CRYSTALLISATION CONTROL METHOD

The present invention relates to a method of controlling crystallisation using enthalpy data from the crystallisation process.

Enthalpy (joules) is a measurement of heat. When a compound dissolved in a solution crystallises out as a solid, heat is liberated. The quantity of heat liberated is directly related to the mass of crystals formed. Power (Watts) in the context of heat represents the rate of heat release (or gain). The rate of heat released is directly related to the rate of crystal formation.

Batch crystallisation is a technique which is used in the manufacture of many types of speciality and bulk chemicals, pharmaceuticals and food products. The vessel used for the crystallisation is called a batch reactor. It comprises of a tank with a cooling/heating jacket around it and/or internal heating/cooling elements.

This technique may also be used for continuous crystallisers.

Crystallisation occurs when a solution containing a solute is cooled or concentrated (by evaporation). When the concentration of dissolved solute exceeds the saturation concentration, the solute precipitates in the form of crystals. The process of forming new crystals in this way is referred to as nucleation. Alternatively the solute may precipitate out on existing crystals causing them to grow in size. In practice, there is a condition referred to as super saturation where a solute can remain in solution at concentrations which are higher than the saturation concentration. Nucleation can be initiated by increasing the level super of saturation or by other means such as the introduction of seed crystals in a saturated solution.

The two ubiquitous problems of industrial crystallisation are; poor particle size distribution and crystals of the wrong structure or morphology. Crystals of different structures can behave differently and therefore the wrong structure is undesirable where crystal structure has a functional purpose (e.g. where structure affects the rate of uptake of drugs in the body). Poor particle size distribution in the crystals is also undesirable because it is less inclined to mix uniformly with other powders in blended materials. The

presence of very small crystals can also be undesirable because it contributes to slow and/or inefficient liquid solid separation. Small particles in the product can also lead to higher dust losses in the dry product.

Batch crystallisation is a process which takes place in stages. The first stage is concentration or cooling of an unsaturated solution. This is followed by the formation of crystals (nucleation) in a saturated or super saturated solution. Nucleation is achieved by cooling (or evaporation) or a combination of cooling (or evaporation) plus some other factor (such as the addition of seed crystals). Once nucleation has occurred, the crystals are grown by a process of further cooling (or evaporation). The quality of the final product is determined by how well the crystallisation process is controlled. Immediately after nucleation, the crystals are very small and therefore the rate at which they can grow is relatively slow. If the user attempts to cool or evaporate the product too quickly at this stage, there will be a tendency to promote further nucleation. This is undesirable since it leads to the final product having crystals of different age (and hence different size).

The following description applies to crystallisation by cooling but it could equally apply (with some minor changes) to crystallisation by evaporation. Alternatively it could be used to control crystallisation by other means. An example of this would be the addition of anti solvent.

In the case of crystallisation by cooling, the traditional method of control has been to follow a predefined cooling curve. The cooling method is established using experience and by running test batches.

This invention provides a technique for controlling batch crystallisation by using process analytical data. The essence of this technique is that the progress of the crystallisation is monitored using an analytical instrument. This provides information which permits the cooling (or heating in the case of crystallisation by evaporation) rate to be controlled using real time data.

The crystallisation in this method is controlled by a computer using real time process analytical data. The operator pre-defines the control strategy. There are three main stages as follows:

• Initial cooling

• Pre nucleation cooling

• Post nucleation cooling (multiple steps)

For the purpose of this invention, post nucleation cooling control is the most important phase. Although the initial cooling and pre nucleation cooling can be controlled by the method of the present invention they could be controlled by a simple recipe or in some circumstances controlled manually.

In operation of the present invention the operator will set up the controller in the following way:

1. Set initial cooling rate

The operator will input the following control parameter into the controller:

1. The required rate of cooling or heating (in terms of either rate of change of process temperature, cooling power or any other means of regulating cooling power). The required rate of heating or cooling can be referred to as the heating or cooling set point.

2. Set ore nucleation cooling rate

The operator will input the following control parameters into the controller:

(a) The required temperature at which the cooling rate switches to the pre nucleation cooling rate.

(b) The required rate of cooling (in terms of either rate of change of process temperature, cooling power or any other means of regulating cooling power)

3. Set conditions for nucleation detection

The operator will input into the controller a parameter to detect the onset of crystallisation based on such parameters as:

Change in process turbidity Change in process enthalpy

Change in heat power output by the process Fixed temperature value Change in process temperature Fixed time interval from the start of cooling Other changes (such as visual detection, capacitance etc)

4. Set conditions for 1 ^s ' post nucleation stage control

The operator will input the following control parameters into the controller:

(a) The required rate of cooling (in terms of either rate of change of process temperature, cooling power or any other means of regulating cooling power)

The start of the 1 ^st post nucleation cooling step is triggered by the nucleation detection event.

5. Set conditions for 2 ^nd post nucleation cooling step

The operator will input the following control parameters into the controller:

(a) The required enthalpy value for triggering the start of the 2 ^nd post nucleation cooling step.

(b) The required rate of cooling (in terms of either rate of change of process temperature, cooling power or any other means of regulating cooling power).

6 to n. Set conditions for post nucleation cooling steps 3 to n.

The operator will choose the number of post nucleation cooling steps and set up the controller for each one as per step 5. Each step will be triggered by a specific enthalpy value.

The enthalpy values described above refer to the enthalpy of crystallisation at any given time (If this value is divided by the theoretical heat of crystallisation, this gives the mass of crystals formed).

The sequence is illustrated by the attached drawing.

Other considerations

In some cases, the pre-nucleation cooling and the nucleation event may not form part of the automated control sequence. The operator may use a combination of manual and automated control operations to control the crystallisation.

The rate at which crystallisation can be achieved without causing secondary nucleation is linked (amongst other things) to the rate of cooling employed and the surface area available for growing crystals on. Thus as the crystals become larger, the speed of crystallisation can be increased. Enthalpy data can be used to estimate crystal size (assuming the number of crystals can be estimated) since enthalpy is directly related to mass of crystals formed. Thus, as a further improvement, where the relationship between crystal size and surface area can be defined by a mathematical expression, the preferred cooling rate can be automatically estimated by the control system. This will assume that other factors (e.g. permissible heat flux at the heat transfer surface) and relationships are recognised and taken account of.

A variation on the technique for the post nucleation control is to adjust the cooling rate continuously (rather than in steps) using the enthalpy (of crystallisation) value as a real time control parameter for regulating cooling rate.

The process of measuring heat is referred to as calorimetry. The process enthalpy can be calculated from the cumulative value of the process power. The process power can be measured with instruments and calculated according to the following equation:

q = ms.Cps.(ti-to) (W)

Where q = process heating (or cooling) power (W) ms = mass flow of heat transfer fluid (kg.s-1)

Cps = specific heat of heat transfer fluid (J.kg-1.K-1) ti = inlet temperature of heat transfer fluid (K) to = outlet temperature of heat transfer fluid (K)

Enthalpy is determined from cumulative value of q over time

The process enthalpy can calculated from the cumulative value of the process power. The process power can be monitored with instruments and calculated according to the following equation:

q = U.A(T-t) (W)

Where q = process heating (or cooling) power (W)

U = overall heat transfer coefficient (W.m-2.K-1) A = heat transfer area (m2)

T = process temperature (K) t = jacket temperature (K)

There are other methods for measuring heat which can also be used. Our preferred method of monitoring heat change and rate of heat change is by heat balance measurement (as per the first equation) using a vessel with a multi-element jacket as described in WO 2004/017007 or more preferably in a vessel with a variable area jacket as described in GB 2374949. With any heat measuring method, corrections must also be made for such phenomena as ambient heat losses/gains, heat gains or losses from mechanical or electrical components, system temperature changes or any other heat gain or loss which is not due to crystallization.

The controller will have automatically selected a desired precipitation/crystallisation rate (based on enthalpy). The preferred method of controlling the desired

precipitation/crystallisation rate is by direct feed back measurement where power (which can also be expressed in terms of grams per second of crystallization) is the measured (by calorimetry) process variable. Typically the mechanism for regulating the cooling rate will be one of the following (but it may also be another type of mechanism such as a controller on an electrical heater):

A valve to regulate the cooling (or heating) fluid inlet temperature A valve to regulate the cooling (or heating) fluid outlet temperature (by varying the flow of heat transfer fluid) A valve to regulate the cooling (or heating) jacket area (Coflux control)

This technique of crystallisation control is ideal for laboratory batch crystallisers up to 10 litres for research and development purposes.

This technique of crystallization control is suitable for batch crystalliser of 100 to 1,000 litres.

This technique of crystallization control is suitable for batch crystalliser of 1 ,000 to 10,000 litres.

This technique of crystallization control is suitable for batch crystalliser of 10,000 to 20,000 litres.

This technique of crystallization control is suitable for batch crystalliser of 20,000 or larger.

This technique of crystallization control is suitable for manufacturing fine chemicals.

This technique of crystallization control is suitable for manufacturing pharmaceutical compounds.

This technique of crystallization control is suitable for manufacturing food products.

This technique of crystallization control is suitable for reactors which use variable area controlled cooling/heating systems.

This technique of crystallization control is suitable for reactors which use constant flux controlled cooling/heating systems.