# Difference between revisions of "Geometrically Mediated Breakup of Drops in Microfluidic Devices"

(→Summary) |
|||

(84 intermediate revisions by 2 users not shown) | |||

Line 1: | Line 1: | ||

+ | Original entry: Zach Wissner-Gross, APPHY 226, Spring 2009 | ||

+ | == Overview == | ||

+ | Link et al. describe two techniques for breaking up droplets in an emulsion using microfluidic devices. The first method makes use of a "T-junction," in which a droplet flows down a channel that abruptly bifurcates into two channels in an orthogonal direction. The second method involves placing an obstacle in the droplet's path. Also notable is the article's analysis of the conditions under which a droplet encountering a T-junction will actually break up. | ||

− | == | + | == Using T-junctions == |

− | + | The process of creating droplets in the first place was previously described by Steve Quake and coworkers [http://prola.aps.org.ezp-prod1.hul.harvard.edu/abstract/PRL/v86/i18/p4163_1], who mixed two immiscible liquids in a similar T-junction microfluidic circuit, made from PDMS using soft lithography [http://en.wikipedia.org/wiki/Soft_lithography]. Link and coworkers here used oil (hexadecane) and water to make the emulsion, including a surfactant to lower the surface tension between the two liquids in order to stabilize droplets. | |

− | == | + | Water droplets in the oil emulsion were flowed toward the T-junction, where they encountered a stagnation point at the intersection (Figure 1). Under certain conditions, the droplets would simply flow into one of the two side channels -- the authors mention that unbroken droplets in fact precisely alternate which channel they choose, although this phenomenon appears non-trivial to me. Perhaps when a given droplet selects one channel, it increases the fluidic resistance of that channel so that the next droplet favors the other channel. But more interestingly, under different conditions, oil flowing behind the droplet will pinch closed the surface, thus breaking the drop into two smaller droplets. |

+ | |||

+ | |||

+ | Pratomo's comment: That the droplets flow into one of the two side channels alternatively is not | ||

+ | explained the the paper. My guess is similar to Zach's, that a higher number of droplets in a channel | ||

+ | causes a higher head pressure in that particular channel. This phenomena is evident again in the | ||

+ | slight discrepancy between the relative sidearm lengths and relative volumes of the droplets on the | ||

+ | second page. | ||

+ | |||

+ | |||

+ | [[Image:Wiki_figure1.png|frame|Figure 1: The top set of images (a-e) shows a droplet that will remain intact, while the bottom set (f-j) shows a droplet that will ultimately break up.]] | ||

+ | |||

+ | == When breakup occurs == | ||

+ | The real physics of this paper occurs when the authors attempt to discern the conditions under which water droplets break up at the T-junction. Breakup takes place due to Rayleigh-Plateau instability [http://en.wikipedia.org/wiki/Plateau-Rayleigh_instability], which predicts the breakup of a cylindrical column of liquid when its length exceeds its circumference. The authors define this geometric ratio as the extension <math>\epsilon</math> of a droplet: | ||

+ | |||

+ | <center><math>\epsilon=\frac{l}{\pi w}</math></center> | ||

+ | |||

+ | where <math>l</math> and <math>w</math> are the length and diameter of the drop, respectively. Thus, when <math>\epsilon>1</math>, a drop will typically bifurcate. The authors state that a water drop with initial extension <math>\epsilon_0>1</math> (defined by the initial length <math>l_0</math> and width <math>w_0</math>) moving toward the T-junction will not break up due to the stabilizing effects of the surrounding walls -- but such stability disappears when the drop reaches the stagnation point at the T-junction. | ||

+ | |||

+ | Due to the deformation that occurs at the T-junction, drops that would originally have been stable can deform to the point that their extension reaches a maximum value of <math>\epsilon_{\text{e}}>1</math> (with length <math>l_{\text{e}}</math> and width <math>w_{\text{e}}</math>). The authors then derive whether a water drop will break up as a function of <math>\epsilon_0</math> and the properties of the emulsion and fluid flow as follows: | ||

+ | |||

+ | First they define the capillary number <math>C</math> of the flow as the dimensionless ratio of the viscous stress to the interfacial surface tension <math>\sigma</math> in the emulsion: | ||

+ | |||

+ | <center><math>C=\frac{\eta v}{\sigma}</math></center> | ||

+ | |||

+ | where <math>\eta</math> is the viscosity of the surrounding medium (oil in this case) and <math>v</math> is the velocity of the flow toward the T-junction. Citing the previous work of Yiftah Navot [http://scitation.aip.org.ezp-prod1.hul.harvard.edu/getabs/servlet/GetabsServlet?prog=normal&id=PHFLE6000011000005000990000001&idtype=cvips&gifs=yes], the authors state that | ||

+ | |||

+ | <center><math>C\propto (l_{\text{e}}-l_0)^2</math></center> | ||

+ | |||

+ | Now they wish to find the critical capillary number <math>C_{\text{cr}}</math> at which the extended droplet becomes unstable, i.e., when | ||

+ | |||

+ | <center><math>\epsilon_{\text{e}}=\frac{l_{\text{e}}}{\pi w_{\text{e}}}=1</math></center> | ||

+ | |||

+ | with the additional constraint of conservation of volume | ||

+ | |||

+ | <center><math>l_0 w_0^2=l_{\text{e}} w_{\text{e}}^2 </math>.</center> | ||

+ | |||

+ | We can combine the last two relations to see that | ||

+ | |||

+ | <center><math>l_{\text{e}}=(\pi^2 l_0 w_0^2)^{1/3}=l_0 \epsilon_0^{-2/3}</math></center> | ||

+ | |||

+ | and hence that | ||

+ | |||

+ | <center><math>C_{\text{cr}} \propto l_0(\epsilon_0^{-2/3}-1)^2 \propto \epsilon_0(\epsilon_0^{-2/3}-1)^2</math>.</center> | ||

+ | |||

+ | Finally, the authors show a rather remarkable plot of <math>\epsilon_0</math> versus <math>C</math> that they created by using different droplet sizes and different velocities (Figure 2). The transition between drops that break and those that do not behaves very much as the above theory predicts (with a constant of proportionality very close to unity!). | ||

+ | |||

+ | |||

+ | Pratomo's comment: I thought it is worthwhile to mention that the author also showed a series of T | ||

+ | junctions by which drops (slugs) are broken up eight times. | ||

+ | |||

+ | |||

+ | [[Image:Wiki_figure2.png|frame|Figure 2: Droplet breakup for various values of initial extension and capillary number.]] | ||

+ | |||

+ | == Using isolated obstacles == | ||

+ | Toward the end of the article, the authors show off some neat droplet bifurcation they can induce using isolated obstacles in the middle of their channels (Figure 3). | ||

+ | [[Image:Wiki_figure3.png|frame|Figure 3: Droplet bifurcation using obstacles in the microchannels.]] | ||

+ | The main application the authors suggest involves reliably generating small homogeneous drops (i.e., drops with a low polydispersity [http://en.wikipedia.org/wiki/Polydispersity]) with no moving parts. However, what I find to be a more exciting possibility, especially after seeing the incredibly repeatable emulsion patterns of Figure 3, is using these structures as solid building blocks via photopolymerization [http://www.nature.com.ezp-prod1.hul.harvard.edu/nmat/journal/v7/n7/full/nmat2208.html]. While UV masks are currently used to generate complex geometries, there might be a use for precise yet complex droplet formation via T-junctions and/or isolated obstacles prior to photopolymerization. | ||

+ | |||

+ | |||

+ | Pratomo's comment: This is a promising technique where the limits of droplet sizes in droplet | ||

+ | generation techniques (flow focusing, T-junction, etc) are extendable by the using either of the two | ||

+ | techniques mentioned above. Since the breakup utilizing T-junctions is highly monodisperse, its | ||

+ | products are useful in a variety of applications such as ultrasound contrast agent, in which highly | ||

+ | monodisperse agents are necessary. |

## Latest revision as of 02:33, 24 August 2009

Original entry: Zach Wissner-Gross, APPHY 226, Spring 2009

## Overview

Link et al. describe two techniques for breaking up droplets in an emulsion using microfluidic devices. The first method makes use of a "T-junction," in which a droplet flows down a channel that abruptly bifurcates into two channels in an orthogonal direction. The second method involves placing an obstacle in the droplet's path. Also notable is the article's analysis of the conditions under which a droplet encountering a T-junction will actually break up.

## Using T-junctions

The process of creating droplets in the first place was previously described by Steve Quake and coworkers [1], who mixed two immiscible liquids in a similar T-junction microfluidic circuit, made from PDMS using soft lithography [2]. Link and coworkers here used oil (hexadecane) and water to make the emulsion, including a surfactant to lower the surface tension between the two liquids in order to stabilize droplets.

Water droplets in the oil emulsion were flowed toward the T-junction, where they encountered a stagnation point at the intersection (Figure 1). Under certain conditions, the droplets would simply flow into one of the two side channels -- the authors mention that unbroken droplets in fact precisely alternate which channel they choose, although this phenomenon appears non-trivial to me. Perhaps when a given droplet selects one channel, it increases the fluidic resistance of that channel so that the next droplet favors the other channel. But more interestingly, under different conditions, oil flowing behind the droplet will pinch closed the surface, thus breaking the drop into two smaller droplets.

Pratomo's comment: That the droplets flow into one of the two side channels alternatively is not explained the the paper. My guess is similar to Zach's, that a higher number of droplets in a channel causes a higher head pressure in that particular channel. This phenomena is evident again in the slight discrepancy between the relative sidearm lengths and relative volumes of the droplets on the second page.

## When breakup occurs

The real physics of this paper occurs when the authors attempt to discern the conditions under which water droplets break up at the T-junction. Breakup takes place due to Rayleigh-Plateau instability [3], which predicts the breakup of a cylindrical column of liquid when its length exceeds its circumference. The authors define this geometric ratio as the extension <math>\epsilon</math> of a droplet:

where <math>l</math> and <math>w</math> are the length and diameter of the drop, respectively. Thus, when <math>\epsilon>1</math>, a drop will typically bifurcate. The authors state that a water drop with initial extension <math>\epsilon_0>1</math> (defined by the initial length <math>l_0</math> and width <math>w_0</math>) moving toward the T-junction will not break up due to the stabilizing effects of the surrounding walls -- but such stability disappears when the drop reaches the stagnation point at the T-junction.

Due to the deformation that occurs at the T-junction, drops that would originally have been stable can deform to the point that their extension reaches a maximum value of <math>\epsilon_{\text{e}}>1</math> (with length <math>l_{\text{e}}</math> and width <math>w_{\text{e}}</math>). The authors then derive whether a water drop will break up as a function of <math>\epsilon_0</math> and the properties of the emulsion and fluid flow as follows:

First they define the capillary number <math>C</math> of the flow as the dimensionless ratio of the viscous stress to the interfacial surface tension <math>\sigma</math> in the emulsion:

where <math>\eta</math> is the viscosity of the surrounding medium (oil in this case) and <math>v</math> is the velocity of the flow toward the T-junction. Citing the previous work of Yiftah Navot [4], the authors state that

Now they wish to find the critical capillary number <math>C_{\text{cr}}</math> at which the extended droplet becomes unstable, i.e., when

with the additional constraint of conservation of volume

We can combine the last two relations to see that

and hence that

Finally, the authors show a rather remarkable plot of <math>\epsilon_0</math> versus <math>C</math> that they created by using different droplet sizes and different velocities (Figure 2). The transition between drops that break and those that do not behaves very much as the above theory predicts (with a constant of proportionality very close to unity!).

Pratomo's comment: I thought it is worthwhile to mention that the author also showed a series of T junctions by which drops (slugs) are broken up eight times.

## Using isolated obstacles

Toward the end of the article, the authors show off some neat droplet bifurcation they can induce using isolated obstacles in the middle of their channels (Figure 3).

The main application the authors suggest involves reliably generating small homogeneous drops (i.e., drops with a low polydispersity [5]) with no moving parts. However, what I find to be a more exciting possibility, especially after seeing the incredibly repeatable emulsion patterns of Figure 3, is using these structures as solid building blocks via photopolymerization [6]. While UV masks are currently used to generate complex geometries, there might be a use for precise yet complex droplet formation via T-junctions and/or isolated obstacles prior to photopolymerization.

Pratomo's comment: This is a promising technique where the limits of droplet sizes in droplet generation techniques (flow focusing, T-junction, etc) are extendable by the using either of the two techniques mentioned above. Since the breakup utilizing T-junctions is highly monodisperse, its products are useful in a variety of applications such as ultrasound contrast agent, in which highly monodisperse agents are necessary.