Monograph 645 of the United States Pharmacopeia has a direct impact worldwide on water analysis and quality in the pharmaceutical manufacturing industries
Monograph 645 of the United States Pharmacopeia (USP) is American legislation that has been in force a few short years.
It has a direct impact worldwide on water analysis and quality in the pharmaceutical manufacturing industries.
Water is the world's most abundant solvent and naturally is one of the primary vehicles for drug administration in pharmaceutical applications.
Water for injection and purified water need to be rigorously tested for trace impurities.
Before the updated monographs, WFI and PW tests for chloride, sulphate, calcium, ammonia, and carbon dioxide were done off-line in laboratories. For example, the test for chloride ion was carefully to take a grab sample, neutralise it with nitric acid and then to add silver nitrate.
Depending on the chloride content, the sample became more or less cloudy.
To determine whether the sample of WFI or PW exceeded the chloride limits laid down, the turbidity of the sample was compared against a standard solution.
Naturally this is a very labour-intensive method, and when added to the requisite tests for sulphate, calcium, ammonia, and carbon dioxide, proved highly inefficient.
Being a pass/fail only methodology, it also gave no indication of how far the sample was from the acceptable limit.
An alternative method was needed to maintain or improve the existing water quality, improving the reliability of testing, reducing the number of manual tests, permitting qualitative in-line control, and ensuring quality standards continue to apply in WFI and PW production methods destined for the USA.
The alternative - electrolytic conductivity measurement.
Enter the measurement of electrolytic conductivity. This is a well-established means of measuring water quality or the concentration of a chemical solution.
The technique involves measuring the ability of a liquid to pass an electrical current. It is employed in industries as diverse as detecting trace impurities in power generation steam turbine cycles, in semiconductor etch rinse duty, food and beverage industry control of product/water interfaces and detergents strength monitoring, and general chemical industry concentration measurements. Conductivity measurement has evolved over the years to provide some very specific industry techniques so that it is now a very simple to use method backed by highly sophisticated technology. It is interesting therefore to see why the United States Pharmacopeia chose to request devices without these advances. Firstly it is necessary to review how conductivity is measured. There are two basic requirements: a means of passing, or inducing, a small alternating current through a precise volume of liquid to be measured - this function is carried out by the conductivity measuring cell; and an instrument capable of supplying power to the conductivity measuring cell and measuring small changes of electric current passing through the solution, thereby providing means of indication, recording or transmission of a current proportional to the electrolytic conductivity.
As opposed to electron flow in 'metallic' conductors, current in liquids is transported by ions.
To measure conductivity it is necessary to have a conductivity meter and a measuring cell.
The cell embodies the electrodes and provides a fixed geometry between the electrodes and the sample.
In view of the wide variety of samples suitable for measurement and control by their conductivity, and the wide range of measurable conductivities, many types of cell are available with different geometries, electrode sizes and cell materials.
One major requirement of USP Monograph 645 is that the cell constant is confirmed.
This is an onerous task not fully appreciated by the user.
In relatively high conductivities, say over 500mS/cm (microsiemens/cm - this is the standard unit of measurement for conductivity), it is comparatively easy to source a calibration solution and verify the validity of measurement.
It is even possible to trace it back to national standards.
Below this level, when one approaches the sub-10mS/cm range of WFI and PW samples, it is impossible.
No realistic calibration solution is available.
Even if it were, it would be grossly unstable, absorbing carbon dioxide from the atmosphere and rapidly increasing its apparent conductivity value.
The solution is for instrumentation suppliers to substantiate the validity of their cell constants by comparing to traceable transfer standards. During manufacture consistency of the engineering tolerances is controlled via calibrated dimensional gauges and statistical process control.
The in-house primary standard conductivity cells are traceably calibrated on a regular schedule using a certified National Institute of Standards and Technology (Nist) conductivity solution transfer standard.
In-house secondary standard cells for the production line are then calibrated from these primary standard cells.
Cells from the production batches are calibration checked against secondary standard cells to ensure conformity of manufactured cell constant (K) values to the certified Nist standard.
These are measurements not taken lightly and must be made under strict regimes.
Measurements are made at elevated levels in a temperature-controlled environment and readings compared with a traceably validated standard cell.
The result is a sensor that has been verified to be within the +/-2% tolerance demanded by Monograph 645.
If the supplier is effective, the actual tolerance will be better, at +/-1% of K. Some suppliers offer a correction factor to input into the instrument enabling correction of manufacturing inefficiencies. Others are guaranteed to be within tolerance without any further adjustment.
A similar but altogether easier requirement is validation of the instrument's accuracy and resolution to better than 0.1mS/cm.
Modern microprocessors and signal processing technology mean this kind of performance is readily achievable.
By replacing the cell with precision resistances traceable to Nist, meter calibration is simply performed. Re-calibrations, or rather re-validations, of the cell constant and instrument performance are generally recommended on an annual basis.
This is usually performed back at base by the instrument supplier.
There are a significant number of users who prefer to do this themselves and have invested in traceable, portable verifying systems.
These very handy tools provide a valuable service but they are not without their problems, as will become evident further on.
Conductivity measurement essentially concerns the mobility of ions through aqueous media.
These mobilities are greatly affected by temperature and result in a range of specific temperature coefficients ranging from 1-3% per C.
Naturally, as heat is applied dissolved ions become more mobile.
The ubiquitous water molecule springs another surprise.
Water is readily dissociated into highly conductive H+ and OH- ions.
These small and highly mobile ions have high coefficients up to 5% per C, since the effect of temperature results in increased dissociation.
Instrumentation suppliers have strived to provide ever more accurate temperature compensation mechanisms for specific contaminants.
These have proved a boon to users in semiconductor manufacture or power stations where contaminants are well defined.
The instrument measures the raw conductivity that is, not temperature-compensated, and the process temperature.
Since the behaviour of pure water is known, its contribution is extracted from the determined value.
The resulting measurement is compensated with a dedicated curve specific for a particular contaminant.
The compensated conductivities for concomitant species, ie, pure water and contaminant, are re-combined to provide an accurate measurement. Against this backdrop the pharmaceutical industry adopted conductivity as a method of detecting levels of contaminant in purified water and water for injection.
Monograph 645 was established to supersede testing of trace contaminants.
But which coefficient should one use for chloride or sodium? The answer, perhaps surprisingly, is none.
Rather than adopt a specific temperature coefficient, the solution was to measure a raw conductivity only.
The pragmatic solution was to select the least conductive impurity at a range of temperatures and to determine the maximum allowable conductivity for the maximum concentration permissible by the USP.
The supposition being that as long as one operates below a certain band of values for low conductivity species, then one will also be below the level of higher conducting impurities.
Since the sodium chloride curve enshrined in IEC746 doesn't always give the lowest conductivity permissible by USP, this curve cannot be used as a pass/fail limit.
Instead, USP adopted a composite table of its own.
The so-called Stage 1 test requires that a manual sample is tested for temperature and the conductivity limit is checked against the table above.
If a measured temperature is not shown, the next lowest is selected.
Automation enables an in-line Stage 1 test whereby the instrument not only displays but also outputs both parameters for permanent record.
An in-line sensor is often mounted between the exit of a reverse osmosis system and the inlet to a pure water storage tank.
Water is delivered to and from the process from this tank.
To ensure rapid detection of non-compliance, a fast re-circulating loop is also employed, in which another validated measurement is installed.
More sophisticated modern transmitters also incorporate the USP test limits and permit an error band to be set so that the user can operate up to a pre-set margin of safety.
An example would be that the device is configured to alarm at 10% below the maximum permissible.
Such devices can also provide dual validated outputs of conductivity and temperature to facilitate a permanent QA record to be taken.
We can now return to the issue of portable checking devices.
It is by now evident that when a continuous sample is taken from a process for passing through a validation unit, there is a risk of temperature errors.
If there is a significant delta between process and sampled temperatures, there could equally be a corresponding measured conductivity error. This may not be due to actual differences in levels of contaminants, but simply the effect of temperature.
The solution is to keep sample lines short.
It should also be noted that carbon dioxide has a propensity to infiltrate manual samples, such that it will quickly raise the background conductivity.
The tendency is for manual samples to read higher than expected, particularly at low contaminant levels.
It is always better to keep the conductivity cell enclosed in the measuring solution by performing in-line measurements.
Quite often a manual sample will be erroneous by the time it has reached the laboratory, purely because of ingress from the atmosphere.
The great affinity of carbon dioxide should also make the validation unit user wary. Not only should sample lines be kept short, but the material of transport line should be considered.
Either stainless steel or nylon should be used.
Other plastic lines such as PTFE have a tendency to permit the passage of carbon dioxide through the walls and raise the local conductivity.
When installing conductivity cells, the recommended orientation is with sensing electrodes pointing downwards and attached at the top of an elbow.
Alternatively, side on connection is possible with the exit below.
Either way, effective in-line cleaning is assured by pointing the sensor element in the direction of cleaning agents. Conductivity cells are unaffected by their orientation per se however they are affected by the collection of air which serves to depress the apparent measured value.
Installation in the branch off a line could exacerbate this situation and makes it more difficult to guarantee effective sensor cleaning.
When all care has been taken, a perfectly successful Stage 1 test is accomplished.
If the sample fails, then further testing is required.
Stages 2 and 3 are off-line and involve actively encouraging ingress of carbon dioxide.
A sample is taken and stirred vigorously until a stable conductivity value is determined.
This should be below a maximum of 2.1mS/cm at 25C to pass Stage 2.
Failing that, the subsequent Stage 3 involves further adjustment of the ionic strength to target pHs and look up against a check-up table.
The use of raw conductivity measurements has provided the industry with a successful method of determining contaminant control in WFI and PW applications. Measurements can now be performed in-line, improving water quality, improving reliability of tests, simplifying the procedure and giving an insight into the degree of purity rather than merely pass or fail.
