This is a qualitative, sensi-
tive test for oxalic acid with no interference.21 The result was
positive, proving convincingly that oxalic acid is present in the
sample.
other hand, the peak should not contain signiÐcant Br MA as
2
no increase in Br AcA can be observed in Fig. 5. We suggest
2
Consequences for the negative feedback loops:
Mechanistic possibilities
The identiÐcation of oxalic acid as a Ñow-through interme-
diate in the BZ reaction can help to clarify some open prob-
lems regarding the mechanism. According to our present
knowledge that mechanism includes one positive and two
negative feedback loops. The mechanism of the positive feed-
back is well known from the FKN mechanism2 which is the
autocatalytic oxidation of Ce3` by acidic bromate with
bromous acid (HBrO ) and bromine dioxide radical (BrO ~)
2
2
as autocatalytic intermediates.
The remaining problems of the mechanism are related to
the negative feedback loops. The Ðrst negative feedback loop
via bromide was also known from the beginning.1,2 Bromide
is a so-called control intermediate because it reacts quickly
with bromous acid, the Ðrst autocatalytic intermediate. Thus a
bromide inÑow can control or even interrupt the autocatalytic
cycle. While the control role of bromide was clear the source
of this bromide was more problematic. One source is the
Ce4`Èbromomalonic acid reaction the mechanism and the
products of which were clariÐed just recently.10 Nevertheless,
experiments by Koros and coworkers22 indicated that some
other bromide sources should also be present in the classical
BZ system.
Fig. 4 Chromatogram of a sample enriched with oxalic acid and
other multivalent acids. The reaction was quenched with NH OH 2
4
min after the start.
As was mentioned already, the second negative feedback
loop was discovered later, about 10 years ago.14 Here organic
free radicals play the control role as they react with bromine
dioxide, the second autocatalytic intermediate. In this loop the
source of both the organic and the inorganic radicals was
clear but there was no information about the products of
these radicalÈradical reactions.
The present results can shed some light on the unsolved
problems of the negative feedback loops. Let us summarize
again our experimental observations relevant to these prob-
lems: (i) Our most important Ðnding is that oxalic acid is a
Fig. 5 Chromatogram of the same sample as in Fig. 4 but after heat
treatment.
Phys. Chem. Chem. Phys., 2000, 2, 4023È4028
4027

View Article Online
Ñow-through intermediate of the BZ reaction. (ii) We found
that its concentration grows in the induction period when the
autocatalytic reaction is ““switched onÏÏ, in spite of the fact
that in this period both Ce4` and HOBr concentrations are
high and these oxidants react with oxalic acid rapidly. (iii) It
acids can be important intermediates of both the bromide and
the radical controlled feedback loops.
Finally we remark that oxalic and/or mesoxalic acid might
be the organic substrate in the photoreduction of bathoferriin
by visible light observed in a BZ system and reported recently
by Toth et al.25
was observed that rate of the CO evolution in the preinduc-
2
tion period is relatively low but it increases rapidly when the
autocatalytic reaction starts.
These observations strongly support the conclusion that
Acknowledgement
both CO and oxalic acid are products of the radicalÈradical
This work was partially supported by OTKA (T-030110),
FKFP (0287/1997) grants, by the Deutsche Forschungsge-
meinschaft and the Fonds der Chemischen Industrie and the
ESF Program: ““ReactorÏÏ. M. Wittmann acknowledges the
support of the Bolyai Research Stipendium. The authors
2
reactions. Moreover, as in the beginning of the induction
period the majority of the organic radicals are malonyl rad-
icals, CO and oxalic acid should be decomposition products
2
of malonyl bromite. Most probably this decomposition route
follows reaction (DR2a) i.e. in the Ðrst step it gives mesoxalic
and hypobromous acids. A possible mechanism would be the
following hydrogen ion assisted rearrangement of the chemical
bonds:
thank M. Kozak for her help in the CO measurements.
2
References
1
2
A. M. Zhabotinsky, Dokl. Akad. Nauk. SSSR, 1964, 157, 392.
R. J. Field, E. Koros and R. M. Noyes, J. Am. Chem. Soc., 1972,
94, 8649.
Oscillations and T raveling W aves in Chemical Systems, ed. R. J.
Field and M. Burger, Wiley, New York, 1985.
3
(DR2a)
4
5
P. Ruo† and R. M. Noyes, J. Chem. Phys., 1986, 84, 1413.
Z. Noszticzius, W. D. McCormick and H. L. Swinney, J. Phys.
Chem., 1989, 93, 2796.
Nevertheless, identiÐcation of the characteristic Ðrst interme-
diate mesoxalic acid in this reaction mixture was not possible
because the oxidation of mesoxalic acid by Ce4` is very fast
(more than two orders of magnitude faster than that of oxalic
acid12).
6
7
L. Gyorgyi, R. J. Field, Z. Noszticzius, W. D. McCormick and
H. L. Swinney, J. Phys. Chem., 1992, 96, 1228.
Sz. Nagygyory, M. Wittmann, Sz. Pinter, A. Visegrady, A.
Dancso, Nguyen Bich Thuy, Z. Noszticzius, L. Hegedus and
H.-D. Forsterling, J. Phys. Chem. A, 1999, 103, 4885.
Y. Gao, H.-D. Forsterling, Z. Noszticzius and B. Meyer, J. Phys.
Chem., 1994, 98, 8377.
A. Sirimungkala, H.-D. Forsterling and Z. Noszticzius, J. Phys.
Chem., 1996, 100, 3051.
Another possibility would be some complex rearrangement
reaction giving oxalic acid, CO and bromide ions directly:
8
9
2
10 J. Oslonovitch, H.-D. Forsterling, M. Wittmann and Z. Nosztic-
zius, J. Phys. Chem. A, 1998, 102, 922.
(DR3)
11 I. Szalai, H.-D. Forsterling and Z. Noszticzius, J. Phys. Chem. A,
1998, 102, 3118.
While we cannot exclude this route completely its contribu-
tion cannot be signiÐcant because this would eliminate the
possibility of the so-called ““non-bromide controlledÏÏ oscil-
lations.15 On the other hand, even if the contribution of reac-
tion (DR3) to the overall process is minor it can play a role as
an additional bromide source.
12 L. Hegedus, H.-D. Forsterling, M. Wittmann and Z. Noszticzius,
J. Phys. Chem. A, in the press.
13 L. Gyorgyi, T. Turanyi and R. J. Field, J. Phys. Chem., 1990, 94,
7162.
14 H.-D. Forsterling and Z. Noszticzius, J. Phys. Chem., 1989, 93,
2740.
In later stages of the induction period the situation becomes
more complex as bromo- and dibromomalonic acids and their
radicals can appear in the reaction mixture. A route similar to
(DR2a) is not possible, thus hydrolysis should play a major
role here. Hydrolysis of bromomalonyl malonate in the Ðrst
step yields bromotartronic acid as well as bromous acid and
in the next step bromotartronic acid hydrolyses rapidly, yield-
ing mesoxalic acid and bromide:17,23
15 H.-D. Forsterling, Sz. Muranyi and Z. Noszticzius, J. Phys.
Chem., 1990, 94, 2915.
16 Z. Noszticzius, Z. Bodnar, L. Garamszegi and M. Wittmann, J.
Phys. Chem., 1991, 95, 6575.
17 A. Sirimungkala, H.-D. Forsterling, V. Dlask and R. J. Field, J.
Phys. Chem. A, 1999, 103, 1038.
18 The delay time of the system is the time required for the carrier
gas to transport the CO from the reactor through the meth-
2
anizer and the resulting methane to the FID.
19 H.-D. Forsterling, H. Idstein, R. Pachl and H. Schreiber, Z.
Naturforsch. A, 1984, 39, 993.
20 The lengths of the preinduction period and the real induction
period, determined in separate spectrophotometric measure-
ments, are 25 and 90 s, respectively. In Fig. 1 the preinduction
time corresponds roughly to the time between the Ðrst maximum
and minimum while the real induction period is approximately
the time between the Ðrst minimum and the second maximum.
21 F. Feigl and V. Anger, Spot T ests in Organic Analysis, Elsevier,
Amsterdam, 1966, p. 457.
Thus, as we can see, with the appearance of bromomalonic
acid the bromide controlled and the radical controlled feed-
back loops become mixed: the radicalÈradical reaction Ðnally
produces a bromide ion, and this connects the two di†erent
loops.
Furthermore, it is interesting to mention that oxalic acid
itself is a substrate of the BZ reaction24 and that the hypo-
bromous acidÈoxalic acid reaction also provides an additional
bromide source. It is probable that mesoxalic acid can play a
similar role. Thus we can conclude that oxalic and mesoxalic
22 M. Varga, L. Gyorgyi and E. Koros, J. Am. Chem. Soc., 1985,
107, 4780.
23 Y. Gao and H.-D. Forsterling, J. Phys. Chem., 1995, 99, 8638.
24 Z. Noszticzius and J. Bodiss, J. Am. Chem. Soc., 1979, 101, 3177.
25 R. Toth, V. Gaspar, A. Belmonte, M. OÏConnell, A. Taylor and
S. K. Scott, Phys. Chem. Chem. Phys., 2000, 2, 413.
4028
Phys. Chem. Chem. Phys., 2000, 2, 4023È4028