3) Estimation of optimum temperature in cell extract preparation & role of power density
With a valid heat transfer model in place, we can next estimate the maximum temperature threshold during cell lysis for optimum yield from the common E. coli BL21 DE3 star strain. This can be estimated using the published data by Kwon et al. (Kwon & Jewett, 2015) where relative CFPS yields were measured from 144 samples of extract prepared by sonication at different volumes (1500, 1000, 700, 500, 300, and 100 µL) and input energy; other parameters such as sonication amplitude, tip size, pulse on duration, and pulse off duration were held constant at 50%, 3 mm, 10 s, and 10 s respectively and the tube was cooled by an ice bath, starting at 277K. From this, a contour map was produced by plotting energy added (x-axis), volume of sample (y-axis) and yield (z-axis) (reproduced from original data in Fig 5a). This plot shows the effect of sample volume and energy added on yield.
To convert this data to temperature, we first needed to determine the power input for each of the sample volumes. We empirically determined the power at each of the measured volumes using the same sonicator (Qsonica 125) set at 50% amplitude (Supplement Fig 6,7), observing that the power input at a given amplitude setting on the sonicator is dependent on the volume of the liquid. This measured input power was then divided by the sample volume to get the volumetric heat source used in the heat transfer model (Qg term, discussed in methods). The pulse time (10 s on and off), cooling condition and all associated geometric and thermal parameters were supplied as model inputs according to the simulation outlined in the methods section. Using the model, the temperature is estimated over time for each volume. The peak temperature in a cycle of on-off pulses is observed at the last second of the on pulse (Fig 3, top of each saw tooth wave). From the temperature vs. time data, we select the temperature at each peak and convert the sonication time data (cumulative of pulse on time only) to energy by multiplying with the effective power value. By this we obtain temperature vs. energy data and interpolate to determine the temperature value for a given input energy value (over the range of 0 to 1600 Joules used by Kwon et al., see Supplement table 16 for example calculation). This data is then used to create a temperature matrix (supplementary table 17,18) and this data is superimposed as contour lines on the same yield plot from Kwon et al. (Fig 4a). Temperature and normalized yield for individual sample volumes are also plotted in Supplementary Figure 8.