Reactor D

2.4 Reactor D

Reactor D (Figure 2.14) was developed in chapter 5 in order automate the screening and opti-misation of RAFT polymerisations. Only homogeneous polymerisations were performed in this chapter thus a stainless steel tubular reactor (PDMAm synthesis: 2.1 mm I.D, 5 mL /PtBuAm synthesis: 0.7 mm I.D, 2 mL) was sufficient. In order to control the reaction temperature a Eurotherm 3210 controller fitted with 2 Elmatic Max K cartridge was used. All modules in the platform (pumps, temperature controller, NMR and GPC) were connected to a PC and using MATLAB software allowed for alteration of reaction conditions and automated analysis to de-termine conversion, molecular weight and dispersity. The NMR protocol used was the same as for Reactor C. In order to perform GPC analysis a switching valve was installed between the reactor outlet and the NMR allowing for samples to be injected into the GPC for analysis.

Figure 2.14: Reactor D used in chapter 5 to perform automated reaction screening and self-optimisation of RAFT polymerisations.

2.4.1 Online GPC analysis

Most commercially available GPC systems are quite costly and often have proprietary software which can make integration of systems into a separate automated platform complex. Therefore, a custom GPC system was developed for use in the continuous platform. The GPC consists of 5 parts (Figure 2.15): a HPLC pump (Jasco PU-980), a sample injector (6 Port VICI-EHMA), a guard column (Agilent 5 µm), a separation column (Agilent Rapide M) and a detector (Knauer K2301). Often multiple separation columns are used to achieve a higher resolution chromatogram. Although, this comes with an increase in acquisition time which would reduce the “real-time” quality of data being obtained therefore only one separation column was used.

2.4. Reactor D

Figure 2.15: GPC setup developed for use in Reactor D to analyse polymer molecular weight and dispersity

2.4.1.1 Calibrating Injection Volume

Samples were taken from the reactor stream via an automated sampling valve attached to both the reaction and GPC eluent streams. As the reaction stream exited the flow reactor it filled a sample loop attached to the valve (Figure 2.16a). As the valve switched into the GPC eluent stream, reaction solution was pushed out of the sample loop and into the GPC eluent stream (Figure 2.16b).

Figure 2.16: The position of the switching valve in either the a) loading state or b) the injection state. When acquiring a sample for GPC analysis the valve will switch to the loading state for a very short period of time (100 ms).

Therefore, the amount of sample injected into the GPC from the reactor stream was determined by the flow rate of the GPC eluent stream. The volume of material injected into the GPC at various flow rates was then determined using a tracer dye (Sudan III) of known concentration (Figure 2.17).

2.4. Reactor D

Figure 2.17: a) UV-Vis spectra obtained for Sudan III showing a λ_max at 520 nm and b) Calibration curve obtained for Sudan III in DMF at 520 nm used to determine the amount of material injected to online GPC

The sample loop was loaded with a tracer dye and switching was performed with flow rates set at 1, 2 and 3 mL min^-1 and samples were collected and analysed by UV-Vis spectroscopy. By comparing the concentration of dye before and after switching the volume of material injected could was determined to be 1.5, 3 and 5 µl respectively.

2.4.1.2 GPC Protocol for PDMAm Synthesis

For analysis of the RAFT polymerisation of DMAm performed in Chapter 5 DMF GPC was used to acquire chromatograms in 3 minutes. However, continuous use of this setup led to significant peak broadening. Continuous collection of chromatograms for a pre-made PDMAm₁₀₀ (M_n = 10,000 g mol^-1, Đ = 1.17) indicated broadening was worsening as the number of chromatograms collected increased (Figure 2.18a). A subsequent increase in dispersity from 1.5 to over 2.0 was observed as the number of chromatograms obtained increases (Figure 2.18b).

2.4. Reactor D

Figure 2.18: a) GPC chromatograms and b) dispersities obtained continuously over 70 min-utes for PDMAm₁₀₀ on online platform. As more samples are taken (black to red) increased broadening of chromatograms is observed. 5 µL of sample was injected per measurement.

One possible cause of column degradation was material overload. In order to minimise the amount of material injected whilst still clearly detecting polymer signals, the GPC flow rate was reduced to 1.0 mL min^-1for sample injection. No significant broadening over time was observed when using this reduced injection volume (Figure 2.19a). Dispersities for the chromatograms obtained using a reduced injection volume were more consistent with (Đ = 1.20 - 1.25) with small variations caused by baseline noise (Figure 2.19b).

Figure 2.19: a) GPC chromatograms and b) dispersities obtained continuously over 70 minutes for PDMAm₁₀₀on online platform. As more samples are taken (black to red) no chromatogram broadening is observed. 1.5 µL of sample was injected per measurement.

A well-defined PDMAm₂₀₀ standard was then analysed to ascertain whether reducing the injec-tion volume had a similar effect at higher molecular weight. As with the PDMAm₁₀₀ samples the obtained chromatograms (Figure 2.20a) showed little variation over multiple runs. How-ever, there was a significant variation in the measured dispersities (Đ = 1.6 - 2.1) for these chromatograms (Figure 2.20b).

2.4. Reactor D

Figure 2.20: a) GPC chromatograms and b) dispersities obtained continuously over 70 min-utes for PDMAm₂₀₀ on our online platform. As more samples are taken (black to red) no chromatogram broadening is observed. 1.5 µL of sample was injected per measurement.

Whilst reducing the injection volume improved the consistency of chromatograms, the columns had already been significantly altered from consistent use. A series of PMMA standards, used to calculate molecular weight, were compared over time (Figure 2.21). The gap in elution time between 4 kg mol^-1and 265 kg mol^-1had decreased from 40 seconds to 30 seconds. The shorter range of elution times led to a larger change in M_n, and therefore dispersity, with respect to time. The shortening of elution times was likely due to a change in the pore size distribution of the column. Due to the reduced elution time observed in the column it was deemed unsuitable for further use and was replaced.

Figure 2.21: Calibration curves obtained for online GPC system highlighting the shortening elution time between high molecular weight polymers after heavy use.

As well as reducing the injection volume, further steps were also taken to try to reduce column