Oxidation in Solution

1. Oxidative Reactions
1.1. Oxidation Mechanisms
l.2 Quantitative Considerations in Oxidations in Two Phases
l. 3 Simple Oxidation in Solurion in a Closed System
L4 Oxidation in an Open Systern
., 1.5 Oxidative Autocaralysis
-· More Cornplex Models
2.1 Stabiliry Prediction by Fractional Lives
Antioxidants
A
"T • Other Work
Re tcrences

1. OXIDATIYE REACTIO:SS

Tbe 1987 Stability Guidelines state that "It is also suggested that
the following conditions be evaluated in stability studies on solutions or
suspensions of the bulk drug substances .... High oxygen atmosphere,"
Oxidation reactions are relatively rare in pharmaceutical dosage
forros as a main reaction. Sorne oxidation probably takes place in many
cases and rcsults in small amounts of unidentified degradation producís.
When an oxidation reaction is one of the rnain reactions, then it is a

122

~ .. "' _ ...... - --
 123
Oxidation in Solution

serious matter, and formulating around such a problem can be di.fñ.cult.

Notable examples are líquíd vitamin A products (alth~ugh, srríctly
speaking, they are not solutions). Other examples will be cired shortly.

1. 1 OXIDATION MECHA~IS,tS

Oxidation, as the word implies, is an interaction between drug

substance, A, and oxygen, and the net reaction would be

A+ Oi-> products (4.1)

However oxidation reactions are usuall v the sum of a senes of reactions

(at times chain reactions) and these ~ start with one particular reaction
(the initiation reaction) which usually does not involve molecular
oxygen. Frequently oxidations are catalyzed by metal ions M T+; an
example of this is the oxidation mechanisms reponed for captopril
(Wallace and Schriesheim 1962; Wallace and Schriesheirn; 1963, Wallace
et al. 1965; Timmins et al. 1982; Lee and Notari, 1987). Captopril contains
a thiol group and will be symbolized as ASH below:
2ASH + 2M++-> 2{AS·} + ::!H+ + 2~1+ (4.2)
2{AS·}-> AS·SA (4.3)
2M+ + 02 -> 2M++- + 022- (,-..1,...~;1\
()z2- + 2H+ -> H20 + 0.5 02 (4.5)

where {·} denotes free radical. Hence, the overall reaction is:

2ASH + 0.5 Ch -> AS·AS + H20 {4.6)

It is noted that there is no net consumption of metal ion. (This larter has
been chosen as being divalent in the above, but could have other
valences, depending on metal in question). There is, however a
consumption of oxygen (otherwise there would be no oxidation). U~o
exhaustion of oxygen, the oxidation would cease. n
. If o~ygen is abundant, then, of course, all the drug can. decompose,
and m th1s case, as shall seen below, the kinetics can be simply first
order, since [02] becomes (vinually) constant.
Whcn captopril decomposes without presence of metals, it
undergoes an aurooxidation:
2ASH <-> 2AS- +2 H+
2As-+ 202-> 2{AS·} + 2{02·}- (4.7)
2As- + 2{()z·} -> 2{AS·} + 2()z2- (4.8)
2 {AS·} -> ASSA (4~9)
(4.10)
 J.
Chapter - 4
124 -
1
Oi2- + 2H+ -> H20 + 0.5 02
·(4.11)· i
1

So the total reaction is:

2ASH + 0.5 02 -> ASSA + H20 (4.12)

Most oxidations in pharmaceutics occur in aqueous solution, and are a

function of the oxygen dissolved in the aqueous phase. In practice it is
desired to minimize the decomposition and it is a practice to remove
oxygen from such systems by nitrogen flooding (Brown and L~eson,
1969). Depending on the method used for the removal there can still be
quite a bit of oxygen left in solution. Simple flooding is not effective and
to completely remove oxygen by boiling is rather difficult, and since
oxygen has a low molecular weight,' it can still be left present in
sufficient (molar) excess to allow considerable oxidation. Hence, to free
solutions of oxygen, nitrogen should be bubbled through them (the very
best way of driving out dissolved oxygen). The use of antioxidants
(bisulfite, ascorbic acid) is advocated where possible, in the case of
oxygen sensitive drug substances in solution.
In practice, complexing the heavy metals (e.g. by use of etbylene
diamine tetraacetic acid) is often employed, since, as seen in the
reactions sequences shown, metal ions are deleterious to drug stabili ty
in oxidative situations.

1.2 QUANTITATIVE CONSIDERATIONS IN OXIDATIONS IN

T\VO PHASES
.
Most · ox.idation problems in the pharmaceutical sciences occurs in
s~·stems. (Iiquid ~r solid) which are in contact with a gas phase . (usually
air) which cont~ms. oxygen. As far as the methodology for following a
~ecay, th~ monitonng ~f parent compound and decomposition products
ts what · 1~ usually car:1ed ~ut. If it is necessary to monitor oxygen
concentranons, then Wmkler s · method (Winkler, 1888, Novaczyk et al
1993) may be used, although it is cumbersome. Oxygen electrodes ma '
~so _be used, but ~bese are only good below 45ºC. Finally, potentiometri~
trtran,onbs are possi le. For work in the gas phase • Ram an spectroscopy

_ ~a; e used. Most of what follows will concentrate on the

disappearance of the parent drug.
Henry's law states, in the case of oxygen/water systems, that

Po2 = K' Xo2 • K*C

(4.13)
. where C is molar concentration and X 1 · ·
d K mo e fraction of dissolved oxygen,
an . and K* are Henry's law constants. F di ...
or or mary atmospheric
 12S·
Oxidation in Solution

·· f K' e o ygen is 3.3!~

couditions. p02 =
0.22 atm. At 25°C the value. o ior x Id be
l O 7 atm, i.e. the mole fraction of oxygen d1ssolved water . wou .
0.22/(3.3 107) = 6.6 10-9. The molar concentration correspondmg to this
is given by

Xo2 = Co21'[55.5+Co2] = Coi/55.5

i.e. Co2 is about 3.7• l Q-7 molar. Equation 4.13 is often expressed in
inverse forro, 1.e.
( 4.14)
C=KP

will be u sed in the

where K = l(K*, and both presentation forms
following.

\ 1

Gaseous .
\ Oxygen Vapor Pressure = P, Vol u me=
'

V1

Dissolved Oxygen Ooncentration = C, Volume = V2

Oxygen .~ \ " :

FIGURE 4.1 Nomene l at ure use d reor kinetic schemes, -

1

'
1 If a solution of a drug substanc_e (A) is oxygen sensitive, then

A+üi-~B
. . ·. (4.15)
where B represents oxidation · products · In t h e fiirst ca . ·
assume d th at the system is closed (Fig 4 l) h se cons1dered, it is
e m 3' an~ that the liquid volume is V2 c~3 .' n.:
at the head space is V 1
oxygen in the ensemble is the number of mol e tot~ number of moles of
the number of moles, n2, dissolved. es, nj m the gas phase plus

.-
 ·11
l
Chapter .4 .· ·
'·
.
1
~

The gas law gives the value for ni:

(4.16)
\ ,; \ .\ ('

wbere p in the following is u~&erstood to be the oxygen pressure in the

atmosphere.
(4.17) .

so that the total number of moles, N, is given by

(4.18)

lntroducing Eqs. 4.13 and 4.16 gives

(4.19)

The oxidation is either (a) simple or (b) catalytic or (e)

autocatalytic.

1. 3 SI!\1PLE OXIDATION IN SOLUTION IN A CLOSED SYSTEM

The simple case will be treated first (Franchini et al., 1993. 1994 ).
lt fo1lows from the general second order reaction equation that

d[A]/dt = - k2[AJ C (4.20)

where [AJ is concentration of drug, k2 is the second order rate constant

and C is the oxygen concentration in solution at time t. If the inicial
concenrrations are denoted by subscript zero, and the concentration of
decomposition product is Y molar at time t, then

{A}= Ao-Y
(4.21)
and
d[A] • -dY
(4.22)
so that now

dY/dt ª k2f Ao - Y] C • k2Aofl - g·Y] C

(4.23)
where ·
 12 7 ,
Oxidatinn in Solution

(4.24)
g = 1/A0

The number of moles decomposed is V 2 Y so that the total number

of oxygen molecules, N(t), at time t is given by:

( 4.25)
~
~ It is assumed that the distribution ot oxygen between head space and

•
liquid is rapid, so that at time t
• •
-t C = K P(t) ( 4.26)

~ where P(t) is the oxygen pressure remaming at time t. Mass-balancing

•
~- of total oxygen gives:

t .
t Introducing 'Eq, 4.26 into this gives:
t
t

• Solving for C gives the expression ,

•
C = q[l - aY] ( 4.29)

•
where

q = N0/(V2+(V1/RTK)} (4.30)

•
•• a= V2f(V2+(V 1/RTK)} ( 4.31)

:a
and

Introducing this into Eq. 4.23 grves

dY/dt = q·k2A0[1 - gY][l - aY] (4.32)

This equation c~ be solved (in closed form as well as by graphical

integration) and gives curves of the_ t:~e shown in Fig. 4.2. Examples of
this type · of curve are · the . ox1~anon of promethazine at pH 3 .2

• (Underberg, 1978)_ and the oxidation of propildazine (Ventura et al

A0 is large and fairly constant during the reactíon, then

1981). This latter is "shown in Fig. 4.3. ·'
 If

•"
-, ·t-,

•
(4.33)

',.
:~-~·-~~--.A--.'·"·- .,. . .. · . ,._,_" , . . , . ~...... · ---------·-·-· ... ··~-- .. --------

•
•
•
"'
12 8-. Chapter 4

lntroducing Eq. 4.29 then gives:

dY/dt= k2Aoq[1 - a Y] ( 4.34)

This can be rewritten

(4.35)

which integrates to •

ln[(l/a) _ Y] = -k2A0qat + ln[(l/a)] (4.36)

If a is ciose to unity, then this pre d i. cts first order kinetics, as is _f~r

instance the case in work reporte d b y Kassam aentd alL. ee(s1o9n72) (1f9or69)a.scorTthnec
acid Wan and Hwang (1969) and by Brown
. 'results ~ of the 'Iatter being shown in Fig. 4.4.

100

80

>
...•..
-
•
o
.o....

20..L.-_.. ,, ,,._..,... ..----.---,

o - 20 40
Time

FIG{.;RE 4.2 Profile of [1-Y] versus t according to Eq. 4.32.

It should finally be added that there are cases (which will be

discussed in Chapter 7), wherc oxidation leads to zero order profiles
(Franchini and Carstensen, 1995). These are cases where the drug is
amphiphilic and where one of the non-micellar n-mers (An) oxidizes. In
.this case therc is an excess of oxygen, and the concentration of the
, reacting species, An, is ·constant (since it will be replenished by , the
micellar species ) .
.. _
 ~
'<\

Oxidation in Solution 129

e:
o
·.;:: O Aerobic
~ 0.25 o Anaerobic
~
u,

o 1
~ : as ¡ sp hs ¡ 100

0.00 "1'---.---.,.......~...-.---.------..¡..,..,....,.......... ,

"": Time(h~) l. :
, b/ ·2 <,, !
L
_,'.,< ~ .Ó
FIGURE 4.3 fl
bxidation of propildazine. Graph constructed from data
by Venture et al., (1981 ).

3.0

~ ¡e;-;-¡
~ -~--

•
•
~
1.0 -+--....--.----.--or------.-.....--..,.-...--
0 2 4

•
Hours

•
- '•,.
•
t

t
- .,
t
F tem layered wíth nitrogen and . (B) an unprotected system (which
I extrapolates to zero time at 21 % oxy gen as it should), . Graphs
G constructed from data published by Brown and Leeson (1969). The Ieast
U =
squares fit lines _are: A: y 2.71 - .0.067x and B: y = 3.24 ; 0.36x.
R
E

4
.
4
H
e
a
d
s
p
a
c
e
o
x
y
g
e
n
c
o
n
t
e
n
t
v
e
r
s
u
s
ti
m
e
i
n
(
A
)
a
s
y
s
 1
j

Chapter ~:,4 · :
. 13'0-.

Cases where the oxidation is photochemically driven will often

· become zero order. An example of this is the work by Asker al. et
. (1985), whose data are" shown in Fig.' 4.5. It should be noted that the
data are probably more suitably considered S-shaped curves, · such . as
shall be discussed shortly.

100

"O
' 'ü
<(.)
75
:e....
' . ·~ o
o
(/'J
''

50
<
-e:
Q)
No Methionine

....
() 25
Q)
a,
,} ·O
... t' o .2 4. , s a l o 12
'
'
· Exposure Time (Hours) .

FIGURE 4.5 Photolytic oxidation of ascorbic acid solutions Graph

const:uc.ted from data by Asker et al. (1985). Least squares fit: · 10 mcer9o
methionine: y = 99.38 - 7.64x (R = 0.97) and no methionine· , = 104.19
- 4.67 (R=0.95). . )

1.4 OXID~ TION IN AN OPEN SYSTEM

\ ~ - 4 o~ . ~
• 'I 1. 4. i ~-·"" ,;..r.- 'k,. ,

. . It is always wórthwhile determining k2 in a se ar . .

m an open system. This is done by bubblin ox p ate expenment
solution of drug, so that e is the saturation cg ygen. through an aqueous
at all times: . oncentrat1on of oxygen (Csat)

dNdt = -k2[A}Csat = - k2[A]KP

(4.37)
,. Since the amount of A present is the . ori inal
. th~ ~ amount decomposed, Y*' it follows that: amount of drug, Ao * less
'.,.
Oxidation in Solution 131 "

dY*/dt = k2[Ao * - Y*]KP ( 4.3 8)

r
·'
-lt is noted here that the A* and Y* values -are V2 times concentrauons.
Eq. ·4.40 integrates to ·

ln[A * - Y*] = -(k2KP) t + ln[A0 *] (4.39)

or

ln[l - (Y*/ Ao *)] = ln[fraction retained] = .; (k2KP) t (4.40)

where the initial condition that X =

O at t = O has been invoked. Eq.
4.39 , predicts that (a) the reaction is first order and (b) that the
observed rate constant is:

( 4.41)
,/
where e is the (now virtually constant) oxygen concentration in the
aqueous phase. The situation described holdsc '."for ~y~ iituation where
the oxygen concentration changes so little.: that it becomes almost
constant and the reaction is pseudo first arder. '>11~ ,, ,- • 1.c.- ( ·< c.

e 1.0
o
·.;:

-....
e:
(])
o
0.5

O.O
e:
o
(J

~-
Q)

:J
--0.5

-1.0
O 25 mUséc

-o
(1)
C] >30 mUsac/

.5 ·1.5

-2.0
o 10 20
Minutes

\
-~
t. '1
\ •.
~\) \'i. '

FIGURE 4.6 . Sulfite oxidation as a function of rate of bubbling. The

=
least squares fit lines are: 25 mg/mL: Y 0.75 - 0.13 x and >30 mg/mL:
y = 0.734 - 0.19 x. Graph constructed from data by Schroeter (1963).
132 Chapter 4

It has tacitly been assumed that the equilibrium between gaseous

and dissolved oxygen is instantaneous, but ar low agitation it rnay be
diffusion controlled. A good example of this is the oxidation of sulfites
reponed by Schroeter (1963).
lt can
be seen from-the data in Fig. 4.6 that at the lowest bubbling
_ speed. . the oxygen s'upply' does .!!Q!_ keep up . with the decomposition. In
general, the observed rate constant should be a function of oxygen
concentration, which in this model is constant. It can, of course, be
varied by having different mixtures of oxygen and inert gas bubbling
through the mixture and in this case Eq. 4.41 becomes:

ln[kobs) = ln[C] + ln[k2] 1 )

(4.42) ., y
\ ,·Jt·••lv"
1,,1.>'~'h·~'·) 11. l '•.'

Plotting either ln[kobs] versus ln[C] or [kobsl versus. C should give straight
line plots. In the former case the slope would have to be unity. An
. example of this is (Fig. 4.7) the oxidation of ascorbic acid (Kassam et al.,
1972) (where, however, the slope is not unity).

~ 21
--.....
'·•
:c,1...
ti')
>-

--
C':
~
V
::::..
e
·-
_ •·
' 1 •
••w
~-1----.- .......--,.--..,.........,-_,.. ,. ,
e 1 2
ln[C], mg/L
. \ 1))-. .
. \ .f \ ..' j¡

-l L
FIGURE 4.7 Effect of oxygen concentration on oxiid ati·on rate, Figure
constructed from data by Kassem (1972). If p.lotted Iinearly the equation
=
for the trace would be: y -3.93 +0.41 x. It is shown parabolicall_\·; y =
-3.6 - 0.04 X + 0.125 x2.

1.5 OXIDATIVE AUTOCATALYSIS {

I' "-i
In auto~atalysis. the rate constant, k2, is a function of y . The rnost
all-encompassing equation of this type is the Ng-Equation, Eq·. 4
.43 . (Ng,
 133
Oxida tion in Solution
.
~ ,, • J

1975) which will be dealt with at another point. Eq.· 4.23 assumed that
the reaction rate was proportional to drug concenrra~ion ~d react~t
concentration. Where this is in deviance from expenmenta1 results ir
may be assumed that it is dependen! on powers of these, i.e.

dY /dt = k2 yn(l - Y)P (4.43)

In many cases, the exponents are simply unity, and hence Eq. 4.-+3
becomes:

( 4.44) .

where, here, Y denotes' fraction decomposed. Profiles of this type are

shown in Fig. 4.8. c~'·flº )~ r'.i' ·?

,_.,e.. . There is an initial apparent lag time. and then a precipitous drcp.
} Often, in real systems, it is difficult to duplicare the induction tir;q~ point,
.: t*, at which the drop commences. The FDA has often qiioted this
situation as one which makes the Agency leery about extrapolations,
because it is possible to get a good linear fit to the data at times t < t*,
• and these would extrapolate to high retention values, which in the case
cited, would be incorrect. Eq. 4.44 is rearranged to:
~

•
dY /(Y (1 - Y}) = k2t (4.45)

t To integrate Eq. 4.45 it is recalled that

• l/(Y(l - Y}) = (l/Y) + 1/(1 -- Y) ( 4.46)

• so that Eq. 4.45 becomes

•
dln[Y] - dln[l - Y] = k2t (4.47)
which integrates to:

•• ln[Y/(l - Y)] = k2(t - t¡)

(4.48)

--
",.'
where t¡ is the point in time where the inflection pcinr
may be expressed as:

- Y = exp[k2t]/{ l+exp[k2t]}
- occurs. Eq. -k48

,. 1

This type profile is comparable to a Hend

an inverse S. and it would have the a
· , ': ·
erson-Hasselbach curve. -i.e. 15
po~er funct~onal dependence of k on p~e~ance shown in Fig. 4.8.
15
If
(4.49)

~ . typical lag time curve will

generate.
~
~ ..... l ~
larger
than one,
then rhe
 Chapter 4
134 /
/
In the preamble to this section on oxidation, it was stated that
oxidation was not the _ order of the day in pharmaceutical system~/ This is
exemplified 'by the number of publications. In J.Pharm.Scl., e.g., for
the period 1965 to J 990 there are but sorne 30 publications on the .; ,
subject, and sev~rál"" of these do not déat~ 'with the .kinetics, but rather" i- ·
J

with the (very important) problem of proper assígnmént of degradation

products. Sorne of the articles in literature (beyond those quoted
already) are listed below. ·
Repta and Beltagy (1981), reported on the oxidation of ó-seleno
guanosine in aqueous solution.. Underberg (1978) reponed on the
oxidative degradation of prornethazine., ~- Duchene et al. (1986) have
reviewed the effect of cycledextrins complexes on drug oxidative
stability .. Szejtli et al. (1980) and Szejtli and Bollan (1980) demonstrated
that fer vitamin D3 the cyclodextrine complex reacted less rapidly with
oxygen. Swarbrick and Rhodes (1965) showecfP that the maximum
I oxidation rate of linoleic acid in aqueous systerns containing the surfac-
tant Brij 35 was a slightly decreasing, linear curve.
I
75

........
>.
..,....
1

..._. 50
o
.o
.,....
25
k = 0.1

o+----r-~.-....,...._,.~~...:;::::::;:::::::-.._
o 25 50 75 100
Time, t (arbitrary units)

FIG. URE 4.8 Trace of (Eq. 4 5"l) h f .

e.xp(-kt)/(l+exp{-kt}] for
· .t e
- unction
various values of k2. The profiles of the f
_for k = 0.01 and 0.005, d uncuon have been generated
an a starting value of 100.
•
.~

-.
.,.
~10RE COMPLEX ~10DELS

·. · . A . ·more . general . sem1· -empirical ·

.. autox1dauv~ reactions (invoking s equanon for the treatment of
. i s-. . orne power dependence of y and
..... (1-Y)
. t,
,,
 .
Oxidation in. Solution
6
135

ln (yn,:(l - Y)P} = -k'(t - ti) (4.50)

. A more manageable form of thís equancn is:

1n ( Ya./(1 - Y)} = -k(t - ti) ( 4.51)

w he r e
'•
!l. = '~' /p
!l. 1 • (4.52)

an d

q = n/p -

~ca
~
:J
:¡)
......
._, 30
c..

o 50
u
o
o
-
o
u
.....
40 ·- .... . . "'. ,

..o. 20
o
a 100. 200
: Time (days) ·
. . '

FI Gt.;RE 4.9 Decomposition of retinoic acid ar 37ºC. Graph constructed

from data reponed by Tan, Melzer and Lindenbaum (1993).

The problem exists as to how to treat Eq. ~.51 and obtain paramecers
from the untransformed Y versus t cquation. since a. and p and k are not
known. The reason for recasting Eq. -+.50 is to reduce the number of
iterants to one. La~gt numb~ of i~erants in ¡¡ W.º,~~l, are suip~ft:. because e-8-iJ
good but not quite ngorous first tstimates may li::u:t to seamda.n¡ minima arui ·t
l •
is always a good practice to reduce the iterants to as freo as possible. E q. w ,

4_5¡
, ),-'
_,. .....
" .
......,.. """''"'"
' -

.... 1--r' .....

. , ·~. .. '
---=
 Chapter 4
136

only requires one iterant, so that data treatment is not all that
questi onable .1 r' ~ •••• - ..

But even then is it necessarv to ha\;e a reasonable estimate. ~{ the

. . ir. •

iterant, q, before starting the iteration procedure. Such an estímate can

be obtained . (and -. in obtaining . the · estimare, often the iteration is
unnecessary). by the considerations to. follow. Consider .the data (Fig.
4.9) reported by Tan et al. (1993). .
Manual tri al ~
and error will often gi ve reasonable estimates of
.

iterants (in this case q), and the value q =

0.3, arrived at in a few tr1es,
gives good results, as shown in Fig. 4.1 O. This value could then be used
in computer iteration, as a first estímate.

...... 2
-X1

--
~
M
o
<x
..._. o
e

o 100 200 -, 300

,, Time (days)

FIGURE 4.10 _ · Treatment of dati reportee by Tan et

2.33 using q = 0.3. al. (1963) by Eq .

. ."
The least squares fit lines are: .
. •'
37ºC: ln[x0.3/(1. - x)]. = .::.i.192 + O.Ol6()¡
·e 4:54)
so~c: ln[x0.3/(1 - x)] = -0.881+ 0.0613t
(4.55)
7o:ic: ln[x0.3/(1 - x)] = -0.654+ 0.307t
(4.56)

1 Iteration wilh. more than one iteram alw . .

. nnnnnum obtamcd by iteration is . a) s )caves the questiou of
_of . this .. is shown in Chapter 3. a pnrnary or a secondary minirnum. whAenther the
example
Oxidation in Solution 137

When the slopes (the rate constants, in unns of day+) are plotted versus
absolute inverse temperature, a good Arrhenius fit is obtained as shown
in Fig. .+.11. It is observed that the activaticn energy is about 18
k Cal/mole, which is a reasonable figure.

-3

-4

-5-+-~--~--~---~---~--~--~-----.
2.9 3.0 3.1 3.2 3.3

1000/T

FIGL'RE 4.11 Arrhenius plot of data in Fig. 4.10.

2 .1 ST ABILITY PREDICTION BY FRACTIONAL LIVES

It has been shown in Chapter 2 that fer zero and first order
reactions it is possible to asse~'sr:.- and extrapolare · stability hi gh trJfu
temperarure data by use of fractional lives.
Tan et al. (1993) have extended this principie to oxidations, and
y oshioka et al. (1994) have extended it to =complex reactions -- following
neither of the more describable orders. They - consider reactions where
the .fracrion decomposed is a second degree {-l~~)'_D9!I!_i_ª1_ in time, t. and
(more frequently) a triple exponential formation, i.e .
.
x = Aexp(at) + Bexp(bt) + Cexp(ct) (4.57)
or

x ~a+ bt + ct2 (4.58)·

Gv ,..,,,J..,-3 -~
" :;i
;!''(_ 1 ...._.¡ c>--
I
When treating _ decomposition of projeins by ei th er _ of these
equations, they obtain best pararneters and héríce can calculate the point
in time where x = 0.1 (which is usually denoted t9o). Tuey th
en s h ow by
this empirical mode, that in the cases srudied -

. ( "'!.
138 Chapter 4

ln[t90] = Qff +q (4.59)

lt is noted that the two presentation modes .for the appearance of

x with time do not contain what is usually thought Óf as a rate constant.
Hence;··unliké~ tlie' ·treatment of zero and first order equations, there is no
way of "invoking" a ternperature dependence on k or t90. The method
has great pracrical value. ~ . . .
. To the contr ary, Tan et al. (1993) attempted to obtain an
explanation for the fact that fractional lives (from B = 0.1 to 0.9) give
linear plots, but they found that the lines change "activation energ y".
The argument put forth by Tan et al. (1993) is the following: the
decomposition reaction may be characterized by the differential
equation:

dx/0(x) = kdt
• (4.60)

where ó(x) is a function which is not "simple". If this is integrated (if it

can be integrated), then the integral is denoted F(x), i.e.

Fíx) = kt (4.61)

or

, F(B) = ktg (4.62)

where B is the fractional life. By now invoking that

~ = Zexp(--Ea/RT} ( 4. 6 3)

it follows that

F(B) = Z{ exp(--Ea!RT}t.s
(4.64)
o; .' : • .• • -e ¡ .. j •
• t .
1,,
(4.65).
)}rr, pro forma.. the .~ir>~ar-ity. of ln[ts] . versus u/n ~o Id .
,,+' r·?roven .. :. - l~ -shoul d b:~ noted that integrals of the t e d: /0(appear to be
VJ.
,
o~
.....
solved 1n• ·~., l:l'tuw;;i sucn as to h ave only one "k-value"
· · ,.,...~ - ·· .. ~· YP ~ · x) can rarelv•·
; If '·. on~ ~~'k(!S thc cxample by Yoshioka
• .l al. (1994.) et
polynomial in t, then Oüe would have to .express dx/d - where x is a
not or t. Here ·> I t as a function · of , x,
x =a+ bt + ct2
(4.66)
 139
Oxidation in Solution
\ 1

"·' ,.

lf this is differentiated so that dx/dt be .expressed as a function of X, .. not

of t.:' then

dx/dt = b + et. (4.67)

. . - . - . ., '

To. eliminate t, Eq. 4.66 is solved with respect to t. and

t = -b/2c ± [(b2/4c2) - {(a·- x)/c}] 11~

If
[(a - x)/b ]2 << (b2/4c2) - {(a - x)/c}
... ' "•

then
..
t = (a - x)/b (4. 70)

so

dx/dt = b + {(a - x)/b}c =A+ Bx (4.71)

~ ' •• ¡. 1.) •• ,. ~

where

., A = b[l+(a/c)] t. : ~; • r. . \ 4 . 72 )

and
..
"
.z .,
B = -c/b ~{ .'.':.,. • 1
··(4.73)
rJá ~ ~~¡./" .

It is noted that even with a fairly simple function (second arder

polynomial) it is not possible to obtain an equation of the type of Eq.
4.60. where there is a definitive "k". At best 'one rnight expect B to
follow an ,Arrhenius equation.
Henté the method only, has 't heo r ett cal . basis . if dx/0(x) 'can be
presented in · a form, and in .such ·a way that' the raie cons tant, . k, ·is· a
meaningful quantity. ~~~-t"°C>-'J< · · ·' •
.4.../7 : · · '· : .
The method, ne-térrJzeless, has gr eat -practicat importance b
· · · · ·
<it al l ows exrrapo l ations zn.,, a simp 1 e· manner, · . · .. • . · . - .• ec· a, u. . se

,, . ', ..
. ·' t•
rpiill'--

... '• ...

Chapter 4
140

»
'í ~ ¡ - ....

' - "'
. ..._,, í.'
2
......

. o
-2 i..-...i...........i.-"'---"---'--- .......---..

2.9 3.0 3.1 3.2 3.3 3.4

1000/T
•
FIGURE 4.12 Decomposition of a-chymotrypsin tablet. lnltcn] plotted in
Arrhenius fashion. The neat decomposition is not an integer arder.
Graph constructed from data published by Yoshioka et al. (1994).

1 '
The data bv · Yoshioka are shown in Fi1!. 4.12.. Again, the "activation
energies" are high, •b~t no theoretical importance sAould be placed on them.

3• A...~TIOXIDAl\75

Antioxidants work by consuming . oxygen at a faster rate than the

drug substance reacts with oxygen, and in such cases they will protect
the drug substance until they are completely used up.
This means the use. of antioxidants irnposes a lag time upon the
decomposition profile of the drug.
An example is the work by Pudapeddi (1992), where it was shown
that sulfites act in a quantitátive manner. o'~¿¿v\he sulfites are
consumed, the oxidation starts.
Antioxidants can also act by interfering in e.g. the reactions
- schemes shown in Eqs. 4.2-4.11 so that the , oxidative pathw ay is
,. interrupted. In · such a case they will, thernselves, be regenerated, • and
will function in a manner independent of elapsed ti~e. The most
common antioxidants used are ascorbic acid, BHA, BHT and sodium
sulfite. The use of ethvlene diamine tetraacetic acid as a chelator has
already been menfioried;
Chakrabarrí et al., (1993) have shown that hydroquinone,
butylated deoxycholate, · and ascorbyl palmitate stabilized hamycin (a
polyene · antifungal antibioticj.
Oxidation in Solution · 1 ..i 1

~. OTHER _\VO RK

As mentioned, until recenrly, the number of rep~r_ts on oxidation í

lll the pharmaceutical literature were scarce,1 ''out the gap is being filled. l.'

For instance, Bosca et al., (1992) have described the oxidatíve

decarboxylation of naproxen,/ and Bosca et al. (l 992b) have described
the photochemical byproducts. Vargas et al. (1992} have described the
photochemical oxidation in light of nifedipine. / Tereoka et al. (1993)
have described the oxidation of ranitidine in acerare buffer.j-
Franchini and Carstensen (1994) have studied the auto-oxidation
(in the presence of trace quanities of heavy metals) of the oxygen-
sens itive compound B-[2-(2-carboxyethyl)-thio-2]-[8-2-phenylacetyl 1-
benzenepropanoic acid. This is denoted A in the following.

o 0.6M60=c
~ 0.8 M 60;;C
l. o o.8 :\11s=c
D 1 M93=c
_, 6. 0.4 :-..193=c
-~
:.,) 0.8 Theory

-
,..:...,,) 0.6
(

e
o 0.4
-·
-
~

.
~ : .. ~ '.
0.2 -
-
.' .v •
r•

,
-
o.o ,.,
o 1
,..,
» . ...
Reduced Time. t/t0:5

• r

FIGURE 4.13 Graph constructed frorn data . published by. Franchini

and Carstensen(1994 ). In the legend Yl denotes molarity of phospháre
buffer. The pH employed was. 8~0 and the concenrration of drug was o.2
mg/mL.
. \,
\.Jf' .
~ 5'·''
The degradation. is an oxidation leading ro a . 1- id
· · c ac1id.
clllilam1 The mecham.sm that applies to the " d· su ro.x.i ··e . and a
. s:
 ecomposition i
• r •" -·

A+A->A2 with cquilibrium constant K

, --- - .·rr
.. ' _.. . i ,,)t,J .. ,. ...... ..... • • .. ,. . .... . .. ._ ,. _

1 ' .. . ... ... . .. - .. ...

 Chapter 4
142

and
with rate constant k2

The ensuing rate .bqu~tion is fourth order in Aobs·. The authors studied
the decomposition at · a series of buffer concentrauons and temperarures
and sorne of their data are shown in Fig. 4.13. They showed t~at the
reaction · was subject to kinetic salt effect, but that buffer catalysis was
absent.

REFERENCES

Asker, A.F., Canady, D. and Cobb, C., (1985), Drug. Dev.Ind. Pharm.,
11:2109.
Asker, A. F., and Larose, M., (1987), Drug. Dev. lnd. Pharm.,13:2239.
Bosca, F., Martinez-Manez, R., Miranda, M., Primo, J., Soto, J. and V ano, L.,
(1992a), J.Pharm.Sci., 81: 479.
Bosca, F., Miranda, M. and Vargas, F., (1992b), J. Pharm. Sci., 81:181.
Chakrabarti, P.K, Harindran, J., Saraf, P.G .• Wamburkar, M.N., (1993), D r u g
oe« lnd. Pharm.,19:2595.
Duchene, D., Vaution, C. and Glomot, F., (1986), Drug.Dev.lnd.Pharm.,
12:2193.
Franchini. M. and Carstensen, J.T., (1994), Int. J. Pharmaceutics, 111: 153.
Franchini, M., Unvala, H. and Carstensen, J.T., (1993).J.Pharm.Sci., 82:550
Franchini, 1\-1. and- Carstensen, J.T., (1995), J.Pharm.Sci., 83:Submitted.
Kassem, M.A., Kassem, A.A., Aromar, H.O., (1972), Pharm.Acta Helv.,
47:97.
Kassem, M.A .• Kassem, A.A.; Ammar, R.O., (1969), Pharm.Acta Helv
44: 667. . .,
!'\g, W.L .. (1975), Aust J. Chem., 28:1169.
Nowaczyk, F.J., Jr. Schnaar:, R.L., Ofner, C.~1.III and Wigent, R .. J.,(1993),
Pharm. Res., 10: 30.:i.
Pudipeddi, M., Alexander, K .• Parker, G .A. and Carstensen, J. T. 992) o
Drug Dev. Ind -. Pharm .., 18:2135. ' '
Repta, A. J. and Beltagy, Y. A., (1981), J. Pharm. Sci., 70:635.
Schroeter, L., (1963), J. Pharm. Sci., 52:559
Sw~~rick.J. and Rhodes, C. T., (1965), J. Pharm. Sci. 54·903
Sze~tl~. J. and Bollan, E., (1980), Starke, 32:386: ' · ·
Szejtli, J .. Bolla·Pusztai, E., Szabo p and F T
35: 779. ' · erenczy, ·• (1980), Pharmazie,
Tan, X, Meltzer, N. and Lindenbaum S (1993) J
Analysis. 11 (9): 817. ' ·• • · Pharm and Biomed.
Tereoka, R., Otsuka, M. and Matsuda y ( .
Underbera W. J M (1978) J Pha • .•. 1993), J.Pharm. Sci., 82:60,1.
º' ·• ·• • · rm.Sci.,67:1133 ..