In this two-part article Nigel Fletcher, principal consultant at Foster Wheeler Energy's pharmaceutical division, aims to take the pain out of CIP implementation for multi-purpose API plants. Here he considers the cleaning procedure and materials
In this two-part article Nigel Fletcher, principal consultant at Foster Wheeler Energy's pharmaceutical division, aims to take the pain out of CIP implementation for multi-purpose API plants. Here he considers the cleaning procedure and materials.
There are many traditional approaches to the protocols for Clean-In-Place (CIP) and, in the main, they are very successful at cleaning API units. Failures tend to creep in where sites try to make one CIP fit all, make one 'solvent' the universal cleaning agent or try to take shortcuts for environmental, safety or economic reasons. These issues can be avoided but a systematic approach is needed, along with small changes to cleaning techniques.
As many modern API facilities are multi-purpose this can seem a challenge but the methods outlined here in the Foster Wheeler CIP ‘roadmap’, provide a technique that can take much of the pain out of the CIP process.
The roadmap (see figure 1) is a plan with various nodes, which the cleaning protocol designer needs to visit and resolve to produce a successful result. In this way, the cleaning process can be built up systematically step by step. Ideally, this is addressed at the plant design stage; if not, problems are usually seen later on.
The basic formula of rinse/wash/ rinse/final rinse/(optional) dry is so well proven it is difficult to suggest major improvements. The key is to ensure that each step is carried out correctly and that there is a mechanism to deal with failures.
Typically cleaning failures are blamed on:
While there may be an element of fault in any of these items it is rare that anyone asks the basic question: “Is the cleaning method right for this process?” It is always assumed that whoever wrote the protocol must have got it right. The case study opposite illustrates what can and does happen.
Before considering the steps in the roadmap, it must be realised that every material to be removed or cleaned away has to be taken through the roadmap. This may initially seem a massive task when handling hundreds of reagents or raw materials, but a few minutes of consideration will show that it is not the insuperable task it seems.
First, most substances are soluble in a solvent or water and the few that are not, such as activated charcoal, are well known and can be dealt with in other ways. Second, the number of solvents commonly used in the pharmaceutical industry is small – probably fewer than 20 – and most sites have 10–15 of these available in bulk form. It is more than likely that the material will be soluble in one of these. Third, if conventional dissolution does not work then mechanical cleaning will; this is considered later.
Cleaning is all about three things: dissolution, dislodgement and suspension. If you cannot dissolve the material then you must dislodge it or mechanically remove it from the surfaces to be cleaned. It is not always necessary to dissolve every speck of “dirt”, as simple rinsing can remove a lot of material and this is where suspension comes into play. If the dirt is dislodged or simply rinsed off the surface then it has to be carried out of the equipment and this can happen only if it is suspended in the cleaning fluid. If at least two of these three mechanisms do not occur then the equipment or surfaces will not be cleaned.
Solvents and reagents
So the starting point is to look at data about the product to find out: what will dissolve the product; its physical forms during the processing; and the intermediates that are its necessary precursors. The data will also tell you about reagents, their forms and their expected fate. In many reactions, one or more reagents are added in excess to drive the product reaction forward.
The data and associated description of the process given by the research chemists will give information useful for cleaning, such as references to ‘crusting’ or solids precipitation. This data must be analysed for information on the physical chemistry of the product and intermediates and the chemistry of the reactions.
In the case of physical chemistry, this data will refer to what the product and inter-mediates will dissolve in, how fast they will dissolve, the solubility in g/litre and solid information on particle size and density.
The chemical information gathered can be used to determine if there is a way of decomposing the product/intermediate chemically: for example, turning a base into an acid salt by adding a solution of hydrochloric acid, or whether the material will break down if mixed with hypochlorite. This can be very valuable to reduce the toxicity or potency of the material or break it down into chemicals that are soluble in water or the common organic solvents.
At this point it is important to define the terminology used. Cleaning refers to a process of removing all residues or materials from a surface so that the surface is essentially free of all “foreign” substances. Sometimes the term “decontamination” is used to mean cleaning. However, decontamination means removal of a contaminant by various techniques. At the end of decontamination, the surface may still have material on it. Arguably, the target should be to get the surface truly clean so there is no risk whatsoever to the next process. This is a choice that has to be made when designing the ‘cleaning’ protocol.
Armed with this information the selection of the cleaning agent can be undertaken. Remember the solvents used in the process itself have been chosen for one of two reasons; either the product is soluble in it or the product is not soluble in it but side products are, as in solvent exchanges. This means that the first review should consider these solvents and tabulate the information in a manner similar to that in table 1.
In the case study, for the final step of the process, any residual intermediate can be removed with water but final product and reagent residues will not be removed and will require an organic solvent. Methanol would clearly be the best choice as both reagent and product are soluble in it. It may be necessary to use two cleaning agents successively. In the case study, hot methanol should be used first to remove the final product and the reagent; subsequently water can be used to remove both the intermediate and residual methanol.
This method then has to be applied to all process steps and a selection of cleaning agents identified. But this is not the end of the matter. If we had selected dichloro-methane for removal of the intermediate, for example, we would also know that we might have to use a large volume of it, as the intermediate is only slightly soluble.
This would require the protocol writer to estimate the amount of material to be removed, thus ensuring sufficient solvent was charged as part of the cleaning process. This could influence the whole protocol if excessive volumes of waste have to be dealt with. It would also influence how this volume was introduced into the equipment.
Continuing with the case study – if it were estimated that 1–2kg of intermediate were crusted on the walls of the 3,000-litre equipment, at least 2,000 litres of solvent would have to be introduced to dissolve the residue. In fact, this would produce a saturated solution and is not a workable proposition for a cleaning protocol.
Typically, achieving 50% saturation in the solution is more than enough but in the case study, this would mean putting 4,000 litres into a 3,000 litre item of equipment. This is clearly difficult and so techniques to address the problem will have to be considered, such as two 50% washes (dissolution) or high pressure jetting to achieve mechanical removal. There are many other techniques but each situation should be judged individually, taking into account the availability of the cleaning equipment, access, operator availability etc.
Methanol does not present undue health problems, nor is it particularly dangerous to handle. It would also present only a modest environmental problem for disposal. Conversely, if dichloromethane had been selected, then the HSE considerations would be very different.
While this is a simplistic example, this analysis must be carried out for all the cleaning agents selected; it might mean that the best cleaning agent needs to be replaced by a safer or more environmentally acceptable solvent. Also the cost of the (spent) cleaning agent has to be taken into account as well as its final disposal cost.
The next step is the design of the protocol. Many manufacturers have templates for protocols for other purposes and it is important that the cleaning protocol follows the same style so it can be lodged in the site's document system, will be familiar to the operators and can be copied and reused where possible.
Next, the cleaning objectives have to be set to identify what is an acceptable level of cleanliness. Much has been written about acceptable residual levels and whether the dose-based method, contamination level or dilution method in the next batch is used.
Selection is entirely dependent on companies’ preferred QA/QC practices. Whatever method is used, residual
levels have to be calculated so that a precise statement can be put in the protocol regarding the permitted level in the final rinse sample or swab test. This number is the final goal of the cleaning operation. Set this value too low and there will be many cleaning failures; too high and there would be an unacceptable risk to the patient – a realistic level has to be set.
The second point to make about the target for the cleaning is that this is the intended end result. That is to say, at the beginning of the cleaning the equipment will contain far more material than is acceptable. Let us say it contains 20 times too much material and the cleaning protocol has to reduce it by at least a factor of 20. This means that during the cleaning it will drop from 20 times too much to, say, 15 times too much and then eight times and so on until the target is achieved. This also means that the progress of the cleaning can be monitored and the protocol designed to take this into account.
For example, if we were to sample the end of the first rinse and analyse it, we might find that the residue level was down to 15 times too much. Suppose we were to do this when carrying out an approved protocol but discovered that the residue level was still 18 times too much, we could simply have a statement in the protocol that said if the residue level was above 15 times too much then repeat the rinse and sample again. A second failure would have to be addressed by a supervisor but if this second rinse achieved the correct level then the protocol could proceed.
What is being proposed is that a series of test points are placed at suitable points in the protocol to stop it if the test point values for residual levels are not met and allow a choice to be made. If this is not done, the whole cleaning procedure would be carried out before the problem was identified and the cleaning failure would require a repeat of the whole procedure. These test points would have to be validated with the protocol but would at least allow the manufacturer to be confident that the cleaning process progressed smoothly along a predictable track to a successful conclusion.
We now have two elements of the protocol: the cleaning agents and the cleaning target, together with a set of cleaning test points. Next, we will consider the order in which the cleaning will be carried out.