Here, we report a unique microfluidic technique that utilizes a membrane filter and plug-in tubes to remove oil and pack water-in-oil droplets for controlled incubation of droplet-based assays. This technique could be modularly incorporated into most droplet-generation devices without a need to alter the original designs. Our results show that removing excess oil to form tightly packed droplets allows for extended and controllable incubation for droplets traveling in microchannels. The efficiency of this technique was evaluated and confirmed using a time-dependent enzyme assay with a fluorometric readout. The system is also readily generalizable to control inter-droplet distance, crucial for studying droplet communication and pattern formation.

Despite the increasing popularity of high-throughput droplet microfluidics in analyzing low-input targets such as single molecules, cells, or microorganisms in isolated pL–nL volumes,1–4 the limited channel length and high flow rates of droplet-generation devices5,6 make it hard to achieve the extended incubation times required for most biological assays. To extend the droplet-dwelling/incubation time, techniques for on-chip7,8 and off-chip9,10 droplet incubation have been developed. Off-chip incubation techniques often require droplets to be manually collected into a microtube and then reinjected into a second microfluidic device; they are thus suited for reactions requiring relatively long incubation time but not sensitive to the time taken for transferring samples and leading to a broad randomization of droplets upon reinjection into the second device. On-chip methods of incubation offer a streamlined and order-preserving method of time delay, but the period is often limited by constraints on device size since time is volume in a continuously flowing microfluidic device to less than 15 min.

Here, we report a method incorporating plug-in tubes on microfluidic devices for efficient oil extraction and time-controlled droplet incubation, which compensates the advantages and disadvantages of the on-chip and off-chip incubation techniques aforementioned. The key component is a thin polytetrafluoroethylene (PTFE) membrane (Fig. 1) embedded underneath a microchannel, allowing fluorinated oil to be filtered and drained through a plug-in tube while leaving aqueous droplets flowing over on the top, forming packed droplets for subsequent incubation and detection. The technique is unique in construction and offers distinct superiorities over other droplet-based oil-removal designs: (i) the plug-in tube can be inserted at any downstream location with no modification to the original device required; (ii) it does not rely on internal channel networks to extract oil8,10,11 or external reservoir to store droplets,12,13 reducing the complexity of fluid manipulation; (iii) the filter membrane separates the main channel and draining tube, preventing droplets from being lost into the oil extraction channel;10,11 and (iv) the incubation time can be controlled using plug-in tubes of different lengths or diameters with no need to modify channel length or flow rates. We apply this technique to dye-labeled droplets and a droplet-based enzyme assay to measure enzymatic activities at variable incubation times from 3 to 20 min. The factors that affect oil-removal efficiency/droplet-packing density and droplet incubation performance were also investigated.

FIG. 1.

Concept demonstration of the oil-removal and droplet-packing device. (a) Schematic showing oil removed through a membrane filter in a droplet-generation device. (b) Top- and (c) side-view of the oil-removal module. (d) An experimental demonstration of oil removal and droplet packing in a microfluidic device. The flow rates of oil and black dye are 2 and 1.5 μl/min; the outlet of the tube is left open (the oil is drained by gravity).

FIG. 1.

Concept demonstration of the oil-removal and droplet-packing device. (a) Schematic showing oil removed through a membrane filter in a droplet-generation device. (b) Top- and (c) side-view of the oil-removal module. (d) An experimental demonstration of oil removal and droplet packing in a microfluidic device. The flow rates of oil and black dye are 2 and 1.5 μl/min; the outlet of the tube is left open (the oil is drained by gravity).

Close modal

The key component to fabricate the oil-removal module is a hydrophobic Fluoropore PTFE membrane filter with a 0.22-μm pore size (FGLP02500, EMD Millipore, Billerica, MA, USA). Microbore PTFE tubings are used for device and fluidic connections (SK-06417-11/21/31, Cole Parmer, Vernon Hills, IL, USA). Other regular materials, in addition to all chemicals and instrument, can be found in the supplementary material.

A thick polydimethylsiloxane (PDMS) slab (3–5 mm) with microfluidic channels (40 μm in depth) was first fabricated via a soft photolithography method and then bonded to a glass slide spin-coated with PDMS. The access ports were punched by 18- and 20-gauge needles (Jensen Global, Inc., Santa Barbara, CA, USA), including the one for the oil draining tube. As illustrated in Figs. 1(a)1(d), a 3 × 3 mm2 piece of a filter membrane was placed flat on top of the draining hole and pushed to the bottom of the channel by a blunt PTFE tube (1.5-cm long), forming a circular membrane that separates the channel from the tube and allows oil to be filtered through the 0.22-μm pores into the tube outlet while keeping aqueous droplets (typically 100 μm in diameter) remained flowing in the channel. The diameter of the filtering chamber was defined by the outer diameter of the inserted tube, which was 1.07 mm, and the depth was made to be the same as the channel height.

The oil-removal device was initially tested by leaving the outlet of the draining tube open. Flipping the device over, the heavier oil (ρ = 1.614g/ml) could be efficiently removed by gravity under the tested flow rate [Fig. 1(d)]. To further investigate the effect of the oil-withdrawal rate on oil-removal efficiency, an additional syringe pump was used to drain the oil (see a representative experiment in Movie S1 of the supplementary material), enabling a quantitative tuning of the droplet-packing density.

The effect of the oil-withdrawal rate on oil-removal efficiency was first tested at a fixed droplet-generation rate (2 and 1 μl/min for oil and water infusion). The oil-withdrawal rate was then increased (indicated as D) from 0.0 to 2.1 μl/min in steps of 0.3 μl/min. When D is low, the droplets with the remaining oil exit from the oil-removal module into a narrower channel, resulting in irregular spaces between unpacked droplets [Figs. 2(a) and 2(b)]. Droplets packing increased as D increased [Fig. 2(c)] until it approached the oil infusion rate (indicated as O), where a fully packed droplet population was generated and observed in the chamber [Fig. 2(d)]. The interval length between droplets was measured before (I1) and after (I2) the oil removing process, and it was found that I2 linearly decreases as the oil-withdrawal rate D increases [I2 = −114.3D + 221.1 with r2 = 0.9858, Fig. 2(e), blue line]. The oil-removal efficiency, defined as E% = (1 − I2/I1) × 100%, was inversely proportional to I2 and increased with the oil-withdrawal rate [Fig. 2(e), black line]. Both experimental measurements of I2 (filled blue squares) and E% (filled black circles) match their respective theoretical values (empty triangles) calculated based on the ratios of the oil withdraw rate D and the infusion rate O, which give: I2 = I1 × (1 − D/O) and E% = D/O × 100%. It should be noted that the oil-removal efficiency of 100% (when the neighboring droplets are closely packed) does not mean that the oil is removed completely. A minimal amount of oil remains outside of the droplets to maintain droplet separation. Further increasing the oil-withdrawal rate causes droplets to deform, break up, and merge when they transition between channels in different dimensions.

FIG. 2.

The effect of the oil-withdrawal rate on oil-removal efficiency at a fixed droplet-generation rate (O = 2.0 and W = 1.0). A series of images at (a) D = 0.0, (b) D = 0.6, (c) D = 1.2, and (d) D = 1.8. O, W, and D represent the flow rate of oil infusion, water infusion, and oil withdrawal, unit: μl/min. (e) The plot of oil-removal efficiency E (black circles) and droplet space I2 (blue squares, averaged from ∼60 drops) vs oil withdrawing rate D. The hollow triangles represent theoretical values.

FIG. 2.

The effect of the oil-withdrawal rate on oil-removal efficiency at a fixed droplet-generation rate (O = 2.0 and W = 1.0). A series of images at (a) D = 0.0, (b) D = 0.6, (c) D = 1.2, and (d) D = 1.8. O, W, and D represent the flow rate of oil infusion, water infusion, and oil withdrawal, unit: μl/min. (e) The plot of oil-removal efficiency E (black circles) and droplet space I2 (blue squares, averaged from ∼60 drops) vs oil withdrawing rate D. The hollow triangles represent theoretical values.

Close modal

The water infusion rate was then altered to change the volume fraction of water to investigate the effect on the oil-removal efficiency under fixed oil infusion and withdrawal rates. When the volume fraction of aqueous and oil phases decreased from 1.0 to 0.25, there was no significant difference in the oil-removal efficiency, but slightly less packed droplets were observed at lower volume fraction (Fig. S1 in the supplementary material). The same trend was discovered upon increasing the flow rates (Fig. S2 in the supplementary material). It follows that lower packing efficiencies may be observed under extreme flow conditions.

These findings indicate that the oil-removal efficiency is largely dictated by the oil-withdrawal rate. Near matching of the oil-withdrawal and infusion rates, more than 95% of the oil can be removed from the droplet train, results in highly packed droplets that can be controllably incubated, as described below.

Droplet incubation in tubing has been reported previously,6,14 but it has not been shown to allow precise incubation on account of loosely packed droplets within the tubing, which can then have differential transport rates within the tubing. To demonstrate improved droplet incubation afforded by droplet packing, a K-junction device15 was used to tag ∼1000 droplets with a black dye by applying an electrical droplet injection potential for 10 s [Fig. 3(a)]. The subsequent plug of packed droplets is then incubated within the tubing [Fig. 3(b)]. Although there is still some droplet diffusion, a large majority of the dyed droplets (>99%) are confined within a 20-s zone [Fig. 3(c)]. This occurs because the droplets are tightly packed, which limits dispersion within the external incubation tubing (Movie S2 in the supplementary material), with the observed dispersion likely due to droplets rearrangement at the intersection of two channels having different dimensions or at curved features within the channels.16 In contrast, droplets that have not been tightly packed via controlled oil removal show a high degree of dispersion, with large observed gaps of dyed droplets within the undyed background droplets (Movie S3 in the supplementary material). This homogenous droplet incubation time is important for steps in which precise timing of in-droplet chemistries needs to be maintained, including application such as reagent injection,17 droplet fusion, or synchronization.18,19 The described oil extraction module is valuable for restoring tight droplet dispersion (Fig. S3 in the supplementary material), making this approach versatile for many different potential applications.

FIG. 3.

(a) Droplets tagged with a black dye using a K-junction device. (b) Stitched picture showing black droplets traveling in a connecting tubing within a confined zone after oil removal. (c) Contrast adjusted image of (b) for grayscale evaluation of the droplet distribution in the tubing. The flow rate of the droplets in the tubing: 1.6 μl/min. The arrow indicates the flow direction. Scale bars: 500 μm.

FIG. 3.

(a) Droplets tagged with a black dye using a K-junction device. (b) Stitched picture showing black droplets traveling in a connecting tubing within a confined zone after oil removal. (c) Contrast adjusted image of (b) for grayscale evaluation of the droplet distribution in the tubing. The flow rate of the droplets in the tubing: 1.6 μl/min. The arrow indicates the flow direction. Scale bars: 500 μm.

Close modal

To demonstrate the flexibility of altering incubation time with packed droplets, we bridged a droplet-packing device [Fig. 4(a) (i)] to a detection device [Fig. 4(a) (iii)] with another piece of external plugged-in tubing. Two aqueous streams containing enzyme and substrate solutions, respectively, were combined into droplets, with subsequent oil removal to form packed droplets (Movie S4 in the supplementary material). The enzyme and substrate-containing droplets were then incubated for varying amounts of time flowing through tubing that bridged the packing and detection devices [Fig. 4(a) (ii)]. Reaction incubation time was varied by using bridging tubing with different diameters and lengths (see Table S1 in the supplementary material) to achieve reactions ranging from 3 to 20 min.

FIG. 4.

The connection of two devices via a bridging tubing for a droplet-based enzymatic assay. (a) Pictures showing packed droplets (i) formed after oil removal, (ii) incubated in a 5-cm long tubing (above the focus plane), and (iii) reinjected and respaced in a droplet detection device for the measurement of enzymatic activities. Flow rate: O = 2.0, W = 1.0, and D = 1.8 μl/min. (b) Enzymatic activities measured in photon counts after a different incubation time: blue, 3 min; red, 6 min; dark green, 10 min; and light green, 20 min. Flow rate: O = 2.0, W1 (substrate) = 0.5, W2 (enzyme) = 0.5, and D = 1.8 μl/min. (c) Enzymatic activity (averaged from ∼100 droplets) vs incubation time.

FIG. 4.

The connection of two devices via a bridging tubing for a droplet-based enzymatic assay. (a) Pictures showing packed droplets (i) formed after oil removal, (ii) incubated in a 5-cm long tubing (above the focus plane), and (iii) reinjected and respaced in a droplet detection device for the measurement of enzymatic activities. Flow rate: O = 2.0, W = 1.0, and D = 1.8 μl/min. (b) Enzymatic activities measured in photon counts after a different incubation time: blue, 3 min; red, 6 min; dark green, 10 min; and light green, 20 min. Flow rate: O = 2.0, W1 (substrate) = 0.5, W2 (enzyme) = 0.5, and D = 1.8 μl/min. (c) Enzymatic activity (averaged from ∼100 droplets) vs incubation time.

Close modal

The enzymatic assay performed to show the homogeneity of reactions within packed droplets even during long incubation times was the conversion of the non-fluorescent substrate resorufin β-d-galactopyranoside to fluorescent resorufin via a reaction with a β-galactosidase enzyme. The devices were assembled on a confocal microscope with 561 nm excitation and a detection bandpass filter with a 40 nm transmission window centered at 593 nm (additional information can be found in the supplementary material). In this way, the fluorescent resorufin generated by the enzymatic conversion of the substrate was detected in individual droplets after respacing them downstream of Device II [Fig. 4(a)]. Figure 4(b) shows four independent trace lines of droplet intensity after droplets were incubated for various times from 3 to 20 min (a full 10-s acquisition was recorded and presented in Fig. S4 of the supplementary material). Each trace represents the fluorescence intensity of droplets after incubation for the specified times. As expected, the average fluorescent intensity increases with increasing reaction time owing to more complete substrate conversion [Fig. 4(c)]. Importantly, the width of the plugs of fluorescent droplets retains their temporal homogeneity even up to the 20-min reaction time. This is due to the tightly packed droplets that travel through the incubation tubing that are not able to spread out during transit. In fact, the relative standard deviations of the height of all of these traces is less than 4.5%, implying that the droplets have kept their generation order in a packed zone while traveling in the incubation tubing. By comparing droplets spanning with and without oil removal shown in Fig. 3 and Movie S3 of the supplementary material, it is expected that a nearly fivefold expansion of the droplets zone without oil being first removed. This expansion could induce a larger variation for droplet traveling/incubation time, which, for a time-sensitive kinetic assay shown in Fig. 4(c), could, in theory, result in a wide distribution of fluorescence intensity. In this way, we demonstrate the value of oil removal and tight droplet packing as a way to achieve highly homogenous droplet reaction times even during long incubations needed for any common enzymatic reactions. Although we only demonstrated incubation up to 20 min for this specific assay, further extension of incubation time can be achieved by stopping the flow after droplets are tightly packed in a tubing until a desired period of incubation, ranging from minutes to a couple of months,20 is reached.

A simple oil-removal module is described that uses plugged-in tubing and a PTFE membrane filter that allows for tight and controllable packing of aqueous droplets. These droplets can then be transported to other devices via external tubing, which can allow for a lengthy droplet incubation period that would be difficult to achieve within a microfluidic device because of the size required and subsequent back-pressure that would accompany such on-chip incubation. Controllable droplet packing is demonstrated by varying the rate of oil removal in a way that is efficient and robust across different droplet-generation parameters. This simple module will find utility in facilitating multistep assays that require transferring droplets between devices or for allowing extended, homogenous droplet incubation by maintaining tight droplet packing. This capability was demonstrated through variable time enzymatic assays that showed maintenance of droplet plug integrity even during long reaction times. This extended, homogenous incubation could also facilitate studies of droplet-to-droplet exchange in which the distance between droplets plays an important role that can be a source of noise.21,22 One such application currently under investigation in our laboratory involves incubation and communication between droplet-based artificial cells.23–25 

See the supplementary material for Chemicals, Materials, and Instrument; incubation time with different tubings (Table S1); effects of volume fraction (Fig. S1) and droplet-generation rate (Fig. S2) on oil-removal efficiency; periodicity restoring from reinjected droplets (Fig. S3) and full traces of droplet detection of the enzyme assay (Fig. S4); and video recordings of droplet packing with (Movies S1, S2, and S4) and without oil removal (Movie S3).

We gratefully acknowledge financial support from the National Institutes of Health (Grant Nos. CA191186 and NIGMS-R35GM119688), the National Science Foundation (Early Career Grant No. 1553031; MCB No. 1817909), and the Alfred P. Sloan Foundation. The authors also wish to thank the Single Molecule Analysis in Real-Time (SMART) Center of the University of Michigan, seeded by NSF MRI-R2-ID Award No. DBI-0959823 to Nils G. Walter, as well as J. Damon Hoff for training and use of Alba v5 confocal spectroscopy.

The data that support the findings of this study are available within the article.

1.
D. K.
Kang
,
M.
Monsur Ali
,
K.
Zhang
,
E. J.
Pone
, and
W.
Zhao
,
TrAC Trends Anal. Chem.
58
,
145
153
(
2014
).
2.
H. N.
Joensson
and
H.
Andersson-Svahn
,
Angew. Chem. Int. Ed.
51
,
12176
12192
(
2012
).
3.
N.
Shembekar
,
C.
Chaipan
,
R.
Utharala
, and
C. A.
Merten
,
Lab Chip
16
,
1314
1331
(
2016
).
4.
Y.
Ding
,
J.
Choo
, and
A. J.
deMello
,
Microfluid. Nanofluid.
21
,
58
(
2017
).
5.
M.
Sun
and
Q.
Fang
,
Lab Chip
10
,
2864
2868
(
2010
).
6.
J. J.
Agresti
,
E.
Antipov
,
A. R.
Abate
,
K.
Ahn
,
A. C.
Rowat
,
J. C.
Baret
,
M.
Marquez
,
A. M.
Klibanov
,
A. D.
Griffiths
, and
D. A.
Weitz
,
Proc. Natl. Acad. Sci. U.S.A.
107
,
4004
4009
(
2010
).
7.
H.
Song
and
R. F.
Ismagilov
,
J. Am. Chem. Soc.
125
,
14613
14619
(
2003
).
8.
L.
Frenz
,
K.
Blank
,
E.
Brouzes
, and
A. D.
Griffiths
,
Lab Chip
9
,
1344
1348
(
2009
).
9.
E.
Brouzes
,
M.
Medkova
,
N.
Savenelli
,
D.
Marran
,
M.
Twardowski
,
J. B.
Hutchison
,
J. M.
Rothberg
,
D. R.
Link
,
N.
Perrimon
, and
M. L.
Samuels
,
Proc. Natl. Acad. Sci. U.S.A.
106
,
14195
14200
(
2009
).
10.
B. E.
Debs
,
R.
Utharala
,
I. V.
Balyasnikova
,
A. D.
Griffiths
, and
C. A.
Merten
,
Proc. Natl. Acad. Sci. U.S.A.
109
,
11570
11575
(
2012
).
11.
J. R.
Haliburton
,
S. C.
Kim
,
I. C.
Clark
,
R. A.
Sperling
,
D. A.
Weitz
, and
A. R.
Abate
,
Biomicrofluidics
11
,
034111
(
2017
).
12.
S. M.
Bjork
,
S. L.
Sjostrom
,
H.
Andersson-Svahn
, and
H. N.
Joensson
,
Biomicrofluidics
9
,
044128
(
2015
).
13.
J.
Dai
,
H.
Soo Kim
,
A. R.
Guzman
,
W. B.
Shim
, and
A.
Han
,
RSC Adv.
6
,
20516
20519
(
2016
).
14.
P. A.
Romero
,
T. M.
Tran
, and
A. R.
Abate
,
Proc. Natl. Acad. Sci. U.S.A.
112
,
7159
7164
(
2015
).
15.
S. R.
Doonan
and
R. C.
Bailey
,
Anal. Chem.
89
,
4091
4099
(
2017
).
16.
E.
Surenjav
,
S.
Herminghaus
,
C.
Priest
, and
R.
Seemann
,
Appl. Phys. Lett.
95
,
154104
(
2009
).
17.
A. R.
Abate
,
T.
Hung
,
P.
Mary
,
J. J.
Agresti
, and
D. A.
Weitz
,
Proc. Natl. Acad. Sci. U.S.A.
107
,
19163
19166
(
2010
).
18.
L.
Mazutis
,
J. C.
Baret
, and
A. D.
Griffiths
,
Lab Chip
9
,
2665
2672
(
2009
).
19.
M.
Lee
,
J. W.
Collins
,
D. M.
Aubrecht
,
R. A.
Sperling
,
L.
Solomon
,
J. W.
Ha
,
G. R.
Yi
,
D. A.
Weitz
, and
V. N.
Manoharan
,
Lab Chip
14
,
509
513
(
2013
).
20.
L.
Li
,
D.
Mustafi
,
Q.
Fu
,
V.
Tereshko
,
D. L.
Chen
,
J. D.
Tice
, and
R. F.
Ismagilov
,
Proc. Natl. Acad. Sci. U.S.A.
103
,
19243
19248
(
2006
).
21.
M.
Weitz
,
A.
Mückl
,
K.
Kapsner
,
R.
Berg
,
A.
Meyer
, and
F. C.
Simmel
,
J. Am. Chem. Soc.
136
,
72
75
(
2014
).
22.
P.
Gruner
,
B.
Riechers
,
B.
Semin
,
J.
Lim
,
A.
Johnston
,
K.
Short
, and
J.-C.
Baret
,
Nat. Commun.
7
,
10392
(
2016
).
23.
Y.
Guan
,
Z.
Li
,
S.
Wang
,
P. M.
Barnes
,
X.
Liu
,
H.
Xu
,
M.
Jin
,
A. P.
Liu
, and
Q.
Yang
,
eLife
7
,
e33549
(
2018
).
24.
Y.
Guan
,
S.
Wang
,
M.
Jin
,
H.
Xu
, and
Q.
Yang
,
J. Vis. Exp.
139
,
e58240
(
2018
).
25.
M.
Sun
,
Z.
Li
,
S.
Wang
,
G.
Maryu
, and
Q.
Yang
,
Anal. Chem.
91
,
9813
9818
(
2019
).

Supplementary Material