Johnson Matthey Technol. Rev., 2016, 60, (4), 277
Palladium Impurity Removal from Active Pharmaceutical Ingredient Process Streams
A method for scale-up
- By Stephanie Phillips* and Duncan Holdsworth
- Johnson Matthey Plc, Orchard Road, Royston, Hertfordshire, SG8 5HE, UK
- Pasi Kauppinen**
- Johnson Matthey Finland Oy, Autokatu 6, FI-20380 Turku, Finland
- and Carl Mac Namara
- Johnson Matthey Plc, PO Box 1, Billingham, Cleveland, TS23 1LB, UK
In this article, we will look at palladium impurity removal from active pharmaceutical ingredient (API) process streams using metal scavengers and the drivers for the implementation of such processes. The article will review some of the available scavengers and detail how Johnson Matthey approaches the trial work and the methods used for screening, optimisation and scale-up of the scavenger process. It will outline the steps taken to ensure smooth transfer of the metal impurity removal process from lab to plant. This will include Johnson Matthey data from batch isotherm, kinetic and fixed bed trials and the application of mathematical models for performance characterisation and scale-up, which all feed into the final system design. Performance data for a number of the Johnson Matthey range of scavengers will be referenced both in batch and cartridge systems and the benefits of using the scavengers in a cartridge system will be presented.
Batch Isotherms and Kinetic Studies
The need to remove palladium (Pd) from API process streams is driven by International Conference on Harmonisation (ICH) Q3D guidelines which dictate the permissible levels of Pd allowed in the final drug product. The platinum group metals (pgms), including Pd as well as platinum, rhodium and ruthenium are generally considered as route-dependent human toxicants (ICH Classification 2b), (1), therefore the limits for these metals are low – 10 μg g–1 as an oral concentration in the drug product, drug substance or excipient. This oral concentration is converted into a permitted daily exposure limit for each platinum group metal of 100 μg d–1, which is based on an assumed worst-case dosage level of 10 g d–1. If the dosage rate is known to be lower, the permitted oral concentration may be increased. For the pgms, it may also be an option to use a component approach, whereby the overall level of the element in the excipient would be considered. As the pgms would only come from a catalyst source, the overall level of the element in the excipient could then be determined. For a chemist developing or manufacturing the API, a target of 10 μg g–1 or even lower may still be put in place to avoid process re-works (1).
Despite the challenges of removal, the use of homogeneous Pd catalysts offers many benefits over classical, often stoichiometric reaction protocols (2–4). It is often possible to achieve a reduction of the number of steps in a multi-step synthesis of a target compound, as well as increasing overall yields, by employing Pd catalysis. In order to achieve the benefits offered by the catalysts, there will be a subsequent need for Pd removal (4–10).
Numerous Pd removal methods are available to the chemist and the optimal method is typically selected based on cost, time, quality and ease of implementation. When considering which Pd removal method to use, the default is often to use standard methods such as distillation, crystallisation or carbon adsorption to remove the Pd (10, 11). All are efficient methods of separation; however, the process costs, and their robustness, may be challenged during scale-up. Equally, consideration needs to be given to the time taken to run the process and any impacts that the removal method may have on product yield.
Scavengers provide an excellent solution for Pd removal issues, particularly at low metal levels (6, 8). Johnson Matthey’s screening programme ensures that the best scavenger is selected based on the reaction conditions, but also that the process is robust and that consistent removal of Pd is achieved, irrespective of process scale.
There are a range of different scavengers available on the market for product purification (5–8). A significant portion of these are based on a silica support, however others, such as those based on polymer resins and polymer fibres (5, 7) are also available. The Johnson Matthey range includes all three classes of support materials and details of the core product types are listed in Table I, with specific product details provided in Table II.
Below is an outline of the steps involved in achieving a robust, long-term metal removal solution. This full screening package can be completed by Johnson Matthey or at the customer site with technical support provided.
The extent to which a scavenger can remove Pd to the required levels is critical. The initial trials within the screening programme look at a number of scavengers for Pd removal efficiency. Conditions can then be optimised with respect to the amount of scavenger, the temperature and whether a mix of scavengers with different functionalities should be used together. Time taken for Pd removal is considered within the batch isotherm, and kinetic and fixed bed studies are carried out to generate the performance data required to model, scale-up and design an optimised process for full-scale application.
The Screening Programme: Initial Screening
The initial screen is used as a feasibility study to determine the most suitable scavenger for Pd impurity removal from a particular sample or application. All scavengers used in the screen are commercially available and all screening is completed on representative process samples. Where solid samples are provided, these are dissolved in an appropriate solvent.
Prior to carrying out any screening, customer process data is required to ensure optimal results taking into account any constraints relating to operating conditions.
Initial screening can be carried out using a variety of equipment, with the two preferred options being: (a) roller mixer machine; or (b) flask with overhead stirrer to allow best mixing of and contact with the scavengers. A hot plate, with a magnetic stirrer or a Radleys CarouselTM, have also been used, with care being taken to ensure suitable contact with the scavenger material.
Initial Screening: Experimental
All solutions are filtered prior to screening to ensure homogeneous sampling. A sample of mother liquor is retained for analysis. Typically 1 w/v% scavenger is allowed for Pd concentrations up to 500 mg l–1. However, if a single element (or the total pgm concentration) has a higher concentration than this, the amount of scavenger is increased. For solutions with a pgm concentration in the region of 500–1000 ppm, for example, 3 w/v% is recommended for initial screening. During optimised screening, the amount of scavenger will be adjusted based on knowledge of the API process and the type of Pd catalyst used during it. Consideration will also be given to the presence of other reagents and metals which may be present and could compete for sites on the scavenger.
Initial screening is completed at room temperature for 2 h. Approximately six to eight scavengers are included in the initial screening. All solution samples are analysed using inductively coupled plasma-optical emission spectroscopy (ICP-OES) or atomic absorption spectroscopy (AAS) to determine the Pd concentration pre- and post-screening (12).
In the examples shown below, an organic process solution used in an API manufacturing process with an oncology indication was the selected material for the screening programme. This was a more challenging Pd removal project, both in terms of the ease with which the Pd was removed and the low initial concentrations of the Pd in solution. The final target level was 10 μg g –1 in the API, equating to >1 mg l–1 in the solution.
Initial Screening: Results
Results from this screening can be presented in a number of ways. In this case they are shown as:
(a) Table III, where the initial raw data is given and the difference in concentration between the original solution and the solution following scavenger treatment is shown. Conversion of this to the actual concentration in the API stream is given in the final column; and
(b) Figure 1, where the scavenger performance is shown from left to right in terms of optimal performance.
In this example the solid API, with 98 μg g–1 Pd present as an impurity, was diluted with 15 volumes of process solvent prior to analysis. The analytical results on the solution were therefore multiplied, based on this dilution factor, and taking into consideration the density of the solvent to give:
(a) the initial Pd levels present in the API; and
(b) the Pd levels present after scavenger treatment.
At this stage of the screening, the final Pd levels following scavenger treatment were all above the target required by the process, necessitating further optimisation. For most screening, the best three performing scavengers from the initial screening are taken through to the optimisation stage.
Based on the results from the initial screening, three scavengers were taken through to the optimisation stage: QuadraPure® TU (thiourea functionalised polymer resin), QuadraSil® MP (mercapto propyl functionalised silica) and QuadraSil® AP (amino propyl functionalised silica). During this phase, testing focused on modifying the amount of scavenger used, the effects of temperature increase on the scavenging performance and possible use of a multi-scavenger system. Time taken for recovery was determined during the later kinetic trials.
Based on the Pd removal rates from the first screen, the amount of scavenger used was modified accordingly. As a 0.5 w/v% addition (with respect to volume of process solution) did not yield complete metal removal, the amount of scavenger added was increased to 1 w/v% and 2 w/v% (Figure 2).
Typically the higher temperature would be:
(b) 10ºC below the boiling point of the solvent, or
(c) dictated by the process conditions or API stability.
From these results, the optimal conditions were found to be 2 w/v% of QuadraSil® MP for 2 h at room temperature to give 0.5 mg l–1 in the solution, which equated to 5.65 μg g–1 in the API. This was below the required upper limit of 10 μg g–1. No performance improvements were seen by increasing the temperature and, from a processing point of view, room temperature was preferable. It was also deemed unnecessary to test a mixture of scavengers as the targeted Pd removal was achieved with a single scavenger.
At this stage in the screening programme, the scavenger and the conditions under which that scavenger will be best utilised have been determined using a fixed time under batch conditions. The decision for the process chemist now is whether the process will be transferred to the pilot plant in batch or in a cartridge system. To support this decision, a batch isotherm will be run followed by a kinetic trial for cartridge systems. This data is then utilised in a first principles modelling system and the actual and theoretical data compared before the system is defined.
During this phase of the screening, we are looking at:
(a) the theoretical maximum metal loadings which can be achieved for the target scavenging process. Data from this stage will help to determine the weight or volume of scavenger required in the larger scale process
(b) the breakthrough point (where the metal is no longer being removed by the scavenger) to determine at what point the chemist or engineer stops the process
(c) the kinetics of the adsorption process, considering the various mass transfer resistances which control the Pd from the liquid phase to the solid phase. This will indicate what contact time is required to ensure the metal content is reduced to the required levels.
For this project, a kinetic trial in batch was considered for the three best performing scavengers, shown as raw data in Figure 5. Due to the more challenging nature of the solution, kinetics were slower than is often seen, with target recovery in 4 h. This was within process requirements. Note that kinetics for Pd removal are typically fast with up to 99% recovery often being seen within 10 min. Factors that can affect the metal uptake are:
(a) nature of the Pd catalyst (in terms of accessibility of the Pd species)
(b) presence of other reagents that can compete with the metal uptake, and
(c) reactions within the API itself.
A batch isotherm trial was carried out to determine the maximum loading capacity of the material as well as investigating how changing initial concentrations of the feed solution may affect this capacity (Figure 6).
Prior to designing a full-scale cartridge system, a column trial is carried out in which feed solution is flowed through a fixed bed of the ion exchange material and the outlet concentration from the column is measured over time. An outlet concentration of 0 mg l–1 indicates that all of the Pd has been recovered from the feed solution. An outlet concentration above 0 mg l–1 indicates that the material has almost reached its maximum loading capacity and that the feed flow should be stopped. The time before the outlet concentration rises above 0 mg l–1 can be increased by increasing the column size (Figure 7).
First Principles Modelling
The results obtained from laboratory isotherm and column trials are used to estimate several equilibrium and kinetic parameters which describe the adsorption of Pd from solution. The parameters are part of the models described in Equations (i) and (ii) (13, 14) and by fitting these models to the experimental data the values of these parameters are estimated.
Equation (i) is fitted to the results shown in Figure 6 to determine K and qt, the material equilibrium constant and maximum capacity respectively. The results of this fitting are shown in Figure 8. An R2 value of 0.99 was obtained indicating the fitting is very good.
Equation (ii) is fitted to the results shown in Figure 7 to determine the kinetics parameters of adsorption (contained within the ∂q /∂t parameter in the equation which represents the rate of adsorption). The results of this fitting are shown in Figure 9. Again a very high R2 value of 0.99 was obtained.
Once these model parameters have been estimated, Equations (i) and (ii) can be used to determine the performance of the QuadraSil® MP material as a function of time, feed flow rate, concentration and the size or scale of the column through which the feed is flowed, effectively allowing scale-up and optimisation of the process design to ensure full Pd recovery, as further discussed in the next section.
The ultimate aim of scale-up is to take the understanding obtained from the small-scale scavenger performance and propose a system that can achieve 100% scavenger utilisation at the end of the system life, i.e. the entire scavenger mass should be loaded to its maximum capacity at the point of scavenger change out. In reality this is an impossible task as, for example, in a column system axial diffusion, radial velocity gradients, near-wall effects and slow kinetics inevitably lead to broadening of the mass transfer zone and non-ideal behaviour (15, 16). However, a robust engineered solution should aim to get as close to full utilisation of the scavenger as is possible for a given set of conditions as this helps to minimise the installed scavenger volume and, ultimately, the process cost. For a batch system, the challenge is even greater, which drives Johnson Matthey to recommend a cartridge system wherever possible. Scavenger systems using cartridges also offer the added benefit of containment when a potent API is being manufactured.
Optimisation can be achieved by manipulating variables such as the liquid linear velocity, the column diameter and the aspect ratio of the scavenger bed (17), while being mindful of customer requirements that could place limits on the proposed design. Such requirements might include the need for the system to be either a once-through or a recycle system; the processing time requirements; the need for a reusable or disposable system depending on the potency of the API; and the operability of the system. At this point the scavenger system will be well-defined, leaving only the mechanical design and the selection of the final conditions and materials of construction to satisfy the needs of the process, the relevant regulations and ultimately the characterisation of a safe and robust system.
At this final stage, all of the data from the screening programme is collated and used in the design of the final system used during piloting. Examples of a scalable system, the sealed flow cartridge system, are shown in Figure 10.
The use of a screening programme has been demonstrated and the steps required outlined in this article along with the results obtained, in order to design a robust plant-scale recovery process. Whether the process will be run in batch or a cartridge system, the method is essential in ensuring that optimal use of the scavenger is achieved while attaining Pd removal targets. For challenging processes, such as the one highlighted here where the nature of the process solution is complex and the initial metal levels are low (in the diluted form), the screening process becomes even more valuable.
Johnson Matthey provides screening and scale-up services to the pharmaceutical industry and has become expert through its application of knowledge in this area. To date, numerous plant-scale scavenging projects are running globally.
“ICH Harmonised Guideline, Guideline for Elemental Impurities, Q3D”, Current Step 4 Version, International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, 16th December, 2014 : http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q3D/Q3D_Step_4.pdf (Accessed on 26th August 2016)
C. C. C. Johansson Seechurn, M. O. Kitching, T. J. Colacot and V. Snieckus, Angew. Chem. Int. Ed., 2012, 51, (21), 5062 LINK http://dx.doi.org/10.1002/anie.201107017
J. Magano, ‘Large-Scale Applications of Transition Metal Removal Techniques in the Manufacture of Pharmaceuticals’, in “Transition Metal-Catalyzed Couplings in Process Chemistry: Case Studies from the Pharmaceutical Industry”, eds. J. Magano, J. R. Dunetz, Wiley-VCH Verlag, Weinheim, Germany, 2013, p. 313 LINK http://www.wiley.com/WileyCDA/WileyTitle/productCd-3527332790.html
E. J. Flahive, B. L. Ewanicki, N. W. Sach, S. A. O’Neill-Slawecki, N. S. Stankovic, S. Yu, S. M. Guinness and J. Dunn, Org. Process Res. Dev., 2008, 12, (4), 637 LINK http://dx.doi.org/10.1021/op600280g
S. Phillips and P. Kauppinen, Platinum Metals Rev., 2010, 54, (1), 69 LINK https://www.technology.matthey.com/article/54/1/69-70/
G. Reginato, P. Sadler and R. D. Wilkes, Org. Process Res. Dev., 2011, 15, (6), 1396 LINK http://dx.doi.org/10.1021/op200127y
J. Frankham and P. Kauppinen, Platinum Metals Rev., 2010, 54, (3), 200 LINK https://www.technology.matthey.com/article/54/3/200-202/
C. J. Welch, J. Albaneze-Walker, W. R. Leonard, M. Biba, J. DaSilva, D. Henderson, B. Laing, D. J. Mathre, S. Spencer, X. Bu and T. Wang, Org. Process Res. Dev., 2005, 9, (2), 198 LINK http://dx.doi.org/10.1021/op049764f
M. J. Girgis, L. E. Kuczynski, S. M. Berberena, C. A. Boyd, P. L. Kubinski, M. L. Scherholz, D. E. Drinkwater, X. Shen, S. Babiak and B. G. Lefebvre, Org. Process Res. Dev., 2008, 12, (6), 1209 LINK http://dx.doi.org/10.1021/op800161v
C. E. Garrett and K. Prasad, Adv. Synth. Catal., 2004, 346, (8), 889 LINK http://dx.doi.org/10.1002/adsc.200404071
J.-P. Huang, X.-X. Chen, S.-X. Gu, L. Zhao, W.-X. Chen and F.-E. Chen, Org. Process Res. Dev., 2010, 14, (4), 939 LINK http://dx.doi.org/10.1021/op100094p
G. Lecornet, ‘Analysis of Elemental Impurities in Drug Products using the Thermo Scientific iCAP 7600 ICP-OES Duo’, Application Note 43149, Thermo Fisher Scientific Inc, Massachusetts, USA, 2016
C. E. Borba, G. H. F. Santos and E. A. Silva, Chem. Eng. J., 2012, 189–190, 49 LINK http://dx.doi.org/10.1016/j.cej.2012.02.017
R. B. Garcia-Reyes and J. R. Rangel-Mendez, Bioresource Technol., 2010, 101, (21), 8099 LINK http://dx.doi.org/10.1016/j.biortech.2010.06.020
J. B. Butt, “Reaction Kinetics and Reactor Design”, Second Edition, Marcel Dekker Inc, New York, USA, 2000, pp. 332 – 333
V. J. Inglezakis and S. G. Poulopoulos, ‘Heterogeneous Processes and Reactor Analysis’, in “Adsorption, Ion Exchange and Catalysis”, First Edition, Elsevier, Amsterdam, The Netherlands, 2006, pp. 148 – 160
V. J. Inglezakis and S. G. Poulopoulos, ‘Reactors Scale-up’, in “Adsorption, Ion Exchange and Catalysis”, First Edition, Elsevier, Amsterdam, The Netherlands, 2006, pp. 531 – 539
Stephanie Phillips is the Market Specialist for Johnson Matthey’s Advanced Ion Exchange (AIX) business at Royston, UK. Stephanie has worked in Sales and Marketing since 2001 and has been involved with the development and marketing of the Smopex® range of products since that time. Stephanie has a BSc in Chemistry and an MSc in Analytical Chemistry.
Duncan Holdsworth is a Senior Process Engineer in the Johnson Matthey AIX business and is based in Royston, UK. Duncan has worked as a Process Engineer within the various Johnson Matthey business divisions since 2010. He most recently joined AIX where he manages the scale up, design, fabrication and commissioning of the various process units operating in both the pharmaceutical and industrial chemicals markets. Duncan gained his MEng in Chemical Engineering from the University of Sheffield, UK.
Pasi Kauppinen gained his PhD in University of Oulu, Finland, and works currently as a Principal Scientist in AIX at the Johnson Matthey plant in Turku, Finland. In this role he is focused on developing AIX processes and products globally
Carl Mac Namara is a Process Engineer within the Johnson Matthey Water Technologies group, Chilton, UK. He obtained his MEng in Chemical Engineering from Cork Institute of Technology, Ireland, and an Engineering Doctorate from the University of Birmingham, UK. His doctorate and post-doctoral projects were based in Procter & Gamble’s Newcastle Innovation Centre, UK, where he carried out fundamental research on textile cleaning processes. His current role is focused on modelling and developing new water treatment technologies all the way from R&D through to commercial stages.