Instrumentation for pH monitoring in the process industries leaves something to be desired, says Rob Pastushak, senior technical supervisor of pharmaceutical manufacturing for Raylo Chemicals (Edmonton, Canada), a member of Degussas Exclusive Synthesis and Catalyst Business Unit. There arent a lot of instruments out there that can withstand intense pH ranges, he says. Most sensors for pharmaceutical applications are water-based, but the products were working with are usually isolated from aqueous-organic mixtures. The result, Pastushak says, is that most probes lack the robustness, accuracy and compatibility required for key chemical processes.
In Raylos active pharmaceutical ingredient manufacturing, pH is of course a critical processing parameter (CPP) and processes must be validated for extremely tight pH tolerances. With no reliable instrumentation to measure and track pH, maintaining strict process control is extremely difficult.
Such was the case with the mixing of an API intermediate for a successful drug substance manufactured by Degussa. The mixing involves water, two organic solvents and a strong mineral acid. At the start of the process, enough acid is charged into the vessel to achieve a 0.6 pH. Water is then added to dilute the organic level to allow pH measurement and to control impurities. Organic solvent is then used for the extraction of organic impurities. From this point, caustic is added to the aqueous phase incrementally until a narrow window just below a pH of 4 is achieved, at which point the product precipitates out. The mixture is agitated to achieve maximum yield and then proceeds to filtration.
Our approach to pH adjustment used to be to charge a little bit of caustic, not trust what the field instrument said because it was usually off by 1 or 2 pH units, and then grab a subsample and measure it in the lab, Pastushak recalls. Depending on the reading, you would either continue to add caustic or back-adjust with acid if you overshoot the target pH range. Once the target pH is achieved, you then proceed to the next stage in the process.
This elaborate adjustment process often took 24 to 36 hours, with dozens of samples obtained and measured on a benchtop pH meter in the laboratory. When you process 3,000 to 5,000 liters and add 5 to 10 kilos of caustic solution at a time, it might take 20 to 40 lab tests to ensure that the pH is right, notes Pastushak. This just killed our production efficiency.
An equally perplexing issue was that the pH probes Raylo was using could not withstand the harsh chemicals being used. The manufacturer would often go through two or three probes for each batch. At $600 or so a pop, this was clearly unacceptable. The probes we were using were mainly designed for the pulp and paper industry, or solely for aqueous-based applications, Pastushak says. They werent really designed for an aqueous-organic mixture, let alone aggressive chemicals.
Raylo identified Teflon as the ideal material it was looking for, but few such probes were on the market. The next best thing was Ryton, a hard polymer one grade below Teflon and offered by Invensys Process Systems, Measurements & Instruments Division (Foxboro, Mass.), in its Foxboro 871PH Series sensors. Pastushaks team found that the Ryton probe withstood aggressive reagents and solvents.
But there was still the O-ring issue. Invensys fashioned a ring made of Kalrez, a flexible elastomer, which did the trick. Each ring may cost in the neighborhood of $200, Pastushak notes, but in three years of use he has not had to replace one ring.
The probe has proven to be accurate to +/- 0.03 pH units, well within target limits. The sensitivity of the probe was excellent, says Pastushak. It was giving us to 0.01 pH unit readings, where actual adjustment range was plus or minus 0.2.
The API mixing includes a recirculation loop through which the slurry mix travels through via a double-diaphragm Teflon-lined pump, before it is introduced back into the mixing vessel. The pH probe plugs into a T-adaptor midway through the loop. Readings for pH are collected in seconds.
Inputs and outputs are relayed to an HMI linked to a Moore Control Systems (Katy, Texas) APACS distributed control system, which connects hard pH data to a main APACS module in the center of the plant. Supervisors can monitor the pH data via a local alphanumeric graphical display at the same time that operators view the human machine interface (HMI) in the reactor suite or from remote interfaces.
Despite the fact that pneumatic valve operations, process setpoint changes and alarms are controlled by the Moore system, operators and supervisors still control much of the process. Its a necessary balance because the chemical processes and equipment are so complex that we need to have people interacting with the equipment and physically making setpoint changes and adjustments continually to ensure that they are in the stated ranges as described in the batch operating instructions, Pastushak says.
For the pH adjusting, the operator controls the flow of caustic being charged to the vessel. Could the process be completely automated? Its theoretically possible, Pastushak says. Youd have to have a sophisticated logic program to measure pH and adjust the flow of acid or base, but you have to assess it from a business case point of view to justify the expense.
Calibration of probes is done on the plant floor prior to the start of each batch. An instrument technician uses a series of buffer solutions to calibrate the probe through the local control panel interface. The calibration is valid for 24 hours.
Through the calibrations, were able to get reproducible results to +/- .03 pH units within the process setting, Pastushak says. Thats incredible for a pH probe.
Despite the reliability of the on-line probe, Raylo maintains a backup unit in the laboratory to cross-check floor readings. In this type of environment, secondary checks are key, says Pastushak. Thats ingrained in our cGMP philosophy.
We have 95% full field control, and then a 5% confirmation element in the lab, he continues. The number of samples are drastically reduced, from upwards of 40 to just one or two. From a safety perspective, reducing the sampling means reducing operator exposures.
The result has been shorter cycle times, and huge cost savings. From 24 to 36 hours, the pH adjustment can be done now in 8 to12 hours, he says. That correlates to about 50 extra processing days in the plant per year from a reduced processing time. That allows you to make more batches of the same product, or make room for additional products. In other words, theres no small payoff for aggressively monitoring pH.