Page 16 - manuscript_ijb05583

A 1:1 ratio yielded a sample stream width of 400 µm regardless of the fluid pressure,

demonstrated at G) 30 mbar H) 40 mbar I) 50 mbar, scale bar = 400 µm. J) Quantification of
stream widths at varying flow rate ratios. K) Traveling trajectory of BBs tracked via

TrackMate, microscaffold in flow circled in red, ROI in yellow, trajectory in blue. L)
Travelling speeds of BBs quantified at 20 mbar, 40 mbar and 60 mbar.

Optical detection system
A detection system was built to identify the single buckyballs (BBs) in flow-based on the

fluorescent properties of the photoinitiator 4,4′-Bis(diethylamino)benzophenone in the ZrHyb

microscaffolds. The optical detection system for in-line detection of buckyballs in flow was
built as shown in Figure 5a, overall following the fundamental optical setup of a widefield

fluorescence microscope. The fluorescent emission spectrum of the microscaffolds produced
produced from ZrHyb was first determined to optimize the wavelength used in the optical

detection system. The absorption maximum of the photoinitiator 4,4′-
21
Bis(diethylamino)benzophenone was previously reported to be at 360 nm . At an excitation
of 360 nm, BBs in 1-propanol showed an emission maximum at 530 nm (Figure 5b). Therefore,

an LED with a wavelength of 370 - 380 nm was used to excite passing microscaffolds. Passing
microscaffolds emitted light at a wavelength of 530 nm as a response to the excitation light,

which was sensed by the photodiode and registered in the main control program as a short-term
increase in voltage. This change in voltage was then used to differentiate between the empty

channel with carrier liquid (background signal) and an intact microscaffold (peak signal).

Signals in between the two values were further interpreted as scaffolds with structural defects
or debris, again depending on the intensity of the signal. Thresholds for these signals were

determined experimentally and summarized in Figure 5c. First, the background signal by the
LED was determined to be around 262 ± 1 mV. With the sorting device and 1-propanol added,

the background signal slightly increased to 292.33 ± 1 mV. A batch of intact microscaffolds

resulted in a mean detection value of 749 ± 59 mV. In comparison, a batch of half-buckyballs
was printed to simulate manufacturing defects, which showed a mean detection value of 422 ±

24 mV (halves, Figure 5).
Following these results, 0.70 V was used as a threshold value for the sorting algorithm to

identify a passing scaffold as intact. Further, values above 0.4 V were used as thresholds to

identify “debris” in the fluidic channel. Values below 0.4 V were ignored as background signal.
These values were used as thresholds for sorting buckyballs in all subsequent experiments.

15