`Printed in Great Britain.
`
`0009—2509/88 $3.UU+U.00
`Pergamon Press plc
`
`Design of Reaction Systems for Specialty Organic Chemicals
`
`Edward L. Paul
`
`Merck Sharp & Dohme Research Laboratories,
`Rahway, New Jersey 07065 (USA)
`
`requires utilization of chemical
`specialty organic chemicals
`for
`reaction systems
`Design of
`reaction engineering principles for
`a wide variety of kinetic problems. Kinetic analysis must
`include breakdown of the overall reaction into definable components in order to identify parallel
`and/or consecutive reactions
`that
`result
`in lay-product
`formation.
`Once
`identified, methods
`of minimizing by-product formation can be developed.
`
`Examples are described of complex reaction systems which have required development of specialized
`procedures to minimize by-product formation.
`Each example represents a different kinetic problem
`and method of solution. Emphasis is placed on the close interaction between chemists and chemical
`engineers during laboratory development
`and plant
`reaction system design to achieve successful
`commercial operation.
`
`I.
`
`Introduction
`
`requires application
`for specialty organics
`reaction systems
`and scale-up of
`Development
`of
`the chemical
`reaction engineering discipline to solve a wide range of problems. While
`defying systematic categorization because of
`their variety,
`these reaction systems may
`be broadly characterized according to their kinetic complexity as will be developed in
`the discussions below.
`
`the
`the industry can be regarded as complex in terms of
`Plant design in this segment of
`chemistry of
`the larger molecules
`and the number of steps
`to complete the synthesis of
`a
`specialty chemical.
`Another obvious generalization is that
`the volume of production
`is modest
`in comparison
`to
`the
`heavy
`chemical
`industry,
`thereby allowing effective
`utilization of batch and semi-continuous reactor systems
`instead of continuous operations.
`Indeed,
`the use of continuous systems may be dictated not on throughput or other economic
`grounds but
`rather on kinetic restrictions which preclude batch or semi-batch operations
`because of scale-up considerations.
`Thus, while batch operations may be economically viable
`because of
`limited production requirements
`and even desirable for plant versatility in
`multi-product utilization,
`the use of continuous or semi-continuous systems may be required
`to achieve satisfactory kinetic scale-up and in some cases to minimize in-process inventory
`of potentially hazardous reagents.
`
`The purpose of this paper is to investigate the kinetic characteristics of specialty organic
`chemical reaction systems
`to accomplish successful scale-up from laboratory through pilot
`plant
`to plant operations.
`Specific complex reaction systems
`that have required special
`design considerations to achieve successful
`scale—up will be described.
`The analysis of
`each system will
`include an outline of the kinetic models involved,
`the reasons that special
`designs are necessary,
`and the specific operational
`and equipment design considerations
`that were applied to achieve successful scale-up.
`
`II. General Scale-up Considerations
`
`the individual reaction systems that have been chosen
`Before getting into the specifics of
`as models, a few observations on the scale-up of batch reactions in general may be in order.
`
`reactions require no special design or operational considerations once the reacting
`Many
`system has
`been established and
`its
`requirements determined.
`For
`these
`reactions
`a
`laboratory scale sensitivity analysis
`and pilot plant evaluation may be
`sufficient
`to
`demonstrate
`the
`feasibility of
`successful direct
`scale-up
`to
`production
`operation.
`Successful scale-up can be defined as plant operation that achieves the same conversion,
`selectivity,
`and product distribution as defined in the laboratory.
`Reactor design is
`then accomplished through direct volume scale-up permitting utilization of standard batch
`reactor
`configurations.
`These
`simple
`cases
`require
`that
`those
`parameters which
`are
`inherently different on direct volume
`scale-up are not significant
`in terms of changing
`the course of
`the reaction within the scale-up factor
`required.
`These variables include
`the following:
`
`1773
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 1 of 10
`Page 1 of 10
`
`
`
`1774
`
`EDWARD L. PAUL
`
`0
`
`0
`
`0
`
`transfer or vapor-liquid
`limit heat
`Reduction in surface area/volume ratio does not
`temperature maintenance
`limits are
`characteristics
`so
`that heat-up,
`cool-down,
`or
`achievable with standard equipment
`and gas or vapor dissolution and/or evolution are
`not
`limiting.
`
`Sensitivity to mixing (i.e. circulation time,
`etc.) does not affect reactor performance.
`
`shear, mass
`
`transfer between phases,
`
`Time of addition of a reactant and/or
`a significant variable.
`
`removal of a product
`
`in semi-batch mode is not
`
`While these considerations are well known, it is sometimes difficult in laboratory evaluation
`to arrive at definitive conclusions
`for
`individual
`reactions regarding their response to
`these parameters because the responses may be masked or not
`separable from the overall
`results.
`It may be
`informative during the
`laboratory development phase
`to attempt
`to
`categorize the various possible
`factors
`that may disguise the
`true kinetics. Attempts
`at characterization of reactions leads directly into the main body of
`this paper
`in which
`more complex kinetic systems are considered as
`those which are affected by any or all of
`these scale-up considerations.
`
`III.
`
`Complex Kinetics
`
`the design of reaction
`The opportunity for chemical engineers to influence the outcome of
`"Design for Multiple
`systems
`is
`emphasized
`by Levenspiel
`(Ref.
`1)
`in his
`chapter on
`Reactions".
`He points out
`that most systems can be reduced to an analysis of combinations
`of parallel and series reactions. More complex reactions obviously provide more formidable
`technical challenges
`to both chemists
`and chemical engineers
`and the interdependence of
`chemistry and reactor design requires close integration of
`the development skills of both
`disciplines.
`Indeed,
`the possibility always exists that
`a
`reaction system.
`that
`can be
`operated on a
`laboratory scale is judged to be unfeasible for successful operation on a
`plant
`scale because of
`insufficient understanding of
`the system.
`Such an extreme result
`could invalidate an otherwise elegant synthesis necessitating development of a less favorable
`alternative and inducing some loss of confidence in our chemical colleagues for our design
`abilities.
`
`these colleagues could dismiss as obviously unscalable an otherwise
`It is also possible that
`attractive reaction system because of incomplete understanding of the potential contributions
`of chemical engineers by the creative application of chemical reaction engineering principles
`and methods.
`
`Developmental strategy must be focused on defining the kinetic relationships of the reaction
`system so that
`the strictly chemical
`issues can be addressed and separated from the scale-up
`issues.
`This
`type of analysis can lead to a
`further broad characterization of complex
`reaction systems for purposes of this discussion as follows:
`
`0
`
`0
`
`0
`
`reactions that require resolution of kinetic problems in order to be run successfully
`in the laboratory;
`
`reactions which can be run successfully in the laboratory but which require special
`plant design considerations and equipment;
`
`reactions which pose both special laboratory and scale-up problems.
`
`the role
`reaction system,
`studies of any individual
`During early laboratory development
`of kinetics may not
`be obvious since a kinetically feasible maximum selectivity may as
`yet be unknowu. Kinetic complications are, of course.
`implicated when actual selectivities
`fall short of values that can be reasonably expected. Therefore, assignment of a reaction
`system to one of the categories described above may be difficult without some early scale-up
`experience to determine response to changes in scale of conversion, selectivity, and product
`distribution.
`
`is even
`that of unfeasibly low selectivity even in the laboratory -
`The first category -
`more difficult
`to identify since discrimination between
`inherently low selectivity and
`failure to control a parallel or consecutive reaction may not be possible in these early
`stages when a kinetic model has not been established.
`
`The first example to be discussed is drawn from this category of complex reactions where
`identification of
`the impact of a consecutive reaction leads to development of a solution
`to achieve
`its minimization.
`This kinetic solution is essential
`for
`achievement
`of
`reasonable selectivity on a laboratory scale and is,
`therefore, primarily a chemical problem.
`Successful
`implementation of
`the method
`for kinetic control
`in production then depends
`on successful scale'up of the revised laboratory system.
`
`IPR2020-00770
`|PR2020-00770
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 2 of 10
`Page 2 of 10
`
`
`
`Design of reaction systems for Specialty organic chemicals
`
`1775
`
`IV.
`
`Examples of Complex Kinetic Systems
`
`Example 1
`
`is generated
`that
`(HCI)
`a reaction system in which a by-product
`is of
`The first example
`by the primary reaction would decompose both the desired product and the starting material
`to give essentially no yield unless its concentration were controlled.
`In addition,
`the
`actual selectivity as well as
`the conversion rate is a
`function of
`the method and extent
`of this control.
`The chemistry is shown in Figs.
`1 and 2 and the kinetic system summarized
`in Fig. 3. This chemistry is discussed extensively by Weinstock (Ref. 2).
`
`FiGUHE 1
`
`OVERALL CHEMISTRY
`
`FIGUREZ
`
`'NTERMEDIATES
`
`l
`
`sI
`
`CH COCI
`2
`
`0
`
`“0cm
`rr ,
`0 N\
`
`I
`
`CHSO—sozw
`A
`
`COOCHZDCHg
`
`|cS
`
`|
`
`H2043
`9C“:
`1:
`0 V4 Ho
`l
`0 k
`S
`“2500'
`
`no.
`cm—Q—sozm Gommnm
`'
`‘
`a
`R‘
`
`”
`
`,7 H
`
`‘— |
`—>
`
`|
`
`5
`
`CHgocHN
`
`R
`
`OCH,
`.
`I
`DEC
`.
`cm
`‘cH;
`on;
`‘CH,
`CHQAQ—SOZNHLH/
`J
`
`s,
`
`O
`
`HCI
`
`.
`
`flick—Q‘s”:
`COOCHgOCH,
`$2
`
`0
`
`CH3
`
`SOZNHA
`
`n gCHas
`'
`'
`H/
`CHZOCONHZ
`0
`COOCHEOCHa
`COOCHEOCHa
`A
`
`a
`
`mm.
`
`5
`L |
`
` L l
`
`s
`
`CH
`
`N
`.
`2Cu:
`O
`
`N
`
`R"
`
`I
`
`5
`
`o H
`n
`I 9%
`CHZC—N E
`S
`j:-(
`0’ ”52‘ CHQOCONHZ
`COOCHZOC H3
`H
`
`FIGURE 3
`
`TRANSACYLATION KINETICS
`
`A + B
`
`R-+B
`
`_—n~——>
`
`
`R' + HCI
`
`——-——>
`.__._ R"+S1+HCI
`
`H‘ + HGI
`
`q—._.—“—’
`
`51
`
`R" + HCI
`
`..____“’—’
`
`R + B
`
`S‘
`
`———-———>
`
`32 + HCI
`
`The method of mediating the concentrations of HCI below that causing excessive decomposition
`while maintaining its concentration high enough for
`its participation in
`the
`required
`reactions is,
`therefore, critical
`to the success of
`the overall
`scheme.
`The product, R,
`after hydrolysis
`is now one of
`the world's leading parenteral antibiotics and is made
`in
`relatively large volume.
`A feasible, commercially-viable synthesis of
`this compound was
`essential for operation in a manufacturing environment.
`
`and
`(1) acylations to form imides
`taking place:
`reaction types
`There are two distinct
`and
`an
`acid chloride.
`Consecutive
`cleavages
`producing amides
`(2) HCI—promoted
`imide
`decomposition by
`reaction with HCl
`is always proceeding depending on
`the concentration
`of HCI.
`If no method of mediating the HCl concentration was applied,
`the concentration
`of HCl would increase to 0.1M and result
`in complete decomposition of R.
`It was determined
`that an optimum concentration of “0.004)“! is required for imide cleavage.
`
`(3A or 4A) were found to be very effective for this mediation under very
`Molecular sieves
`well-defined conditions.
`The HCl concentration in solution is determined by both the amount
`of sieves used as well as
`their external surface area.
`Thus,
`sieve pellets ('v 4001.!) are
`not satisfactory because of rate-controlling diffusion in the pores whereas powdered sieves
`(1-414) are satisfactory.
`It
`is also apparent
`that
`the rate of HCl
`removal
`is critical
`as well as its actual concentration. This criticality is also underscored by the improvement
`in selectivity that was
`subsequently achieved through development of
`a homogeneous HCl
`mediator,
`trimethyl silyl methyl carbamatc.
`Elimination of
`the mass
`transfer
`resistance
`at
`the sieve surface by the presence in solution of a reagent
`to react directly with HCI
`resulted in a significant yield increase. Comparison of selectivity of R by four different
`methods of HCl mediation is shown in Table IV-l.
`
`cEs BIB-E
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 3 of 10
`Page 3 of 10
`
`
`
`1776
`
`EDWARD L. PAUL
`
`TABLE IV- 1
`
`Method
`
`None
`Distillation
`Molecular Sieves
`
`Homogeneous Scavenger
`
`Relative Selectivity
`
`- Essentially Zero
`-
`"
`"
`
`+
`
`++
`
`Scale-up of both
`in the laboratory.
`the above reaction studies were carried out
`All of
`sieve and homogeneous
`scavenger mediated reactions was
`relatively straight-forward once
`the concentrations
`and reaction were defined in the
`laboratory.
`Successful plant-scale
`operation did require rapid heat-up and cool-down, however,
`to minimize time at other than
`optimum temptature.
`Typical production scale
`reaction kinetic profiles,
`as determined
`by HPLC, are shown in Figs.
`4 and 5 illustrating consistency of overall performance despite
`a significant difference in time to termination of reaction.
`Thus,
`if the run shown in
`Fig. 4 had been allowed to continue as
`long as
`that
`shown in Fig. 5,
`a significant yield
`loss would have been experienced as projected in Fig. 4.
`
`FIGURE 4
`KINETIC PROFILE PRODUCTION PLANT-HPLC
`
` Projecled
`
`». / Overreaction
`\\
`.
`
`0.5
`/V'
`Total Malenal
`
`Balance
` 07
`
`1.0
`
`0.9
`
`o
`O
`‘3
`
`0.5
`
`0,5
`0.0
`
`0.3
`
`0.2
`
`o I
`
`FIGURE 5
`KINETIC PROFILE PRODUCTION PLANT-HPLC
`
`1.0
`
`0.9
`
` /V'
`Toial Material
`
`O
`‘3
`
`0.5
`0.7
`
`0.6
`
`0.5
`0.4
`
`0.3
`
`0.2
`
`6.1
`
`Termmalion
`of Reaclian
`
`
`.
`
`.'\
`
`Termination
`at Readion
`
`
`
`
`
`
`I
`
`2
`
`3
`3
`Time ~ Hrs
`
`I
`5
`
`_ " ‘r
`5
`
`‘l
`
`2
`
`4
`3
`Time ~ HIS
`
`5
`
`5
`
`E x amp 1 e 2
`
`in that successful
`from the first
`second example reaction system is quite different
`The
`laboratory operation was quickly established but scale-up to the pilot plant resulted in
`reduced selectivity.
`The
`laboratory synthesis
`is discussed by Blacklock, et.
`a1
`(Ref.
`3).
`The
`cause of
`scale—up complications
`is over-reaction of primary product bTrapid,
`consecutive reaction with one of
`the starting materials.
`The reaction involves formation
`of
`the dipeptide L-alanyl-L-proline from L-alanine-N-carboxyanhydride and L-proline.
`The
`chemistry is shown in Fig.
`6 and the kinetics in Fig. 7.
`
`FIGUREG
`
`COU PLI NG REACTION L-ALANYL-L- PFIO LINE
`
`HN
`
`0
`
`OH
`
`A
`
`+
`
`o
`H c
`3 \KIL
`O —-—-—>
`NHTf
`0
`
`B
`
`k
`R' _—2—» R+OOZ
`
`o
`HJCVLN
`NH2
`
`_ 602
`
`0
`
`OH
`
`R”
`
`DII
`(2
`
`FIGURE 7
`KINETICS OF THE
`
`COUPLING REACTION
`
`k
`1
`A+B “'- Fl‘
`k2
`R“ ———-———> R+CO2
`
`R3
`n+3 -—————> 31
`
`Fl+B
`
`k
`
`3
`
`‘N
`H30—(
`H30
`NH
`
`k1 £100l/mol. sec.
`
`NH;
`
`O O
`
`OH
`
`k3/k1~0.1
`
`IPR2020-00770
`|PR2020-00770
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 4 of 10
`Page 4 of 10
`
`
`
`Design of reaction systems for specialty organic chemicals
`
`1777
`
`L-alanyl—L-proline will further react with additional L-alanine N-carboxyanhydride as shown.
`The primary rate constant,
`k1,
`is large enough to result
`in completion of
`the primary
`reaction in the order of one
`second.
`The rate of
`less of C02
`to form R is significant
`enough to affect
`the overall kinetics.
`If it were very rapid compared to k1,
`the system
`would demonstrate simple consecutive-competitive kinetics and the selectivity would depend
`only on k3/k1 and be
`independent of addition time of B. However,
`a dependency on this
`addition time has been shown in the laboratory.
`Long addition time (1000 sec.) results
`in a
`reduction in selectivity of
`“-107...
`Rapid addition,
`therefore,
`results in reduced
`opportunity for R to react with B,
`thereby diminishing the importance of k3 on selectivity.
`R
`is less reactive with B than R.
`
`resulted in significant yield reduction and
`to pilot plant equipment
`Scale-up (50—fold)
`increased by-product
`formation compared
`to laboratory reanlts as would be expected from
`the fast kinetics and significant consecutive reaction.
`Furthermore, additional scale-up
`(WZO-fold) was
`required for production-scale operation and an additional
`loss
`in yield
`was anticipated.
`The reduced selectivity of the initial scale-up and the Ora-going definition
`of
`the rapid kinetics of
`the
`system combined with the requirement
`for minimization of
`by-product
`formation necessitated evaluation of
`an alternative reactor
`configuration.
`An inrline mixer was chosen and was successfully developed for production scale operation.
`The
`in-line mixer chosen was
`the Koch static mixer with an L/D ratio of 4.
`The nominal
`residence time of
`the combined Z-liquid phase stream was
`1 sec.
`Reynolds number
`in the
`mixer was 2000 based on empty tube diameter.
`The reactant mol ratio was 0.95-1.0 mol alanine
`NCA/proline .
`
`in both laboratory scale (0.8 cm) and plant scale (2.54 cm)
`the in-line mixer
`Results of
`operation were excellent.
`No change
`in selectivity or product distribution occurred over
`this scale-up so that the expected selectivity was achieved.
`
`issue of
`scale-up raise the
`successful
`for
`its requirements
`reaction system and
`This
`identification of
`reaction systems with potential
`for mixing-related scale up problems
`and selection of mixing devices for proper contacting.
`The kinetics of
`this system, while
`not
`identical,
`are similar
`to the consecutive—competitive reactions
`that have been used
`to study the effect of mixing on the selectivity of reaction systems. These studies include
`the work of Bourne and co-workers
`(Refs. 4, 5, 6, 7)
`in development of
`the diazotization
`reaction sequence
`that has been used so effectively both in the experimental definition
`of
`the micro-mixing problem as well as
`in modeling for prediction of mixing effects.
`It
`has been long recognized that any reacting system in which the
`primary rate is on the
`same time scale as the time required for molecular mixing of
`the reagents is in the regime
`of mixed diffusion/kinetic control.
`
`a specific system is significantly affected
`the product distribution for
`Whether or not
`by mixing depends
`in turn on
`the
`relative magnitudes of
`the rates of other possible
`reactions.
`Finally,
`the
`significance of
`these by-product
`reactions
`in scale-up of
`an
`industrial process depends on their
`impact on the final reaction mixture.
`Three effects
`could be anticipated on scale-up of a mixing sensitive reaction, all of which are potentially
`detrimental.
`The effects are:
`
`0
`0
`0
`
`reduced conversion
`reduced selectivity
`increased impurity levels in reaction products
`
`The negative
`is obviously specific for each reaction system.
`The actual economic impact
`effects on downstream processing in terms of separation of
`increased levels of
`impurities
`and their possible effect on subsequent
`reactions cannot be underestimated.
`Even in the
`case of acid or base additions to change pH in the presence of organic substrates, parallel
`decomposition reactions of
`the substrates with the acid or base can occur
`in the entering
`reagent stream leading to unanticipated loss of substrate.
`
`The negative impact of mixing sensitive reactions on scale-up can be minimized by design
`of
`reagent mixing systems as discussed by many authors
`(Refs. 8, 9, 10,
`ll).
`Bourne and
`Dell'Ava (Ref. 12) have published data on the diazotization reaction on scale-up to $70 1.
`A dependence on addition rate was observed which is attributed to decreased circulation
`which causes a decrease in molar ratio at
`the point of addition.
`
`joint project with a student of Beurne's, Scale-up studies
`in a
`At Merck Sharp 6: Dohme,
`have been extended to 4-000 liters.
`The critical nature of circulation was again observed.
`Power
`requirements on scale‘up are also significantly increased as
`shown in Figs. 8, 9,
`and
`10,
`in which scale-up in three different mixing configurations
`is
`summarized.
`For
`similarly positioned addition points
`in each mixing configuration,
`the power
`required to
`achieve equivalent selectivity and product distribution was greater than would be predicted
`by equal P/V. More data is required to establish a quantitative relationship, however.
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 5 of 10
`Page 5 of 10
`
`
`
`1778
`
`0.!
`
`EDWARD L. PAUL
`
`FIGURES
`SELECTlVlTY VS. RPM
`
`0
`
`0 100 Gal GL Retreat Blade
`El
`1000 Gal GL neueax Blade
`
`23
`as + H
`
`‘
`
`x
`
`FIGURE 9
`SELECTIVITY VS. RPM
`1 000 Gal wnh Variable Push
`Glass Turbine
`A60'
`
`V 60' Reverse
`
`0 so-
`23
`7
`' 2s + n
`
`A
`o
`
`V
`
`O
`
`A
`
`_1
`
`0
`
`,05
`
`
`
`
`‘
`L
`I_
`o
`1
`2
`3
`
`FIGURE 10
`SELECTIVITY VS. RPM
`
`RPM ~ 5e: "
`
`0.1
`
`X
`
`0 100 Gal
`X 1000 66% — 1 Blade
`.8 100a Gal -2 Blades
`28
`
`x— ZS+R
`
` O
`
`2
`
`3
`
`I
`
`RPM ~ sec"
`
`the circulation effect was also pursued by comparison of a standard
`Further evaluation of
`Rushton turbine mixing system using one
`turbine with one
`in which an upper pitched blade
`was
`added to increase circulation.
`A significant
`reduction in "by-product
`formation was
`achieved thereby indicating that
`the normal
`reduction in circulation on scale-up must be
`compensated for
`in order
`to prevent
`reduction in selectivity.
`These
`results are shown
`in Figure 10.
`
`scale-up
`the possibility for decreased selectivity on
`The preceding section focused on
`and the difficulty in compensating for
`this decrease in large vessels.
`The problem was
`overcome for
`the amino acid coupling system described above by utilization of an in-line
`mixing device. Utilization of
`this type of mixer
`- when properly designed to achieve the
`necessary mixing intensity - would appear
`to be
`the safest approach to solve a potential
`sca1e~up problem when compared to increasing the power
`to achieve the necessary mixing
`intensity in a
`large vessel. Other constraints, however, may require the use of
`larger
`vesels and thereby require careful characterization of the optimum mixing system. Evidence
`to date points to the system as
`shown
`in Fig.
`11 as having the greatest potential
`for
`minimizing scale-up losses
`in an otherwise conventional vessel.
`The point of addition
`is critical as well as the agitator configuration.
`
`FIGURE 11
`
`MIXING CONFIGUHATION SEMI-BATCH REACTORS
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 6 of 10
`Page 6 of 10
`
`
`
`Design of reaction systems for specialty organic chemicals
`
`1779
`
`Example 3
`
`(Ref. 13),
`31.
`reaction system which has been described by King, et.
`third sample
`The
`falls in the category of
`requiring a special reactor even in the laboratory in order
`to
`achieve a reasonable yield.
`The cause of
`this sensitivity is the instability of both the
`starting material and product
`in the reaction mixture.
`The chemistry of
`the reaction is
`shown in Fig. 12. These compounds are intermediates in the total synthesis of the parenteral
`antibiotic primaxin described by Pines (Ref. 14).
`
`HGURE12
`
`HYDROLYSIS REACTION
`
`OR
`
`OH
`)\
`
`o
`
`H
`
`k3
`
`OH
`.
`2 ‘
`0
`
`OH—
`
`NH O
`
`8
`
`OH
`
`k1
`B
`
`O I
`
`!OCH
`)x
`
`0/
`
`A
`
`k2
`
`OH-
`
`U
`K
`A+B —1-->- Fl
`k
`A+B —2-—- U
`
`ka
`A+B—>3
`
`the formate portion of A
`The primary reaction itself is a straightforward hydrolysis of
`to its corresponding alcohol, R.
`The
`reaction can be catalyzed by either acid or base.
`Because both the starting material and product are far more stable under acidic conditions
`than basic conditions,
`the original synthetic route utilized acid catalysis with an aqueous
`solution of a mineral acid.
`The organic substrates, however, are soluble only in non-polar
`solvents Ci.e. methylene chloride) and the hydrolysis rate of the two—phase reaction mixture
`was
`impractically slow even with excellent mixing.
`A co-solvent, methanol, was used to
`increase mutual solubility and thereby achieve a satisfactory hydrolysis rate.
`Following
`the reaction,
`the methanol had to be removed by water extraction prior to crystallization
`of
`the product
`by addition of hexane.
`One
`further
`complication that
`is
`typical of
`multi-reaction synthesis is that several reactions are run in series without
`intermediate
`isolation resulting in build-up of
`impurities along with the intermediate.
`In this case,
`these impurities
`accumulated in the methylene chloride feed stream and
`amounted to 75%
`of reaction feed on a solvent-free basis.
`
`large vessels and utilized large
`it required several
`Although the process performed well,
`volumes of solvents requiring additional equipment both for solvent recovery and in meeting
`environmental restraints.
`
`investigated in which the high aqueous
`Development of an alternative reaction system was
`solubility of
`the enolate salt of A and R was exploited.
`Base titration of the B-ketoester
`13 whereas
`portion of
`the compound forms
`the corresponding enolate salt as
`shown in Fig.
`the impurities in the reacting system are incapable of such salt
`formation.
`This provides
`the opportunity to separate A and R from the impurities by aqueous extraction at high pH.
`Thus,
`the potential
`for base catalyzed hydrolysis
`and simultaneous extraction could be
`combined to achieve an integrated reaction and separation system.
`
`FIGURE13
`
`CONCEPTUAL PROCEDURE
`
`
`
`pH12
`
`OH
`/"~
`
`O
`
`on
`
`NH O’
`Na”
`
`O
`
`B—Ketoester
`
`pH 7
`
`Enolate Salt
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 7 of 10
`Page 7 of 10
`
`
`
`1780
`
`EDWARD L. PAUL
`
`12) would determine the
`and base decomposition (Fig.
`relative rates of hydrolysis
`The
`feasibility of
`the
`reaction/extraction system.
`These
`rates were measured independently
`and the ratio was
`found to be sufficiently favorable (kl/k2>100)
`to proceed with design.
`Such a system would have to accomplish the following:
`
`(1)
`
`provide rapid contacting of the two reacting phases
`
`(2)
`
`(3)
`
`(4)
`
`(5)
`
`provide residence time to extract A into the aqueous base
`
`provide residence time to complete hydrolysis of A to R ('h 15 sec.)
`
`limit residence time to minimize base decomposition to U
`
`separate the phases to allow transfer of the aqueous phase into acid
`for neutralization/crystallization
`
`These units are
`a Podbielniak centrifugal extractor.
`chosen was
`reactor/extractor
`The
`feed for extraction and are capable of having two or three
`normally run with countercurrent
`stage efficiency.
`The operation including the reaction zone is shown in Fig. 14 in a cutaway
`side view of the centrifugal rotor.
`The solutions of A plus impurities in methylene chloride
`(heavy phase)
`and aqueous base
`(light phase) Were
`fed countercurrently into appropriate
`a
`ends
`of
`the
`rotor
`and
`extracted/reacted
`in
`concentric
`contacting
`zone.
`The
`mixing-contacting zone between the organic phase containing A and the aqueous base provided
`enough interfacial area and residence time to accomplish the first three requirements listed
`above.
`By maintaining the principal interface,
`the two phases were separated as exit streams
`and the aqueous phase transferred directly into aqueous acid to stop formation of U and
`crystallize R.
`
`FIGURE 14
`
`2-STAGE COUNTERCUFIRENT OPERATION
`
`Aqueous Compound I]
`Enelate Solution Oul
`
`Mecrz compound I
`Feed In
`
`Contacting Zone
`
`Heavy Liquid
`clarivicalio r1 Zen 9
`
` Perlofaled Ele ments
`
`
`Feed Ill
`Ramnane 0u1‘<,._ Aquacus Caustic
`
`Org anic ¢WW
`.\\\\
`Light LiquId
`
`_.A
`Clalificalion Zone
`.K.m ‘0:
`_as“!
`
`
`
`Principal Intervace
`
`
`the aqueous product
`system satisfactorily accomplished the reaction objectives but
`This
`stream was extracting more base-soluble impurities from the feed stream than was anticipated
`thereby reducing product purity.
`The Podbielniak extractor was in effect doing too efficient
`an extraction job by achieving its rated two or
`three stage efficiency.
`The distribution
`coefficient of
`the enolate salt was
`found to increase with increasing pH above the salt
`pKa of 10.5, whereas
`that of
`the other extractables did not.
`Therefore, by raising the
`pH above 10.5 the enolate salt of A could be extracted with one stage (d r1’100)
`to achieve
`more selective extraction.
`To take advantage of this operationally,
`the system was changed
`to include a line mixer to pre-mix the two feed streams and complete the extraction/reaction
`with only one stage of separation.
`The Podbielniak was
`then used as a separator as shown
`in Fig.
`15.
`No
`further extraction occurred Since counterCurrent operation is
`replaced
`by cocurrent separation.
`To provide a
`further
`increase in purity,
`a methylene chloride
`stream is added in the normal entry for
`the heavy phase to achieve some Countercurrent
`back extraction.
`
`FIGURE T5
`MIXED FEEDS SINGLE STAGE WITH BACKWASH
`
`Fresh MeCIz
`FDI Backwash In
`
`Pe r10 rated
`Elemsnls
`
`
`
`
`
`
`M.
`
`
`Ixed Feed In
`Hauinane Out
`(M6012 1: H20) Compound I
`Organic WWI/[$76]?
`
`
`n
`,
`.
`Clari'FCSIiOfl Zone
`Liz:
`I Lian-Id
`
`
`[MI/II
`Backwash
`Zone
`Heavy Liquid
`Clarilicat ion 20 ne
`
`
`Aqueous Enolale
`Solution om Compound [I
`
`Principal lmerIace
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 8 of 10
`Page 8 of 10
`
`
`
`Design of reaction systems for specialty organic chemicals
`
`1781
`
`FIGURE 16
`
`FINAL DESIGN—MIXED FEEDS SINGLE STAGE
`WITH SOLVENT BACKWASH
`
`
`onenmc
`RAF FINATE
`
`wdimuACID
`
`
`
`AOUE0.)5 NAOH
`
`AQUEOUS
`WY”En
`LIOJORS
`
`SOLID
`COMPOUND II
`
`line mixer achieves the required reaction
`The
`The final plant design is shown in Fig. 16.
`conversion and the subsequent separation is completed in a suitaUle time frame to minimize
`base-catalyzed decomposition.
`The expected gain in purity by changing to in-line, one-stage
`mixing and adding a methylene chloride back extraction within the Podbielniak was realized
`(70% increased to 90%).
`
`to characterize than plant
`In this case laboratory development was actually more difficult
`operation because a good small-scale extractor
`to simulate Podbielniak performance is not
`available.
`This case is illustrative,
`therefore, of
`the need to conceptualize full-scale
`process performance and equipment design in the absence of an integrated laboratory model
`and to utilize separate laboratory reaction rate data on the different reactions to design
`the overall reaction scheme.
`
`acid-catalyzed
`the original
`over
`extraction/reaction process
`the
`superiority of
`The
`hydrolysis is readily apparent on comparison of key factors including yield (81% increased
`to 95%),
`reduction in number of solvents,
`and replacement of solvents with aqueous base
`and acid.
`The method is also illustrative of a practical method to mix and separate reactive
`streams that are unstable at processing conditions.
`
`Conclusions
`
`Manufacture of specialty organic chemicals utilizes a wide variety of reactions and reactor
`systems to address the specific needs of each process.
`
`to chemists
`and experimental challenges
`theoretical
`systems present
`these
`Some of
`chemical engineers for interactive development of scalable and operable reactors.
`
`and
`
`resulted from such
`that
`systems
`the actual
`The examples discussed above describe some of
`development programs. Other kinds of systems that could be expected to present significant
`reactor design challenges to this industry include the following:
`
`electrochemistry
`low temperature anion chemistry
`enzyme catalysis
`high dilution reaction
`reaction of macromolecules
`high pressure chemistry
`
`these implies further interactive associations with other disciplines for effective
`Each of
`reduction to practice of these complex systems.
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 9 of 10
`Page 9 of 10
`
`
`
`17 82
`
`EDWARD L. PAUL
`
`Literature Cited
`
`(1)
`
`0. Levenspiel, ”Chemical Reaction Engineering", J. Wiley & Sons, 1962.
`
`(2)
`
`L. M. Weinstock, 1986, Chemistry and Industry, 3, 86.
`
`(3) T. J. Blacklock, R. F. Shuman, J. W. Butcher, W. E. Shearin and J. Budavari.
`in Journal of Organic Chemistry, 1988.
`
`to be published
`
`(4)
`
`J. R. Bourne, F. Kozicki, and P. Rye, 1981, Chem. Eng. Sci., 29, 1643.
`
`(5) w. Angst, J. R. Bourne and P. Dell‘Ava, 1984, Chem. Eng. Sci. 22} 335.
`
`(6)
`
`(7)
`
`(8)
`
`(9)
`
`J. R. Bourne, F. Kozicki, U. Moergeli and P. Rys, 1981, Chem. Eng. Sci., §§, 1655.
`
`J. R. Bourne, 1984,
`
`IChemE Sym. Series 87 (ISCRE 8, Plenary Lecture).
`
`E. L. Paul and R. E. Trebal, 1971, AIGhEJ, 11} 718.
`
`J. C. Middleton, F. Pierce and P. M. Lynch, 1984,
`
`IChemE Sym. Series (ISCRE 8), 239.
`
`(10) O. Bolzern and J. R. Bourne, 1984 IChemE Sym. Series 87(ISCRE 8), 543.
`
`(11)
`
`J. R. Bourne and J. Garcia'Rosas, 198A, Paper 52C AIChE Annual Meeting, San Francisco.
`
`(12)
`
`J. R. Bourne and P. Dell'Ava, 1987, Chem. Eng. Res. Des., 65, 180.
`
`(13) M. L. King, A. L. Forman, C. Orella and S. H. Pines, 1985, Chem. Eng.
`36.
`
`Progress, £1, 5,
`
`(14)
`
`S. H. Pines, "Organic Synthesis Today and Tomorrow", pg. 327, Pergamon Press (1981).
`
`(15)
`
`C. B. Roses, 1969,
`
`I&EC Fundamentals,_§ 361.
`
`IPR2020-00770
`|PR2020-OO77O
`United Therapeutics EX2023
`United Therapeutics EX2023
`Page 10 of 10
`Page 10 of 10
`
`