1 of 32

Evaporation and the coffee-ring effect for non-circular droplets

Madeleine Moore1 & Alex Wray2

1 University of Hull, 2University of Strathclyde

Contact: m.r.moore@hull.ac.uk, alexander.wray@strath.ac.uk

2 of 32

Introduction�Coffee rings

2

A ‘coffee ring’ is the stain left behind after a droplet of coffee dries on a surface.

3 of 32

Introduction�Taking advantage of the coffee ring…

3

Aiding DNA mapping

Jing et al, PNAS 95:8046-8051 (1998)

Strengthening on the microscale

Chung et al, App. Phys. Lett. 105:261901 (2014)

Colloidal patterning

Choi et al, Langmuir 26(14):11690–11698 (2010)

Printing conductors

Layani et al, ACS Nano. 3(11):3547-3542 (2009)

4 of 32

Introduction�… or trying to fight it

4

Sam et al, Laser & Photonic Reviews 8(1):172-179 (2014)

In applications such as (AM)OLED screen manufacture, large arrays of non-circular droplets evaporate to form the RGB pixels.��

Understanding how the droplets evaporate, and the transport of any solute is a critical engineering challenge.

However, even the case of one irregularly-shaped droplet is non-trivial…

5 of 32

Introduction�Droplet geometry

5

Saenz et al, Nature Comm. 8(1):1-9 (2017)

Droplet geometry is known to play a significant role in both evaporation and solute transport.

In the most common evaporation model – diffusive evaporation (which we will discuss shortly) – the evaporation rate is higher near parts of the contact line with high curvature.

Moreover, the coffee-ring effect is enhanced in these regions.

6 of 32

Introduction�Droplet geometry

6

In fact, one can characterize* the shape of the growing coffee-ring explicitly

*Moore, Oliver & Vella, J. Fluid Mech., 940:A38 (2022)

 

solute mass profile

evaporative flux coefficient

local contact angle

gamma distribution PDF

droplet

 

 

solid

One of the major limiting factors in understanding the coffee ring effect is therefore determining how the droplet evaporates.

7 of 32

Problem configuration

7

Consider a droplet of liquid evaporating from a solid substrate into the surrounding air.

 

 

 

 

 

 

 

 

 

 

 

 

8 of 32

Diffusive evaporation

8

For the purposes of this analysis, we shall concentrate on a diffusive evaporation model.

In this process, evaporation is limited by the transport of vapour away from the liquid-air interface.

Convection effects are ignored and the evaporation is assumed to be slow, so that the whole process is quasi-steady.

 

 

 

 

 

The evaporative flux is then given by

 

 

9 of 32

Diffusive evaporation�Exact solutions

9

 

This can be extended to hemispheres, hemi-ellipsoids and elliptical disks (see, for example, Kellogg (1929)), as well as spherical caps (Popov 2005).

This mixed boundary value problems occurs in a broad range of fields, including:

  • evaporation of droplets;
  • microbubble dissolution;
  • capacitance and charge density in electrostatics;
  • imbibation problems in porous media;
  • ice crystal growth…

However, due to its complexity, very few analytical solutions exist.

10 of 32

Diffusive evaporation�Green’s function formulation

10

Beyond these, it is extremely challenging to make any progress.

Analytically, things become intractable, while numerically, the singularity at the contact line is problematic.

A fruitful avenue of pursuit is to seek approximate or asymptotic solutions.

A sensible approximation is to assume that the droplet is thin, which is relevant in many applications – particularly evaporating droplets with a pinned contact line.

 

 

 

 

where

 

 

 

 

 

11 of 32

Diffusive evaporation�Green’s function formulation

11

surface concentration (specified)

evaporative flux (desired)

 

We approach this by making use of Copson’s identity*

 

 

 

*Copson, Proc. Edin. Math. Soc. 8(1):14-19 (1947)

12 of 32

Diffusive evaporation�Green’s function formulation

12

 

 

 

we may seek an asymptotic solution of the form

 

 

13 of 32

Diffusive evaporation�Leading-order solution

13

 

Abel-type

Abel-type

L-type

We have two types of integrals that we can now invert.

L-type: Here we make use of the identity

 

Abel-type: Standard inversion formulae:

 

 

14 of 32

Diffusive evaporation�Leading-order solution

14

 

We may apply these successively to invert the integral.

 

 

 

Iterating, we eventually return leading order solution,

 

which agrees with the classical Weber solution.

15 of 32

Diffusive evaporation�First order solution

15

 

Proceeding to next order, it is helpful to write the leading-order solution in the asymptotically equivalent form

Then, writing

 

shifts singularity to correct location

ensures smoothness at origin

 

we find that

16 of 32

Diffusive evaporation�Solution

16

Repeating this process, it follows* that the evaporative flux is given by

 

*about 20-odd pages of algebra later…

 

 

 

 

17 of 32

Diffusive evaporation�Solution

17

Before validating, an important point to raise is that the leading-order solution

 

was used as an ansatz by Fabrikant (1986)* in his attempt at approximating solutions to this problem.

While this form has the correct singularity at the contact line and may be used to give good approximations of the total evaporation rate for a given flux, it is deficient in other aspects.

Notably, this form of the solution gives contours of the evaporative flux that are scaled versions of the contact line shape:

 

As we shall now see, this is decidedly far from reality.

*Fabrikant, ZAMP 36(4):616-623 (1987)

18 of 32

Validation

18

We validate our asymptotic predictions by comparing to COMSOL simulations of the full mixed-boundary value problem.

 

Note the change in shape of the contours.

19 of 32

Validation

19

 

 

20 of 32

Polygonal drops

20

We motivated this analysis by considering arrays of square (or sometimes hexagonal) droplets seen in the manufacture of (AM)OLED screens.

As these are clearly far from circular, we might expect our analysis to not be applicable.

However, things go surprisingly well…

Consider a regular n-gon,

 

The goal is to find a Fourier expansion of this shape. However, the sharp corners cause problems.

until the maximal curvature does not exceed a certain tolerance. Typically we chose 10 dimensionless units.

 

 

 

 

21 of 32

Polygonal drops

21

We then express the evaporative flux in the form

 

 

We may then truncate the series appropriately – this is shape dependent and depends on the decay of the coefficients – and compare to COMSOL simulations.

22 of 32

Polygonal drops

22

 

23 of 32

Non-polygonal drops

23

We can even extend this idea beyond regular polygons!

Wray & Moore (2023)

Solomon (1964)

Fabrikant (1986)

Aspect ratio

24 of 32

Application to the coffee-ring effect

24

25 of 32

Mathematical model�Flow model

25

When the droplet is thin, the liquid flow is governed by the lubrication equations

 

 

 

 

for

 

 

and the capillary and Bond numbers are defined by

 

 

 

at

 

 

Appropriate boundary conditions are boundedness at the origin alongside

and a specified initial droplet profile.

26 of 32

Mathematical model�Solute model

26

We shall concentrate on the final deposit residue at the contact line, i.e. the total deposit swept to the contact line.

Since time is separable in this problem, the pathlines coincide with the streamlines.

 

Thus, it is straightforward to show that

 

 

 

 

27 of 32

Large droplets�Example: square droplet

27

a)

b)

c)

d)

 

b) Pressure, streamlines and contours of the evaporative flux from the asymptotic model.

c) Residue density – the ‘coffee ring’ from the asymptotic model.

d) Residue density – the ‘coffee ring’ from the COMSOL results for a smoothed square.

28 of 32

Small droplets�Example: triangular droplet

28

 

Saenz et al, Nature Comm. 8(1):1-9 (2017)

 

 

Flow streamlines calculated from the asymptotic model.

Free surface profile from the asymptotic model.

29 of 32

Summary & the future

29

30 of 32

Summary

30

We have derived a novel asymptotic solution for a classical problem in potential theory.

Our asymptotic model can be employed even in scenarios far from the theoretical limit, as we demonstrated by considering polygonal, rectangular and star-shaped droplets.

We demonstrated a particular application of the results for calculating the deposit residue after a solute-laden droplet evaporates.

Our model produces predictions in very good agreement with both COMSOL simulations and experimental data.

31 of 32

Future directions

31

  • Arrays of non-circular droplets
  • More industrial/experimental comparisons
  • Convection effects
  • What happens when the solute is no longer dilute?
  • Non-thin drops / electrodes

There are a number of different future directions, with several already under way.

32 of 32

32

Thank you for listening!

To find out more detail, see our (open access) JFM paper: