As Total Organic Carbon (TOC) analyzer technology improves, industry is more able to utilize it for real time release of clean utilities and for cleaning process risk mitigation. As end users determine and communicate their goals in the use of TOC and other process analytical technologies (PAT) for cleaning systems, the requirements and specification for off the shelf analyzers will evolve to meet those goals. Unfortunately, industry has been somewhat reluctant to adopt PAT until a precedent of regulatory acceptance has been set, resulting in a functional evolution of online instrumentation that has been slow to meet current needs. However, through careful analysis of process requirements and corporate quality and risk management goals, currently available instruments can be selected to meet end user goals and still drive the development of improved analytical instrument and sensor technologies.
Currently, automated CIP systems provide greater assurance that production residues are reliably and reproducibly removed through monitoring and control of process variables that are critical to cleaning process performance rather than forward processing criteria that indicate cleanliness of the equipment. This includes monitoring and control of temperature, detergent concentration, cleaning solution flowrates, and cleaning solution contact times. To provide greater assurance that the cleaning process is reliable, industry relies on empirical data collected during and after multiple successful process trials, thereby validating performance. One of the advantageous characteristics of process analytical technologies (PAT) for the manufacture of therapeutic substances is the capability to provide process information directly relevant to product quality for a particular process step. This information, if provided within an adequate timeframe to allow for critical decisions or the initiation of subsequent process steps is referred to as forward processing criteria (FPC).
Upon completion of the cleaning cycle, most systems only monitor the conductivity of the rinsewater to ensure that cleaning solutions have been adequately removed. Unless the products manufactured in the process equipment are also conductive, monitoring conductivity only provides partial assurance that the equipment is clean and ready for the next production step. For automated cleaning processes, appropriate FPC would indicate that the process equipment is free of not only the cleaning agents introduced to remove production residues, but that production residues have also been removed from the system. Since a large majority of pharmaceutical and biopharmaceutical products are based on organic molecules or proteins, monitoring total organic carbon (TOC) is a step towards ensuring that organic materials from previous production operations have been adequately removed from the system.
As members of a highly regulated industry, manufacturers of pharmaceutical and biopharmaceutical products have not been the first to adopt new technologies until they have been developed and proven. TOC saw adoption primarily by municipal and industrial water treatment entities with less stringent regulatory requirements and different process requirements. Accordingly, with municipalities as the primary market, TOC analyzers were initially developed as instruments to reliably monitor high throughput treatment processes for extended periods of time, where larger changes in monitored TOC were tolerable process perturbations. Pharmaceutical manufacturers, especially those manufacturing parenteral substances have much less tolerance to changes in TOC; a few hundred parts per billion can often impact final product quality. Since TOC would be used to monitor only the final rinse phase of a CIP cycle, the actual operating time of the instrument is much less than that typical of municipal water applications. The final rinse phase of most CIP cycles only amounts to a period of minutes in which critical information must be collected as opposed to the continuous operation of a TOC analyzer employed in a water treatment application. Given that analyzers were first developed for continuous water treatment processes, certain characteristics such as sensitivity to conductivity sensor interference, analytical result lag time, TOC spike recovery, and long-term calibration instability were more tolerable than they would be for the pharmaceutical industry.
Addressing Instrument and System Installation Requirements
Regardless of the exact means employed to oxidize organic compounds, all analyzers completely degrade organic compounds to CO2. The sensor technologies utilized to quantify carbon dioxide are also varied and should be considered when selecting an analyzer for a specific application. The most commonly employed sensors are based on either liquid phase conductivity of CO2 or on gas phase analysis using Non-Dispersive Infra-Red (NDIR) spectrophotometry. Within each category of detector, there are variations that may influence the decision as to which analyzer is most appropriate for a given application but the major consideration is the suitability of the application and analyte stream for aqueous or gas phase analysis. For aqueous streams, conductivity is employed, but is subject not only to false readings from other ionic compounds, but to the dependence of conductivity on temperature.
Analyzers that utilize conductivity are also referred to as conductometric TOC analyzers. This group can be further subdivided into those that directly measure the conductivity of the sample immediately after oxidation, and those that measure the conductivity of the solution after allowing the evolved CO2 to permeate across a selective membrane. The former are termed direct conductometric analyzers, an example of which is shown in Figure 1. Membrane-conductometric analyzers typically split the sample flow into two streams and measure the dissolved CO2 from atmospheric carbon dioxide in the reference stream, and the total dissolved CO2 from the photooxidation reactor in the second stream as shown in Figure 2. Regardless of the type of detector, each must in some way address the variation of ionic conductivity as a function of temperature. Alternatively, opting to analyze the evolved carbon dioxide in the gas phase through the use of an NDIR detector obviates the need to compensate for conductivity temperature dependence.
For conductivity based detection systems in which temperature can affect the measurement of aqueous phase CO2, introduction of the sample directly to the analyzer from the CIP return manifold is a relevant concern since the temperature of the CIP final rinse return flow is often elevated to aid with drying of the cleaned process equipment. While manufacturers do not overlook the fact that conductivity is a temperature dependant measurement, different instrument designs compensate for it in different manners. Temperature variations in the sample stream may be addressed through temperature compensated conductivity sensors, or measurement of raw conductivity data with sampling apparatus that allow for temperature equilibration through ambient dissipation or active heat exchange. For instruments that are less tolerant of sample variations, a heat exchanger may be installed upstream of the instrument in the sample line. In most cases, if an external heat exchanger is necessary, the manufacturer offers a complementary product designed to work with their particular instrument.
Instrument & Process Delays
There are multiple delays that one must be aware of for the implementation of a TOC analyzer in an online capacity. The first is related to the time necessary for the instrument to “come online” and be in a state of readiness for analysis at the appropriate stage of the process. The second delay is the duration from the time that the sample is injected into the instrument to the time that an analytical result is achieved. This time is comprised of the residence time within the instrument as well as ant hold periods associated with reaction of any oxidizable substances, and the time necessary to measure the amount of CO2 generated by the oxidation reaction. The final and last delay is associated with the residence time of the sample in external sampling equipment and lines leading from the CIP return line to the inlet of the instrument.
Online technologies like TOC that perform destructive analyses on a captive sample can exhibit delays on the order of 1 to 5 minutes for the oxidation reaction or equilibration of the instrument sensors before the results are available. While this is somewhat immaterial to a continuous water treatment process, the duration of the final rinse is on the same order as the equilibration. As such, the delay between when the result and the sampling time may be reduced by beginning the equilibration of the instrument prior to the time it is required. For this duration, the instrument will often require that sample is fed to it even though no analytical result will be forthcoming for this time period. This duration will of course be different depending on the manufacturer of the instrument, so some characterization must be completed in advance, but compensation for the equilibration lag may be addressed by knowing when the final rinse phase of the CIP cycle will occur, and integrating a start signal into the CIP system automation that precedes the start time of the final rinse by a duration equal to or greater than the equilibration lag. This lag is further exacerbated by the sampling equipment configuration. Longer sampling lines lead to a greater residence time before the sample is delivered to the analyzer. Regardless, by removing the delays traditionally associated with sampling manual sampling, storage and offline analysis, critical to quality decisions may be made on a more immediate basis. The immediacy of critical data is the first step in reducing risk and allows for less down time between manufacturing campaigns that would otherwise be spent waiting for analytical results.
The general configuration employed in one test case is pictured in Figure 3. In this instance, a separate reservoir of high purity water is pumped to the analyzer to allow the instrument to equilibrate. The pump was necessary not only to deliver water from the reservoir, but to provide motive force for the fluid in the CP return line since it was under vacuum (to assist with return flow). The pump could of course be eliminated if the water supply and return line were under positive pressure, or if manufacturers integrated suitable sample pumps into the analyzer system. Similarly, an internal reservoir or equilibration “reagent” source could also avert the need for the manufacturers to supply clean water to the analyzer for equilibration.
Analytical results are typically time-stamped and stored locally on the instrument, but may also be logged by a plantwide control system. This result corresponds to a process event off set by the sample equipment residence time, the instrument residence time, and any intrinsic reaction or measurement time required to obtain that result. The real-time lag may be further minimized by keeping the length of sample lines as short as possible and the flowrate through those lines as high as possible. To further allow for synchronization of the data from the TOC analyzer with data collected by other instrumentation on the skid the analyzer must first either have the capability to synchronize its onboard time with that of the plant Supervisory Control and Data Acquisition (SCADA) system, or export analytical results directly to the SCADA in an appropriate format. Additionally, it is necessary to characterize the total time lag resulting from the residence time of the sample in the sample equipment as well as the residence time within the instrument.
Spike Recovery
Since CIP operations have the potential to have higher concentrations of organic materials in the final rinse, it is advantageous to select an analyzer that is capable of analyzing concentrations in excess of what would be expected for compendia grade waters. Even though a CIP cycle is validated, it does not mean that the initial portions of the final rinse cannot reach values of 10 ppm TOC or greater. At the very least, an analyzer should be able to recover from a concentration spike that exceeds the range of the instrument even if it cannot quantify the spike. For analyzers that add oxidizer to increase the range of the instrument, recovery from an excursion such as this may just involve increasing the oxidizer and acid flowrates for a period of time to purge organic materials from the reaction zone. Other instruments that rely purely on ultraviolet photooxidation without reagents may require a special cleaning regimen.
Additional measures may be taken to protect the instrument by means of the CIP system automation. Most formulated cleaning agents contain organic surfactants as well as conductive ions and are detectible by conductivity and TOC. To ensure that larger concentrations of organic surfactants are not introduced to online TOC analyzers, an automation interlock can prevent opening of the TOC analyzer sample valve unless the rinsewater conductivity is below a certain threshold value. Alternatively, manufacturers have already developed firmware based dilution schemes for automated calibration, so a similar automated dilution strategy could be employed for initial samples to not only expand the range of the instrument, but ensure that a surge of organic matter would not be introduced into the analyzer.
From the standpoint of regulatory compliance and instrument qualification, the requirements of CIP are different than those for compendial water systems; it is not necessary that online TOC analyzers for CIP systems are functionally treated in the same manner as an instrument in the laboratory. Most laboratory instruments serve double duty for the analysis of compendia water and are therefore subject to the regulatory requirements for pharmaceutical waters. The USP requires that TOC analyzers for compendia waters are tested periodically with suitability samples that include standards that are both easy to and difficult to oxidize. Sucrose is typically employed, as a representative of a typical chemical for oxidation. The hard to oxidize material specified in the USP is parabenzoquinone, which while it is more difficult to oxidize than sucrose, may not be the worst case material the analyzer will encounter in solutions rinsed from manufacturing equipment. For this reason, it is recommended that end users review the potential materials in the system, including the cleaning agents and determine a worst case substance to oxidize.
Once qualified and operational, an analyzer on the manufacturing floor typically becomes the responsibility of either the manufacturing or metrology groups. End users in these groups are less likely to have the time to devote to frequent calibration or suitability checks. In order to address this need, systems must be made robust and require little interaction from operators on a day to day basis. As it is, most instruments available exhibit calibration stability for about a year. Likewise, unless there is a system malfunction, most instruments are stable and perform in a repeatable and reproducible manner for system suitability checks. By limiting the amount of technical understanding required for introducing standards, many manufacturers have also addressed this with off the shelf coded products that the instrument can recognize and analyze with a minimum amount of interaction from technicians or end users. Implementing such measures across the industry would greatly simplify the selection process for online CIP analyzers.
Ultimately manufacturers want a product that they can sell for which there is a market, and end users want a product they can purchase that meets all of their needs with a minimum of modifications to accommodate the instrument.
Author Biography
Keith Bader received a bachelors of science degree in Chemical Engineering from the University of Colorado at Boulder in 1995. Keith is the Director of Technical and Quality Services at Hyde Engineering + Consulting, Inc. where he is responsible for the implementation of Technical and Quality documentation and standards for Hyde Engineering + Consulting, Inc. on a corporate wide basis. Keith also provides high level consultation externally to Hyde’s Engineering’s clients in topic areas ranging from strategic quality and validation documentation to design of experiments for supporting studies. Keith has focused in recent years on the implementation of online instrumentation and process analytical technologies for clean-in-place systems, as well as the translation of bench scale cleaning process development data to full scale manufacturing systems.