Droplet Characterization

Due to the relatively low flow rate that fluid moves at in microfluidic chips, outputs offloaded from microfluidic chips typically form droplets as opposed to a continuous stream. Therefore, in order to accurately dispense these droplets into wells, we created a model to predict their volume so that users can control the end sample volume in the well of interest. To develop the model we used a free-body diagram to identify the forces present right before a droplet falls off the nozzle. Using the sum of the forces we were able to model the volume of the droplet. To test the model we ran preliminary experiments to calculate volumes and compared them against the theoretical value given by the model. Because we can’t directly measuring the volume of the droplet, we used the time interval between droplets and the given flow rate to calculate what the experimental volume is.

After comparing the experimental droplet volume to the theoretical droplet volume value at flow rates of 1 mL/hr, 2.5 mL/hr, 5 mL/hr, 7.5 mL/hr we learned two things:

• the theoretical volume from the model was less than the experimental
• this difference, on average, increased with flow rate

The droplet volumes was expected to be consistent across all flow rates and only change depending on tubing geometry and material because as flow rates increase, the interval between droplets decrease as the volume of the droplets should remain constant. We concluded that the model was missing another physical component that would both contribute to the theoretical droplet volume and account for the dependence on flow rate.

order to determine the correction factor we ran a series of experiments to calculate experimental droplet sizes at 15 different flow rates. For statistical significance we measured 30 droplet volumes per flow rate by measuring the time interval between droplets. Using a MATLAB script, the droplet volumes were used to calculate the correction factor for each flow rate. The average correction factor was then calculated to see how they differed between flow rates. The plots blow display the results of our modeling experiments.

Figure 1. Plot of the average experimental droplet volume versus the predicted volume from the static model.

Figure 2. Plot of the correction factors calculate based on the experimental volumes at various flow rates against the average correction factor.

Unfortunately, these experiments did not result in a consistent correction factor that can be used across different flow rates. While we observed an increase in droplet size as flow rate increased, we concluded that either more data was necessary to extrapolate an experimental correction factor or more However, we learned other important information to adjust the current static model. Outside of the 2.5 mL/hr trial, the experimental droplet volumes have low standard deviates and are centered around the average volume size. Because of this we decided to multiply the static model by a factor of 1.33 to account for the difference between average and static model.

References:

1. Zhang, D. F., and H. A. Stone. “Drop Formation in Viscous Flows at a Vertical Capillary Tube.” Physics of Fluids 9, no. 8 (August 1997): 2234–42. https://doi.org/10.1063/1.869346.