U.S. patent application number 11/425895 was filed with the patent office on 2006-10-19 for system and method for determining optimal reaction parameters using continuously running process.
This patent application is currently assigned to Cellular Process Chemistry, Inc.. Invention is credited to Volker Autze, Ansgar Kursawe, Sebastian Oberbeck, Kemal Hunkar Sahin, Thomas Jochen Schwalbe.
Application Number | 20060234381 11/425895 |
Document ID | / |
Family ID | 33300000 |
Filed Date | 2006-10-19 |
United States Patent
Application |
20060234381 |
Kind Code |
A1 |
Schwalbe; Thomas Jochen ; et
al. |
October 19, 2006 |
SYSTEM AND METHOD FOR DETERMINING OPTIMAL REACTION PARAMETERS USING
CONTINUOUSLY RUNNING PROCESS
Abstract
A reaction system enables a plurality of optimization
experiments for a reaction to be performed continuously, to enable
optimal reaction parameters to be determined. Dilution pumps are
included to automatically vary the solvent mixed with reactants so
a concentration of each reactant can be selectively varied. The
reactants are introduced into a reaction module selectively coupled
to residence time chambers or directly to an analytical unit. The
analytical unit determines the yield and/or quality for each
optimization experiment, enabling optimal parameters to be
determined. Residence time chambers can be employed sequentially to
enable total residence time to be varied. The controller performs
as many experiments as required to enable each parameter to be
varied according to a predefined testing program and can redefine a
testing program based on the results from previous experiments. At
least two reaction parameters can be varied according to periodic
functions to further enhance analytical efficiency.
Inventors: |
Schwalbe; Thomas Jochen;
(Brookline, MA) ; Autze; Volker; (Frankfurt am
Main, DE) ; Oberbeck; Sebastian; (Weilburg, DE)
; Kursawe; Ansgar; (Frankfurt, DE) ; Sahin; Kemal
Hunkar; (Pittsburgh, PA) |
Correspondence
Address: |
LAW OFFICES OF RONALD M ANDERSON
600 108TH AVE, NE
SUITE 507
BELLEVUE
WA
98004
US
|
Assignee: |
Cellular Process Chemistry,
Inc.
Cambridge
MA
|
Family ID: |
33300000 |
Appl. No.: |
11/425895 |
Filed: |
June 22, 2006 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10824186 |
Apr 14, 2004 |
7101515 |
|
|
11425895 |
Jun 22, 2006 |
|
|
|
60462860 |
Apr 14, 2003 |
|
|
|
Current U.S.
Class: |
436/34 ;
436/43 |
Current CPC
Class: |
B01J 2219/00698
20130101; G05B 13/024 20130101; B01J 2219/00889 20130101; Y10T
436/12 20150115; Y10T 436/11 20150115; B01J 2219/00909 20130101;
B01J 2219/0059 20130101; B01J 2219/00977 20130101; B01J 2219/00186
20130101; B01J 2219/00873 20130101; B01J 2219/00867 20130101; B01J
2219/00961 20130101; B01J 2219/00984 20130101; B01J 2219/00986
20130101; B01J 2219/00702 20130101; B01J 19/0093 20130101; B01J
2219/00963 20130101; B01J 2219/00166 20130101; B01J 19/0033
20130101 |
Class at
Publication: |
436/034 ;
436/043 |
International
Class: |
G01N 31/10 20060101
G01N031/10 |
Claims
1. A method for using a continuously running system to determine at
least one optimal reaction parameter for a reaction to produce a
desired product, comprising the steps of: (a) identifying at least
one reaction parameter to be varied; (b) for each reaction
parameter, identifying a plurality of values to be assigned to the
reaction parameter; (c) selecting a baseline value for each
reaction parameter from the plurality of values identified for each
reaction parameter; (d) using the baseline values to generate the
desired product in a continuously running reaction system; (e)
determining at least one of a quantity and a quality of the desired
product generated using the baseline values; (f) changing the
baseline value for at least one reaction parameter, thereby
affecting the desired product being produced by the continuously
running system; (g) determining at least one of a quantity and a
quality of the desired product generated using the at least one
baseline value that was changed; and (h) comparing the at least one
of the quantity and the quality of the desired product generated
before changing the at least one of the baseline value with a
corresponding at least one of the quantity and the quality of the
desired product generated after the step of changing, to determine
the at least one reaction parameter responsible for generating the
highest of at least one of the quantity and the quality of the
desired product.
2. The method of claim 1, wherein the plurality of values for each
reaction parameter correspond to upper and lower boundaries.
3. The method of claim 2, wherein the step of changing the baseline
value for at least one reaction parameter comprises the step of
changing the baseline value for at least two reaction parameters
according to a periodic function, wherein each of the at least two
reaction parameters are changed based on different periodic
functions.
4. The method of claim 3, further comprising the step of
determining if data corresponding to at least one of a quantity and
a quality of the desired product generated in the continuously
running system operated with the at least two reaction parameters
that are changed based on different periodic functions indicates
that any ranges for the plurality of values for each reaction
parameter should be changed, and if so, changing the range as
indicated by such data.
5. The method of claim 1, wherein the step of changing the baseline
value for at least one reaction parameter comprises the step of
changing the baseline value for at least one reaction parameter
according to a linear function.
6. The method of claim 1, wherein the step of changing the baseline
value for at least one reaction parameter comprises the step of
changing the baseline value for at least one reaction parameter
according to a predefined pattern.
7. The method of claim 1, wherein the step of changing the baseline
value for at least one reaction parameter comprises the step of
changing the baseline value for at least one reaction parameter
according to a user input.
8. A method for using a continuously running reaction optimization
system to determine at least one optimal reaction parameter for a
reaction to produce a desired product, comprising the steps of: (a)
identifying at least one reaction parameter to be varied; (b)
identifying a baseline value for each reaction parameter; (c) using
the baseline values to generate the desired product in a
continuously running reaction optimization system; (d) determining
at least one of a quantity and a quality of the desired product
generated using the baseline values; (e) changing the baseline
value for at least one reaction parameter, thereby affecting the
desired product being produced by the continuously running system;
(f) determining at least one of a quantity and a quality of the
desired product generated using the at least one baseline value
that was changed; and (g) comparing the at least one of the
quantity and the quality of the desired product generated before
changing the at least one of the baseline value with a
corresponding at least one of the quantity and the quality of the
desired product generated after the step of changing, to determine
the at least one reaction parameter responsible for generating the
highest of at least one of the quantity and the quality of the
desired product.
9. The method of claim 8, wherein the step of changing the baseline
value for at least one reaction parameter comprises the step of
changing the baseline value for at least two reaction parameters
according to a periodic function, wherein each of the at least two
reaction parameters are changed based on different periodic
functions.
10. The method of claim 8, wherein the step of changing the
baseline value for at least one reaction parameter comprises the
step of changing the baseline value for at least one reaction
parameter according to a linear function.
11. The method of claim 10, further comprising the step of
determining if data corresponding to at least one of a quantity and
a quality of the desired product generated in the continuously
running system operated with at least one reaction parameter being
changed according to a linear function indicates any values
corresponding to a linear discontinuity, and if so: (a) for each
value corresponding to a linear discontinuity, defining that value
as a baseline value; and (b) repeating steps (c)-(g).
12. The method of claim 8, wherein the step of changing the
baseline value for at least one reaction parameter comprises the
step of changing the baseline value for at least one reaction
parameter according to a user input.
13. The method of claim 8, wherein the step of changing the
baseline value for at least one reaction parameter comprises the
step of changing the baseline value for at least one reaction
parameter according to a predefined pattern.
14. A method for using a continuously running reaction optimization
system to generate data that can be used to identify at least one
optimal reaction parameter for a reaction to produce a desired
product, comprising the steps of: (a) identifying at least one
reaction parameter to be varied; (b) identifying a baseline value
for each reaction parameter; (c) using the baseline values to
generate the desired product in a continuously running reaction
optimization system; (d) determining at least one of a quantity and
a quality of the desired product generated using the baseline
values; (e) changing the baseline value for at least one reaction
parameter, thereby affecting the desired product being produced by
the continuously running system; and (f) determining at least one
of a quantity and a quality of the desired product generated using
the at least one baseline value that was changed.
15. The method of claim 14, further comprising the step of
comparing the at least one of the quantity and the quality of the
desired product generated before changing the at least one of the
baseline value with a corresponding at least one of the quantity
and the quality of the desired product generated after the step of
changing, to determine the at least one reaction parameter
responsible for generating the highest of at least one of the
quantity and the quality of the desired product.
16. The method of claim 15, wherein the step of changing the
baseline value for at least one reaction parameter comprises at
least one of the steps of: (a) changing the baseline value for at
least one reaction parameter according to a predefined pattern; (b)
changing the baseline value for at least one reaction parameter
according to a periodic function; (c) changing the baseline value
for at least one reaction parameter according to a linear function;
and (d) changing the baseline value for at least one reaction
parameter according to a user input.
17. A method for using a continuously running reaction optimization
system to generate data that can be used to identify at least one
optimal reaction parameter for a reaction to produce a desired
product, comprising the steps of: (a) identifying at least one
reaction parameter to be varied; (b) identifying a baseline value
for each reaction parameter; (c) using the baseline values to
generate the desired product in a continuously running reaction
optimization system, such that data corresponding to at least one
of a quantity and a quality of the desired product generated using
the baseline values is collected; and (d) changing the baseline
value for at least one reaction parameter over time, thereby
affecting the desired product being produced by the continuously
running system, such that data corresponding to at least one of a
quantity and a quality of the desired product generated using the
at least one baseline value that was changed is collected over
time.
18. The method of claim 17, wherein the step of changing the
baseline value for at least one reaction parameter comprises the
step of changing the baseline value for at least one reaction
parameter according to a predefined pattern.
19. The method of claim 17, wherein the step of changing the
baseline value for at least one reaction parameter comprises the
step of changing the baseline value for at least one reaction
parameter according to a periodic function.
20. The method of claim 17, wherein the step of changing the
baseline value for at least one reaction parameter comprises the
step of changing the baseline value for at least one reaction
parameter according to a linear function.
21. The method of claim 17, wherein the step of changing the
baseline value for at least one reaction parameter comprises the
step of changing the baseline value for at least one reaction
parameter according to a user input.
Description
RELATED APPLICATIONS
[0001] This application is a divisional of a prior co-pending U.S.
patent application Ser. No. 10/824,186, filed on Apr. 14, 2004,
which itself is based on a prior provisional patent application
Ser. No. 60/462,860, filed on Apr. 14, 2003, the benefit of the
filing dates of which is hereby claimed under 35 U.S.C. .sctn.
119(e) and 35 U.S.C. .sctn. 120.
FIELD OF THE INVENTION
[0002] This invention generally relates to a chemical processing
apparatus, and more specifically, to a continuously operating
system configured to vary reaction parameters over time in order to
identify optimal reaction parameters.
BACKGROUND OF THE INVENTION
[0003] Apparatus for controlling and optimizing the production of
chemical substances are well known in the prior art. Reaction
parameters affecting the quantity and quality of the product
generated include concentration levels of each reactant,
temperature conditions, flow rates, and residence times. Varying
one or more of the reaction parameters generally results in a
change in product yield. It is therefore advantageous to optimize
such reaction parameters to maximize production and quality.
[0004] A basic prior art optimization procedure is as follows.
Initial reaction conditions from an initial synthesis are used as a
starting point. Using temperature, reaction time and concentrations
at the values determined in the initial synthesis, three
experiments with different equivalents (i.e. different
stoichiometric ratios) are conducted. For example, where the
initial synthesis was based on using a 1:1 ratio of a first
reactant and a second reactant, ratios such as 1:1.1, 1:1.2, 1.1.3;
or 1.1:1, 1.2:1, and 1.3:1 could be employed. In a second set of
experiments, three different temperature conditions are applied. A
third and fourth set of three experiments each are also performed,
changing other variables in each set. After twelve such experiments
have been performed (i.e., four sets of three experiments), the
results are reviewed, and optimized reaction parameters are defined
based on the data collected from the twelve experiments. An
additional set of twelve experiments can then be performed,
similarly varying the optimized parameters defined by the first
series of experiments. In such an optimization procedure, typically
twenty four experiments are required for a first optimization of
reagent equivalents, temperature conditions, reaction time, and
reagent concentration for a given reaction, as each experiment is
repeated to check the reproducibility of the results. One
disadvantage of this approach is that interactions between these
parameters are difficult to quantify.
[0005] Because of this difficulty, process optimization methods
referred to as statistical design experiments have been developed.
The goal of such methods is to model an equation in order to couple
process variables with process results (i.e., the yield of a
reaction). A well known two-value approach requires 2.sup.n+1
experiments, where n is the number of variables. Each variable is
employed at two different values, and an additional experiment is
performed using the mean of each variable (as a control to
determine if the behavior is linear). Typically, every experiment
is repeated to estimate the reproducibility. For the
above-mentioned case, (2.sup.4+1)*2=34 experiments are needed. The
disadvantage of such an empirical approach is the fact that the
success of the optimization is largely based on how well each of
the two values for each variable is selected. Selecting levels that
are close together results in only small improvements in
optimization being achieved; so that it is likely an additional
2.sup.n+1 sets of experiments will be required. Selecting values
that are far apart results in a risk that one or more variables
will exceed a critical parameter, which will significantly affect
yields (such as exceeding a reaction temperature beyond which yield
drops sharply or no reaction takes place). When this result occurs,
the initial set of 2.sup.n+1 experiments are of little value, and
the experiments must be repeated after different values have been
selected.
[0006] Furthermore, if the mean value experiment indicates that non
linear behavior exists, then it is necessary to determine the
impact of quadratic terms. This step can only be assessed by
expanding the design of experiments to 3.sup.n experiments, where
the three values are defined as the lower and upper bounds, as well
as the mean values of these bounds. For the analysis of a
four-parameter system, this approach implies a total of 3.sup.4=81
experiments will be required. Preferably, each set is repeated to
validate reproducibility, so that a total of 162 experiments must
be performed. In practice, some terms and factors in an equation
model are often identical, and it is not unusual for the 81
experiments noted above to be reduced to about 40 experiments
(without the duplication for validation of reproducibility).
[0007] This analysis can be performed efficiently using software
packages that determine the values for each experiment, the order
in which these values should be changed, and evaluate the outcome
to provide a mathematical relationship between the performance
criteria being investigated and the variables to be adjusted to
optimize the performance. Today, equipment for parallel batch
experiments is also available, so that a number of experiments can
be conducted at the same time. These parallel analysis systems are
based on matrices of reaction modules in which the chemicals to be
analyzed are input manually at variable concentrations. Some
reaction conditions, such as temperature, are often identical for
all the vessels being analyzed at any given time due to the
physical dimensions and limitations of the system. The reaction
duration is also generally identical for efficient analysis. Due to
the discrete nature of experimentation, the evaluation at the end
of the experiment has to be performed for all reaction modules
separately, to determine the performance of each system. These
results are analyzed off-line as one data set for a fixed
temperature and reaction duration. Experimentation at different
reaction temperatures requires the generation of another matrix
with the same reactants, and repetition of the experiments at the
new temperature, as well as a new analysis of the collected data.
Once the analysis for concentrations and concentration ratios at
different temperatures is completed, the same set of experiments
can be performed to determine the effect of reaction time on yield.
The repetitiveness of such experiments (i.e., the batch-like
processing) is enforced due to the matrix-like structure of the
parallel reaction vessels.
[0008] While such methods can enable optimized reaction parameters
to be achieved, it would be desirable to provide a method and
apparatus based on optimizing reactions parameters using a
continuously running system, as opposed to using the batch-based
testing of the prior art.
SUMMARY OF THE INVENTION
[0009] The present invention employs a continuously operating
system that enables reaction parameters to be varied over time, to
optimize a chemical reaction. The time to achieve optimization is
thus dramatically reduced compared to the prior art batch-based
optimization techniques discussed above.
[0010] The system employed includes a reaction module (preferably
including a micro reactor, so that minimal reactant volumes are
required), a plurality of residence time chambers, fluid lines
coupling the micro reactor to the residence time chambers, fluid
lines for introducing reactants into the reaction module, and fluid
lines for directing a product exiting the reaction module into
either residence time chambers or to an analytical unit. Pumps are
employed to move fluid through the system, and temperature control
is achieved using heat exchangers. The system is controlled by a
processor, which in one preferred embodiment is implemented using a
personal computer. The analytical unit is configured to analyze
each product produced by the system. Based on the analysis, the
controller identifies the process conditions that provide the
highest yield of product.
[0011] For optimization of reactions, the relative concentrations
of reactants must be varied. Prior art optimization methods
generally require batches of reactants at different concentrations
levels to be prepared before a set of reactions are executed. In
the present invention, dilution pumps are coupled to reactant feed
lines and a solvent supply. The controller can vary the amount of
solvent introduced into a reactant supply line, thereby
automatically varying the concentrations of the reagents. Thus, the
manual preparation of the reagent solutions at different
concentrations of the prior art is eliminated. Not only does
elimination of manually preparing reagent solutions of differing
concentrations save time, but the fact that the reactant supply
vessels need not be physically disconnected from the system
eliminates problems associated with pausing the reaction operation
to change reactant supply vessels. The controller can be configured
to vary concentration randomly, or more preferably, according to a
predefined protocol. The ability to manipulate flow rates of
individual reactants, and the ability to add diluting solvents to
manipulate the concentrations of each reactant enable an infinite
number of combinations of flow rates and reactant concentrations to
be achieved. The flexibility of the reactant pumps and dilution
pumps enables concentration variations to be explored continuously,
whereas in the prior art, after a first set of experiments were
executed, new solutions having different concentrations had to be
prepared before additional optimization experiments could be
performed.
[0012] Each reactant (generally at least two reactants are
employed, although those of ordinary skill in the art will
recognize that other types of reactions can be optimized, such as
those using a single reactant and a catalyst, or three or more
reactants) is introduced into the reaction module, where the
reactants are mixed under the desired temperature conditions, and
the reaction is initiated. The combined reactants are then directed
into a first one of the plurality of residence time chambers. The
mixed reactants are pumped through the residence time chambers for
a period of time sufficient to enable the reaction to be completed.
Adjustments in the residence time can be achieved by modifying the
flow rates of both reactants. The residence time chambers are
employed sequentially, such that mixed reactants/product exiting
one residence time chamber are directed to a downstream residence
time chamber for additional holding time. Residence time can also
be varied by employing no residence time chambers for some
reactions, some residence time chambers for other reactions, and
all residence time chambers for still other reactions.
Significantly, major changes in residence time can be analyzed
efficiently by selectively changing the number of residence time
chambers used for a particular reaction. Routing a product through
one or more residence time chambers is achieved using appropriate
valving. Through intelligent valve switching algorithms,
information on multiple residence times can be obtained
efficiently. In contrast, prior art optimization techniques either
explored incremental changes in residence times, or explored larger
changes by removing or installing residence time chambers, which
generally required bringing the system to a temperature where an
operator can install/remove a residence time chamber, purge both
the heat transfer fluid and reactant/product liquids, reassemble
the system, and heat up the system up to operating conditions
before additional experimentation can be performed.
[0013] The present invention encompasses methods for using such a
continuously operating system for optimization of reaction
parameters, or continuous kinetic parameter evaluations of chemical
reactions. The primary goal of such methods is the efficient and
rapid determination of optimal reaction performance criteria, such
as yield, conversion, and selectivity. Operating conditions such as
temperature, reactant concentrations, reactant concentration
ratios, and residence times can be modified. Using these operating
conditions and the resulting performance data, it is possible to
calculate important chemical reaction parameters, such as
activation energies, and reaction orders for the reactants being
analyzed. Since these parameters are independent of the reactor
used, the information can be used for numerically optimizing the
performance of the reaction in any vessel.
[0014] The optimization experiments can be performed according to
several different protocols. In one embodiment, testing conditions
are predefined, and the system is operated continuously until the
predefined range of each variable is tested. The optimal reaction
conditions providing the maximum performance can be evaluated after
the entire range of parameters has been analyzed and all data has
been collected. In another embodiment, the optimization can be
performed in real-time. In this mode of operation, the performance
information is reviewed as soon as it is obtained, to determine if
new testing conditions can be defined based on the data obtained
from previous experiments. In yet another embodiment, specific
reaction parameters, such as temperature, concentration, and
reactant equivalence are varied according to periodic functions,
while data are continuously collected. These data can then be
reviewed to identify optimal operating conditions. Reactant
equivalence (stoichiometric ratio) can be varied based on changes
to concentrations of individual reactants, as well as by
manipulating flow rates.
BRIEF DESCRIPTION OF THE DRAWING FIGURES
[0015] The foregoing aspects and many of the attendant advantages
of this invention will become more readily appreciated as the same
becomes better understood by reference to the following detailed
description, when taken in conjunction with the accompanying
drawings, wherein:
[0016] FIG. 1A schematically illustrates a continuous flow
optimization system in accord with the present invention;
[0017] FIG. 1B schematically illustrates the system of FIG. 1A with
valves having been manipulated to direct a flow of fluid exiting a
reaction module to an analytical unit, thus bypassing each
residence time chamber;
[0018] FIG. 1C schematically illustrates the system of FIG. 1A with
valves having been manipulated to direct a flow of fluid exiting
the reaction module to a first residence time chamber, and then to
the analytical unit, thus bypassing the second and third residence
time chambers;
[0019] FIG. 1D schematically illustrates the system of FIG. 1A with
valves having been manipulated to direct a flow of fluid exiting
the reaction module to the first residence time chamber, then to
the second residence time chamber, and then to the analytical unit,
thus bypassing the third residence time chamber;
[0020] FIG. 1E schematically illustrates the system of FIG. 1A with
valves having been manipulated to direct a flow of fluid exiting
the reaction module to the first residence time chamber, then to
the second residence time chamber, then to the third residence time
chamber, and finally to the analytical unit, thus achieving a
maximum residence time;
[0021] FIG. 1F schematically illustrates a preferred configuration
for residence time chambers and the reaction module of the system
of FIG. 1A;
[0022] FIG. 2 is a flow chart including the logical steps employed
in a first method for optimizing parameters of a continuously
running system in accord with the present invention;
[0023] FIG. 3 is a flow chart including the logical steps employed
in a second method for optimizing parameters of a continuously
running system in accord with the present invention;
[0024] FIG. 4 is a flow chart including the logical steps employed
in yet another aspect of the present invention, in which reaction
parameters are varied according to periodic functions;
[0025] FIG. 5 is a graph showing the result of varying temperature,
concentration, and reagent parameters according to periodic
functions;
[0026] FIG. 6 is an exemplary graph showing a possible result of
the periodic variation of FIG. 5; and
[0027] FIG. 7 illustrates graphs of temperature and percent yield
versus time, showing how thresholds identified in data collected
from the continuously running optimization system can be used to
determine new limits for variables in further optimization
experiments.
DESCRIPTION OF THE PREFERRED EMBODIMENT
[0028] The present invention employs a continuously operating
system that enables reaction parameters to be varied over time, to
dramatically reduce the time needed for optimizing a chemical
reaction, compared to the prior art batch-based optimization
techniques discussed above. FIG. 1A illustrates the functional
elements of an automatically controlled, continuously running
reaction optimization system 110. A system controller 148 is used
to control the system, including selecting reactant concentrations
and controlling reactant flow rates, solvent flow rates,
temperature conditions, pressure conditions (for systems configured
to vary pressure conditions), and residence times. System
controller 148 preferably comprises a computer or other
programmable computing device; however, it should be understood
that an application specific integrated circuit (ASIC) can
alternatively be beneficially employed for the system controller.
System controller 148 is operatively connected (through connectors
A) to a Reactant A pump 120 (which is associated with selecting a
flow rate for Reactant A), a Reactant B pump 116 (which is
associated with selecting a flow rate for Reactant B), a Solvent A
dilution pump 118a (which is associated with controlling a
concentration of Reactant A), a Solvent B dilution pump 118b (which
is associated with controlling a concentration of Reactant B), a
reaction module 122 (in which Reactant A and Reactant B are
thermally conditioned and mixed), a heat exchanger 150 (for
controlling thermal conditions inside reaction module 122), a
plurality of residence time chamber valves 133, 134, 136, and 138
(which, as will be discussed in greater detail below, enable an
inline analytical unit to receive a fluid from the reaction module,
or from a specific residence time chamber/residence time unit), an
outlet valve 140, a waste container 143, an automatic online
analytical unit 146 (i.e., a detection device), and an optional
throttle valve 142. If desired, additional heat exchangers 152 can
be employed, such that system controller 148 can independently
control the temperature in each residence time chamber.
[0029] Heat exchanger 150 is coupled in fluid communication with
reaction module 122 and employs a temperature-conditioned fluid to
control the temperature within the chemical reactor. Similarly, if
used, optional heat exchangers 152 are coupled in fluid
communication with each residence time chamber 124, 126 and 128, to
employ a temperature-conditioned fluid for controlling the
temperature within each residence time chamber. Alternatively, the
temperature-conditioned fluid from heat exchanger 152 can also be
coupled in fluid communication with each residence time chamber to
control its temperature by circulating the temperature-conditioned
fluid therethrough. However, in such a configuration, the
temperature condition in each residence time chamber cannot readily
be independently varied. Particularly where one optimization
parameter to be explored is providing different thermal conditions
to residence time chambers versus a reaction module, the ability to
independently control thermal conditions in each residence time
chamber is desirable and can be implemented by using a separate
temperature-regulated fluid to control the temperature in the
residence time chamber(s).
[0030] The volumes of Reactant A supply 114 and Reactant B supply
112, as well as the volumes of solvent supply 113, are functions of
the reaction to be optimized. For example, if two parts of Reactant
A need to be mixed with one part of Reactant B, then about twice as
much Reactant B ought to be available. Preferably, sufficient
volumes of each reactant and solvent are provided such that system
110 can be operated continuously until each optimization parameter
has been tested, so the system does not have to be shut down to
re-supply a required fluid. As noted above, many reactions
requiring optimization will be based on combining two different
reactants that react under appropriate conditions to generate a
desired product. However, some reactions are based on exposing a
single reactant to a catalyst under specific conditions to achieve
a desired product, and other reactions are based on combining more
than two reactants to achieve a desired product. Those of ordinary
skill in the art will readily recognize that system 110 can be
simply modified to enable optimization reactions to be continuously
performed for reactions requiring less than, or more than, the two
reactants indicated in FIG. 1A. Thus, the present invention is not
limited to the optimization of reactions utilizing two reactants,
although a significant use of the present invention will likely be
for the optimization of such reactions. Reactant A and Reactant B
will generally be liquids, although one or both of the reactants
can be gaseous. Solid reactants will generally be dissolved or
suspended in a liquid for ease of handling and processing in the
system, before being placed in supply 112 or supply 114.
[0031] The solvent employed must be compatible with Reactant A and
with Reactant B. In most applications, a common solvent will exist
for both Reactant A and Reactant B, and a single solvent supply can
be employed. If necessary, separate solvent supplies can be
provided, with a first solvent being used to dilute Reactant A, and
a second solvent that is different than the first being used to
dilute Reactant B. Solvent in the present invention is employed to
selectively vary the concentration of Reactant A and Reactant B, to
enable the effects of varying the concentrations of the reactants
on product yield and quality to be determined. Those of ordinary
skill in the art will readily recognize that the selection of an
appropriate solvent for a specific reactant (or pairs of reactants)
is well within the skill of the ordinary practitioner of this art.
In addition to being employed to change a relative concentration of
a reactant, the solvent can also be employed to flush the system.
Because the same reagents are being employed in the plurality of
optimization experiments, it is likely that the system will not
need to be flushed between every experiment. Particularly for
optimizations requiring many different optimization experiments, it
may be desirable to periodically flush the system to minimize any
residue from building up in the reaction module, residence time
modules, and fluid lines. Such flushing may be performed more
frequently when the reactants or products have a relatively high
viscosity.
[0032] Pump 120 controls the flow rate of Reactant A, and pump 116
controls the flow rate of Reactant B. When performing optimization
experiments for a specific reaction, flow rates will at times be
held constant (while other parameters are varied), and at times,
the flow rate(s) will be the parameter(s) that are being varied.
Dilution pump 118a is employed to change the relative concentration
of Reactant A, by introducing a solvent into the flow of Reactant A
entering the reactor. Dilution pump 118b is similarly employed to
change the relative concentration of Reactant B, by introducing a
solvent into the flow of Reactant B entering the reactor. When
solvent is added to dilute a reactant and a flow rate for that
reactant is to be held constant, preferably a flow rate of that
reactant will be decreased by an amount required to offset an
increase in flow rate attributable to the solvent. A variety of
different types of pumps can be beneficially employed. Preferably,
each heat exchanger will incorporate its own pump (not separately
shown) to supply the temperature-conditioned heat transfer medium
to the reaction module (or residence time chamber), under the
control of system controller 148.
[0033] While not shown, it is expected that pressure sensors and
filters can optionally be used in association with each pump in
system 110. A signal produced by the pressure sensors will provide
confirmation to system controller 148 that the reactants are
flowing, and the filters can be employed to filter any particulate
matter that may have contaminated Reactant A supply 114, Reactant B
supply 112, solvent supply 113, and/or any heat transfer fluid. In
a preferred embodiment in which the reaction module incorporates a
micro reactor, these filters are particularly important, since the
fluid channels within a micro reactor are characteristically very
small in size. Thus, even relatively small particles can clog these
channels and significantly impair the efficiency of the micro
reactor. Preferably, system controller 148 is programmed to alert a
user to check the filters when pressure sensors indicate a change
in pressure in the system, as such a pressure change may be
indicative of a clogged filter. If desired, system controller 148
can be configured to periodically terminate the flow of reactants
in the system, so that the solvent supply can be used to flush the
system.
[0034] As noted above, reaction module 122 is preferably
implemented as a micro reactor, such that only relatively small
volumes of reagents are required for each optimization experiment.
So long as the volume of product produced is sufficient to enable
accurate analysis, there is no need to generate large volumes of
product during optimization experiments. It should be noted that
FIG. 1A does not attempt to illustrate the fluid paths of reactants
within reaction module 122. Two reactants are directed into
reaction module 122, and either a single desired chemical product
exits the reaction module 122, or alternatively, mixed and
partially reacted reactants exit the reaction module and enter one
or more residence time chambers, to be pumped through for a period
of time to enable the reaction to complete. Thus, it will be
appreciated that reaction module 122 includes at least a mixing
unit, and that thermal conditions inside reaction module 122 are
controlled.
[0035] While in one preferred embodiment, the chemical reactor
within reaction module 122 is a micro reactor, a macro-scale
reactor could alternatively be used in the present invention. Micro
reactors are generally characterized as incorporating fluidic
structures of less than 1 mm in size, especially with respect to
reactant fluid pathways. However, the present invention is not
limited to reaction modules that include a micro reactor, because
it is also contemplated that the reaction module can incorporate a
chemical reactor whose fluidic structures are larger or even
substantially larger in size than the micro-scale fluidic
structures generally associated with micro reactors.
[0036] If required, reaction module 122 will include structural
elements necessary to facilitate the reaction between the two
reactants selected. In some cases, one of the reactants may need to
be exposed to a catalyst for a reaction to be initiated or carried
out efficiently. Other reactions require an electrochemical, a
photochemical, and/or other forms of stimulus. Process parameters
that can be beneficially incorporated into reactors for use in the
present invention include magnetic, piezoresistive, piezoelectric,
shape memory, radioactive, catalytic, optical, electromagnetic, and
electrostatic parameters. Each such parameter is preferably capable
of being controlled by system controller 148.
[0037] Once a quantity of Reactant A and a quantity of Reactant B
(diluted as required using dilution pumps 118a and 118b) have been
introduced into reaction module 122 and have been suitably mixed
and thermally conditioned, the mixed reactants can either be routed
to analytical unit 146 for analysis, or introduced into one of
residence time chambers 124, 126, and/or 128, by manipulating
appropriate one or more of valves 133, 134, 136, and 138. FIGS.
1B-1E, discussed in detail below, illustrate the flow paths for
directing fluid from reaction module 122 to analytical unit 146
without passing through any residence time chamber (FIG. 1B),
directing fluid from reaction module 122 to residence time module
124, and then to analytical unit 146 (bypassing residence time
chambers 126 and 128; FIG. 1C), directing fluid from reaction
module 122 to residence time module 124, then to residence time
chamber 126, and then to analytical unit 146 (bypassing residence
time chamber 128; FIG. 1D), and directing fluid from reaction
module 122 to residence time module 124, then to residence time
chamber 126, then to residence time chamber 128, and then to
analytical unit 146 (thereby achieving a maximum residence time for
a given flow rate; FIG. 1E). Those of ordinary skill in the art
will recognize that other valve configurations can be employed to
achieve the same functionality (i.e., the ability to select a flow
path between the reaction module and the analytical unit, such that
a desired number of residence time units are utilized). One
alternative configuration would be to utilize a fewer number of
multi-port valves. Thus, the specific valve configuration shown is
intended to be exemplary, rather than limiting on the scope of the
present invention. Outlet valve 140 enables controller 148 to
direct fluid either to waster container 143 or toward a product
outlet. The optional throttle valve 142 enables controller 148 to
selectively vary pressure conditions in system 110, by reducing a
flow rate downstream of valve 142 (i.e., toward valve 140), which
causes the pumps upstream of valve 142 (the solvent pumps and the
reagents pumps) to increase the pressure in the reaction module and
any residence time chambers being used.
[0038] Residence time chambers can be used in a variety of ways.
For example, residence time chambers can be used to enable a higher
throughput to be achieved, by selecting a flow rate that causes
partially reacted reagents to be discharged from the reaction
module before the reaction is complete. In such an embodiment,
residence time chambers are used in parallel. The residence time
chambers in the present invention can also be used sequentially
(i.e., wherein a material exits the reaction module and is directed
into a first residence time chamber, and is then subsequently
directed from the first residence time chamber into one or more
additional residence time chambers coupled in series, before being
analyzed by the inline analysis unit). The valving of system 110
enables product/mixed reactants exiting reaction module 122 to be
directed into the analytical unit after passing through residence
time chamber 124 only, after passing through residence time
chambers 124 and 126, or after passing through residence time
chambers 124, 126 and 128. If only one of the residence time
chambers is to be used, then product/mixed reactants exiting the
reaction module are directed into residence time chamber 124, and
then diverted to the analytical unit before such fluid enters the
additional residence time units. As a further example, if all three
residence time chambers are used, then the product/mixed reactants
exiting the reaction module are directed in residence time chamber
124, then into residence time chamber 126, and finally into
residence time chamber 128, and then diverted to the analytical
unit. Using residence time chambers sequentially enables the
present invention to vary residence time as one of the reaction
parameters, so that data can be collected, and an optimal residence
time determined.
[0039] As noted above, using residence time chambers enables more
reactions to be processed per unit time by reaction module 122. The
reagents are mixed and thermally conditioned in reaction module 122
and then transferred to one or more residence time chambers
sequentially. While three residence time chambers are shown, it
should be understood that additional residence time chambers can be
employed. For example, if the reaction to be optimized requires
five minutes for the reaction to complete, and mixing and thermal
conditioning can be achieved in reaction module 122 in one minute,
then four residence time chambers (numbered 1-4), each of which has
a volume sufficient such that at a predefined flow rate it will
take a minute for a fluid to pass through the residence time
chamber, will enable the reagents to be continually introduced into
the reaction module. The first set of reagents (i.e., predetermined
quantities of Reagent A and Reagent B, which can be identical if a
parameter such as temperature is being tested) will be introduced
into reaction module 122, mixed and thermally conditioned, and
directed to residence time chamber 1. The fluid exiting residence
time chamber 1 has been processed for two minutes (one minute in
the reaction module and one minute in residence time chamber 1),
and is directed to residence time chamber 2 (for an additional
minute of processing time). The fluid exiting residence time
chamber 2 is directed to residence time chamber 3 (for yet another
minute of processing time). The fluid exiting residence time
chamber 3 is directed to residence time chamber 4, for the fifth
required minute of processing time. Note that after the initial
fluid exits from residence time chamber 1, additional fluid is
exiting the reaction module and entering residence time chamber 1.
This simplified scenario does not include the times required for
filling and emptying the residence time chambers (or the reaction
module), but it does demonstrate how the use of residence time
chambers enables the system to operate continuously, when the
reaction module itself does not provide the required residence time
for the reaction to complete.
[0040] Adjustments in the residence time can also be achieved by
modifications of the flow rates of the reagents (and solvent, if
employed to vary a reagent concentration). Because pumps generally
exhibit linear behavior over a limited range, reactant flow rates
can be accurately varied only throughout a defined range by
controlling a speed of the pumps. Generally, this range enables
flow rates to be varied over at least one order of magnitude.
Hence, the residence time impact can be analyzed for a factor of
10-20, for a fixed system volume.
[0041] Preferably, heat exchangers 152 are employed to maintain the
same thermal conditions in the residence time chambers as are
present in reaction module 122. As noted above, one optimization
parameter that could be tested is the effect of different residence
time chambers temperatures on product yield and/or quality.
[0042] In one embodiment of the present invention, each residence
time chamber includes a helically-coiled capillary passage, and the
length of the capillary passage controls a residence time of the
reactants in the residence time chamber. In a preferred embodiment
in which reaction module 122 includes a micro reactor, the
capillary passage is of sufficient length to achieve a 45-minute
residence time at a flow rate of one milliliter per minute.
Generally, a residence time of 45 minutes is sufficient for the
majority of most chemical reactions to reach completion. However,
different reactions can require different residence times, and the
residence time chambers must be matched to the requirements of the
reaction being optimized. Furthermore, while capillary passages can
serve as effective residence time chambers, it should be understood
that the specific design of each residence time chamber is not
critical. As long as each residence time chamber provides a
sufficient volume in which the incompletely reacted mixture of
reactants exiting the reaction module can reside until the reaction
is complete, the particular physical configuration of the volume is
not critical.
[0043] Various reactions can be performed in system 110 that are
pressure dependent. For example, reactions involving decreasing
volumes, increasing boiling points, or increasing gas
concentrations in a liquid phase are pressure dependent. Thus, it
may be desirable to enable a reaction to occur at a predefined
pressure. To increase the pressure along a reaction path requires a
throttle at the distal end of the reaction path. Preferably, valve
142 acts as a throttle, so that partially closing the valve causes
pumps 116, 118a, 118b, and 120 to produce higher pressures in the
reaction module (and residence time chambers) in order to maintain
a constant flow rate. Note that valve 142 is optional, because
while the ability to vary pressure conditions is useful, many other
optimization experiments can be performed without changing pressure
conditions (i.e., by varying parameters such as concentration,
stoichiometric ratios, temperature, and residence time).
[0044] With respect to heat exchanger 150, it is preferred that the
heat transfer media used be fluidic in nature. While solid phase
heat transfer media are known in the art (such as silica), assuring
a continual flow of such solid phase heat transfer media through
small passages in heat exchangers can be difficult, and in general,
fluidic heat transfer media are preferred in the present invention.
Preferably, system 110 can control (and measure) thermal conditions
over a range of about -80.degree. C. to about 200.degree. C. While
not shown separately, it is preferable for reaction module 122 to
include a plurality of temperature sensors disposed so as to enable
temperature conditions to be monitored in various selected
locations in the reaction module. Similarly, it is desirable to
also include temperatures sensors at a plurality of locations in
the residence time chambers. These temperatures sensors can also be
employed to measure the temperature of the heat exchange media
entering and leaving reaction modules 122 (and/or the residence
time chambers). Particularly important locations for incorporating
such sensors include those in system 110 where variations and
temperature gradients are expected, due to the release or
absorption of energy as a result of reaction kinetics.
[0045] It should be noted that dilution pumps 118a and 118b are
very important for enabling system 110 to operate continuously and
to vary the relative concentrations of each reactant as required to
complete the optimization reactions. The concentration of Reactant
A and Reactant B can be easily varied (via dilution) by mixing more
or less solvent into the reactant feed stream using the dilution
pumps. Thus, the concentration of either reactant can be adjusted
automatically during the process, instead of requiring manual
preparation of different reagent concentrations as in done in the
prior art. Significantly, the incorporation of the dilution pumps
eliminates the problem of requiring reagent supply vessels to be
changed (to enable manually prepared solutions of varying
concentration to be introduced into the reaction module). In prior
art systems, modifying reactant concentrations is only possible by
either terminating an experiment, or by waiting until the
experiment is complete, and starting a new experiment using
different concentration reactant(s). In contrast, controller 148
can vary the concentration of the reactants continuously during the
performance of the optimization experimentation, either randomly,
or more preferably, in a predefined manner.
[0046] Referring now to analytical unit 146, those of ordinary
skill in the art will recognize that a variety of analytical
devices are available that might be used for this component, and
certain devices are more suited to the detection of a specific
product or to monitor a specific quality of the product than
others. Clearly, analytical unit 146 must be capable of detecting
the desired product of the reaction to be optimized, to enable the
quantity and quality of the product to be determined. Preferably a
quantitative measurement is obtained, although a qualitative
measurement capable of distinguishing between different levels of
quality would be useful as well. The product (from reaction module
122 or one of the residence time chambers) is passed over a
measuring cell, introduced into a measurement device, exposed to
quantum particles, or collected as appropriate for the analytic
unit selected. By selecting a non-destructive analytical technique,
a second analytical unit (not separately shown) can be used to
determine additional information, such as more detailed composition
information (for example, the byproducts that are present in the
desired product). Analytical units based on the following
techniques can be beneficially employed, although it should be
understood that the present invention is not limited only to the
techniques discussed herein. Non-destructive techniques that can be
used include Infrared spectroscopy (including Fourier Transform
techniques), Raman spectroscopy, ultraviolet (UV) spectroscopy, and
nuclear magnetic resonance (NMR) spectroscopy. Destructive testing
techniques that might be used include mass spectroscopy and
separation-based analytic techniques, including high performance
liquid chromatography (HPLC) and gas chromatography (GC).
[0047] FIGS. 1B-1E are based on FIG. 1A, and illustrate various
flow paths enabled by manipulating valves 133, 134, 136, and 138.
The flow paths enabled in each Figure are shown in bold. FIG. 1F
provides details of a particularly preferred configuration of the
residence time chambers and the reaction module. Note that while
FIGS. 1B-1E do not indicate a flow of solvent, it should be
understood that the use of a solvent to dilute Reagent A or Reagent
B does not affect the flow paths enabled by the manipulation of
valves 133, 134, 136 and 138. Each valve (i.e., valves 133, 134,
136, and 138, as well as valves 140 and 142) is controllably
coupled to controller 148. In one embodiment, controller 148 is
configured to selectively actuate valves 133, 134, 136 and 138
according to a predefined pattern to achieve a plurality of
optimization experiments, each with a different residence time
(Table 1, discussed in detail below, describes one such pattern of
11 optimization experiments covering a relatively broad range of
residence times achievable using only three residence time
chambers). A working model based on FIG. 1A employs a reaction
module having a volume of 2 ml, and residence time chambers each
having a volume of 15 ml. Based on a flow rate of 1 ml/min, in the
working model Reactants A and B can be processed for as little as 2
minutes (using no residence time modules), or for as long as 47
minutes (using all three residence time modules). Additional
residence time variations can be achieved by altering the flow
rate. For example, at a flow rate of 10 ml/min, the minimum
residence time in the working model is 0.2 minutes (24 seconds),
and the maximum residence time in the working model is 4.7
minutes.
[0048] It should be noted that the valve configuration (i.e.,
valves 133, 134, 136, and 138) of FIGS. 1A-1F have been
specifically selected for use where the reaction module and each
residence time unit are implemented by stacked plates in which
openings in individual plates define fluid channels. Such a
configuration enables a compact system to be designed, but limits
the ability for valves to be placed in between adjacent residence
time units and in between the reaction module and residence time
chamber 124. Reaction module 122 and residence time modules 124,
126 and 128 are physically stacked on top of one other in such an
implementation, as indicated in FIG. 1F. Separation plates 123a-c
are disposed between the stacked reaction module/residence time
chambers. Each separation plate includes fluid channels enabling
the immediately adjacent elements to be in fluid communication with
each other. Thus, an outlet of reaction module 122 is in fluid
communication with an inlet of residence time chamber 124 via plate
123a, an outlet of residence time chamber 124 is in fluid
communication with an inlet of residence time chamber 126 via plate
123b, and an outlet of residence time chamber 126 is in fluid
communication with an inlet of residence time chamber 128 via plate
123c. The fluid channels in plate 123a are coupled in fluid
communication with valve 133, the fluid channels in plate 123b are
coupled in fluid communication with valve 134, and the fluid
channels in plate 123c are coupled in fluid communication with
valve 136. Before the system is used for a first optimization
reaction, the entire system is flooded with fluid (i.e., the
reaction module, the residence time chambers, and the fluid lines
coupling such elements). The reaction module and the residence time
chambers are always in fluid communication. As a result of pressure
build-up against closed valves (valves 133, 134, 136 and 138),
fluid that is introduced into the system via the solvent pumps and
the reactant pumps flows solely through the open fluid pathways
(pathways which are defined by valves 133, 134, 136 and 138). A
detailed description of the various fluid paths enabled by valves
133, 134, 136 and 138 is provided below. It should be understood
that other valve configurations and other fluid paths can be
employed to achieve the desired functionality of enabling fluid
exiting from the reaction module to be directed to the analytical
unit without passing through each residence time chamber, and to
enable fluid exiting each residence time chamber to be directed to
the analytical unit. Thus the valve configuration of FIGS. 1A-1F is
exemplary, and not intended to limit the present invention.
[0049] In FIG. 1B, valves 133, 134, 136 and 138 are manipulated
such that fluid exiting reaction module 122 bypasses each reaction
module, and proceeds directly to analytical unit 146 (each of FIGS.
1B-1E is based on valve 140 directing fluid toward the product
outlet, as well as valve 142 not being used as a throttle). Based
on a flow rate of 1 ml/min and reaction module 122 having a 2 ml
volume, a residence time of 2 minutes is achieved. Valve 133
includes two ports, and when valve 133 is open (as in FIG. 1B)
fluid from reaction module 122 is able to pass through valve 133
and proceed to valve 134. When valve 133 is closed, fluid from
reaction module 122 flows into the fluid line coupling reaction
module 122 to valve 133, but cannot flow past valve 133. Regardless
of whether valve 133 is open or closed, reaction module 122 is in
fluid communication with residence time chamber 124. When valve 133
is open, and valves 134, 136, and 138 are properly positioned, even
though reaction module 122 is in fluid communication with each
residence time chamber, the only path enabling fluid to reach
analytical unit 146 passes through open valve 133 (as indicated by
the bold lines), and not via a fluid path passing through any
residence time chamber.
[0050] In FIG. 1C, valve 134 is manipulated so that fluid from
residence time chamber 124 is directed to valve 136 (and on to
analytical unit 146). The states of valves 136 and 138 in FIG. 1C
remain unchanged from their respective states in FIG. 1B, so that
residence time chambers 126 and 128 remain bypassed. Thus, the
analytical unit receives fluid that passed through reaction module
122 and residence time chamber 124, but not residence time chambers
126 and 128. Valve 134 includes three ports, and is configured such
that at any one time, two of the three ports are in fluid
communication. In FIGS. 1C-1E, valve 134 is configured to place a
fluid line coupling an outlet of residence time chamber 124 in
fluid communication with a port of valve 136 (as indicated by the
bold lines). Note that regardless of the position valve 133, it is
the position of valve 134 that determines whether fluid from valve
133, or fluid from an outlet of residence time unit 124, is
directed to valve 136. In at least one embodiment, valve 133 is
eliminated, and valve 134 alone determines whether fluid from an
outlet of reaction module 122, or an outlet of residence time
chamber 124, is directed to valve 136. Based on a flow rate of 1
ml/min, reaction module 122 having a 2 ml volume, and each
residence time chamber having a volume of 15 ml, a residence time
of 17 minutes is achieved.
[0051] In FIG. 1D, valve 136 is manipulated so that fluid exiting
residence time chamber 126 is directed to valve 138 (and on toward
analytical unit 146). The state of valve 138 in FIG. 1D remains
unchanged from its state in FIGS. 1B and 1C, so that residence time
chamber 128 remains bypassed. Thus, the analytical unit receives
fluid that passed through reaction module 122, residence time
chamber 124, and residence time chamber 126, but not residence time
chamber 128. Valve 136 includes three ports, and is configured such
that at any one time, two of the three ports are in fluid
communication. In FIGS. 1D-1E, valve 136 is configured to place a
fluid line coupling an outlet of residence time chamber 126 in
fluid communication with a port of valve 138 (as indicated by the
bold lines). Note that regardless of the positions of valves 133
and 134, it is the position of valve 136 that determines whether
fluid from valve 134, or fluid from an outlet of residence time
unit 126, is directed to valve 138. Based on a flow rate of 1
ml/min, reaction module 122 having a 2 ml volume, and each
residence time chamber having a volume of 15 ml, a residence time
of 32 minutes is achieved.
[0052] In FIG. 1E valve 138 is manipulated so that fluid exiting
residence time chamber 128 is directed to analytical unit 146, and
no residence time chamber is bypassed. Valve 138 includes three
ports, and is configured such that at any one time, two of the
three ports are in fluid communication. In FIG. 1E, valve 138 is
configured to place a fluid line coupling an outlet of residence
time chamber 128 in fluid communication with analytical unit 146.
Note that regardless of the positions of valves 133, 134 and 136,
it is the position of valve 138 that determines whether fluid from
valve 136, or fluid from an outlet of residence time unit 128, is
directed to analytical unit 146. Based on a flow rate of 1 ml/min,
reaction module 122 having a 2 ml volume, and each residence time
chamber having a volume of 15 ml, a residence time of 47 minutes is
achieved. Of course, additional residence times (for any of FIGS.
1B-1E) can be achieved by manipulating the flow rate as well (such
as by manipulating the reagent pumps and/or solvent pumps).
[0053] Based on FIGS. 1B-1E, one potential series of optimization
experiments that can be implemented is as follows. First, the
entire system is flushed using a suitable solvent, with each valve
being manipulated through their possible states (thus solvent flows
into waste container 143 as well as the product outlet downstream
of analytical unit 146). This reduces air bubbles in the system
which could lead to unstable signals from the analytical unit
(which in one preferred embodiment is an infrared (IR)
spectrophotometer). The system valves are manipulated to achieve
the configuration shown in FIG. 1B (the reaction module output is
directed to analytical unit 146) and a series of experiments are
conducted using different flow rates (preferably starting with a
highest flow rate, thus a lowest residence time). Valve 134 is
manipulated to achieve the configuration illustrated in FIG. 1C
(i.e., residence time module 124 is being used), and another series
of experiments is conducted using different flow rates (again
preferably starting with a highest flow rate). Valve 136 is
manipulated to achieve the configuration illustrated in FIG. 1D
(i.e., residence time modules 124 and 126 are being used), and
another series of experiments is conducted using different flow
rates (again preferably starting with a highest flow rate).
Finally, valve 138 is manipulated to achieve the configuration
illustrated in FIG. 1E (i.e., all residence time modules are being
used), and another series of experiments is conducted using
different flow rates (again preferably starting with a highest flow
rate). Preferably, after each change of residence volume or flow
rate, the system is allowed to equilibrate (as indicated by a
stable signal from the analytical unit) before the next parameter
is varied.
[0054] FIG. 2 is a flowchart 210 showing the overall logic used in
system 110 of FIG. 1A for continually varying and testing a
plurality of different reaction parameters to identify optimal
reaction parameters for a specific chemical reaction. In a block
212, a first experiment is performed (i.e., Reagent A and Reagent B
are introduced into the reaction module under a defined set of
reaction parameters). In a block 214, the resulting product is
tested and the analytical result (e.g., the product yield or
quality) is recorded. In a block 216, one of the reaction
parameters (such as temperature, Reactant A concentration, Reactant
B concentration, flow rate, or residence time) is varied, and
additional quantities of Reagent A and Reagent B are introduced
into the reaction module. In a block 218, the product is analyzed
and the results recorded.
[0055] A decision block 220 determines if additional parameters
need to be varied. If so, then the logic turns to block 216. If
not, the optimization experiments are completed, and the data
collected can be reviewed to determine one or more optimal
parameters.
[0056] The process in FIG. 2 is based on identifying, in advance,
specific parameters to be varied. For example, before the
optimization procedure is started, it might be determined that the
concentrations of Reagent A and Reagent B will be varied to test a
range of concentrations of +/-20% (or some other desired
percentage) from baseline concentrations, in 5% increments (or some
other desired increment). Temperature conditions can be similarly
varied by +/-50 degrees (or some other desired range) from a
baseline temperature, in 2 degree increments (or some other desired
increment). The baseline values correspond to the initial values
selected for the first experiment. The system automatically varies
each parameter, until all possible combinations and permutations of
the selected variables have been tested.
[0057] The process described by flowchart 310 in FIG. 3 is based on
the logic described above in connection with FIG. 2 and includes
modifications to increase efficiency. In a block 312, a first
experiment is performed using the baseline parameters. In a block
314, the product is tested and the analytical results (such as the
product yield) are recorded. In a block 316, one of the reaction
parameters is varied, and additional quantities of Reagent A and
Reagent B are introduced into the reaction module. In a block 318,
the product is analyzed and the results recorded.
[0058] A decision block 320 determines if additional parameters
need to be varied. If not, the initial optimization experiments are
completed. If so, then in a block 322, the data collected are
evaluated to identify any trends. For example, data collected might
indicate that temperature conditions below a certain level result
in poor yield, and if such a trend is detected, then no additional
low temperature optimization experiments below that level need to
be performed. Thus, in a block 324, new testing conditions are
defined, and the logic returns to block 316 to carry out additional
optimization experiments based on the new testing conditions.
[0059] The logic employed in FIG. 3 is useful when the initial
optimization spans a broad range. For example, the initial
optimization may define testing parameters for varying temperature
over a 100 degree range in 10 degree increments. If a temperature
at the low end of the range is tested, followed by a temperature at
the middle of the range, and then a temperature at the high end of
the range, the results for those experiments can be compared to
determine which temperature (low, medium or high) results in a
better yield. If the middle range is best, then new testing
conditions for temperature can be defined, +/-25% (or some other
logical range that is narrower than the original defined range)
relative to the initially identified best temperature. This
approach tends to reduce the number of optimization experiments
required, because additional optimization experiments based on
temperatures in the low end and high end of the original testing
parameters need not be performed.
[0060] FIG. 4 illustrates a flowchart 410 that can be used for
continuously collecting data from a system 110 in which
parameter(s) is/are varied according to a periodic function. For
example, parameters such as temperature, reactant concentration,
and reactant equivalence can be varied according to periodic
functions while the system is operated continuously. Reactant
equivalence (stoichiometric ratios) can be varied by changing
reactant flow rates, and by changing reactant concentration. For
example, if it is desired to perform an experiment where two parts
of Reactant A are combined with one part of Reactant B, such a
ratio can be achieved using a solution of Reactant A that is twice
as concentrated as a solution of Reactant B (where each reactant is
provided using the same flow rate), or a solution of Reactant A can
be provided using a flow rate that is twice as great as a flow rate
utilized for Reactant B (where each reactant is provided at the
same concentration). Data are continually collected, and the
optimal parameters can be selected from the accumulated data. In a
block 412, the periodic functions are defined, and the reagents are
introduced into the reaction module. As product passes into
analytical unit 146 (see FIG. 1A), data are collected and stored,
as indicated by a block 414. In a block 416, after the system has
been operated continuously for a time sufficient to enable the
parameter(s) to be varied according to the periodic function, the
results are analyzed to identify the optimal value(s) of the
parameter(s).
[0061] FIG. 5 graphically illustrates periodic functions for
reagent equivalence, as indicated by a line 502, temperature, as
indicated by a line 504, and reagent concentration, as indicated by
a line 506. The parameters fluctuate between defined upper and
lower limits. The system is operated until each parameter is able
to complete at least one period. Such periodic function-based
testing is also applicable to periodically varying residence time
as well, by simultaneously varying the flow rates of both reactants
without changing the number or reaction time chambers employed.
[0062] FIG. 6 graphically displays analytical results collected
using the periodic functions of FIG. 5, including a peak 602
clearly indicating the value of the parameter providing the highest
yield. FIG. 6 is thus based on simultaneous analysis of operating
conditions according to FIG. 5.
[0063] FIG. 7 is a graph illustrating how data collected by system
110 can be evaluated to identify new ranges for variables for
additional optimization experiments. Portion 702 of FIG. 7
represents a temperature parameter changing over time from a high
value to a low value. Portion 704 of FIG. 7 shows the signal
amplitude from analytical unit 146 for the same time period. Note
that signal amplitude corresponds to percent yield--the higher the
signal amplitude, the higher the yield. Note that FIG. 7 is based
on changing the value of a parameter based on a linear function (as
opposed to the periodic function of FIG. 5). Where discontinuities
are identified, additional optimization testing can be performed
using parameter values that are changed in smaller increments. It
should be understood that the linear variation illustrated in FIG.
7 is not limited to being implemented only for temperature; values
for other parameters (concentration and residence time, for
example) can also be varied linearly (as well as periodically, as
discussed above).
[0064] At a point 706, a slope of the signal amplitude changes
significantly, and the signal amplitude begins to drop rapidly.
Point 706 corresponds to a point 708 in portion 702. Thus, a
temperature 710 can be identified at the beginning of the sharp
decline in amplitude. That temperature is then selected as the new
lower limit, and additional optimization experiments are performed
based on the old temperature maximum and the new temperature
minimum.
[0065] In addition to the method discussed above in connection with
FIGS. 2, 3, and 4, system 110 can also be used to improve the
efficiency of prior art optimization methods. For example, consider
a statistical design optimization involving a set of 34 reaction
conditions. In the prior art, a plurality of different reactant
supply vessels would be required, with each different concentration
of a reactant requiring a separate vessel. Using system 110, only a
single reactant supply vessel is required for each reactant, since
the solvent supply is employed to enable dilution of the reactants
to be achieved to vary the concentration of the reactant. Using
system 110, an operator can elect to implement an off-line
analytical method (e.g., GC, HPLC), or a real-time method (e.g.,
FT-IR, Raman spectroscopy). If a real-time analytical method is
selected (for implementation by analytical unit 146 of FIG. 1A),
the reaction results are analyzed automatically, and a second set
of optimization reactions can be automatically defined based on the
optimal parameters identified from the first 34 reactions.
[0066] One of the prior art methods for optimization was based on
selecting two values for each reaction parameter, and then
performing an additional experiment based on the mean of each
variable (i.e., the 2.sup.n=1 design discussed in the Background of
the Invention). System 110 can also be used to increase the
efficiency of such an optimization process, as described below.
[0067] In this variation, system 110 is filled with solvent (to
flush the system), and the system is heated to the operating range
required by the high value temperature parameter. If a real-time
product analysis (i.e., using analytical unit 146) is implemented,
parameters for the optimization experiment with the shortest
residence time are implemented, and each reactant is individually
passed through the system, such that the analytical unit collects
data corresponding to each raw material (i.e., each reactant).
[0068] After flushing the system with solvent and cleaning with
solvent, the optimization experiment requiring the highest
concentration of reagents (and the highest excess of reagents) is
introduced into the system, and held for the longest residence time
defined in the reaction parameters to be tested. After
equilibration of the system, the appropriate valve is selected to
employ the residence time chamber providing the shortest residence
time. After the corresponding product has been analyzed, the
valving is manipulated to select both the longest residence time
and the lowest concentration value of the reactants. Because the
system is operating continuously, data corresponding to a value
between the high value and low value is automatically
collected.
[0069] As noted above, parameters can be changed in several ways. A
value for a parameter can be directly set (a step function). A new
value for a parameter can be set by making a series of small
changes between the old value of the parameter and the new value of
the parameter based on a linear function (the ramping shown in FIG.
7.). The value of the parameter can be changed continually based on
a periodic function (as shown in FIG. 5). Changing a parameter
value based on a series of relatively small linear changes enables
linear discontinuities to be investigated. Such linear
discontinuities can arise due to material decomposition (which can
be experienced with increasing temperature) or side reactions
(which can occur when the stoichiometric ratios of the reactants
are varied, as excess reagents may be available for such side
reactions). Whenever such linear discontinuities are identified,
additional optimization experiments wherein parameters are varied
in smaller increments about the discontinuity can be performed, to
better define the conditions associated with such
discontinuities.
[0070] For systems where nonlinear dependence on the operating
parameters is expected, the continuous nature of the optimization
process provided by the present invention implies that the system
will pass through the mean value of each parameter while passing
from the low value of the parameter to the high value of the
parameter (or vice versa). Hence, a selection between 2.sup.n=1
design experiments (based on assumptions of linearity) versus
3.sup.n design experiments (based on assumptions of nonlinearity)
are moot when the system of the present invention is used, because
the continuously operating system already obtains performance
criteria at the intermediate levels of the parameters.
[0071] When system 110 includes at least 2 residence time chambers
for processing the reactants, one can efficiently determine the
high, mean, and low settings for residence times. First, the
real-time analytic output for the system is determined based on the
product being directed through each residence time chamber
sequentially. After the data are collected, the final residence
time chamber is bypassed by selecting the appropriate valve. The
system responds by evaluating the product output of the preceding
residence time chamber, reducing the residence time by a factor
related to the residence time chamber volume and the total system
volume. Because the analytic signal is almost immediately updated,
the entire range of residence times can be evaluated using this
approach. For the high-mean-low type experimental design, this
capability implies that all 3 values for residence time can be
obtained by just switching two valves, and evaluating the
performance for all three residence times very quickly. For the
other variables (concentration, temperature, and equivalence), the
mean can be obtained by selecting one of the high value and the low
value, allowing the system to equilibrate, and collecting data.
Then a midpoint value is selected (i.e., a value between the high
and low values), the system is allowed to equilibrate, and data are
collected. Then the other of the high and low values is selected,
the system is allowed to equilibrate, and data are collected. Such
a process is significantly simpler than the 3.sup.n design
discussed above.
[0072] Collecting real-time analytical data from a continuously
running system offers a unique advantage. As values of the
parameters are changed, data are collected not just for each
individual value selected, but also for every value between the
selected values. The main problems with the statistical approach
used in the prior art batch approach, i.e., that no significant
improvement was identified because a range selected was too narrow,
or no improvement was identified when the reaction fails because
the range selected was too broad, can be automatically avoided, and
the optimization can be carried out without performing useless
experiments.
[0073] Even if in the case of continuously changing reaction
parameters from one value to the other value, there is an
uncertainty as to what specific reaction parameters correspond to
specific data collected by the system, a range of probable
parameters can be readily determined. Then a new optimization based
on that range can be performed to more specifically identify the
optimal parameters.
[0074] A significant improvement over the statistical design
experiments noted above can be achieved in a continuously running
system equipped with a real-time analytical device, if at least two
variables are changed at the same time, where the two variables are
changed between upper and lower values, and the changes are
implemented with different periods (see FIG. 5). Experimental
results for all possible permutations of the variables are thus
rapidly achieved. Analyzing the reaction results (yield of desired
product, amount of undesired product) regarding the time dependency
(i.e., using a Fourier transform analysis) shows the influence of
the different variables on the result (see FIG. 6). Even if the
precise reaction condition experienced by the product being
currently analyzed is not precisely known due to the periodically
changing conditions (i.e., in a non-stationary system), a
reasonable approximation of those conditions can be identified
based on the periodic function controlling the variability and
based on knowledge about the fluidic configuration of the system.
This approximation can then be used as a starting point for
additional optimization reactions that efficiently produce still
better values of the parameters for the optimal reaction.
[0075] Traditional experimental optimization techniques use a
plurality of individual reaction vessels. Each reaction vessel
includes reactants having different concentrations. Each reaction
vessel is operated for a given duration, and the products from each
reaction vessel are analyzed and evaluated. The present invention
enables reaction concentrations to be varied using a single
reaction module, and eliminates the need for each reagent
concentration to be separately prepared. Even more significant are
the time savings the present invention achieves in analyzing the
effects of temperature variations on given reactant concentrations,
as well as the effects of different residence times, within a
single system. Table 1 (below) provides a sample set of experiments
using flow rates and valve settings that cover a broad range of
reaction times with only small variations in flow rates.
Significantly, the quantity of optimization data that can be
obtained by the system of the present invention is an order of
magnitude higher than the quantity of data obtainable by any single
prior art system.
[0076] Another significant advantage of the present invention
relates to the fact that prior art systems for varying temperature
included a plurality of reaction vessels disposed relatively close
to one another, where each reaction vessel was operated at a
different temperature. Temperature difference between reaction
vessels could vary by as much as 300.degree. C., significantly
increasing the complexity of the thermal controls required by the
system. This problem is important, because in the case of even
slightly complex reactions, changes in temperature can cause a
series of side reactions or chain reactions to be initiated, or to
be stopped. Thus, temperature control is very important.
[0077] One goal for determining optimal reaction parameters in the
laboratory is to ensure that the parameters thus determined are
also applicable to systems wherein larger volumes of the desired
product are produced than were produced during laboratory testing.
Due to scale-up complications, optimal reaction conditions
identified using conventional optimization techniques in the
laboratory are not always carried over or applicable to the same
reaction implemented in large scale reactors. Using the system of
the present invention, it is possible to generate relatively large
volumes of material by operating identical systems in parallel.
Using parallel systems to generate production quantities, as
opposed to using a single large reactor, eliminates the problem of
determining optimal conditions in a relatively small reactor, which
are also applicable for producing the desired product in a
relatively large reactor, because the parallel production reactors
are identical to the reactor used in the present invention to
determine the optimal process conditions. TABLE-US-00001 TABLE 1
Exemplary Order of Experiments # of Residence Total Flow Resulting
Experiment Residence Time Volume Rate Residence # Time Units (ml)
(ml/min) Time (min) 1 3 47 1 47.00 2 3 47 2 23.50 3 3 47 3 15.67 4
2 32 3 10.67 5 2 32 4 8.00 6 1 17 3 5.67 7 1 17 4 4.25 8 0 2 1 2.00
9 0 2 2 1.00 10 0 2 3 0.67 11 0 2 4 0.50
[0078] Table 1 indicates the residence times that can be achieved
using three residence time units and flow rates ranging between 1
and 4 ml/min (the linear region for precision pumps). Note that by
implementing only 11 experiments, a relatively broad range of
constantly decreasing residence times (between 47 minutes to 0.5
minutes) can be achieved. The controller automatically switches the
valves required to enable the above residence times to be
achieved.
[0079] Although the present invention has been described in
connection with the preferred form of practicing it and
modifications thereto, those of ordinary skill in the art will
understand that many other modifications can be made to the present
invention within the scope of the claims that follow. Accordingly,
it is not intended that the scope of the invention in any way be
limited by the above description, but instead be determined
entirely by reference to the claims that follow.
* * * * *