---
title: 'Drop-by-Drop: Discrete Liquid Manipulation'
url: https://www.emergentmind.com/topics/drop-by-drop
type: topic
---

# Drop-by-Drop: Discrete Liquid Manipulation

Drop-by-drop denotes the production, deposition, transport, merger, and algorithmic manipulation of liquids as discrete droplets rather than as continuous jets or films. In the literature, this mode of operation appears in blood pinch-off, wetting transitions during gentle deposition, dripping from wettable nozzles and grooved condensing plates, sieve-mediated printing, single-droplet sample preparation on digital microfluidic chips, and drop-drop coalescence [1605.06004], [1010.3795], [1102.4762], [2602.17390], [1910.06063], [1908.09618], [2411.12638]. Across these settings, the relevant control parameters are capillary pressure, inertia, viscosity, visco-elasticity, wetting barriers, geometric confinement, and, in digital microfluidics, graph-theoretic constraints on droplet generation and waste.

## 1. Dripping, pinch-off, and the formation of individual droplets

Kar et al. studied blood-drop breakup from a pendent configuration grown from a \(2\ \mathrm{mm}\) tube at a constant flow rate of \(50\ \mu\mathrm{L/min}\), imaged at \(10{,}000\ \mathrm{fps}\). They reported two distinctive breakup modes. In the incessant neck-collapsing mode, observed for samples with higher haematocrit such as \(\mathrm{HCT}=41.0\%\), the minimum neck radius shrinks continuously until pinch-off without forming a long ligament. In the extended-thread breakup mode, observed for \(\mathrm{HCT}=39.5\%\) and also for \(\mathrm{HCT}=32.3\%, 44.8\%, 50.9\%\), a long slender filament forms before breakup and then thins exponentially [1605.06004].

Their rheological description approximates blood as a shear-thinning power-law fluid,
\[
T=\mu (\nabla u)^n,
\]
with \(n\approx 0.70\text{--}0.78\). Near pinch-off, the dominant balance is among inertial stresses, capillary pressure gradients, and viscous or visco-elastic stresses within the neck region. The capillary pressure and its gradient are written as
\[
p_c \sim \frac{\sigma}{d_{\mathrm{neck}}}, \qquad \nabla p_c \sim \frac{\sigma}{d_{\mathrm{neck}}\,z}.
\]
For incessant neck collapse, the neck diameter follows
\[
d_{\mathrm{neck}}(\tau)\sim A\tau^\alpha, \qquad \tau=t_p-t\to 0^+,
\]
with measured exponents \(\alpha=0.7810\) at \(\mathrm{HCT}=34.3\%\), \(\alpha=0.7422\) at \(37.0\%\), \(\alpha=0.7476\) at \(41.0\%\), and \(\alpha=0.7633\) at \(43.0\%\), giving a mean \(\alpha\approx 0.75\pm 0.02\). For extended-thread breakup, the extensional strain-rate relation
\[
\dot{\epsilon}\equiv -\frac{1}{d_{\mathrm{neck}}}\frac{d\,d_{\mathrm{neck}}}{dt}=\frac{1}{3T_r}
\]
implies
\[
d_{\mathrm{neck}}(\tau)=d_0\exp\!\left[-\frac{\tau}{3T_r}\right].
\]
Reported relaxation times range from \(\sim 0.07\ \mathrm{ms}\) to \(\sim 0.8\ \mathrm{ms}\), with \(T_r\approx 0.07\ \mathrm{ms}\) for \(\mathrm{HCT}=32.3\%\), \(0.27\ \mathrm{ms}\) for \(39.5\%\), and \(0.78\ \mathrm{ms}\) for \(44.8\%\).

Chang, Nave, and Jung examined a different dripping pathway: drop formation from a wettable nozzle. Their experiments used stainless-steel syringe needles of gauge 19, 20, and 21, with inner radii \(R=0.406\ \mathrm{mm}, 0.324\ \mathrm{mm}, 0.292\ \mathrm{mm}\), silicone oil with \(\rho=970\ \mathrm{kg/m^3}\), \(\mu=10\ \mathrm{cSt}\), \(\sigma=20\ \mathrm{mN/m}\), and flow rates \(Q=50\ldots 250\ \mathrm{mL/hr}\). They observed that the droplet initially climbs the outer wall due to surface tension and later falls under gravity as its weight increases [1102.4762].

The governing momentum balance is
\[
\frac{d}{dT}\bigl[M(T)V(T)\bigr]
=
2\pi R\sigma\cos\theta_c
-
6\pi \mu R L(T)V(T)
-
M(T)g,
\]
with \(M(T)=\dot{M}T\), \(\dot{M}=\rho Q\), \(V(T)=dz/dT\), and \(L(T)\approx z(T)\). Two asymptotic solutions were identified. In the early capillary-dominated stage,
\[
z(T)\approx
\Bigl(2\pi \sigma \cos\theta_c\Bigr)^{1/2}\frac{T}{\dot{M}^{1/2}},
\]
so the initial climb is linear in time with slope proportional to \(\mathrm{We}^{-1/2}\). In the late gravity-dominated stage,
\[
z(T)\approx z_m-\tfrac12 g(T-T_m)^2.
\]
The experiments reported log-log slopes of approximately \(-1/2\) for both the early \(\log(dz/dT)\) versus \(\log\mathrm{We}\) relation and the late \(\log|dz/dT|\) versus \(\log\mathrm{Fr}\) relation.

Taken together, these studies show that drop-by-drop generation is not governed by a single breakup law. In blood, the relevant distinction is between power-law neck collapse and exponential elasto-capillary thinning; on a wettable nozzle, the controlling competition is among capillary rise, viscous drag, and gravity.

## 2. Deposition onto textured substrates and deceleration-driven wetting transition

Kwon et al. analyzed “gentle” drop deposition on textured hydrophobic substrates and showed that quasi-static release can still trigger a Cassie–Baxter to Wenzel transition. In their experiments, a syringe-pump produced mono-disperse water droplets with typical diameters on the order of \(1\text{--}2\ \mathrm{mm}\), volumes \(\sim 1\text{--}4\ \mu\mathrm{L}\), and masses \(\sim 1\text{--}4\ \mathrm{mg}\). Initial release velocities satisfied \(U_0\lesssim 1\ \mathrm{cm/s}\), but brief perturbations such as stage vibrations or needle recoil produced accelerations and decelerations of order \(10^2\text{--}10^4\ \mathrm{m/s^2}\). High-speed imaging at \(5{,}000\text{--}20{,}000\ \mathrm{fps}\) with \(5\text{--}10\ \mu\mathrm{m/pixel}\) resolution enabled frame-by-frame extraction of \(z(t)\) and \(a(t)=d^2z/dt^2\) [1010.3795].

The substrate consisted of lithographically patterned silicon wafers bearing vertical pillars of height \(h\sim 10\text{--}20\ \mu\mathrm{m}\), diameter \(d_p\sim 5\text{--}10\ \mu\mathrm{m}\), and pitch \(p\sim 20\text{--}30\ \mu\mathrm{m}\), coated with a fluorosilane monolayer. The flat reference surface had \(\theta_Y\sim 115^\circ\); the textured Cassie–Baxter state exhibited static contact angles \(\theta_{CB}\gtrsim 160^\circ\) and roll-off angles below \(5^\circ\).

The key mechanism is a transient water-hammer pressure,
\[
P_{\mathrm{wh}}=\rho c \Delta U,
\]
compared against the anti-wetting capillary pressure
\[
P_{\mathrm{ca}}\simeq \frac{2\gamma |\cos\theta_Y|}{r_c}.
\]
The transition criterion is
\[
\rho c\Delta U \ge \frac{2\gamma |\cos\theta_Y|}{r_c}.
\]
For water, with \(\rho\approx 1000\ \mathrm{kg/m^3}\) and \(c\approx 1480\ \mathrm{m/s}\), and for micrometer-scale posts with \(r_c\sim 2\ \mu\mathrm{m}\), the capillary barrier is on the order of a few kilopascals. Kwon et al. reported that when decelerations produced \(\Delta U\) above \(\sim 0.05\ \mathrm{m/s}\), corresponding to \(a\sim 5\times 10^3\ \mathrm{m/s^2}\) over \(10\ \mu\mathrm{s}\), the resulting \(P_{\mathrm{wh}}\) of about \(7\ \mathrm{kPa}\) crossed the anti-wetting threshold and triggered impalement.

Their quantitative results distinguish two regimes. Gentle quasi-static deposition with \(a_{\max}<10^2\ \mathrm{m/s^2}\) gave \(P_{\mathrm{wh}}<1\ \mathrm{kPa}\ll P_{\mathrm{ca}}\) and no transition. Perturbed settling events with \(a_{\max}\approx 10^3\text{--}10^4\ \mathrm{m/s^2}\) over \(\Delta t\approx 10\text{--}50\ \mu\mathrm{s}\) yielded \(\Delta U\approx 0.01\text{--}0.2\ \mathrm{m/s}\) and water-hammer pressures from \(1\ \mathrm{kPa}\) to \(30\ \mathrm{kPa}\). Onsets of Wenzel transitions occurred when \(P_{\mathrm{wh}}\) crossed approximately \(5\text{--}10\ \mathrm{kPa}\). The high-speed footage further resolved the chronology: first contact at \(t=0\ \mu\mathrm{s}\), a deceleration-induced profile kink at \(t\approx 50\ \mu\mathrm{s}\), local pore filling by \(t\approx 100\text{--}200\ \mu\mathrm{s}\), and relaxation to a final Wenzel footprint by \(t>500\ \mu\mathrm{s}\), with the contact angle collapsing from \(\sim 160^\circ\) to \(\sim 120\text{--}130^\circ\).

A common misconception is that “gentle” deposition implies negligible forcing. These results show the opposite: even modest macroscopic energies can be concentrated into microsecond pressure spikes large enough to overcome the capillary barrier of a superhydrophobic texture.

## 3. Geometry-defined dripping from condensing surfaces

The 2026 study on controlled dripping from a grooved condensing plate asks whether geometry can replace randomness as the governing mechanism of edge dripping. On a smooth vertical surface, condensation produces sweep drops that grow by coalescence, slide downward unpredictably, and strike the lower edge, where hanging droplets form and detach irregularly. On grooved substrates, by contrast, laser-engraved vertical grooves redirect surface flow into groove-guided drainage and produce localized, steady dripping points [2602.17390].

The relevant geometric parameters are groove spacing \(s\), aspect ratio \(d/w\), and groove orientation. When \(s>R_c\), with critical sweep radius \(R_c\approx 1\ \mathrm{mm}\), sweep drops dominate and hanging droplets remain sparse, impact-driven, and positionally unstable. When \(s\lesssim R_c\), groove-guided transport supplants sweeping; hanging droplets become more numerous and narrower as \(s\) decreases, and their positions lock to basin centers. At very tight spacing \(s\ll R_c\), each groove basin collects less water, leading to fewer dripping sites but highly regular spatial and temporal patterns.

Aspect ratio determines the strength of capillary anchoring. Shallow grooves with \(d/w<0.3\) provide little anchoring and give irregular dripping similar to a smooth face. Intermediate grooves with \(0.3\lesssim d/w\lesssim 1\) produce transitional behavior in which flank droplets are pinned but sweep-drop intrusion persists. Deep, narrow grooves with \(d/w>1\) fully confine surface condensate within the channels; flank droplets span many grooves, remain stable against perturbations, and feed hanging drops in a highly periodic fashion.

Orientation changes the spatial organization of drainage. Parallel vertical grooves distribute drainage sites uniformly along the edge and yield quasi-periodic dripping in multiple bands. Convergent grooves, for example with secondary channels tilted at \(45^\circ\), funnel all condensate in a basin to a single outlet and fully localize dripping points at predetermined positions. In these convergent designs, the dripping point is locked to the collector groove axis within roughly one capillary length,
\[
\lambda=\sqrt{\frac{\sigma}{\rho g}}\approx 2.7\ \mathrm{mm}.
\]

The study also provides a simple condensation–capillarity model for the period \(T\) between successive drops. For a drainage basin of width \(b\) and face height \(L\), the effective condensing area is
\[
A=b\left(L-\tfrac{b}{4}\right).
\]
With condensation rate \(c\), accumulated mass is
\[
m_{\mathrm{accum}}=cAT.
\]
A detached pendant drop has characteristic mass
\[
m_{\mathrm{drop}}=\frac{2(\ell+e)\sigma}{g},
\]
where \(\ell\) is the hanging-drop width and \(e\) the plate thickness. Equating accumulated and detached masses up to an empirical factor \(\alpha\approx 0.25\) yields
\[
T=\alpha\,\frac{2(\ell+e)\sigma}{g\,c\,b\,(L-b/4)}.
\]
Hence,
\[
T\propto \frac{1}{A},
\]
or approximately \(T\propto 1/b\) when \(b\ll L\). Agreement with experiments across basin widths \(b=10, 20, 40\ \mathrm{mm}\) supports the interpretation that each convergent groove acts as an independent capillary attractor.

This work shifts the focus of drop-by-drop control from fluid properties alone to drainage-basin architecture. A plausible implication is that, in condensation-driven systems, deterministic droplet release can be engineered passively by selecting the topology of liquid collection upstream of the detachment edge.

## 4. Drop-by-drop printing through impact and recoil

Modak et al. introduced “Drop Impact Printing,” in which a millimetric parent drop of diameter \(D_0\) impacts a superhydrophobic sieve of pore size \(L\) at velocity \(U_0\). A single droplet is not ejected during initial impact because the dynamic pressure \(p_{\mathrm{dyn}}\approx \rho U_0^2\) remains below the breakthrough pressure \(p_{\mathrm{break}}\approx 4\gamma/L\). Instead, ejection occurs during recoil, when the collapse of a central air cavity creates a local pressure spike \(p_{\mathrm{coll}}\approx \rho U_{\mathrm{coll}}^2\gg \rho U_0^2\), forcing a jet through a pore and producing a single, satellite-free droplet [1910.06063].

Two cavity-mediated modes were reported. In the Impact-Cavity mode, the cavity forms during early spreading and recoil of the parent drop, and its collapse drives the jet. In the Recoil-Cavity mode, liquid that briefly penetrates the mesh recoils back past the sieve, forms a second cavity within the drop, and then ejects a droplet upon collapse. The operating window for satellite-free ejection was \(17\lesssim \mathrm{We}\lesssim 25\), with
\[
\mathrm{We}=\frac{\rho U_0^2 D_0}{\gamma}, \qquad
\mathrm{Re}=\frac{\rho U_0 D_0}{\mu}, \qquad
\mathrm{Oh}_L=\frac{\mu}{\sqrt{\rho\gamma L}}, \qquad
Z=\frac{1}{\mathrm{Oh}_L}.
\]

Measured droplet diameter followed the empirical scaling
\[
D_d \simeq 0.88\,L^{1.07},
\]
so droplet volume satisfies \(V_d\sim \text{constant}\times L^3\) to leading order. The fit held across sieves from \(L=533\ \mu\mathrm{m}\) to \(76\ \mu\mathrm{m}\), excluding the largest mesh that exhibited Impact-Penetration behavior. The platform handled surface tension down to \(32\ \mathrm{mN/m}\), viscosity up to \(33\ \mathrm{mPa\cdot s}\), printable \(Z\)-range \(3\lesssim Z\lesssim 200\), suspensions up to \(71\ \mathrm{wt}\%\ \mathrm{ZrO_2}\), and particles up to \(20\ \mu\mathrm{m}\) in diameter dispensed through a \(76\ \mu\mathrm{m}\) pore, with droplet diameters remaining around \(80\ \mu\mathrm{m}\).

The experimental implementation used commercial Cu meshes with pore sizes \(76\text{--}533\ \mu\mathrm{m}\), roughened with Cu nanowires and silanized to contact angle \(\sim 159^\circ\). Parent drops of \(D_0\sim 2.56\ \mathrm{mm}\) were released from \(2\text{--}5\ \mathrm{cm}\), yielding \(U_0\approx 0.7\text{--}0.8\ \mathrm{m/s}\); the substrate was placed about \(1\ \mathrm{mm}\) below the mesh, and a tilt of about \(5^\circ\) reduced residue buildup. Reported ejection-angle deviation was at most \(5^\circ\), corresponding to positional jitter below \(90\ \mu\mathrm{m}\) at \(1\ \mathrm{mm}\) standoff.

The broader significance is not merely the avoidance of nozzle clogging. The study shows that droplet generation can be delegated to a transient hydrodynamic singularity created by cavity collapse, rather than to steady forcing through a nozzle.

## 5. Single-droplet preparation on digital microfluidic chips

In digital microfluidics, drop-by-drop operation is formalized as manipulation of unit-volume droplets with discrete concentrations. The RPRIS paper considers a target consisting of a single droplet with concentration
\[
c=\frac{a}{2^d}, \qquad a\in\{0,1,\dots,2^d\},
\]
where \(d=(c)\) is the precision. A pure reactant droplet is denoted \(1\), a pure buffer droplet \(0\), and waste is any droplet other than the single required target output [1908.09618].

The computational object is a mixing graph: an acyclic directed graph with source nodes emitting \(0\) or \(1\), internal \(1\text{--}1\) micro-mixers of in-degree \(2\) and out-degree \(2\) whose outputs both have concentration \((a+b)/2\), and sink nodes collecting the outputs. The design goal is to minimize the number of waste sinks while producing a single target droplet of concentration \(t\in(0,1)\).

RPRIS, “Recursive Precision Reduction with Initial Shift,” combines two ideas. The Initial Shift maps \(t\) to a value \(t_0\in[1/4,3/4]\) of smaller effective precision \(d_0=d-\gamma+\sigma\), where \(\gamma\) is the number of equal leading bits of \(t\) and \(\sigma\in\{0,1\}\). Recursive Precision Reduction then lowers precision by \(2\) at each step and reconstructs the original level with a converter that adds at most one waste per back-step. The final undoing of the Initial Shift adds at most \(\gamma\) wastes.

The resulting worst-case guarantee is explicit:
\[
\boxed{(d+\gamma)+2}
\]
waste droplets at most for a target of precision \(d\) and leading-bit count \(\gamma\). Construction size, number of mixers, and total droplet operations are all \(O(d^2)\), and the construction time is \(O(d^2)\).

The experimental comparison covered all \(t\in(0,1)\) with precisions \(d\in\{7,8,15,20\}\) against Min-Mix, DMRW, REMIA, GORMA, and ILP. On average, RPRIS used \(50\%\) fewer wastes than Min-Mix, \(40\%\) fewer than REMIA, \(21\text{--}25\%\) fewer than DMRW, and \(17\%\) fewer than GORMA. Relative to the exact ILP method, which times out for \(d>8\), it incurred only about \(7\%\) extra waste on average. For \(d=15\), the reported average wastes were \(15\) for Min-Mix, \(13.5\) for DMRW, \(12.4\) for GORMA, and \(10.2\) for RPRIS.

This strand of research extends the meaning of drop-by-drop beyond hydrodynamics. Here the droplet is a computational and chemical unit, and the central question is not how a neck pinches off but how discrete mixing operations can realize a desired concentration with minimum waste.

## 6. Drop-drop coalescence and crossover dynamics

Xie et al. studied drop-to-drop coalescence in the crossover between viscous and inertial regimes using high-speed imaging up to about \(2.3\times 10^5\ \mathrm{fps}\) at \(1.1\ \mu\mathrm{m/pixel}\). Their variables are the bridge radius \(R(t)\), the undeformed drop radius \(R_0\), the bridge height scale \(\Delta(t)\), and the elapsed time \(\tau=t-t_0\), where \(t_0\) is determined by fitting early data to \(R=\beta (t-t_0)^\alpha\) and extrapolating [2411.12638].

The characteristic scales are
\[
\tau_v=\frac{\mu R_0}{\sigma}, \qquad
\tau_i=\sqrt{\frac{\rho R_0^3}{\sigma}}, \qquad
\mathrm{Oh}=\frac{\mu}{\sqrt{\rho \sigma R_0}}=\frac{\tau_v}{\tau_i},
\]
together with \(R^*=R/R_0\), \(T_v=\tau/\tau_v\), and \(T_i=\tau/\tau_i\). In the viscous-dominated regime, the bridge is V-shaped with \(\Delta\sim R\), and the balance \(\sigma/\Delta\sim \mu (dR/d\tau)/\Delta\) gives
\[
R^*=D_0 T_v, \qquad D_0\approx 1.
\]
In the inertial-dominated regime, the bridge is U-shaped with \(\Delta\sim R^2/R_0\), and the balance \(\sigma/\Delta\sim \rho (dR/d\tau)^2\) gives
\[
R^*=C_0 T_i^{1/2}, \qquad C_0\approx 1.4.
\]
The intermediate regime exhibits power-law growth with exponent \(n\) between \(1/2\) and \(1\), and the local Reynolds number passes through \(O(1)\).

The central result is a one-parameter Padé-type crossover function,
\[
f(P)=\frac{\sqrt{P}}{A+\sqrt{P}},
\]
with \(A\approx 0.71\), so that
\[
R^*=\frac{\sqrt{P}}{A+\sqrt{P}}\,T_v,
\qquad
\sqrt{P}=\frac{\tau_v}{\sqrt{\tau_i\,\tau}}.
\]
An equivalent form is
\[
R^*=\frac{\mathrm{Oh}\,\sqrt{\tau/\tau_i}}
{\mathrm{Oh}+A\sqrt{\tau/\tau_i}}.
\]
This formulation reproduces the viscous and inertial asymptotes and collapses the authors’ data, spanning viscosities from \(1\) to \(1000\ \mathrm{mPa\cdot s}\) and \(\mathrm{Oh}\) from \(0.0037\) to about \(4\), together with previous experimental results, onto a single master curve. The paper also notes that a leading-order logarithmic correction can capture the very earliest behavior with \(R/R_0<0.03\).

In a drop-by-drop context, coalescence is the inverse of pinch-off: instead of one droplet becoming two, two droplets become one through a bridge whose growth law depends on the same capillary, viscous, and inertial competition that governs breakup.

## 7. Cross-cutting interpretation

Several recurrent themes emerge from these studies. First, discrete droplet behavior is often controlled by transient, localized events rather than by slowly varying global conditions. Water-hammer impalement during deposition depends on a microsecond deceleration impulse; sieve printing relies on collapse of a recoil-generated cavity; blood pinch-off is decided by the local neck rheology; and coalescence is set by the near-neck bridge geometry [1010.3795], [1910.06063], [1605.06004], [2411.12638].

Second, geometry is repeatedly used as a control variable. Wettable nozzles alter the trajectory of a forming drop through exterior wetting; grooved condensers divide the surface into drainage basins and fix release locations; pore size on a sieve sets the emitted droplet volume through \(D_d\simeq 0.88L^{1.07}\); and mixing graphs in digital microfluidics determine how many droplets must be created, transported, and discarded [1102.4762], [2602.17390], [1910.06063], [1908.09618].

Third, apparently similar “drop-by-drop” phenomena can belong to different dynamical classes. Blood breakup can follow a power law or an exponential law depending on whether the neck collapses directly or forms an extended thread. Gentle deposition can preserve the Cassie state or trigger Wenzel impalement depending on whether transient pressure exceeds the anti-wetting threshold. Condensate release can be stochastic on smooth faces yet periodic and localized on convergent grooves. This suggests that discrete droplet handling should be classified by its dominant balance and geometric constraints rather than by macroscopic appearance alone.

For research practice, the most consequential implication is methodological. Across these works, high-speed imaging, scaling arguments, and reduced models convert individual droplet events into measurable laws for \(d_{\mathrm{neck}}(t)\), \(z(t)\), \(T\), \(D_d(L)\), waste bounds, or \(R(t)\). In that sense, drop-by-drop is not a single phenomenon but a unifying experimental and theoretical program for treating the droplet as the elementary unit of fluidic behavior, transport, and design.

Source: https://www.emergentmind.com/topics/drop-by-drop