Benzaldehyde Formation from Aspartame in the Presence of Ascorbic Acid and Transition Metal Catalyst

Article abstract of DOI:10.1021/jf960079k

Benzaldehyde was produced from aspartame in aqueous acidic solutions containing ascorbic acid and Cu(II) or Fe(III) ion. Benzaldehyde was identified in the system by GC-MS. The yield of benzaldehyde decreases dramatically as the pH of the medium increases above 2.0. EDTA and DTPA completely inhibited benzaldehyde production, while desferrioxamine inhibited only the Fe(III)-catalyzed reaction. Benzaldehyde is not produced under anaerobic conditions unless H₂O2 is added to reaction mixtures. H₂O₂ is produced by reduction of atmospheric oxygen under aerobic conditions. Benzaldehyde production was dependent on ascorbic acid concentration, but the yield of benzaldehyde decreased as the concentration of ascorbic acid exceeded that of aspartame. Addition of ethanol to the reaction mixture had little or no effect on benzaldehyde production, suggesting a mechanism that may not involve free hydroxyl radical. A mechanism is proposed for the reaction.

Full text of DOI:10.1021/jf960079k

J. Agric. Food Chem. 1996, 44, 3461
−3466
3461
Ben za ld eh yd e F or m a tion fr om Asp a r ta m e in th e P r esen ce of
Ascor bic Acid a n d Tr a n sition Meta l Ca ta lyst
Glen D. Lawrence* and Dongmei Yuan
Chemistry Department, Long Island University, Brooklyn, New York 11201
Benzaldehyde was produced from aspartame in aqueous acidic solutions containing ascorbic acid
and Cu(II) or Fe(III) ion. Benzaldehyde was identified in the system by GC-MS. The yield of
benzaldehyde decreases dramatically as the pH of the medium increases above 2.0. EDTA and
DTPA completely inhibited benzaldehyde production, while desferrioxamine inhibited only the
Fe(III)-catalyzed reaction. Benzaldehyde is not produced under anaerobic conditions unless H2O2
is added to reaction mixtures. H2O2 is produced by reduction of atmospheric oxygen under aerobic
conditions. Benzaldehyde production was dependent on ascorbic acid concentration, but the yield
of benzaldehyde decreased as the concentration of ascorbic acid exceeded that of aspartame. Addition
of ethanol to the reaction mixture had little or no effect on benzaldehyde production, suggesting a
mechanism that may not involve free hydroxyl radical. A mechanism is proposed for the reaction.
Keyw or d s: Aspartame; benzaldehyde; ascorbic acid; free radical degradation products; copper(II)
INTRODUCTION
possible products of ascorbic acid dependent free radical
degradation of aspartame.
The artificial sweetener aspartame (N-L-R-aspartyl-
L-phenylalanine methyl ester) has been thoroughly
studied for safety and stability (Stegink and Filer, 1984),
and several of its hydrolytic decomposition products
have been identified (Homler, 1984). Because it is a
dipeptide of naturally occurring amino acids (aspartic
acid and phenylalanine), its hydrolysis products or
metabolites, other than methanol, are nontoxic in the
amounts that would be consumed by normal use of this
sweetener. However, little is known regarding its
interaction with other food components, such as natural
nutrients or food additives.
Mazur (1976) has reported the pH stability profile
for aspartame in aqueous solution, identifying among
the hydrolysis products 3-benzyl-2,5-piperazinedione-
6-acetic acid, the diketopiperazine derivative of as-
partame. These products were subsequently identi-
fied in several soft drinks that were stored for 6 or 36
months (Tsang et al., 1985). Stamp and Labuza (1989)
showed that â-aspartylphenylalanine, â-aspartame, and
phenylalanine methyl ester are products of aspartame
degradation in dilute acid solution. Hussein et al.
(1984) found that several food-flavoring aldehydes, such
as benzaldehyde, cinnamaldehyde, and vanillin, will
react with the free amino group of aspartame in ab-
solute alcohol to produce the corresponding Schiff base.
Ascorbic acid (vitamin C) is a natural component of
many foods and is often added to foods and beverages
as a vitamin supplement and antioxidant. Several
studies have addressed the stability of ascorbic acid in
aqueous solution in the presence of other food compo-
nents, including aspartame (Ho et al., 1994; Hsieh and
Harris, 1987, 1991). Earlier work in this laboratory
(Gardner and Lawrence, 1993) showed aerobic solutions
of ascorbic acid could produce benzene from the widely
used food preservative sodium benzoate under condi-
tions that are commonly found in beverages. These
studies have been extended in an attempt to identify
EXPERIMENTAL PROCEDURES
Rea gen ts. Aqueous stock solutions of the following re-
agents were prepared from analytical reagent grade chemicals
using distilled deionized water and stored in the refrigerator:
1-50 mM aspartame (L-aspartyl-L-phenylalanine methyl ester;
Sigma) in 0.05 N HCl; 52 mM phenylalanine (Sigma) in 0.05
N HCl; 9.3 mM aspartylphenylalanine (Asp-Phe; Sigma) in
0.10 M phosphate buffer, pH 2.0; 1.6 mM copper(II) sulfate
pentahydrate (G.F. Smith); 150 mM hydrogen peroxide (Flu-
ka); 8 mM benzyl alcohol (Sigma); 8 mM benzaldehyde
(Sigma); 4 mM sodium benzoate (Sigma); 100 mM ascorbic acid
(Fluka) in 0.10 N HCl; 0.10 M monochloroacetic acid buffer,
pH 2.1-3.3; and 0.10 M sodium dihydrogen phosphate buffer,
pH 1.6-7.7, adjusted to desired pH with NaOH. Aspartame
stock solutions were prepared fresh every few days. Hydrogen
peroxide concentration in stock solutions and reaction mixtures
was determined colorimetrically with saturated TiOSO4 solu-
tion in 2 M H2SO4 (ꢀ ) 717 M-1 cm-1 at 410 nm; Ellis and
Sykes, 1973).
Rea ction Con d ition s. A typical reaction mixture (1.0 mL)
contained 50 mM buffer (buffer components and pH varied),
5.0 mM aspartame, 0.16 mM CuSO4, 8 mM H2O2, and 5.0 mM
ascorbic acid. The pH was checked at the end of a reaction
because the presence of HCl in the aspartame and ascorbic
acid stock solutions caused some shift in pH of the buffer when
these were added to the reaction mixtures. The reaction
mixtures were placed in a 40 °C water bath (except where
indicated) and analyzed by direct injection of an aliquot into
the liquid chromatograph 15 min after the reaction was
initiated (except where indicated). The reaction conditions
were varied to study the effects of pH and concentration of
reagents. In some cases Fe2(SO4)3 was substituted for CuSO4,
and oxygen was purged from some reaction mixtures by
bubbling with water saturated, prepurified N2.
The metal-chelating agents disodium ethylenediaminetet-
raacetic acid (Na2EDTA), diethylenetriaminepentaacetic acid
(DTPA), glycine, and desferrioxamine (desferal mesylate; a gift
of Ciba Pharmaceutical Co.) were added to some reaction
mixtures to determine their effect on the metal catalysts.
Ethanol was added as a competitive hydroxyl radical scavenger
to determine whether “free” hydroxyl radical is involved in
these reactions.
* Author to whom correspondence should be ad-
dressed [e-mail, glawrenc@hornet.liunet.edu; fax, (718)
488-1465].
The concentration of ascorbic acid was followed during the
progress of the reaction by diluting a 20 µL aliquot of the
S0021-8561(96)00079-9 CCC: $12.00
© 1996 American Chemical Society

3462 J. Agric. Food Chem., Vol. 44, No. 11, 1996
Lawrence and Yuan
reaction mixture in 1.98 mL of 0.10 M phosphoric acid at a
given time and absorbance measured at 245 nm (ꢀ ) 7943 M-1
cm-1; Graselli, 1973). Other reactants gave little or no
contribution to this absorbance (aspartame ꢀ245 ) 102 M-1
cm-1; this study). Benzaldehyde produced in the reaction has
a strong absorbance at 245 nm (ꢀ247 ) 11 482 M-1 cm-1
;
Graselli, 1973), and its contribution at any time in the reaction
could be calculated from the chromatographic quantitation.
Other products of the reaction, such as aromatic hydroxylation
products, would absorb at 245 nm, but their contribution is
probably not significant at low pH (judging from chromato-
graphic results). The concentration of hydrogen peroxide was
monitored colorimetrically during the progress of the reaction
as described above.
Liqu id Ch r om a togr a p h y. The LC system consisted of a
Knauer model 64 pump, Rheodyne 7125 injection valve with
20 µL injection loop, and Knauer variable wavelength UV
detector at 255 nm with response recorded on either a Kipp
and Zonen BD41 strip chart recorder or a Hewlett-Packard
3396B integrator. A Hewlett-Packard model 1050 liquid
chromatograph with diode array detector became available at
the end of this study (gift of the Hewlett-Packard Educational
Grants Program) and was used for some analyses. The
analytical column for analysis of aspartame and benzaldehyde
was a 4.6 × 100 mm Spheri-5 (5 µm) cyano column, with 4.6
× 30 mm guard cartridge (Brownlee). The mobile phase was
0.05 M sodium dihydrogen phosphate, 50 µM DTPA containing
8% (v/v) acetonitrile. The mobile phase was filtered through
0.2 µm membrane filters prior to use and pumped at a flow
rate of 1.3 mL/min. Retention times were aspartame, 1.9 min;
benzaldehyde, 2.8 min. These analytes were quantified by
comparison of peak height with known standards. A product
of aspartame degradation was not well resolved from the
aspartame peak and interfered with quantitation by peak area;
consequently peak height was used for quantitation of both
aspartame and benzaldehyde.
The chromatographic system for analysis of benzyl alcohol,
benzoic acid, and benzaldehyde in the presence of aspartame
consisted of a Spheri-5 C8 (octylsilane) column (4.6 × 100
mm) with C8 guard cartridge (Brownlee) and mobile phase
containing 0.05 M sodium dihydrogen phosphate, 0.01%
(w/v) sodium octylsulfate, and 8% (v/v) methanol adjusted to
pH 3.9 with phosphoric acid solution and pumped at a flow
rate of 3.0 mL/min. Retention times were benzyl alcohol, 2.0
min; benzoic acid, 3.1 min; benzaldehyde, 4.3 min; aspartame,
6.3 min.
Ga s Ch r om a togr a p h y-Ma ss Sp ectr om etr y. A Hewlett-
Packard 5890/5971 GC-MS (gift of the Hewlett-Packard
Educational Grants Program) was equipped with a 30 m HP-1
(nonpolar) capillary column. GC parameters as follows: initial
temperature, 100 °C; initial time, 2.0 min; rate, 10 °C/min;
final temperature, 160 °C; final time, 2.0 min; injector tem-
perature, 200 °C. MS parameters as follows: low mass, 50;
high mass, 200; threshold, 150; sampling rate, 2/s. A high
yield of a major product observed in the liquid chromatographic
analysis was obtained by placing the standard reaction
mixture in a distillation apparatus and collecting the distillate
at 85 °C. This product could be extracted into organic solvents,
such as cyclohexane, but attempts to concentrate it by evapo-
ration of the solvent resulted in loss of product as well. A small
(0.5 µL) aliquot of the distillate was injected (split ratio 60:1)
into the GC-MS for analysis. The major gas chromatographic
peak (m/z > 50) was identified as benzaldehyde.
Solid P h a se Extr a ction (SP E). Once benzaldehyde was
identified as a product of aspartame oxidation in the hydroxyl
radical-generating system, attempts were made to isolate and
identify related benzyl derivatives (benzyl alcohol and benzoic
acid). A systematic investigation of SPE adsorbents was
undertaken, using octadecyl (C18), octyl (C8), cyano (CN), and
phenyl packings in 3 mL cartridges (Supelco or Baker).
Reversed-phase packings (C18, C8, phenyl, and cyano) were
rinsed sequentially with water, 50% aqueous methanol, 100%
methanol, 50% methanol, and then water prior to addition of
2 mL of reaction mixture. After the reaction mixture was
loaded, the cartridges were rinsed sequentially with 2 mL of
water, with 2 mL of 5% methanol, and several times with 2
F igu r e 1. Liquid chromatograms of (top) aspartame (5.1 mM)
and benzaldehyde (45 µM) standard, (middle) reaction mixture
containing 0.05 M phosphate buffer, pH 2.0, 5.1 mM aspar-
tame, 0.16 mM CuSO4, 8.4 mM H2O2 and 5.0 mM ascorbic acid
incubated for 30 min at room temperature, and (bottom)
reaction mixture after extraction twice with 0.5 mL of cyclo-
hexane/mL of reaction mixture. Chromatographic conditions
on the Hewlett-Packard 1050 LC as described under Experi-
mental Procedures, detector response at 255 nm.
mL of 50% methanol. Each 2 mL eluent was analyzed by
HPLC for benzyl alcohol, benzaldehyde, benzoic acid, and
aspartame, using the C8 analytical column. The bulk of the
benzyl derivatives was eluted in the first 50% methanol
fraction.
RESULTS
Id en tifyin g Ben za ld eh yd e a s a P r od u ct of As-
p a r ta m e Degr a d a tion . Preliminary studies of the
ascorbic acid-hydrogen peroxide-Cu(II) free radical-
generating system in the presence of aspartame re-
vealed many unidentified peaks in the liquid chromato-
grams of the reaction mixtures. There was a noticeably
large peak under acidic reaction conditions with a
retention time close to that for aspartame that disap-
peared almost completely upon extraction of the reaction
mixture with nonpolar organic solvents (Figure 1). An
attempt to evaporate the organic solvent and dissolve
the extracted material in mobile phase resulted in loss
of the unidentified product, suggesting it was a volatile
compound. Even partial evaporation of organic solvent
resulted in a nearly linear loss of the product. Conse-
quently, solid phase extraction with octadecyl or octyl
SPE cartridges showed that a very nonpolar product

Benzaldehyde from Aspartame Free Radical Oxidation
J. Agric. Food Chem., Vol. 44, No. 11, 1996 3463
F igu r e 2. Rate of disappearance of reactants and formation
of benzaldehyde in reaction mixtures containing 0.05 M
phosphate buffer, pH 2.6, 5.0 mM aspartame, 0.16 mM CuSO4,
8.4 mM H2O2, and 5.0 mM ascorbic acid, incubated at 40 °C.
Ascorbic acid (9) and H2O2 (2) were measured as described in
the Experimental Section; aspartame (0) and benzaldehyde
(×) were determined chromatographically.
F igu r e 3. Dependence of benzaldehyde production on pH of
the medium. Each reaction mixture contained 0.05 M phos-
phate buffer at pH indicated; other conditions were as in
Figure 1, middle, except reaction mixtures were incubated for
15 min at 40 °C.
Ta ble 1. Ben za ld eh yd e P r od u ction fr om Asp a r ta m e
u n d er Va r yin g Rea ction Con d ition s
could be partially purified but always contained some
detectable impurities in the eluent containing this
product.
[benzaldehyde]
produced (µM)
additions to reaction mixture
complete reaction mixture
110
111
103
nd
nd
120
66
Since this product was found to be volatile, attempts
to collect it in concentrated form by distillation of the
aqueous reaction mixture were successful. A highly
concentrated solution of the unknown product of interest
was obtained in the 85 °C distillate, devoid of noticeable
impurities. This distillate was analyzed by GC-MS,
and the single peak (m/z > 50) gave a mass spectrum
that closely matched that of benzaldehyde. A standard
solution of benzaldehyde was found to give a peak in
the liquid chromatogram with the same retention time
as that of the unknown peak of interest in the free
radical reaction mixture containing aspartame. A stan-
dard addition of benzaldehyde to the reaction mixture
further indicated the unknown peak of interest was
benzaldehyde. The UV absorbance spectrum of the
peak, using a diode array detector, indicated benzalde-
hyde was the product.
Op tim u m Con d ition s for Ben za ld eh yd e P r od u c-
tion . A time dependence study for the reaction (Figure
2) indicates there is a rapid phase for benzaldehyde
production at pH 2.6 that is complete in about 30
min at room temperature or within 15 min at 40 °C.
There is a slower continuous production of benzaldehyde
and consequent disappearance of aspartame over sev-
eral hours, although this latter phase accounts for a
relatively small amount of the overall reaction. Sub-
sequently, reaction mixtures stood for 15 min at 40 °C
or for 30 min at room temperature for analysis of
products as reaction conditions varied. The initial rapid
phase of the reaction is complete before the concentra-
tions of ascorbic acid and hydrogen peroxide are com-
pletely diminished. The H2O2 and ascorbic acid remain-
ing in the reaction mixture continue to decrease at a
relatively slow rate (Figure 2). This may be due to
chelation of the Cu(II) ion by one of the products (see
Discussion).
anaerobic conditions
+ 32 mM ethanol
+ 500 µM DTPA
+ 200 µM Na2EDTA
+ 500 µM glycine
in 0.05 M chloroacetate buffer, pH 2.3a
in 0.05 M glycine buffer, pH 2.1a
with 4.6 mM Asp-Phe instead
of aspartameb
30
166
with 5.2 mM Phe instead of aspartameb
-H2O2, aerobicc
61
18
nd
15
nd
nd
nd
6
-H2O2, anaerobicc
no metals addedd
+ 880 µM desferrioxamine
+ 500 µM DTPA
+ 500 µM Na2EDTA
+ 500 µM Glycine
with varying [Cu(II)] addede
+ 16 µM Cu(II)
52
82
98
120
+ 48 µM Cu(II)
+ 96 µM Cu(II)
+ 316 µM Cu(II)
with added Fe(III) in place of Cu(II)d
67.5 µM Fe2(SO4)3; [Fe(III)] ) 135 µM
+ 200 µM Na2EDTA
18
nd
16
9
+ 32 mM ethanol
in 0.05 M chloroacetate buffer, pH 2.3a
a
Each reaction mixture contained 0.05 M phosphate buffer, pH
2.3 (except those with footnote a), 5.0 mM aspartame (except those
with footnote b), 8 mM H2O2 (except those with footnote c), 168
µM CuSO4 (except those with footnotes d and e), and 5 mM
ascorbic acid. Reaction mixtures were incubated in a sealed vial
for 15 min at 40 °C prior to removal of an aliquot for direct
injection in the liquid chromatograph. All reactions performed in
duplicate or triplicate. nd ) not detectable.
Under acidic conditions, there was a much greater
yield of benzaldehyde when phosphoric acid/phosphate
buffer was used compared to chloroacetic acid or glycine
buffers (Table 1). The latter buffers may act as radical
scavengers or bind the Cu(II) ion, inactivating or chang-
ing the relative reactivity of this catalyst for free radical
production. Phosphate and glycine buffers were pre-
pared with constant ionic strength (I ) 0.10 M with
added NaClO4) in the acidic pH range and found to give
the same yield of benzaldehyde as those without added
NaClO4 at any given pH (data not shown), indicating
the decrease in yield of this product with increasing pH
was not due to a nonspecific ion effect.
The production of benzaldehyde from aspartame in
the free radical-generating system showed a strong
dependence on pH and buffer composition. There was
a sharp decrease in yield of benzaldehyde as pH of the
reaction mixture increased from 2 to 3 (see Figure 3).
Only trace amounts of benzaldehyde were produced in
the neutral pH range, with a broad, late eluting peak
(unidentified) appearing in the liquid chromatograms
in the higher pH range.

3464 J. Agric. Food Chem., Vol. 44, No. 11, 1996
Lawrence and Yuan
mixture, there was a significant decrease (-76%) in
benzaldehyde production (Table 1).
The addition of 32 mM ethanol to the reaction mixture
had no effect on benzaldehyde production. Ethanol is
a competitive OH• scavenger that would be expected to
diminish products of OH• attack at this concentration
in the reaction mixture. The lack of inhibition of
benzaldehyde production by ethanol suggests a mech-
anism that may involve an aspartame-Cu ion complex
that undergoes an intramolecular, site-specific attack
on aspartame.
DISCUSSION
The present study shows that benzaldehyde is among
the products of aspartame degradation in the presence
of ascorbic acid autoxidation, catalyzed by Cu(II) or
Fe(III) under acidic conditions. Addition of hydrogen
peroxide to the system augments the production of
benzaldehyde, whereas metal ion-chelating agents
strongly inhibit the reaction. We have not found any
report of benzaldehyde as a decomposition product of
aspartame in the literature, and its identification as a
major product under these conditions was quite unex-
pected.
Buettner (1986) found complete oxidation of ascorbic
acid (0.1 mM) in 15 min in air-saturated solutions at
pH 7. All chelating agents used in that study resulted
in inhibition of the copper ion-catalyzed autoxidation
of ascorbic acid, although EDTA augmented the iron ion-
catalyzed reaction. Hsieh and Harris (1991) found that
aspartame would augment the copper ion-catalyzed
autoxidation of ascorbic acid, although they did not
report measuring any aspartame decomposition prod-
ucts. The present study has identified a new decom-
position product of aspartame in solutions containing
ascorbic acid under conditions that might prevail in
foods.
F igu r e 4. Dependence of benzaldehyde production on initial
ascorbic acid concentration. Conditions were the same as in
Figure 1, middle, except concentration of ascorbic acid was
varied. Reaction mixtures were incubated for 15 min at 40 °C.
There is a nonlinear dependence on initial hydrogen
peroxide concentration in the reaction mixture that may
be partially due to the limiting amount of ascorbic acid
present or to binding of the Cu(II) catalyst to products.
A significant amount of benzaldehyde was produced
when hydrogen peroxide was omitted (see Table 1). This
production of benzaldehyde in the absence of hydro-
gen peroxide could be eliminated by deaerating the
reaction mixture with nitrogen. This indicates hydro-
gen peroxide could be formed from ascorbic acid de-
pendent reduction of ambient oxygen in the solution.
Purging the complete reaction mixture with nitrogen
(i.e., when hydrogen peroxide was present) had little or
no effect on benzaldehyde production (see Table 1).
There is a complex dependence of the reaction on
ascorbic acid concentration (Figure 4). The yield of
benzaldehyde increases with increasing concentration
of ascorbic acid in the reaction mixture up to 5 mM but
then decreases with increasing concentration of ascorbic
acid in excess of 5 mM. This can be explained by the
fact that the concentration of aspartame in the reaction
mixture was 5 mM, and ascorbic acid can compete with
aspartame as a free radical scavenger in this system.
It should be noted that there was not detectable ben-
zaldehyde production when ascorbic acid was absent
from the reaction mixture, and hydrogen peroxide was
present, even after 24 h at room temperature. Although
Cu(II) catalyzes the disproportionation of H2O2, this rate
is extremely slow in acidic media. There was no
detectable loss of H2O2 in 3 h in 0.05 M phosphate buf-
fer at pH 2.6, in either the presence or absence of
aspartame.
When Cu(II) ion was omitted from the reaction
mixture, there was a dramatic decrease in benzaldehyde
production, but it was still measurable (Table 1). This
background production of benzaldehyde in the absence
of any added metal ion catalyst was likely due to the
presence of trace amounts of iron in deionized, distilled
water and reagents. When desferrioxamine, a strong
Fe(III) ion chelator, was added to the Cu(II)-deficient
reaction mixture, benzaldehyde was not detectable.
Addition of EDTA or DTPA, two other metal ion
chelators, also completely inhibited benzaldehyde pro-
duction, whether Cu(II) was added or omitted (Table
1). This indicates these latter chelating agents are very
effective at inhibiting both Fe(III) and Cu(II) ions from
catalyzing the title reaction. Addition of 500 µM glycine
to reaction mixtures containing 168 µM CuSO4 had no
significant effect on benzaldehyde production. However,
when 50 mM glycine was used to buffer the reaction
When ascorbic acid was omitted from reaction mix-
tures, there was no detectable formation of benzalde-
hyde from aspartame, even when hydrogen peroxide and
Cu(II) were present. Cu(II) is an effective catalyst for
the disproportionation of H2O2 by the following scheme
(Gutteridge and Wilkins, 1983):
Cu(II) + H2O2 f Cu(I) + HO2• + H+
(1)
(2)
(3)
(4)
-
HO2• f H+ + O2
Cu(II) + O2- f Cu(I) + O2
Cu(I) + H2O2 f Cu(II) + OH• + OH-
However, in this study it was found that there was no
detectable decomposition of H2O2 by Cu(II) ion in acidic
media over several hours. There was no detectable
production of benzaldehyde even after 24 h in the
absence of ascorbic acid.
The presence of ascorbic acid in the reaction mixture
results in reduction of Cu(II) to Cu(I) in the system (eq
5), which would facilitate reduction of H2O2 to OH• (or
Cu(OH)2+ or CuO+, vide infra) by a copper-catalyzed
Fenton reaction (eq 4).
Cu(II) + H2Asc f Cu(I) + HAsc•
(5)
It is likely that the Cu(II) present in the reaction
mixtures is coordinated to aspartame, which may alter

Benzaldehyde from Aspartame Free Radical Oxidation
J. Agric. Food Chem., Vol. 44, No. 11, 1996 3465
the rates and specificities of these reactions relative to
the aquated Cu(II) ion. A Cu(II)-aspartame complex
has been used for resolution of amino acid enantiomers
by capillary electrophoresis (Gozel et al., 1986) and
HPLC (Gilon et al., 1979). Cu(II) readily forms 1:1 and
1:2 complexes with aspartame, with the 1:1 complex
predominating below pH 4 (Aihara et al., 1992). It
appears that the coordination of Cu(II) results in a
ligand-directed, site-specific free radical attack on the
aspartame, since high concentrations of ethanol, a
noncoordinating OH• scavenger, had no effect on ben-
zaldehyde production. Furthermore, the effect of vary-
ing buffer ions (phosphate, chloroacetate, and glycine,
Table 1) shows that stronger coordinating ligands
inhibit benzaldehyde production.
Gutteridge and Wilkins (1983) showed that proteins,
in general, inhibit Cu(II)-mediated (but not Fe(III)-
mediated) OH• generation from H2O2 disproportionation
and subsequent degradation of deoxyribose. However,
the Cu(II) ions appeared to bind to the proteins and
generate OH• in a site-specific attack on the protein
(sparing the deoxyribose). Czapski et al. (1983) have
shown a similar ascorbic acid (or superoxide) dependent,
Cu(II)-catalyzed site-specific damage to proteins in the
presence of H2O2, which is not inhibited by hydroxyl
radical scavengers. The nature of the oxidizing radical
species in these systems has been proposed to be a Cu-
(II)‚OH• species rather than free OH• for reactions
involving Cu(I) ion with H2O2. The pKa for the probable
Cu(OH)2+ has been estimated to be less than 3.5, from
the product yield observed for the Cu(II)-H2O2 reaction
with methanol (J ohnson et al., 1985). There is a similar
decrease in benzaldehyde yield in the pH range of 2-3.5,
indicating a pKa near 2.5 for one of the reactive species
(possibly a Cu-aspartame-active oxygen intermediate).
Snook and Hamilton (1974) have shown that benzal-
dehyde is the major product of OH• attack on 1-phen-
ylalkanols (PhCHOHR) when R was isopropyl or tert-
butyl, indicating the radical-stabilizing effect of the
substituent alkyl groups favored cleavage, whereas a
methyl group favored oxidation to acetophenone. OH•
attack on dipeptides results in H• abstraction from the
C-H bond adjacent to the amide nitrogen rather than
near the protonated amino terminal (Taniguchi et al.,
1970).
We propose a similar mechanism, whereby a Cu(II)-
OH complex attack on aspartame results in H• abstrac-
tion from the R-CH of the phenylalanyl moiety followed
by scission to yield the benzyl radical and an oxidized
peptide backbone (aspartyldehydroglycine methyl ester,
ADGME), which is highly stabilized by a conjugated
π-system (Figure 5). There is a slow, continual forma-
tion of benzaldehyde over several hours, with conse-
quent decrease in the three major reactants after the
initial rapid burst of product. It was initially suspected
that ascorbic acid was completely diminished after the
initial burst of product, and excess hydrogen peroxide
was accounting for the slow phase of the reaction.
However, lack of any benzaldehyde production in the
absence of ascorbic acid suggested this was not so.
Measurement of ascorbic acid in the reaction mixture
indicated there was still a significant amount of ascorbic
acid left when the initial rapid phase of the reaction was
over. Consequently, chelation of Cu(II) ion by one of
the products of the reaction (ADGME?) is proposed to
be the cause of this slowing of the reaction.
F igu r e 5. Proposed reaction scheme.
the yield with aspartame. This may be due to less
π-stabilization of the nonaromatic product in the phen-
ylalanine reaction. Asp-Phe augmented the production
of benzaldehyde (+51%); the nonaromatic product of this
scavenger would have greater π-stabilization than as-
partame due to increased resonance in the carboxylate
group. Addition of H2O to the benzyl radical yields
benzyl alcohol, which would be rapidly oxidized by
Cu(II) to benzaldehyde under the reaction conditions.
A concerted mechanism for scission and oxidation of the
benzyl radical may be implied, since coordination of the
Cu(II) catalyst by the nonaromatic product has been
invoked. Benzaldehyde is not readily oxidized to ben-
zoic acid by Cu(II) (March, 1968).
Walling (1975) has proposed a phenyl (ring-centered)
radical cation in equilibrium with the hydroxycyclo-
hexadienyl radical in acidic media, in the case of OH•
attack on phenylacetic acid. It was suggested the
phenyl radical cation intermediate can undergo decar-
boxylation to yield benzyl radical, which is oxidized to
benzyl alcohol and ultimately benzaldehyde in the
presence of Cu(II) ion. However, the hydroxycyclo-
hexadienyl radical intermediate can undergo oxidation
to yield the corresponding phenol in the presence of
Cu(II) and O2. Phenolic products from OH• attack on
aspartame have been identified under neutral condi-
tions (unpublished results). The pH of the medium
strongly influences the distribution of products in this
reaction, i.e., benzaldehyde vs phenolic products. There
appears to be a decrease in yield of phenolic products
below pH 3 (in conjunction with the increased benzal-
dehyde yield).
Attempts to measure benzyl alcohol and benzoic acid
in the reaction mixtures showed there was little if any
of these products formed. There was no benzyl alcohol
detected in the GC-MS analysis of the distilled reaction
mixture. The octyl (C8) column described in the Ex-
perimental Section gave good resolution of these benzyl
derivatives and aspartame, but chromatograms of the
When phenylalanine replaced aspartame as the scav-
enger, the yield of benzaldehyde was only about 55% of

3466 J. Agric. Food Chem., Vol. 44, No. 11, 1996
Lawrence and Yuan
(II)-aspartame support electrolyte. Anal. Chem. 1987, 59,
44-49.
Graselli, J . G. CRC Atlas of Spectral Data and Physical
Constants for Organic Compounds; CRC Press: Cleveland,
OH, 1973.
Gutteridge, J . M. C.; Wilkins, S. Copper-dependent hydroxyl
radical damage to ascorbic acid: Formation of a thiobar-
bituric acid-reactive product. FEBS Lett. 1982, 137, 327-
330.
Ho, A. H. L.; Puri, A.; Sugden, J . K. Effect of sweetening agents
on the light stability of aqueous solutions of L-ascorbic acid.
Int. J . Pharm. 1994, 107, 199-203.
Homler, B. E. Properties and stability of aspartame. Food
Technol. 1984, 38 (J uly), 50-55.
Hsieh, Y. P.; Harris, N. D. Oxidation of ascorbic acid in copper-
catalyzed sucrose solutions. J . Food Sci. 1987, 52, 1384-1386.
Hsieh, Y. P.; Harris, N. D. Destructive effect of aspartame on
ascorbic acid in Cu-catalyzed solutions. J . Food Sci. 1991,
56, 14-16.
Hussein, M. M.; D’Amelia, R. P.; Manz, A. L.; J acin, H.; Chen,
W.-T. C. Determination of reactivity of aspartame with
flavor aldehydes by gas chromatography, HPLC and GPC.
J . Food Sci. 1984, 49, 520-524.
J ohnson, G. R. A.; Nazhat, N. B.; Saadalla-Nazhat, R. A.
Reaction of aquocopper(I) ion with hydrogen peroxide:
evidence against hydroxyl free radical formation. J . Chem.
Soc., Chem. Commun. 1985, 407-408.
March, J . Advanced Organic Chemistry; McGraw-Hill Co.:
New York, 1968; p 863.
Mazur, R. H. Aspartame - a sweet surprise. J . Toxicol. Environ.
Health 1976, 2, 243-249.
Snook, M. E.; Hamilton, G. A. Oxidation and fragmentation
of some phenyl-substituted alcohols and ethers by peroxy-
disulfate and Fenton's reagent. J . Am. Chem. Soc. 1974, 96,
860-869.
Stamp, J . A.; Labuza, T. P. Mass spectrometric determination
of aspartame decomposition products. Evidence for â-isomer
formation in solution. J . Food Addit. Contam. 1989, 6, 397-
414.
Stegink, L. D.; Filer, L. J ., J r. Aspartame: Physiology and
Biochemistry; Marcel Dekker: New York, 1984; Vol. 12.
Taniguchi, H.; Hatano, H.; Hasegawa, H.; Maruyama, T. Free-
radical intermediates in the reaction of the hydroxyl radical
with amino acid derivatives and related compounds. J . Phys.
Chem. 1970, 74, 3063-3065.
Tsang, W.-S.; Clarke, M. A.; Parrish, F. W. Determination of
aspartame and its breakdown products in soft drinks by
reverse-phase chrmoatography with UV detection. J . Agric.
Food Chem. 1985, 33, 734-738.
Walling, C. Fenton’s reagent revisited. Acc. Chem. Res. 1975,
8, 125-131.
reaction mixtures monitored at 220 nm contained many
peaks, and trace amounts of benzyl alcohol and benzoic
acid could have been obscured by other peaks. It was
estimated by standard addition that as little as 30 µM
benzyl alcohol or benzoic acid could be detected in these
reaction mixtures, but this level was not observed. It
should be noted that UV detection for benzaldehyde is
possible at 254 nm with a limit of detection near 1 µM,
whereas benzyl alcohol has no significant absorption at
wavelengths > 225 nm, and benzoic acid absorbs only
weakly above 230 nm.
This study shows that aspartame, in the presence of
ascorbic acid and a transition metal catalyst, such as
Cu(II) or Fe(III), under aerobic conditions can produce
benzaldehyde via a free radical attack on the aspartame.
Although benzaldehyde is a commonly used flavoring
agent (almond flavoring), and the title reaction would
have little or no significant impact on public health,
these results show that ascorbic acid, in the presence
of trace amounts of metal catalysts, can initiate some
very interesting chemical reactions in commonly used
food additives.
ACKNOWLEDGMENT
G.D.L. thanks Long Island University for release time
from teaching duties to perform this research.
LITERATURE CITED
Aihara, M.; Tanaka, R.; Fujimoto, M.; Takehara, K. Speciation
studies of some metal complexes with L-aspartyl-L-phen-
ylalanine methyl ester by potentiometry. Anal. Scis. 1992,
8, 755-759.
Buettner, G. R. Ascorbate autoxidation in the presence of iron
and copper chelates. Free Radical Res. Commun. 1986, 1,
349-353.
Czapski, G.; Aronovitch, J .; Samuni, A.; Chevion, M. The
sensitization of the toxicity of superoxide and vitamin C by
copper and iron: a site specific mechanism. In Oxy Radicals
and Their Scavenger Systems. Vol. I. Molecular Aspects;
Cohen, G., Greenwald, R. A., Eds.; Elsevier Biomedical:
New York, 1983; pp 111-115.
Ellis, J . D.; Sykes, A. G. Kinetic studies on vanadium(II)-
titanium(IV) and titanium(III)-vanadium(IV) redox reac-
tions in aqueous solutions. J . Chem. Soc., Dalton Trans.
1973, 537-543.
Gardner, L. K.; Lawrence, G. D. Benzene production from
decarboxylation of benzoic acid in the presence of ascorbic
acid and a transition metal catalyst. J . Agric. Food Chem.
1993, 40, 693-695.
Gilon, C.; Leshem, R.; Tapuhi, Y.; Grushka, E. Reversed phase
chromatographic resolution of amino acid enantiomers with
metal-aspartame eluants. J . Am. Chem. Soc. 1979, 101,
7612-7613.
Gozel, P.; Gassmann, E.; Michelsen, H.; Zare, R. N. Electro-
kinetic resolution of amino acid enantiomers with copper-
Received for review February 2, 1996. Revised manuscript
received August 2, 1996. Accepted August 20, 1996.X
J F960079K
X Abstract published in Advance ACS Abstracts, Oc-
tober 1, 1996.