Anda di halaman 1dari 11

(This is a sample cover image for this issue. The actual cover is not yet available at this time.

This article appeared in a journal published by Elsevier. The attached


copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy


Chemical Engineering Science 73 (2012) 2029

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Science


journal homepage: www.elsevier.com/locate/ces

Thermodynamic time invariances for dual kinetic experiments: Nonlinear


single reactions and more
D. Constales a,b,n, G.S. Yablonsky c,d, G.B. Marin b
a

Department of Mathematical Analysis, Ghent University, Galglaan 2, B-9000 Ghent, Belgium


Laboratory for Chemical Technology, Ghent University, Krijgslaan 281 (S5), B-9000 Ghent, Belgium
c
Parks College, Department of Chemistry, Saint Louis University, 3450 Lindell Blvd, Saint Louis, MO 63103, USA
d
The Langmuir Research Institute, 106 Crimson Oaks Court, Lake Saint Louis, MO 63367, USA
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 27 September 2011
Received in revised form
5 January 2012
Accepted 11 January 2012
Available online 21 January 2012

For dual kinetic experiments, i.e., experiments performed in batch or steady-state plug ow reactors
from reciprocal initial conditions, thermodynamic (space-)time invariances for all reversible single
reactions of the rst and second order have been found explicitly.
In all analyzed cases, quotient-like functions of concentrations can be dened which equal the
equilibrium constant of the reaction during the whole course of the experiment, and not only at the
end, i.e., under equilibrium conditions. The obtained invariances can be used as simple ngerprints for
distinguishing the types of reactions.
For multistep reactions: (a) a similar invariance was obtained for the two-step catalytic reaction
(single route complex reaction) under pseudo-steady-state assumption; (b) for the two-step nonsteady-state reaction, an approximation for the thermodynamic invariance was found to be valid in two
domains, (1) at the very beginning of the reaction (2) at the end of the reaction, near equilibrium
conditions. We hypothesize that such two-domain validity of the thermodynamic invariance is a
general feature of dual kinetic experiments performed in complex chemical systems.
& 2012 Elsevier Ltd. All rights reserved.

Keywords:
Kinetics
Transient response
Thermodynamics process
Onsager relations
Reaction engineering
Chemical reactors

1. Introduction
The study of constraints and their inuence on dynamic
behavior is of central importance to chemical kinetics. These
constraints consist of mass balance and relationships between
rate coefcients, see the classical papers of Onsager (1931a,b).
Time invariance for the ratio of concentration dependencies in socalled dual kinetic experiments has been found recently by Yablonsky
et al. (2011a,b). This result, obtained for batch reactors, carries
through for steady-state plug ow reactors in which the space-time
is analogous to the astronomic time of a batch reactor. The simplest
example of such dual experiments is given by the rst order
reversible reaction A#B. In the rst experiment, we start from a
reactor composition of pure A and measure the (space-)time evolution of the concentration of B, C BA t. In its dual experiment, we start
from pure B and measure the evolution C AB t of the concentration of
A. Duality is a symmetric relation: if the second experiment is dual to
the rst then the rst experiment is also dual to the second.
It was shown theoretically, Yablonsky et al. (2011a,b), that in such
dual experiments the ratio of concentration dependencies which

n
Corresponding author at: Department of Mathematical Analysis, Ghent
University, Galglaan 2, B-9000 Ghent, Belgium. Tel.: 32 9 264 49 54;
fax: 32 9 264 49 87.
E-mail address: Denis.Constales@UGent.be (D. Constales).

0009-2509/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2012.01.017

start from symmetric initial conditions is equal to their equilibrium


constant: at any moment in time t 40, C BA t=C AB t K eq .
This was established not only for the simple reversible rst
order reaction A#B but also for all networks of monomolecular
(linear) chemical reactions and for any nonlinear network linearized near the point of detailed balance (see, e.g, Helrich, 2009).
The main feature of this phenomenon is that the well-known
equilibrium relationship K eq B=A, where Keq is the equilibrium
constant, B and A are concentrations of the product and the
reactant respectively, is valid not only at the equilibrium point,
but at any moment of the dual experiments.
Similar, but more complicated invariances of thermodynamic
type have been announced for the nonlinear single reactions
2A#B and 2A#2B, Yablonsky et al. (2011a). However, these
relationships were not yet described in detail.
In Yablonsky et al. (2011b) the general theory of this phenomenon was developed, under two assumptions regarding the network of chemical transformations:

 the substances involved in the dual experiments are connected




by a sequence of reversible reactions, i.e., they are strongly


connected substances,
the Onsager relations hold (see Helrich, 2009).

Within this theory, the time-invariances of thermodynamic type


were interpreted as follows. The shift in time operator is symmetric

Author's personal copy


D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

in the entropic inner product, see Yablonsky et al. (2011b). Its


symmetry allows to formulate the described relationships as symmetry relations between the observables and the initial data.
A similar theory of thermodynamic invariances was developed
for the description of Temporal Analysis of Products (TAP) pulse
response experimental data in Constales et al. (2011). It was
applied to analyze the experimental data of dual TAP-experiments in studies of the watergas shift (WGS) reaction over iron
containing catalyst, CO H2 O#CO2 H2 . The data were obtained
using a thin-zone TAP reactor (TZTR). The effect predicted
theoretically was successfully justied: the ratio of non-steadystate exit uxes, CO2 obtained from pulsed CO and CO obtained
from pulsed CO2 respectively, duly corrected for the presence of
inert zones anking the thin catalyst zone, was constant in the
frequency domain for each temperature.
The goal of this paper is to present the latest achievements in
the development of this theory:
1. to describe systematically all the thermodynamic time invariances that relate to all types of reversible single reactions1
(rst and second order).
2. To present examples of the approach for nding the timeinvariance if the mechanism consists of more than a single
nonlinear reversible reaction. The simplest cases analyzed will
relate to the typical two-step catalytic mechanism under
pseudo-steady-state assumptions, and the two-step mechanism including one nonlinear and one linear step.

(iii) 2A#2B, C A C B 1; A B#2C, C A C B C C 1 and again


C B C A D; A B#C D, C A C B C C C D 1 and in this case
there are constant differences for each side of the reaction,
viz C B C A D , C D C C D .
Note that for convenience we use dimensionless concentrations.

3. Thermodynamic invariances
3.1. First-rst order reversible reaction A#B
This result was presented in a short note by Yablonsky et al.
(2011a). An example is the isomerization reaction tC4 H8 #
cC4 H8 .
By elementary mathematical techniques, for t Z 0,


C AA t

k k expk k t
,


k k

C BA t

k 1expk k t
,


k k

C AB t

k 1expk k t
,


k k

C BB t

k k expk k t
:


k k

Consequently,

2. Classication of reactions
We use a two-term classication of reversible reactions, based
on the kinetic orders of the forward and backward reactions,
respectively. Then, all possible reactions up to and including
second order can be presented as follows:
1. One rst-rst order reaction, A#B.
2. Two second-rst order reactions 2A#B, A B#C.
3. Three second-second order reactions 2A#2B, A B#2C,
A B#C D.
Hence there are only six basic single reactions. In this list, all
reactions are omitted that can be obtained by renaming substances in or reversing of the mentioned ones. For instance, the
reaction A#B C is not mentioned because it is an instance of
A B#C.
By denition, every single reaction
X
X
aj Aj # bj Bj ,
1
j

21

where the Aj and Bj are molecules of reactants and products, and

aj and bj are their stoichiometric coefcients respectively, is

K eq

C BA t
,
C AB t

where by C AA and C BA we denote the time evolutions of the


concentrations of A and B, respectively, when starting from pure
A, i.e., CA 1, CB 0 at t0. Similarly, C AB and C BB denote the time
evolutions of the concentrations of A and B, respectively, for the
initial composition CA 0, CB 1. Then C AA and C BA are called the A
trajectories, corresponding to the rst of the dual experiments,
and C AB and C BB the B trajectories, corresponding to the second.
Knowing the value of the equilibrium constant and one
trajectory therefore allows us to determine the other trajectories.
Application to trajectory analysis. If we plot C BA t vs. C BB t and
write C B,eq K eq =1 K eq ,
C BA tC B,eq K eq C BB tC B,eq ,

i.e., a straight line with slope K eq from C BB ,C BA 1; 0 to


C B,eq ,C B,eq . In particular, at all times the lower deviation
C B,eq C BA t is Keq times the upper deviation C BB tC B,eq . See
Fig. 1.
This and further plots can be interpreted as special phase
trajectories for dual experiments.

characterized by the relationship




1 dC A1
1 dC A2
1 dC B1
1 dC B2



...:
a1 dt
a2 dt
b1 dt
b2 dt

3.2. Second-rst order, 2A#B


2

Using this relationship, it is easy to show that all reactions listed


satisfy so-called kinetic balances,
(i) A#B, C A C B 1;
(ii) 2A#B, C A 2C B 1; A B#C, C A C B 2C C 1 and additionally, the difference of CA and CB is constant, so we write
C B C A D;
1
The time-invariance for the rst order reaction was described previously. In
this text, it will be given only for completeness.

Examples are the dimerization reaction 2CH3 #C2 H6 and


association reactions 2H#H2 , 2O#O2 and 2OH #H2 O2 as well.
The kinetic balance is
C A t 2C B t 1:
In order to reach the same equilibrium concentrations, the initial
compositions of both dual experiments must have the same
absolute amount of matter. Hence, we choose to start the rst
experiment for the reactant (1,0), and the other from the product
(0,1/2). The concentration of the product will be diluted. In both
cases the kinetic balance C A 2C B 1 will be fullled.

Author's personal copy


22

D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

C AB t

Keq
0.125
0.25
0.5
1.0
2.0
4.0
8.0
16.0

0.8

Furthermore, C B,eq 1=2sec y1=sec y 1 1=2tan2 y=2


and if we plot C BA t vs. C BB t 1C AB t they lie on a branch of a
rectangular hyperbola with axes parallel to the coordinate axes,
the horizontal one being C BA 1=2, center at C BB ,C BA C B,eq 1=
sec y1,C B,eq 1=sec y 1 12 2 cot2 y, 12, from the point
(1,0) to C B,eq ,C B,eq , i.e.,



1
1
C BB C B,eq 
cot2 y:
6
C BA C B,eq 
sec y 1
sec y1

CBA

0.6

0.4

In particular, for large t the lower deviation C B,eq C BA t is 2C B,eq


times the upper deviation C BB tC B,eq . See Fig. 2.
Introducing homogeneous coordinates x,y,z such that
C BB x=z and C BA y=z, Eq. (5) can be rewritten as

0.2

yz K eq zx=2zy=2,

which is the equation of a conic section family dened by the


following four points in homogeneous coordinates:

0.2

0.4

0.6

0.8

C BB
-

Fig. 1. Trajectories C BA vs C BB for AB.

The nonlinear differential equation is

(a)
(b)
(c)
(d)

the initial point (1, 0, 1),


the point at vertical innity (0, 1, 0),
the point at horizontal innity (1, 0, 0),
a point innitesimally close to (c) at vertical level 1/2,
1, E=2, E.

The point (d) ensures that the horizontal asymptote is xed as


C BA 1=2. Adding to this set the equilibrium point C B,eq ,C B,eq ,1
denes the hyperbola uniquely.

dC A t


2k C 2A t k 1C A t,
dt
which can be solved by reduction to quadratures (see Appendix)
as
!
r
r

k
1
k
8  1 tanh tk
8  1
2
k
k
!,
r
C AA t r



k
k
1
k
8  1
8  1 4  1 tanh tk
2
k
k
k

3.3. Second-second order, 2A#2B


Examples are any isomerization reactions of the second order.
The kinetic balance is
C A t C B t 1,

!
r

k
1
k
2  tanh tk
8  1
2
k
k
!,
r
C BA t r



k
k
1
k
8  1 4  1 tanh tk
8  1
2
k
k
k

0.5

Keq
0.125
0.25
0.5
1.0
2.0
4.0
8.0
16.0

0.4

!
r

1
k
8  1
2tanh tk
2
k
!:
r
C AB t r

k
1
k
8  1
8  1 tanh tk
2
k
k

CBA

0.3

Consequently,
C BA t
:
C AA tC AB t

0.2

Application: geometric description of trajectories.


Introducing
the
p

parameter y by K eq 18 tan2 y, we nd that 8K eq 1 sec y and


4K eq 1 1 sec2 y=2. Then

0.1

K eq

2
:

1 sec y cothtk1 sec y=2

C AA t

21 sec y cothtk1 sec y=2


,

1 sec2 y 2 sec y cothtk1 sec y=2

1
tan2 y
2
,
C BA t

1 sec2 y 2 sec y cothtk1 sec y=2

0.1

0.2

0.3

0.4

CBB
-

Fig. 2. Trajectories C BA vs C BB for 2AB.

0.5

Author's personal copy


D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

23

which is the usual total balance and offers no difculties for the
initial values (1,0) and (0,1). The differential equation

Geometrically, and in homogeneous coordinates, Eq. (8) can be


rewritten as

dC A t


2k C 2A t 2k 1C A t2 ,
dt

xy K eq zxzy,

so that the hyperbolas for different values of Keq form a conic


section family dened by the following four points in homogeneous coordinates:

can be reduced to quadratures, so that


C AA t

1
r

q ,

k

k k1
1
 tanh2t
k

(a)
(b)
(c)
(d)

p
k

k k
 tanh2t
k
r

C BA t
,

p
k

tanh2t
k
k

1

k

3.4. Second-rst order, A B#C

r
q :

k

1
k k1
tanh2t
k

Consequently,
K eq

C BA tC BB t
:
C AA tC AB t

Application toptrajectory
If we plot C BA t vs. C BB t and
panalysis.

write C B,eq K eq =1 K eq , they lie on a branch of a rectangular hyperbola with axes parallel to the coordinate axes, center
at C BB ,C BA K eq =K eq1 ,K eq =K eq1 ,



K eq
K eq
K eq
C BB t

,
9
C BA t
K eq 1
K eq 1
K eq 12
from C BB ,C BA 1; 0 to C B,eq ,C B,eq . In particular, for large t the
lower deviation C B,eq C BA t equals the upper deviation
C BB tC B,eq . See Fig. 3.

K eq

0.125
0.25
0.5
1.0
2.0
4.0
8.0
16.0

0.8

0.4

0.2

An example is O H#OH.
In this case, there are two balances to satisfy: as before, the
total balance, which is C A t C B t 2C C t 1, but also the
difference C B tC A t will be constant during the reaction. Without loss of generality, we assume that C A rC B (i.e., that A is the
reactant with minimal concentration) and write D for the difference, so that C B t C A t D.
This case is essentially different from the previous ones, in that the
forward reaction involves two different substances, whereas before
we only had a single reactant and a single product. Consequently, the
dual experiments will have specic features: the same reactant has to
be chosen as the nonminimal one in both dual experiments.
To determine the trajectories, we shall require
(a) that they both involve the same total amount of matter,
(b) that each of the pair of dual experiments starts from a
maximal possible value of concentration, CA in the case of
the A trajectory, CC in that of the C trajectory. This choice is
natural, starting the dual experiments from different extreme
conditions in A and C,
(c) Furthermore, each experiment, A and C, has a D value. We
require them to be the same.
These requirements together are essential to tune the dual experiments and obtain the elegant result that will be shown presently.
For the A trajectory we accordingly choose the maximal initial
concentration CA that satises C B C A D, the kinetic balance of
this reaction, and C A C B 2C C 1, the total balance: it is clearly
1D=2 and therefore the initial composition will be
C AA 0,C BA 0,C C A 0 1D=2,1 D=2; 0. Similarly, the maximal initial concentration CC that satises C B C A D and
C A C B 2C C 1 is 1D=2 and the initial values are therefore
C AC 0,C BC 0,C C C 0 0, D,1D=2 for the C trajectory experiment. In both experiments the value of D is the same, and
C BA C AA C BC C AC D; furthermore in both experiments the
total C A C B 2C C 1, thus providing full coherence between
the dual experiments. In the special case D 0 a stoichiometrical
composition CA CB occurs.
These initial conditions lead to an elegant time-invariance in
this case, as obtained by reduction to quadratures, in the form

CBA

0.6

the initial point (1, 0, 1),


its symmetric counterpart (0, 1, 1),
the point at vertical innity (0, 1, 0),
the point at horizontal innity (1, 0, 0).

Adding to this set the equilibrium point C B,eq ,C B,eq ,1 denes the
hyperbola uniquely (and that point is then one of its vertices). For
K eq 1 the hyperbola degenerates into C BA 1C BB and the line at
innity.

r

p
k

k k
tanh2t
k
r
C AB t

p ,
k

k k
1
tanh2t
k
C BB t

K eq

10

0.2

0.4

0.6

0.8

CBB
-

Fig. 3. Trajectories C BA vs C BB for 2A2B.

C C A t
:
C AC tC BA t

11

When D 0 the time-invariance of Eq. (5) for 2A#C is recovered


as a special case. In fact, we then have stoichiometric equality of A
and B at all times.

Author's personal copy


24

D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

If we interpret the invariance geometrically, with homogeneous coordinates for C C C x=z, C C A y=z,
C C A K eq C AC C BA , K eq 1D=2C C C 1 D=2C C A ,

12

leads to the conic section family equation


yz K eq xz1D=2yz1 D=2,

13

which are rectangular hyperbolas passing through the following


points:
(a)
(b)
(c)
(d)

the initial point 1D=2, 0, 1,


the point at vertical innity (0, 1, 0),
the point at horizontal innity (1, 0, 0),
a point innitesimally close to (c) at vertical level 1 D=2,
1, E1 D=2, E.

The point (d) ensures that the horizontal asymptote is xed as


C C A 1 D=2. Adding to this set the equilibrium point
C C,eq ,C C,eq ,1 denes the hyperbola uniquely.
3.5. Second-second order, A B#2C
An example is H H2 O2 #2OH. The kinetic balance is
C A t C B t C C t 1. Again, we assume without loss of generality that A is the reactant of minimal concentration, i.e., C A rC B ,
and write D for their constant difference, so that C B t C A t D.
In the dual experiments, the extreme initial conditions are
C AA 0,C BA 0,C C A 0 1D=2,1 D=2; 0 and C AC 0,C BC 0,
C C C 0 0, D, 1 D, respectively.
By reduction to quadratures, the time-invariance is
K eq

C C A tC C C t
:
C AC tC BA t

C C A tC DC t
:
C AC tC BA t

17

When e.g. D 0 a stoichiometric composition (of A and B)


occurs; then the time-invariance of Eq. (14) for A B#2C is
recovered as a special case.
We illustrate the validity of this result in Fig. 4 for the values


k 2 s1 , k 10 s1 , D 0:3, D 0:2; all concentrations
occurring in (17) are plotted there. All these trajectories change
in time, but the expression in (17) is invariant, and equals the
equilibrium constant. Comparing (17) with (3), we can interpret
C DC =C BA as a correction factor.
If we once more interpret the invariance geometrically, with
homogeneous coordinates for C C C x=z, C C A y=z,
C C A C DC C C A D C C C K eq C AC C BA
K eq 1D D =2C C C 1 D D =2C C A ,

18

leads to the conic section family equation


D xy K eq xz1D D =2yz1 D D =2,

19

which are rectangular hyperbolas passing through the following


points:
(a)
(b)
(c)
(d)

the initial point, 1D D =2; 0,1,


a nonphysical point D ,1 D D =2; 1,
the point at vertical innity (0, 1, 0),
the point at horizontal innity (1, 0, 0).

15

Adding to this set the equilibrium point C C,eq ,C C,eq ,1 denes the
hyperbola uniquely. For K eq 1 the hyperbola degenerates into
the line

16

1 D D C C C 1D D C C A 1D 2 D 2 =2

leads to the conic section family equation


xy K eq xz1Dyz1 D=4,

K eq

14

When D 0, the special stoichiometric case, the time-invariance


of Eq. (8) for 2A#2C is recovered as a special case.
If we again interpret the invariance geometrically, with homogeneous coordinates for C C C x=z, C C A y=z,
C C A C C C K eq C AC C BA , K eq 1DC C C 1 DC C A =4,

write D and D for the constant differences, so that C B t


C A t D and C D t C C t D . To be physical, D D o 1.
When D D 0, the special stoichiometric case occurs.
In this pair of dual experiments, the extreme initial conditions
are
C AA 0,C BA 0,C C A 0,C DA 0 1D D =2,1 D D =

2; 0, D and C AC 0,C BC 0,C C C 0,C DC 0 0, D ,1D D =
2,1D D =2, respectively.
The time-invariance, obtained by reduction to quadratures, is
in this case

which are rectangular hyperbolas passing through the following


points:
(a) the initial point 1D,0; 1,
(b) a nonphysical point corresponding to values C C C 0, i.e., C C C
at the initial point associated to C C A , and C C C 1 D, i.e., the
nonphysical initial point C AA D, C BA 0, C C A 1 D; in
homogeneous coordinates this is 0; 1 D,1,
(c) the point at vertical innity (0, 1, 0),
(d) the point at horizontal innity (1, 0, 0).
Adding to this set the equilibrium point C C,eq ,C C,eq ,1 denes the
hyperbola uniquely. For K eq 1 the hyperbola degenerates into
the line C C C =1D C C A =1 D 1 through (a) and (b), and the
line at innity.

20

through (a) and (b), and the line at innity.


In all cases analyzed from 3.1 to 3.6, the phase trajectory
related to dual experiments (3), (5), (8), (11), (14), (17) are

CC A
C DC

0.8

CAC
CBA

0.6

(CCA CDC)/(CAC CBA)

0.4
0.2

3.6. Second-second order, A B#C D


An example is O3 OH#HO2 O2 . The total balance is
C A t C B t C C t C D t 1. There are now two supplementary
balances, under the form of constant concentration differences,
one on the left and one on the right. We assume without loss of
generality that A and C are reactants of minimal concentration on
the left and right, respectively; then C A r C B and C C r C D , and we

0.1

0.2

0.3

0.4

0.5 0.6
time [s]

0.7

0.8

0.9

Fig. 4. Trajectories C C A , C DC , C AC and C BA , along with the invariance


C C A C DC =C AC C BA for A BC D, when k 2 s1 , k 10 s1 , D 0:3,
D 0:2.

Author's personal copy


D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

hyperbolas, where the lines in the rst-order reversible reaction,


3.1, are a degenerate case.

3.7. Combination of rst-rst and second-second order, A $


B and

k1

2A $
2B

k2

dC B t


k1 K eq 1C B tC B t 2k2 K 2eq 1C B t2 C 2B t,
dt

21

and the time invariant can be expressed as


k1 C AB tk1 C BA t


2K 2eq C AA tC AB tC BA tC BB t
k1
:

k2
K eq C AB tC BA t

0.6
0.4
0.2
0

0.2

0.4

0.6

0.8

k1

k2

k1

k2

25

where A and B are reactants and products, respectively, Z and AZ


are surface species, Z is the active site and AZ is an adsorbed
molecule of A. This mechanism is linear, i.e., in any reaction only
one molecule of the surface substance participates. The surface
species fulll the active site conservation law C Z C AZ C t , where
CZ and CAZ are the surface concentrations of Z and AZ respectively,
and Ct is the total number of active sites per unit of catalyst area.
In this case, the complex reaction is isomerization with an overall
reaction A#B. Under the pseudo-steady-state assumption the
rate of every surface species Z, AZ generation is approximately
equal to its consumption. Then both surface species concentrations can be eliminated. As a result, the kinetic equation describing the temporal evolution of this system will be written as (see
e.g. Marin and Yablonsky, 2011)
 

26

where CA and CB are the concentrations of the gas substances A


and B, respectively. The balance is C A t C B t 1. We write


K 1 k1 =k1 and K 2 k2 =k2 for the equilibrium constants of both


steps, and L k1 =k2 ; then K eq K 1 K 2 and


1
1
1 WW 1 aebt ,
27
C AB t
1 K 1 K 2
a
1
1 K 1 K 2


1
K 1 K 2  WW 1 K 1 K 2 aebt ,

k1

1
ln
C AB t C BA t

AZ#B Z,

1.8

k2

Eliminating time, we obtain the invariant equation

24

1.6

23

A Z#AZ,

1.4

Fig. 5. Trajectories C AB and C BA for A Z $


AZ, AZ $
B Z, along with the invariant


k
k


ln1C BA =C B,eq =1C AB =C A,eq =C AB C BA , 1when k1 2 8 s1 , k2 4 s1 , k1 4 s1 ,

k2 1 s1 .

The simplest two-step mechanism of a catalytic reaction


(TemkinBoudart mechanism) consists of the following reactions:

dC A t
dC B
k k C tk1 k2 C B t

1 2 A
,


dt
dt
k1 C A t k2 k1 k2 C B t

1
1.2
time [s]

22

4. Single-route catalytic mechanism, A Z $


AZ,AZ $
BZ



C BA t

CBA
ln((1-CBA /CB,eq)/(1-CAB /CA,eq))/(CAB+CBA)

2k2 C AA tC AB tk2 C BA tC BB t

or equivalently, in terms of the equilibrium constant, as




CAB

0.8

The kinetic balance is again C A t C B t 1. We assume


detailed balance, so that the equilibrium constants satisfy


k2 =k2 k1 =k1 2 . We shall denote by Keq the equilibrium


constant of the linear reaction, i.e., k1 =k1 .
The differential equation is

1

1.4
1.2

k1

k2

25

C B t
1 A
C B,eq
1 K 1 K 2 K 1 L1
,

C AB t
K 2 L1 K 1 K 1 K 2
1
C A,eq

31

where C A,eq 1=1 K 1 K 2 and C B,eq K 1 K 2 =1 K 1 K 2 . In Fig. 5

this result is illustrated for the values k1 8 s1 , k2 4 s1 ,




1
1
k1 4 s , k2 1 s . The function on the left-hand side of (31)
is constant in time, even though the concentrations that occur in


it vary. As a special case, when K 1 L 1, i.e., when k1 k2 , the
right-hand side vanishes and the linear case is recovered.
k1

k2

k1

k2

C$
D
5. Two step mechanism, A B $


For this mechanism, we consider a triple experiment consisting of three trajectories, dened using the principle of maximal
initial concentration
1. the A trajectory,
C AA 0,C BA 0,C C A 0,C DA 0 1D=2,1 D=2; 0,0;

32

2. the C trajectory,
C AC 0,C BC 0,C C C 0,C DC 0 0, D,1D=2; 0;

33

3. the D trajectory,
C AD 0,C BD 0,C C D 0,C DD 0 0, D,0,1D=2:

34

The kinetic equations are then


dC A t


k1 C A tC B t k1 C C t,
dt

28
dC B t


k1 C A tC B t k1 C C t,
dt

where

K 1 L1
,
K 2 L1 K 1 K 1 K 2

29

dC C t


k1 C A tC B tk1 k2 C C t k2 C D t,
dt

k2 L1 K 1 K 2 2
:
K 2 L1 K 1 K 1 K 2

30

dC D t


k2 C C tk2 C D t:
dt

35

Author's personal copy


26

D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

k1

Let us consider the reaction A B $


C separately rst. The


CCA

k1

CAC

thermodynamic time-invariance would be


C C A t
K 1,
C AC tC BA t

36

(first term in Eq. (41): CCA /(CACCBA))/K1


(first two terms in Eq. (41))/K1

where K 1 k1 =k1 is its equilibrium constant. For the same


reaction as a part of the two-step complex mechanism, Eq. (36)
will be valid in the domain of large time values, near the
equilibrium of the complex reaction, due to the principle of
detailed balance.
Intuitively, the same relationship has to be valid in the domain
of small time values, in which the inuence of the second reaction
on the rate behavior is insignicant. This can be checked using the
power series solutions to the kinetic equations, by successive
derivations and substitution of initial values. Retaining the rst
two nonzero terms in the power series we nd that

0.5

k1

1.02

1 D k1 1D2
t ...

C BA t
4
2

1.01

and therefore

0.99

37

and clearly this tends to K1 as t-0, conrming our intuition. To


improve the expression for K1 we shall correct for the second
term in (37). From the further power series expansions
k k 1D2 2 k1 k2 1D2 k1 k2 k1 k2 3
t 
t ...
C DA t 1 2
8
24

 

 

k1 k2 1D 2 k1 k2 1Dk1 D k2 k1 k2 3
t 
t ...
4
12

C C D t

k2 1D
k 1Dk2 k1 k2 2
t 2
t ...
2
4

C DC t

k2 1D
k 1Dk2 k1 k2 2
t 2
t ...
2
4

0.97
0.96
0.95
0.94
0

10
time [s]

15

20

t ...

concentration dependencies only



2
C C A t
C DA tC C D tC AC t
3 C DC t
K1

1 . . .
C AC tC BA t 2 C AC tC AA t C AD tC DC tC C A t
41

38

In view of Eq. (38), for small values of time, this Onsager-like


quotient is close to 1. For large values of time, both numerator
and denominator tend to C D,eq C C,eq C A,eq , so the quotient also tends
to 1. Now we use (38) to correct for the second term in (37),
relying on the additional power series expansion
1D

2

(first two terms in Eq. (41))/K1

C DA tC C D tC AC t
k 1D
t ...
1 1
C AD tC DC tC C A t
6

C AA t

k2

Fig. 7. Ratio of the estimate C C A =C AC C BA to K 1 k1 =k1 and its rst-order


correction as in Fig. 6, but enlarged.

we can deduce the power series expansion of the Onsager-like


quotient

k1 1D2

20

0.98

0.93

C AD t

15

(first term in Eq. (41): CCA /(CACCBA))/K1

C C A t
k
k 2 k2 1D 2
1  1
t ...

C AC tC BA t
12k1
k1

10
time [s]

k1 1D
k 1Dk1 D k1 k2 2
t 1
t ...
2
4

Fig. 6. Trajectories C C A , C AC and C BA for A B $


C$
D, along with the ratio of the


k
k


estimate C C A =C AC C BA to K 1 k1 =k1 and 1 its 2 rst-order correction, when


1
1
1
1
k1 1 s , k1 0:2 s , k2 0:5 s , k2 0:2 s and D 0:1.

k 1D2
k 1D2 k1 k1 k2 2
C C A t 1
t 1
t ...
4
8
C AC t

CBA

1.5

39

to obtain precisely

2

C DA tC C D tC AC t
3 C DC t
k 2 k2 1D 2
1 1
t ...

2 C AC tC AA t C AD tC DC tC C A t
12k1
40
No exact time invariance expressing K1 in terms of concentrations is known to us at this time, but we can interpret these
power series to propose the following expansion in terms of

We have yet to determine the domain of validity of such


approximations. We hypothesize that such results holds for
general systems of reactions. In Figs. 6 and 7 an example of these
trajectories is plotted.

6. Discussion
In this section, we discuss the novelty of the obtained results,
comparing them with the concepts traditionally accepted in
chemical kinetics.
In contemporary chemical physics, the essential difference
between equilibrium chemical thermodynamics (ECT) and chemical kinetics is always stressed. In ECT, time is not considered;
the typical problem is calculating the chemical composition of
mixture that reacts in a closed system for an innitely long time.
In opposite, chemical kinetics is about the temporal evolution of
the chemical composition of reacting mixtures.

Author's personal copy


D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

The following concepts of chemical kinetics hold:

 It is impossible to present an expression for a non-equilibrium

chemical system based on its description under equilibrium


conditions, except for some relations describing the kinetic
behavior in the linear vicinity of equilibrium (see, e.g., Boudart,
1968).
In closed chemical systems, only linear time-invariants, i.e.,
constant linear combinations of chemical concentrations, can
be observed, and these invariants can be of only two types.
The rst type is the mass conservation law, and the corresponding coefcients are linear combinations of the numbers
of given atoms in the sequence of compounds.
The second type is a so-called kinetic balance, which is caused
by the specic reaction scheme. For the second order reaction
A B#C, the kinetic balances are expressed as follows:
C A C C const and C B C C const. The coefcients in these
kinetic balances are stoichiometric coefcients. The difference
C A C B is equal to the initial excess of reactant concentration
A over reactant concentration B.
There is a substantial difference between these two linear
invariants. The kinetic behavior is always subject to the massbalance linear invariants (mass conservation laws), which do
not depend on the reaction scheme and corresponding kinetic
models. As for the kinetic linear invariants, they may be
observed or not depending on the specic features of the
detailed mechanism, Marin and Yablonsky (2011).

6.1. A new type of time invariant


What is the signicance of our results in comparison with these
traditional concepts? In our theoretical analysis, we found an
experimental procedure (a thought experiment) in which a new
type of chemical time invariant is observed. For the simple reaction
A#B, the value of this invariant C BA t=C AB t is the well-known
equilibrium constant. The invariant is constructed based on the data
of non-equilibrium experiments performed from different initial
conditions of the reacting mixture. We call such experiments dual.
In another case, for the reversible reaction A B#C, three
concentration time dependencies are measured: the temporal
changes of B and C, starting from the mixture A and B in the
absence of C, and the temporal change of A, starting from the

27

maximum amount of substance C and in the absence of A. Then,


combining these dependencies we obtained the expression
C C A t=C AC tC BA t. This is a new time invariant. From our
analysis, its value coincides with the equilibrium constant
K eq C C,eq =C A,eq C B,eq , where C A,eq , C B,eq and C C,eq are the equilibrium concentrations of the substances A, B and C, respectively.
Similar results are obtained for all types of single rst and
second order reversible reactions (see Table 1).

6.2. Dual experiments: setup and constraints


The dual experiments start from different initial conditions
(IC). There are two important requirements for performing these
experiments:

 The IC have to meet the fundamental requirement that for

both experiments the values of the equilibrium concentrations


are the same. For this purpose, in both dual experiments we
have to use the same amounts of atoms. If the atomic
composition of the reacting substances is not presented, the
same total amount of substance mass has to be taken in both
experiments. This requirement of equal total mass is obligatory for the dual experiments.
In both cases, the IC have to have extreme values within the
given balances. For reversible reactions with one reactant and
one product (i.e., A#B; 2A#2B; 2A#B dual experiments are
performed starting from the pure reactant A or pure reactant B.
For nonlinear reversible reactions with two reactants and one
product, or two reactants and two products (i.e., A B#C,
A B#2C, A B#C D), in both experiments the initial difference between reactant concentrations has to be same. E.g., in
one dual experiment for the reaction A B#C, starting from
the mixture A B, we consider that substance B has an excess
compared with substance A. Therefore, the initial concentration difference is C B C A D. In the dual experiment, starting
from the substance C, we have to maintain the same difference
D. For this purpose, we choose as initial concentrations of the
reactants CA 0, C B D.

Further initial concentration differences for the studied reactions


are presented in Table 1.
Summing up, reformulating the constraints imposed on the dual
experiments, we select the special domain of chemical composition

Table 1
Time invariances for each of the reactions considered.

k1

A$
B

k1

K eq

C BA t
k1

C AB t
k1

K eq

C BA t
k1

C AA tC AB t
k1

K eq

C BA tC BB t
k1

k1
C AA tC AB t

k1

B
2A $

k1

k1

2B
2A $

k1

k1

k2

k1

k2

B, 2A $
2B
A$



K eq

k1
 ,
k1

k2
2
 K eq
k2

k1

AB $
C C B C A D Z 0

k1

k1

2C C B C A D Z 0
AB $

k1

K eq

C C A t
k1

k1
C AC tC BA t

K eq

C C A tC C C t
k1

k1
C AC tC BA t

K eq

C C A tC DC t
k1

k1
C AC tC BA t

k1

C D C B C A D Z 0 C D C C D Z 0
AB $

k1

2K eq C AA tC AB tC BA tC BB t
k1

K eq C AB tC BA t
k2

k1

k2

k1

k2

AZ $
B Z K1
AZ $



k1
 ,
k1

K2

k2
 ,
k2

k1

k2

1 K 1 K 2 K 1 L1
1

ln
K 2 L1 K 1 K 1 K 2
C AB t C BA t

C B t
1 A
C B,eq
C AB t
1
C A,eq

Author's personal copy


28

D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

values having given differences in the initial concentrations of the


reacting substances (a given excess in chemical language).
6.3. Kinetic ngerprints: time invariants of the equilibrium
constant type
In chemical kinetics, many ngerprints for distinguishing
irreversible simple reactions are known. For single reactions,
ln C is linear vs. t, 1=C is linear vs. t, C is linear vs. t are
traditional ngerprints of irreversible reactions of rst, second
and zero order, respectively, where C is the concentration of the
chemical substance.
Based on our theoretical results, we can propose new ngerprints for distinguishing the type of reversible reaction. All
invariants of equilibrium type presented in Table 1 can be used
for recognizing the type of reaction, because every reaction is
characterized by a specic time invariant of equilibrium constant
type (see Table 1).
In the simplest case of an isomerization reaction with two
substances A (reactant) and B (product), we need to measure two
concentration dependencies C BA t, obtained from pure A, and
C AB t, obtained from pure B. If the ratio of these dependencies (3)
does not depend on time, i.e., expression (3) is a time invariant,
this can be considered as a ngerprint of the rst order reversible
reaction A#B.
If it does not work, we use four concentration dependencies,
C AA t, C AB t, C BA t and C BB t and check another time invariant,
see Eq. (8). If this ts, it is a ngerprint of the second order
reversible reaction 2A#2B, and we can conclude that this
reversible isomerization reaction is governed by a second order
dependence.
For testing the reaction, A B#C D. we need the four concentration dependencies, C C A t, C DC t, C AC t, C BA t. If the expression in (17) remains constant, it is a ngerprint of this reaction.
6.4. Kinetic ngerprints via analysis of phase trajectories
Our analysis also provides the possibility to use special phase
trajectories for distinguishing the types of reaction.
A linear dependence between the trajectories C BA t and C BB t
is a ngerprint of the reaction A#B, see Fig. 1.
The reactions 2A#B and 2A#2B correspond to Figs. 2 and 3.
The dependence is no longer linear in these, but the curves do
form in each case a family of conic sections, so that coincidence of
data points with one of them is a ngerprint of the corresponding
reaction. The two families are highly distinct: for instance,
inspection of their tangent near equilibrium provides a downward angle of 451 as a ngerprint for 2A#2B.
Using the phase trajectories is more visual in nature, the
equilibrium constant tests more numerical. Both, however, are
essentially equivalent.

7. Conclusions
In this paper, we have established thermodynamic timeinvariances for basic single reactions governed by the mass-action
law, one for the linear case and ve for nonlinear cases. These are
summarized in Table 1. Third-order and autocatalytic reactions
will be treated in the next paper. As examples of complex
multistep reactions, we have considered the two-step catalytic
cycle A Z#AZ#B Z and the two-step reaction with nonlinear
rst step, A B#C#D.
The applications of these results are as follows. First, it is now
easy to recognize a type of reaction by comparing relationships
between the observed temporal dependencies of concentrations

with the equations presented in Table 1. The obtained timeinvariances can be used as simple ngerprints for distinguishing
the types of reactions.
Secondly, knowing the type of reaction the value of the
equilibrium constant and only few temporal concentration
dependencies that start from given initial conditions, we are able
to calculate temporal concentration dependencies for any initial
conditions. For the reaction A#B, we have to know in advance
only one temporal concentration dependence. For the reaction
A B#C we have to know two independent temporal dependencies. It is hypothesized that generally the number of temporal
dependencies which have to know for this purpose has to be
equal to the number of unknown parameters.
Our analysis shows that obtaining these thermodynamic timeinvariances is possible only when starting from special initial
conditions, which are governed by the total balance of substances
for reactions with one reactant and one product, A#B, 2A#B and
2A#2B, and by total balance and additional kinetic balances that
are specic for reactions involving two different species, A B#C
or A B#2C or A B#C D. Elegant expressions are obtained
only when a maximal initial concentration is chosen for each
trajectory. The dependence of dual experiments under other
initial conditions will be discussed separately.
As for the complex multistep reactions, the examples presented in our paper are both promising and challenging. From one
side, the analyzed example of non-equilibrium behavior of the
single-route catalytic reaction is promising. Apparently, many
results obtained for the single reactions can be transferred into
results for single route reactions with linear mechanism, under
the assumption that the non-equilibrium reaction occurs under
pseudo-steady-state conditions regarding surface intermediates
and that its mechanism is linear, i.e., in any reaction only one
molecule of the surface substance participates. From the other
side, the results obtained for the multistep reaction with a
nonlinear step, A B#C#D, show that we are at the beginning
of studying time-invariances for nonlinear multistep mechanisms. What is known now is that the combination of non-steadystate concentration dependencies obtained in the course of a pair
of dual experiments can be expressed as a series where the rst
term is the equilibrium constant. Estimating the equilibrium
constant by the combination of concentration dependencies
works both at the very beginning of the process (small values of
time) and at the end (near equilibrium). We have yet to determine, however, the domain of validity of these approximations.
We also hypothesize that this result holds for general systems of
reactions.

Nomenclatue
Roman
A, B, C, D, symbols denoting chemical species
E
CX
where X A,B,C, . . ., dimensionless concentration of
species X, [1]
C X,eq
equilibrium value of CX, [1]
CX(t)
time evolution of CX, [1]
C X Y t
the time evolution CX(t) occurring in an experiment
corresponding to the Y trajectory, [1]


rate constants, forward resp. backward, in reaction
ki , kj
numbered i resp. j [1/s]
Keq
equilibrium constant, [1]
L
ratio of the backward rate constants in two steps,


k1 =k2
t
time [s]

Author's personal copy


D. Constales et al. / Chemical Engineering Science 73 (2012) 2029

Lambert W function. W 1 denotes its inverse, i.e.,

x,y,z

W 1 x xex
homogeneous coordinates in the plane, [1]

D, D ,
D

46

Similarly, the C trajectory, dened by the initial value C AC 0 0, is


obtained from
Z C A t
C
dC A
t:
47
f C A
0

Greek letters

ai , bj
a, b

obtained from
Z C A t
A
dC A
t:
1D=2 f C A

29

stoichiometric coefcients, [1]


abbreviations dened in Eqs. (29) and (30), [1]
constant differences of concentrations, [1]

Eliminating time then yields the basis of the time invariance


Z C A t
Z C A t
C
A
dC A
dC A

:
48
f C A
1D=2 f C A
0
Using this equation to solve for Keq leads to an expression in C AC t
and C C A t that is valid for all times t 40

Acknowledgments
The authors wish to thank the three anonymous reviewers for
their most valuable comments. Financial support from BOF/GOA
01GA0405 of Ghent University and from the Long Term Structural
Methusalem Funding by the Flemish Government is gratefully
acknowledged.

K eq

1D=2C AA t
,
C AC tC AA t D

49

which can be rewritten as


K eq

C C A t
,
C AC tC BA t

50

where D no longer occurs explicitly.


Appendix A
A.2. Lambert function W
A.1. Reduction to quadratures
The ordinary differential equations for the kinetic modeling
occurring in this paper can all be solved using the technique of
reduction to quadratures. To explain it in detail, we consider the
reaction A B#C: the kinetic equations are
dC A t


k C A tC B t k C C t,
dt

42

dC B t


k C A tC B t k C C t,
dt

43

dC C t


k C A tC B tk C C t:
dt

44

References

The balance C A C B D allows to rewrite C B t C A t D, and the


balance C A C B 2C C 1 then allows to rewrite C C t 1D=
2C A t. Substituting these in (42) we obtain the single ordinary
differential equation for CA(t)
dC A t


k C A tC A t D k 1D=2C A t,
dt

The Lambert function required in this paper is dened as


follows: let x 41, and xex y, then y Wx. This function is
somewhat similar to the logarithm (for which ex y means that
y ln x) but is a separate special function with unique properties,
e.g., allowing to solve equations of the type x ln x y. In the
body of this paper, we have written W 1 x for its inverse, even
though this is the simpler expression xex, to underline the inverse
relationship between this expression and the Lambert function.
Full details can be found in Corless et al. (1996).

45

and we shall write f C A t for its right-hand side, where f is a


polynomial of degree two. To solve dC A t=dt f C A t by reduction to quadratures, we rewrite it as dC A =f C A dt and integrate
both sides from the initial to the current time, so that for instance
the A trajectory, dened by the initial value C AA 0 1D=2, is

Boudart, M., 1968. Kinetics of Chemical Processes. Prentice-Hall.


Constales, D., Yablonsky, G.S., Galvita, V.V., Marin, G.B., 2011. Thermodynamic
time-invariances: theory of TAP pulse-response experiments. Chem. Eng. Sci.
66, 46834689.
Corless, R.M., Gonnet, G.H., Jeffrey, D.J., Knuth, D.E., 1996. On the Lambert W
function. Adv. Comput. Math. 5, 329359.
Helrich, C.S., 2009. Modern Thermodynamics with Statistical Mechanics. Springer.
Marin, G.B., Yablonsky, G.S., 2011. Kinetics of Chemical Reactions: Decoding
Complexity. J. WileyVCB.
Onsager, L., 1931a. Reciprocal relations in irreversible processes. I. Phys. Rev. 37,
405426.
Onsager, L., 1931b. Reciprocal relations in irreversible processes. II. Phys. Rev. 38,
22652279.
Yablonsky, G.S., Constales, D., Marin, G.B., 2011a. Equilibrium relationships for
non-equilibrium chemical dependencies. Chem. Eng. Sci. 66, 111114.
Yablonsky, G.S., Gorban, A.N., Constales, D., Galvita, V.V., Marin, G.B., 2011b.
Reciprocal relations between kinetic curves. Europhys. Lett. 93, 20004.

Anda mungkin juga menyukai