The light intensity at the backside of the culture was detected with a PAR photon
flux density sensor that was connected to an ADAM-5000 data acquisition card. The
light transmitted through the culture (and consequently the turbidity) was evaluated
every minute and compared to the set point, i.e. the amount of light transmitted
24 h after the start of light stress (see above). This set point was controlled via a Lab
View virtual instrument running on a PC (Lab View 7.1, National Instruments), by
automatically switching on or off the feed medium pump. The culture volume was
maintained constant by continuous removal of culture broth via a horizontal outlet
tube connected to a long vertical tube. This vertical tube had one end (bended
upwards) in the culture broth and the top end was situated high above the surface
level. This way a constant level of the aqueous phase could be realised even with an
organic phase on top.
At the same time as the turbidostat was started, 0.5 L organic phase (dodecane,
referred to as Ldod) was added to the reactor to apply a two-phase system. The
sparging with nitrogen gas of the reactor was stopped and the dodecane was sparged
through the reactor to provide mixing and extraction. The dodecane was pumped
from the top layer to the bottom inlet by means of a gear pump (Verdergear, VG-
045.8), viton tubing and stainless steel connectors. The sparging rate of the organic
phase (0.286 Ldod LRV−1 min−1) was calculated based on the volumetric sparging rate
applied by Hejazi et al. [2] of 0.200 Ldod min−1 for a system of 0.7 L of aqueous phase.
As a reference we performed the turbidostat also without the combination
with in situ extraction. The flat-panel photobioreactor was operated as described
by Lamers et al. [6] with minor modifications and set-points for temperature, aeration,
pH as described above. The reactor was inoculated at a lower cell concentration
of 7 × 107 cells L−1. This time both during the cultivation period and the turbidostat
period nitrogen gas was sparged (0.6 L min−1) through the reactor. This was done via
needles at the bottom of the reactor to create an improved flow pattern compared
to the first run. The total volume of the aqueous phase was now 2.5 L (LRV).