RESEARCH ARTICLE

Laser induced phosphorescence imaging for the investigationof evaporating liquid flows

Alexandros Charogiannis • Frank Beyrau

Received: 31 October 2012 / Revised: 3 April 2013 / Accepted: 8 April 2013 / Published online: 14 May 2013

� Springer-Verlag Berlin Heidelberg 2013

Abstract The phosphorescence properties of liquid and

gaseous acetone, following excitation at 308 nm, are stud-

ied and utilized in order to overcome two main challenges

of two-phase flow laser induced fluorescence imaging: the

large fluorescence intensity disparity between the two

phases and the ensuing effect of halation. This is achieved

on account of the different phosphorescence decay rates of

the liquid and vapour phases, which allow for a more

favourable signal ratio to be obtained. The benefits of

visualizing the phosphorescence emission, instead of the

fluorescence, are demonstrated by droplet stream experi-

ments set up in different bath gases, at 1 atm and 297 K.

The liquid–vapour interface can be accurately located,

while the vapour surrounding the droplets is clearly visu-

alized without any halation interference. The vapour phase

phosphorescence signal was calibrated in order to quantify

the vapour concentration around an evaporating droplet

stream, and the results are compared to laser induced fluo-

rescence images collected in the present study and results

found in the literature. The effect of halation in the fluo-

rescence images is shown to extend as far as 10 droplet

diameters away from the interface for 161 lm droplets,

resulting in a significant overprediction of acetone vapour

mole fractions in that region. The vapour profile obtained by

laser induced phosphorescence (LIP) imaging agrees with

data found in the literature, for which a halation correction

on fluorescence images was successfully performed. The

demonstrated LIP technique for simultaneous vapour and

liquid phase visualisation is only applicable to oxygen-free

environments, as even trace quantities of oxygen com-

pletely quench the vapour phase phosphorescence emission.

Abbreviations

Dd Average droplet diameter (lm)

E Excitation energy (mJ)

E0 Maximum excitation energy (mJ)

na Acetone number density (molecules cm-3)

na0 Saturated acetone number density (molecules cm-3)

S Fluorescence or phosphorescence signal (counts)

Sd Average inter-droplet separation (lm)

S0 Prompt fluorescence signal (counts)

t Laser sheet thickness (lm)

xa Acetone mole fraction

Xd Distance from droplet centre (lm)

Dt Delay relative to the start of fluorescence emission

(ns)

1 Introduction

Laser diagnostics are well-established powerful tools for

experimental investigations of reactive and non-reactive

flows. Laser induced fluorescence (LIF) is the most com-

monly used technique for planar measurements of con-

centrations, such as the amount of exhaust gas or of the

evaporated fuel in internal combustion engines (Loffler

et al. 2010; Williams et al. 2010). For fuel vaporization

studies, the excited species can be either the evaporating

liquid itself, or a co-evaporating tracer, for example, ace-

tone or 3-pentanone, which is added to a non-fluorescing

parent fuel (Thurber and Hanson 2001; Rothamer et al.

This article is part of the Topical Collection on Application of Laser

Techniques to Fluid Mechanics 2012.

A. Charogiannis � F. Beyrau (&)

Mechanical Engineering Department, Imperial College London,

Exhibition Road, London SW7 2AZ, UK

e-mail: f.beyrau@imperial.ac.uk

123

Exp Fluids (2013) 54:1518

DOI 10.1007/s00348-013-1518-2

2009). Acetone, in particular, has been widely employed

due to its low toxicity, broadband absorption in the UV and

high vapour pressure, which allows for high seeding con-

centrations already at ambient temperatures. In two-phase

flows, such as evaporating sprays, LIF can be used to

visualize both vapour and liquid phases. Acetone is often

employed in evaporating droplet stream or liquid jet

experiments as its physical and spectral properties facilitate

such studies (Bazile and Stepowski 1995; Connon et al.

1997).

Despite the fact that acetone fluorescence dependencies

on flow properties such as temperature, pressure and

composition are very well investigated (Thurber and

Hanson 1999; Thurber et al. 1998; Braeuer et al. 2006), few

data are available regarding the triplet state phosphores-

cence emission. At this point, it is helpful to introduce the

processes involved in the excitation–deexcitation mecha-

nism of acetone (Schulz and Sick 2005), most clearly

described by means of a Jablonski diagram (Fig. 1). The

absorption of a photon leads to electronic excitation from

the ground electronic state S0 to the first excited singlet

state S1, attributed to the symmetry-forbidden p� n

transition. Following excitation to S1, the first excited

triplet state T1 is readily populated by inter-system cross-

ing. This process describes the transfer of population

between states of different multiplicity and in acetone

occurs very rapidly and with almost 100 % efficiency.

Alternatively, the excited molecule can emit fluorescence

(radiative transition to a state of the same multiplicity) and

return to S0, vibrationally relax to a lower vibrational level,

or decay non-radiatively by internal conversion which is,

however, rather inefficient for the particular tracer. From

T1, the molecule can vibrationally relax, decay radiatively

to S0 by emitting phosphorescence (radiative transition to a

state of different multiplicity), give up its energy to another

ground state molecule via collisionally induced quenching

(mainly self-quenching or oxygen quenching) or give up its

energy to another excited triplet state molecule via a pro-

cess called triplet–triplet annihilation. In the latter case, a

triplet–triplet interaction is described according to which

one molecule will decay non-radiatively to S0, whilst its

excitation energy will be consumed by the other. This

process has previously been accounted for substantially

reducing the phosphorescence lifetime of high concentra-

tion biacetyl vapour (Badcock et al. 1972). Through this

process, the observed phosphorescence quantum yield

depends not only on the collision frequency and efficiency

of collision events with other species (such as oxygen) or

non-excited acetone molecules (through self-quenching),

but also on the excited state population. In that way, the

efficiency of triplet–triplet annihilation, and hence the

phosphorescence quantum yield, will depend on the exci-

tation energy, number density and time available for such

collision events to occur.

For an excitation wavelength of 313 nm at near ambient

conditions, the fluorescence quantum yield of acetone is of

the order of 10-3 (Halpern and Ware 1971; Heicklen 1959;

Koch et al. 2004), with almost all the excited state popu-

lation being transferred to the triplet state T1. The fluo-

rescence lifetime is very short [around 2.4 ns (Halpern and

Ware 1971)] and at near-atmospheric pressures, relatively

unaffected by quenching (Thurber and Hanson 1999).

Despite the fact that these values have been obtained from

acetone vapour experiments, they are equally applicable in

describing the fluorescence lifetime and quantum yield of

liquid acetone as well, as the fluorescence emission from

both phases is limited by extremely efficient inter-system

crossing. In contrast, the phosphorescence emission from

T1 to S0 is slow, due to violation of the spin selection rule,

and strongly subjected to oxygen quenching. In a colli-

sionless environment, the phosphorescence lifetime of

acetone vapour is 200 ls (Groh et al. 1953; Kaskan and

Duncan 1950), whilst at ambient pressure, molecular col-

lisions reduce the lifetime to just a few microseconds (Hu

and Koochesfahani 2002). Virtually, no phosphorescence is

observed in the presence of even minute quantities of

oxygen (Hu and Koochesfahani 2002; Weckenmann et al.

2011).

The phosphorescence properties of liquid acetone have

been assessed by Seitzman and co-workers (Ritchie and

Seitzman 2004; Tran et al. 2005). They reported that

removing any oxygen dissolved in liquid acetone can

greatly increase the phosphorescence lifetime; for example,

following excitation at 266 nm, an increase from 175 ns to

nearly 1 ls was observed. Slightly higher values were

presented for 285 nm excitation, suggesting that the initial

population of lower vibrational levels correlates with

Fig. 1 Jablonski diagram illustrating the radiative and non-radiative

processes of acetone following light absorption. Radiative processes

are shown as straight lines and non-radiative processes as wavy lines

Page 2 of 15 Exp Fluids (2013) 54:1518

123

higher phosphorescence efficiency. In addition, longer

detection wavelengths were shown to display longer life-

times, suggesting that the phosphorescence spectrum shifts

to the red with increasing delay from the initial population

of T1. It should finally be noted that, in addition to oxygen

quenching from any dissolved oxygen, the phosphores-

cence lifetime of liquid acetone is substantially shorter

compared to that of the vapour due to more vigorous self-

quenching (Tran et al. 2006). Despite the fact that this type

of interaction is not efficient (Lozano 1992), the number of

collisions occurring in the liquid is sufficiently large to

render self-quenching a strong deexcitation channel.

As a consequence of the limited availability of acetone

phosphorescence emission data in the literature, the first

aim of this study is to characterize and compare the

emission decays of vapour and liquid acetone, excited at

308 nm, in air and nitrogen bath gases. These measure-

ments reveal the potential for better visualizing acetone

vapour concentrations in the presence of liquid droplets by

addressing two major shortcomings associated with two-

phase flow acetone LIF; the extreme fluorescence intensity

disparity between the two phases and the ensuing intense

halation that obstructs the location of the liquid–vapour

interface and introduces a significant error in the vapour

phase measurement in the vicinity of droplets. Therefore,

visualization experiments of evaporative and non-evapo-

rative acetone droplet streams and their surrounding vapour

fields, by both laser induced fluorescence and laser induced

phosphorescence (LIP), will be carried out in order to

demonstrate the advantages of monitoring the latter. The

final objective of this study is the quantification of the

vapour field around an evaporating droplet stream, by

means of both optical techniques, in an attempt to quanti-

tatively address the suitability of LIP for two-phase flow

imaging.

2 Simultaneous vapour and liquid phase acetone LIF

A major challenge in visualizing the vapour field near

droplets by acetone LIF is that the liquid phase fluores-

cence is more than two orders of magnitude stronger than

the gas phase (Ritchie and Seitzman 2001), mainly as a

consequence of the large number density disparity between

the two phases. Since the commonly utilized intensified

CCD cameras have a limited dynamic range, the experi-

mental parameters are typically adjusted so that the intense

liquid phase signal does not quite saturate the detector.

This, in turn, results to a vapour phase signal which is close

to the noise level of the camera. An additional problem in

discriminating between the two phases is intense halation.

The very strong signal from the liquid introduces a cross-

talk with the neighbouring pixels (which contain the weak

vapour phase signal) and results in an apparent increase in

the spatial extent of the liquid or in an overprediction of the

vapour phase concentration near the interface (Orain et al.

2005). It has been reported, for example, that reliable

measurement of vapour concentrations is not feasible to

within 1 droplet diameter from the droplet surface for 235

and 122 lm droplets (Sahu 2011). Other studies report that

for 230 lm droplets, the extent of halation is of the order of

1–1.5 droplet diameters (Orain et al. 2005). Since halation

is a detector-induced artefact, a non-intensified camera or

EMCCD sensitive to the spectral range of acetone fluo-

rescence would suffer less from the particular effect.

However, intensified cameras are commonly employed in

studies such as the ones cited above, because of their short

exposure times and their superior performance at low light

levels (of the vapour phase signal). In addition, non-

intensified cameras suffer from substantially higher jitter.

In tackling halation, the practice of establishing a cut-off

signal to distinguish between the vapour phase and liquid

phase plus halation signals has been introduced (Bazile and

Stepowski 1995). Other researchers have adopted this

method (Hardalupas et al. 2010; Ammigan and Clack

2009), most commonly by calibrating the saturated acetone

vapour signal independently. The location in the vicinity of

a droplet where the collected signal drops to the saturated

vapour fluorescence level is, however, shifted away from

the droplet surface due to halation. Consequently, the

spatial extent of the droplet is overestimated. Alternatively,

the use of Mie scattering images from ethanol droplets in

order to obtain the radial profile of halation, and thus the

location of the interface, has also been proposed and

practiced (Orain et al. 2005). Frackowiak et al. (2009) have

adopted an entirely different approach in locating the

liquid–vapour interface. They have shown that by masking

the liquid phase fluorescence, an uncontaminated vapour

phase signal up to a certain distance away from the liquid–

vapour interface can be obtained. The fluorescence emis-

sion from the droplet and its subsequent collection by the

detector is then modelled by means of a numerical code,

based on the Lorenz–Mie theory and geometric optics.

Knowledge of the droplet size and acquisition of both

masked and unmasked images of the droplet stream allow

for extrapolation of the vapour field in the proximity of the

droplets, as well as accurate location of the interface. This

method, however, presents some severe drawbacks. Apart

from the inherent complexity of its setup (which necessi-

tates the utilization of a mask) and the application of

numerical methods in order to correct for halation effects,

this tactic cannot be applied in a measurement field where

the droplets are not arranged so that the liquid phase can be

effectively masked, for example, in a spray.

Another method for determining the location of the

liquid–vapour interface of acetone droplets in air was

Exp Fluids (2013) 54:1518 Page 3 of 15

123

proposed by Seitzman and co-workers (Ritchie and Seitz-

man 2004; Tran et al. 2005), based on the simultaneous

phosphorescence and fluorescence imaging of liquid and

vapour acetone. The combined technique utilizes the

phosphorescence signal from liquid acetone to locate the

boundaries of the liquid phase and the fluorescence signal

from the vapour phase to measure the concentration. Since

even trace amounts of oxygen can quench the vapour phase

phosphorescence very efficiently, the liquid phosphores-

cence measurement provides an ‘‘uncontaminated’’ liquid

phase signal (Tran 2008). Despite offering an effective way

to discriminate between the two phases, the challenges

associated with the vapour phase concentration measure-

ment, namely the low signal intensity and halation con-

tamination, are not addressed by this strategy.

In order to demonstrate the aforementioned effects of

halation and large signal disparity between the two phases,

a single fluorescence image of an acetone droplet stream in

saturated acetone vapour in air is presented with two dif-

ferent dynamic ranges (Fig. 2c, d). The excitation wave-

length was 308 nm. For comparison, the droplet stream

was also imaged using diffuse white light illumination

from the background (Fig. 2b), while experimental

parameters relevant to the presented images are included in

Fig. 2a; the laser sheet thickness t is approximately

400 lm, the average droplet diameter Dd is 330 lm and the

average inter-droplet separation Sd is approximately 1 mm.

In Fig. 2c, the droplet stream fluorescence image is pre-

sented with a high dynamic range, and only the liquid

phase is visible because of the weak vapour phase fluo-

rescence. The same image is presented in Fig. 2d with a

low dynamic range, allowing both vapour phase and

halation to be visible as well.

The challenges induced by halation in determining the

location of the liquid–vapour interface and the evaporated

tracer concentration near the droplets are evident in

Fig. 2d. The droplets appear to be approximately 1.5 times

larger than they actually are (500 lm instead of 330 lm),

whilst the vapour regions surrounding the droplets appear

brighter, suggesting locally higher acetone vapour con-

centrations. However, since the image is obtained in a

saturated acetone vapour environment, the vapour phase

concentration is spatially constant, and this apparent

increase is purely an artefact attributed to halation. The

same effect can also be observed in the ‘‘shadow’’ region to

the right and in the immediate vicinity of the droplets. Due

to strong absorption of the laser pulse by the liquid phase,

the fluorescence intensity of saturated vapour acetone is

expected to drop to very low levels but still remain effec-

tively constant. However, a locally brighter signal

extending to approximately one droplet diameter away

from the apparent droplet surface is once more observed.

An uncontaminated vapour acetone signal corresponding to

the real fluorescence intensity within the droplet shadow

area can be obtained further away from the droplet surface.

In an attempt to overcome these challenges and to provide

an improved imaging approach, a series of experiments

aimed at characterizing the fluorescence and phosphores-

cence emission from both phases were set-up and will be

presented in the following sections.

3 Experimental set-up

Three sets of experimental measurements will be pre-

sented; the temporal emission decay curves of liquid and

vapour acetone in air and nitrogen, a comparative evalua-

tion of fluorescence and phosphorescence images of a

droplet stream and the surrounding vapour and a quanti-

tative measurement of the vapour field around an evapo-

rating droplet stream by both LIF and LIP techniques. All

experiments described in this paper were carried out at

297 K and 1 atm. The experimental setup employed in all

three studies is presented in the Fig. 3. A Radiant Dyes

Fig. 2 Results from a stream of acetone droplets in saturated acetone

vapour in air. a Schematic of the illuminated droplet stream. b White

light image of a droplet stream under diffuse background illumination.

c Fluorescence image of the stream with 0–1,000 counts dynamic range.

d The same fluorescence image presented with 0–100 counts dynamic

range

Page 4 of 15 Exp Fluids (2013) 54:1518

123

XeCl excimer laser (308 nm) was used for excitation of the

liquid and vapour phase tracer set up inside a flow cell

equipped with optical access. Fluorescence and phospho-

rescence images were collected using a CCD camera

(LaVision Imager Intense) attached to an image intensifier

with an S20 multi-alkali photocathode and a P43 phosphor

screen, and equipped with an 85 mm f/1.4D IF Nikon

Nikkor lens. The linearity of the combined intensifier–CCD

camera system was examined using a set of calibrated

neutral density filters and found to be very good. Three

neutral density filters of different optical density (OD) were

used to prevent saturation of the collected signal when

necessary. The transmission and optical density charac-

teristics of the filters are 20 % (OD 0.7), 1 % (OD 2.0) and

0.1 % (OD 3.0), respectively. The spectral flatness of these

filters has been verified experimentally over the relevant

range. The camera lens was blocking the excitation

wavelength very effectively, and hence, no additional filter

had to be used to prevent the collection of any elastic

scattering of the excitation pulse.

A beam sampler was placed at 45� to the plane of

propagation of the laser sheet, directing 10 % of the

excitation energy to a pyroelectric energy sensor (ES245C

from Thorlabs) connected to a power metre (PM100D from

Thorlabs). Pulse energies were recorded for all collected

images. A cylindrical f = 500 mm plano-convex lens was

used to focus the laser sheet down to a thickness of

400 lm, while the height of the laser sheet was adjusted to

25 mm by means of an iris and an f = 500 mm plano-

concave lens. A custom-made droplet generator equipped

with a piezo-electric oscillator was set-up on top of the cell.

The droplet generator electronics are comprised of a

Thurby Thandar Instruments TG210 function generator and

an amplifier, allowing for different frequency signals to be

applied to the piezo-electric crystal with adjustable inten-

sity. Depending on the acetone flow rate to the droplet

generator and the input signal frequency, droplet streams of

different droplet sizes and different inter-droplet separa-

tions can be obtained. The droplet generator was supplied

with liquid acetone via a PTFE line from a pressure vessel,

where bulk acetone was pressurized using nitrogen (1.5 bar

absolute pressure). Four gas lines were used for purging the

cell with nitrogen, saturated acetone vapour in air or sat-

urated acetone vapour in nitrogen. The acetone vapour was

supplied from a seeder purged with air or nitrogen. When

purging the cell, approximately 5 min were allowed for the

cell to fill with the respective gas.

When conducting the liquid phase phosphorescence

decay measurements, a smooth continuous stream rather

than a droplet stream was set-up by switching off the piezo-

control electronics (frequency generator and amplifier) and

adjusting the nitrogen pressure to the vessel. Three pin-

holes of 600, 200 and 100 lm diameter were used, the first

in the emission decay experiments and the other two in the

droplet stream experiments. Initially, a set-up employing a

quartz cuvette filled with liquid acetone was considered,

but was then abandoned. Seitzman and co-workers (Ritchie

and Seitzman 2004; Tran et al. 2005) have previously

attempted to measure the phosphorescence lifetime of

liquid acetone in a similar manner by illuminating bulk

acetone in a cuvette and have reported that due to the

presence of an intense background signal emanating from

the cuvette itself, an unrealistic measurement of the liquid

phase phosphorescence lifetime of 39.5 ls (Ritchie and

Seitzman 2004) was obtained. The particular phenomenon

was not fully apparent in the current attempts made to

characterize liquid acetone phosphorescence in a cuvette;

however, photodissociation of acetone triplets occurred and

was clearly visible in the phosphorescence images, as the

photodissociation products, possibly acetyl radicals form-

ing biacetyl molecules, contributed strongly to the col-

lected phosphorescence signals. The same behaviour has

been reported regarding vapour acetone phosphorescence

measurements in static cells (Lozano 1992). It is therefore

strongly advisable to employ flow cells, in the case of the

vapour, or streams in the case of the liquid, when per-

forming similar measurements.

4 Results and discussion

4.1 Emission decay measurements

The emission decay of acetone vapour in air and nitrogen

was examined by purging the flow cell with the respective

gases and imaging the emission over a 3 ls range. The

excitation energy was set to 20.5 mJ per pulse, and theFig. 3 Schematic of the optical set-up used throughout the emission

decay and droplet stream imaging experiments

Exp Fluids (2013) 54:1518 Page 5 of 15

123

detection delay was varied from 0 to 2.4 ls after the start

of fluorescence emission. The IRO gate and gain were set

to 200 ns and 70 %, necessitating the use of an OD 2.0

neutral density filter for the fluorescence measurements.

The first set of phosphorescence images was collected at a

delay of 200 ns after the start of fluorescence emission in

order to ensure the fluorescence signal would not affect the

phosphorescence measurements through electronic jitter.

The rest of the phosphorescence measurements were per-

formed in 100 ns steps throughout the examined delay

range. Each data point on the decay curves corresponds to a

50 image average of a 350 9 200 pixel region of interest.

This corresponds to a 9.62 9 5.5 mm viewing area. All

data points were normalized to the fluorescence inten-

sity, and their position on the plot relative to the x-axis

corresponds to the IRO delay setting plus half of the gate

time.

In examining the emission decay of liquid acetone, three

different cases were considered; a liquid acetone stream in

air, a liquid acetone stream in nitrogen bath gas and a

previously nitrogen-purged liquid acetone stream in nitro-

gen bath gas. The reason for conducting the third experi-

ment is that the phosphorescence intensity and

phosphorescence lifetime of liquid acetone have been

shown to increase as a result of the removal of dissolved

oxygen from the liquid. In this study, acetone degassing

was performed by repeatedly bubbling nitrogen through the

bulk acetone supply vessel via a pipe equipped with a

sintered end cap. In the case of liquid acetone in nitrogen,

the diffusion of oxygen out of the liquid is ongoing within

the imaging region and, as will be shown later, causes a

noticeable variation in phosphorescence intensity along the

stream. This particular experimental condition is conse-

quently not very well-defined, but the obtained direct

imaging evidence of the effect of oxygen diffusion out of

the liquid stream is considered worth presenting.

The emission decay measurements of liquid acetone

were performed by setting up a 600 lm diameter liquid

stream and collecting images in 100 ns steps using 70 %

intensifier gain. For the nitrogen-purged liquid acetone in

nitrogen bath gas, data were collected in 200 ns steps since

the decay time is substantially longer. As before, each data

point on the decay curves corresponds to a 50 image

average, and its position on the plot relative to the x-axis

corresponds to the IRO delay setting plus half of the gate

time. A 200 ns exposure time starting 100 ns before the

laser pulse was used to collect the prompt fluorescence

signals. The gate time for all phosphorescence measure-

ments was 200 ns. A list of the neutral density filters

employed in the fluorescence and phosphorescence exper-

iments shown in Figs. 4 and 5, along with the phospho-

rescence delay settings applied in both liquid and vapour

acetone studies, is provided in the Table 1. All liquid phase

measurements were performed using the same pulse energy

(20.5 mJ).

4.2 Acetone vapour emission decay

The normalized acetone vapour in air and acetone vapour

in nitrogen emission decay curves are presented in Fig. 4.

The measured fluorescence intensities agree to within 2 %,

an outcome that fits the theoretical description of vapour

acetone deexcitation presented in the introduction. For

acetone vapour in air, no signal could be collected at

200 ns delay, an effect attributed to very effective oxygen

quenching of the triplet emission. In contrast, the early

phosphorescence signal of acetone vapour in nitrogen is

clearly detectable, although approximately two orders of

magnitude are lower than the prompt fluorescence signal,

and drops by one order of magnitude over the following

Fig. 4 Normalized vapour acetone emission decay in nitrogen and air

Fig. 5 Normalized nitrogen-purged liquid acetone in nitrogen bath

gas, liquid acetone in nitrogen bath gas, liquid acetone in air bath gas,

vapour acetone in nitrogen bath gas and vapour acetone in air

emission decay curves

Page 6 of 15 Exp Fluids (2013) 54:1518

123

3 ls. Both emission decays have recently been examined

for the same delay range, following excitation at 320 nm

(Weckenmann et al. 2011). Although a direct comparison

cannot be made between the current and the aforemen-

tioned study on account of both the higher excitation

wavelength and the lower gate (100 ns) employed in the

latter, it should be noted that very similar trends are

observed amongst the two studies for both decays.

As it has already been noted, vapour acetone phospho-

rescence decays exponentially when measured in a

molecular beam (collision-free environment) with a quoted

lifetime of 200 ls (Groh et al. 1953; Kaskan and Duncan

1950). At ambient conditions, however, where molecular

collisions are frequent, the phosphorescence spectrum

shifts to longer wavelengths with increasing detection

delay (Tran et al. 2005). This trend is attributed to phos-

phorescence emission from lower vibrational levels popu-

lated by vibrational relaxation. Vibrational relaxation

occurs more readily for liquid acetone than for vapour,

suggesting that the particular effect could be even more

pronounced for the latter. The phosphorescence decay of

acetone vapour in nitrogen is consequently expected to be

comprised of a series of exponential components corre-

sponding to deexcitation from different vibrational levels

with different lifetimes. Hence, it is not practical to fit a

single exponential function to the entire data range. For the

purposes of this study, however, the range of interest is

limited to the first few microseconds of the decay, where

the phosphorescence is more intense and the obtained

signals are useful for imaging purposes. Hence, a single

exponential has been fitted to this range with a lifetime of

920 ns.

4.3 Liquid acetone emission decay

Normalized emission decay curves for liquid acetone in air

and nitrogen bath gas and nitrogen-purged liquid acetone in

nitrogen bath gas are presented in Fig. 5; for comparison,

the vapour phase emission decay in nitrogen and the

vapour phase fluorescence in air are also included. All data

points are normalized to the maximum liquid phase fluo-

rescence signal, that of liquid acetone in air. It should be

noted that all three liquid phase fluorescence signals agree

to within 5 % of each other, reflecting the fact that the

effect of oxygen quenching is insignificant in this case. In

fact, liquid acetone fluorescence has been shown to be

independent of bath gas composition for pressures up to

15 bar (Tran et al. 2006). The intensity of acetone vapour

fluorescence in nitrogen is two orders of magnitude lower

than that of liquid acetone in air, and, as noted before,

almost identical to the fluorescence of acetone vapour in

air. At 300 ns after the start of fluorescence emission, the

intensity of liquid acetone phosphorescence in air is only

2.7 times higher than the vapour in nitrogen, whilst 100 ns

later, the vapour phosphorescence is already more intense.

The nitrogen-purged liquid acetone phosphorescence

remains higher than the vapour for almost the entire delay

range; however, the liquid to vapour signal ratio drops from

around 100 for the fluorescence to only around 10 at 300 ns

delay and to 2 at 1 ls delay. The liquid acetone in nitrogen

decay lies between these two cases.

The phosphorescence lifetime of liquid acetone in air is

73 ns, whilst that of nitrogen-purged liquid acetone in

nitrogen bath gas is 214 ns: a threefold increase over the

unpurged case. Longer purging could result in slightly

longer lifetimes; however, the phosphorescence intensity

along the nitrogen-purged stream is reasonably constant,

suggesting that the liquid is nearly depleted of all oxygen.

This is clearly not the case for unpurged liquid acetone in

nitrogen bath gas where the effect of oxygen diffusion out

of the stream can be clearly visualized in the phosphores-

cence images. The removal of any dissolved oxygen along

the stream at a given exposure time results in an increase in

phosphorescence intensity the further the stream travels

inside the cell. In order to demonstrate this effect, two

average images are presented (Fig. 6), one showing the

phosphorescence of a stream of liquid acetone in air and

the other in nitrogen bath gas. Both images were acquired

at 300 ns delay after the start of fluorescence emission with

200 ns gate. Close inspection of the liquid acetone in air

bath gas phosphorescence image reveals that the emission

intensity distribution along the stream is fairly constant. In

comparison, the emission intensity of liquid acetone in

nitrogen bath gas already starts off at a higher value (since

Table 1 List of the optical set-up parameters used in the emission decay measurements (Figs. 4, 5)

Emission decay Fluorescence filters Phosphorescence delay range and filters

Vapour acetone in air OD 2 –

Vapour acetone in N2 OD 2 200–2,900 ns in 100 ns steps

Liquid acetone in air OD 3 and OD 0.7 200–900 ns in 100 ns steps (OD 0.7 for 200 ns delay)

Liquid acetone in N2 OD 3 and OD 0.7 200–1,400 ns in 100 ns steps (OD 0.7 for 200 ns delay)

N2-purged liquid acetone in N2 OD 3 and OD 0.7 200–2,400 ns in 200 ns steps (OD 0.7 for 200 ns and 400 ns delays)

The fluorescence and phosphorescence gate and gain settings (200 ns, 70 %) were identical for all decay measurements. The phosphorescence

delay values are expressed relative to the start of fluorescence emission

Exp Fluids (2013) 54:1518 Page 7 of 15

123

the topmost part of the imaged area corresponds to a

location approximately 1 cm downstream of the droplet

generator pinhole) and increases all the way to the bottom

of the image. This effect comes from oxygen which is

initially dissolved in the liquid acetone, successively dif-

fusing out of the liquid in the nitrogen bath gas

environment.

5 Two-phase flow imaging by acetone laser induced

phosphorescence

For the purposes of two-phase flow imaging using ace-

tone phosphorescence, and additionally for providing a

well-defined boundary condition (with respect to the

dissolved oxygen content within the liquid), the use of

nitrogen-purged liquid acetone in a nitrogen atmosphere

is considered to be the optimal choice; the signal dis-

parity between the two phases is large enough to clearly

identify the liquid–vapour interface but small enough to

prevent the emergence of any halation. In order to

demonstrate the improvements of this technique for two-

phase flow visualization, two cases are examined. Firstly,

a non-evaporating acetone droplet stream (330 lm aver-

age droplet diameter) is set-up in a saturated acetone

vapour environment. The fluorescence is initially col-

lected, and the same instantaneous image is presented in

Fig. 7 with two dynamic ranges (0–1,000 counts in

Fig. 7b and 0–100 counts in Fig. 7c) as the intense

halation and large signal disparity between the two pha-

ses prevent the saturated vapour and real droplet sizes to

be shown simultaneously. Phosphorescence, in this

example, is collected over 400 ns, starting off at 300 ns

after the fluorescence, with a gain of 70 %. A single

dynamic range suffices as the signal disparity is suffi-

ciently reduced, and no halation is observed (Fig. 7d). A

lower gain setting (50 %) along with an OD 2.0 filter was

employed in the fluorescence image collection, whilst the

excitation energy in both cases was set to 23.5 mJ. Both

fluorescence and phosphorescence images have been

background-subtracted and corrected for laser sheet

inhomogeneities and dark noise. For comparison, a white

light image of the droplets with diffuse background

illumination is also presented, indicating their real sizes

and inter-droplet distances (Fig. 7a).

The effects of halation have already been discussed

earlier; an apparent increase in vapour phase concentration

around the droplets is observed, along with an increase in

the spatial extent of the droplets by at least 50 % (Fig. 7c).

When increasing the dynamic range, the glow around the

droplets subsides, but the vapour phase is now completely

indiscernible due its low fluorescence signal level

(Fig. 7b). Instead, when monitoring the phosphorescence,

no halation around the droplets is observed and the satu-

rated vapour is clearly visible and discernible from the

liquid (Fig. 7d). The droplets sizes also match those of the

shadow image (Fig. 7a). The liquid phase intensities range

between 1,000 and 2,000 counts for both fluorescence and

phosphorescence images, whilst the vapour phase fluores-

cence to the left of the stream has an average intensity of

19 counts and the phosphorescence an average intensity of

420 counts.

The two imaging techniques are also compared under

evaporating conditions, more relevant to two-phase flow

investigations. In Fig. 8, fluorescence and phosphoresce

images of a free-falling monodisperse droplet stream in

pure nitrogen bath gas are presented. The same excitation

energy, collection optics settings and dynamic ranges as for

Fig. 7 are used. In this case, it is impossible to evaluate the

vapour phase signal from the fluorescence image (Fig. 8c)

as the majority of the evaporated tracer lies within the

spatial extent of halation. In contrast when monitoring

the phosphorescence, a vapour cloud forming around the

droplet stream is clearly observed.

6 Quantitative two-phase flow imaging

Following the comparative analysis of acetone two-phase

flow imaging by laser induced fluorescence and phospho-

rescence on a qualitative basis, a quantitative examination

is hereby presented. While the relevant experiments were

once again carried out around an isothermal evaporative

droplet stream, an additional step was necessary, namely

the calibration of both laser induced fluorescence and

phosphorescence signals with acetone vapour concentra-

tion and excitation energy.

Fig. 6 Phosphorescence images from a liquid acetone stream at

300 ns delay a in air and b in nitrogen. The phosphorescence signal

strength is constant along the length of the liquid stream set up in air,

whereas in nitrogen, the phosphorescence increases further down-

stream due to ongoing removal of dissolved oxygen and the

subsequent reduction of oxygen quenching

Page 8 of 15 Exp Fluids (2013) 54:1518

123

6.1 Vapour phase concentration and excitation energy

calibration

It has been firmly established that for low fluences, vapour

acetone fluorescence varies linearly with both excitation

energy and vapour concentration (Lozano 1992; Bryant

et al. 2000). This, however, may not be the case for LIP, as

collisions with either ground state or excited triplet state

acetone molecules result to quenching (through self-

quenching or triplet–triplet annihilation). The probability

that such a collision event will occur is clearly time-

dependent, and hence, both energy and vapour phase

concentration calibrations will apply to particular delay and

gate settings. In comparison, the excited singlet state is

extremely short-lived due to rapid inter-system crossing,

and hence, laser induced fluorescence is hardly affected by

molecular collisions at ambient pressure.

The excitation energy calibration experiments were

performed by saturating a quartz cuvette with acetone

vapour in nitrogen and varying the excitation energy

between 7.8 and 22.1 mJ. The cuvette was employed in

this calibration in order to prevent the large energy drop

occurring within the flow cell when the latter is saturated

with acetone vapour, due to strong absorption of the laser

beam prior to the imaging region. This energy drop would

substantially limit the energy exciting the tracer within the

imaging region, and thus, the obtained calibration would

not be applicable to an evaporating droplet stream, such

as the one presented in Fig. 8 (in this case, the laser beam

propagates unobstructed until the middle of the cell where

it encounters and excites the droplet stream and sur-

rounding vapour). Following both dark current and

background signal corrections, 100 images were collected

and averaged at each energy setting. Vapour acetone

fluorescence was collected using a delay of -100 ns rel-

ative to the start of emission, a gate of 500 ns, a gain of

70 % and an OD 2.0 filter. The vapour phase radiative

decay results (Fig. 4) indicate that the contribution from

any collected phosphorescence when using the particular

delay and gate settings will be insignificant. Vapour phase

phosphorescence was collected by setting the delay to

200 ns after the start of fluorescence emission, maintain-

ing the 500 ns gate and 70 % gain settings, and removing

the neutral density filter. The average fluorescence and

phosphorescence signals, as well as excitation energies,

have been normalized to their respective maxima and are

presented in Fig. 9. The linear relationship between fluo-

rescence and excitation energy was confirmed in these

measurements. Laser induced phosphorescence, however,

was shown to vary non-linearly with excitation energy,

and the obtained data are better fitted with a second-order

polynomial function.

Fig. 7 Results from a non-evaporating stream of acetone droplets in

saturated acetone vapour in nitrogen. a White light image of a droplet

stream under diffuse background illumination. b Fluorescence image

of the stream with 0–1,000 counts dynamic range. c The same

fluorescence image presented with 0–100 counts dynamic range.

d Phosphorescence image of the droplet stream

Fig. 8 Results from a stream of evaporating acetone droplets in pure

nitrogen. a White light image of the droplet stream under diffuse

background illumination. b Fluorescence image of the stream with

0–1,000 counts dynamic range. c The same fluorescence image

presented with 0–100 counts dynamic range. d Phosphorescence

image of the droplet stream

Exp Fluids (2013) 54:1518 Page 9 of 15

123

Acetone vapour laser induced fluorescence and phos-

phorescence were also calibrated against acetone vapour

concentration. For this purpose, the flow cell was satu-

rated with known acetone vapour concentrations, obtained

by adjusting the flow rates of two independent gas lines,

mixed prior to entering the cell. One line provided satu-

rated acetone vapour in nitrogen (with the acetone num-

ber density corresponding to the saturated value at

ambient temperature and pressure), whilst the second

provided pure nitrogen. An extensive account of how the

acetone number density can be calculated by employing

such a set-up can be found in the literature (Bryant et al.

2000). It should also be noted that in performing these

measurements, a correction for laser sheet absorption

within the cell was performed in accordance with the

Beer–Lambert Law. The collected fluorescence and

phosphorescence signals were corrected for the energy

drop across the cell using the previously derived fluo-

rescence and phosphorescence against excitation energy

relationships (Fig. 9). The collection optics settings are

identical to those applied in performing the respective

fluorescence and phosphorescence against excitation

energy calibrations. For both fluorescence and phospho-

rescence experiments, 100 images were collected and

averaged at each number density setting. The corrected

average signals, as well as number densities, were then

normalized to their respective maxima and plotted in

Fig. 10. The results indicate that whereas the fluorescence

signal varies linearly with number density, LIP is again

better fitted to a second-order polynomial function. The

excitation energy was kept constant at 20.5 mJ in the

particular calibration, as well as the following imaging

experiments.

The observed non-linearities in the vapour phase

phosphorescence calibration measurements with respect to

both excitation energy and tracer concentration are

attributed, as has been indicated earlier, to the competition

between the higher absorption rate, and hence increased

triplet state population rate, and the corresponding

increase in the collisional quenching rate through triplet–

triplet annihilation. No explicit account of the particular

deexcitation mechanism has been found in the literature

for acetone; however, for biacetyl, the rate constant of

triplet–triplet annihilation is approximately three orders of

magnitude higher than oxygen quenching (Badcock et al.

1972; Sidebottom et al. 1972). For aromatics in solution,

triplet–triplet annihilation displays almost the same rate

constant as that of molecular diffusion (Turro et al. 2010).

The saturation-like behaviour observed at both high laser

fluences (Fig. 9) and high tracer concentrations (Fig. 10)

signifies the onset of triplet–triplet annihilation as a strong

deexcitation channel. The lower fluence/tracer concentra-

tion regions indicate that for lower triplet state population

densities, the vapour phase phosphorescence signal

behaviour approaches linearity. Phosphorescence lifetime

measurements conducted for three different laser fluences

(255, 120 and 9.6 mJ/cm2) support the findings of the

calibration measurements; the observed phosphorescence

lifetime increases from approximately 850 ns to 1.2 ls

and 4 ls, respectively, as a consequence of the reduction

of the triplet–triplet annihilation rate. In the context of this

study, lower excitation energy could have been used in

order to perform the quantitative analysis within the

regime for which the effects of triplet–triplet annihilation

are not so prominent; however, this would be done at the

expense of more noisy measurements, an outcome that is

clearly unwanted. In order to minimize temporal varia-

tions in the spatial excitation beam profile, a beam

homogenizer (Pfadler et al. 2006) will be used in the

future.

Fig. 9 a Normalized fluorescence signal plotted against normalized

excitation energy. b Normalized phosphorescence signal plotted over

the same excitation energy range. The maximum pulse energy is

22.1 mJ corresponding to a fluence of 221 mJ/cm2 (the laser sheet

thickness and height are 400 lm and 2.5 cm, respectively)

Fig. 10 a Normalized fluorescence signal plotted against normalized

number density energy. b Normalized phosphorescence signal plotted

over the same number density range. The pulse energy is 20.5 mJ

corresponding to a fluence of 205 mJ/cm2. The saturated acetone

vapour number density is 6.9116 9 1018 molecules/cm3 at an ambient

temperature and pressure of 297.15 K and 1 bar, respectively

Page 10 of 15 Exp Fluids (2013) 54:1518

123

6.2 Quantitative acetone vapour measurements

Both laser induced fluorescence and phosphorescence

techniques were applied in order to determine the vapour

distribution around an evaporative droplet stream set up in

nitrogen bath gas at 297 K. Prior to performing the two-

phase flow experiment, a saturated acetone vapour in

nitrogen flow was introduced to the cell and a 100 image

average of the saturated acetone vapour in nitrogen fluo-

rescence signal was obtained. This signal was corrected for

absorption effects through the cell and then used as the cut-

off level between the vapour phase and liquid phase plus

halation signals, following the procedure introduced by

Bazile and Stepowski (1995) and later adopted by other

researchers (Hardalupas et al. 2010; Ammigan and Clack

2009). Subsequently, the vapour cloud forming around the

droplets was quantified using this measurement and the

previously obtained number density calibration.

In monitoring the phosphorescence, the optical set-up

used in the calibration experiments was kept unaltered.

Prior to introducing the droplet stream, 100 images of sat-

urated acetone vapour in nitrogen phosphorescence were

collected and averaged. The resulting phosphorescence

signal was used to distinguish between the liquid and

vapour phases in the same way as with the fluorescence. The

vapour field around the droplet stream was quantified by

means of this threshold and the number density calibration.

Dark current and background contributions were subtracted

from both fluorescence and phosphorescence images, while

white image corrections were also performed. Finally, both

fluorescence and phosphorescence images were corrected

for pulse energy variation by means of the energy calibra-

tion curves. The resulting quantitative images, along with a

shadow image of the droplet stream obtained under diffuse

background illumination, are presented in Fig. 11.

In producing the particular droplet stream, the droplet

generator was equipped with a 100 lm pinhole. The flow

rate through the droplet generator was 3.3 ml/s, whilst the

resonance frequency of the generator electronics was set to

25 kHz; the respective theoretically calculated average

droplet size was 161 lm (Pergamalis 2002). Measured

diameters from both shadow and phosphorescence images

are virtually identical and in good agreement with the

calculations. The fluorescence image, instead, is plagued

by strong halation that forces the location of the liquid–

vapour interface away from the actual droplet surfaces.

Hence, it provides no realistic information regarding the

location of the liquid–vapour interface or droplet sizes. The

vapour distribution obtained from the phosphorescence

image suggests that a low concentration (mole fraction of

up to 0.05) vapour tunnel surrounds the droplet stream,

while adjacent and in between the droplets, the acetone

mole fraction rises to beyond 0.10. A saturated or near-

saturated acetone vapour concentration (mole fraction of

approximately 0.28) is only encountered immediately

adjacent to the droplets. In order to examine more closely

the extent of halation in the fluorescence image and com-

pare the vapour fields obtained by both imaging techniques

amongst each other, and with those obtained by other

researchers, radial profiles of acetone vapour distributions

emanating from droplet surfaces have been collected and

averaged over 10 droplets. As the droplet size and shape

showed periodic variations with respect to the distance

between an individual droplet and the droplet generator tip,

equidistant droplets from 10 different fluorescence and

phosphorescence images were selected. The results are

presented in Fig. 12. The radial profile obtained from the

161 lm droplet stream phosphorescence images suggests

that the vapour tunnel formed around the droplet stream

extends radially to approximately 10 droplet diameters

away from the droplet surfaces and represents the radial

diffusion of acetone vapour. In the immediate vicinity of

the droplets, the vapour mole fraction increases rapidly

to the saturated value. The profile obtained from the

Fig. 11 Quantitative mole fraction results from a stream of evapo-

rating acetone droplets in nitrogen. a Schematic of the illuminated

droplet stream. b White light image of the droplet stream under

diffuse background illumination. c Fluorescence image of the droplet

stream. d Phosphorescence image of the droplet stream

Exp Fluids (2013) 54:1518 Page 11 of 15

123

fluorescence images indicates that a vapour phase mea-

surement is not feasible to within 1.5 droplet diameter

away from the droplet surfaces. At greater distances, the

effect of halation results in an overestimation of the entire

vapour field which extends to nearly 10 droplet diameters.

Orain et al. (2005) have studied the radial profile of the

acetone vapour tunnel forming around a 230 lm mono-

disperse droplet stream at 298 K. They present results with

and without a halation correction; the former obtained by

quantifying the spatial extent and intensity of halation by

means of the Mie scattering signal from ethanol droplets.

From the uncorrected profile, mole fractions can be first

evaluated at a distance of slightly more than half a droplet

diameter away from the droplet surfaces, whilst the mole

fraction distribution coincides with the one obtained from

the corrected images at a distance slightly larger than 2

droplet diameters away from the droplet surfaces. The

spatial extent of halation is shorter than the one observed in

the present study, and this can be attributed to the fol-

lowing two reasons. Firstly, the inter-droplet separation

(measured between the centres of two adjacent droplets) in

the present study is approximately 2.5 droplet diameters,

whereas in the case of Orain and co-workers, it is 3.5. As a

result, the radial distribution measurement across a par-

ticular droplet suffers less severely from halation contri-

butions from the other two neighbouring droplets.

Secondly, the implemented halation correction is incom-

plete, as the authors admit. This is also evident from the

fact that the first mole fraction measurement is not attained

immediately adjacent to the droplet surface but approxi-

mately 0.2 droplet diameters away from it. Comparing the

corrected profile with the one obtained presently by means

of the LIP technique, it also becomes apparent that the

former is not completely free of halation; despite the fact

that the overall trend is very similar, it appears that the

corrected profile consistently overestimates the acetone

vapour mole fraction measurement. However, both the

experimental and numerical results, presented in another

study by Castanet et al. (2007) for 170 lm droplets injected

in ambient air, agree very well with the mole fraction

profile obtained by LIP imaging. In an attempt to account

for halation effects, Castanet and co-workers have suc-

cessfully utilized the technique previously introduced by

Orain and co-workers.

Another quantitative account of the radial acetone

vapour distribution around a droplet stream is provided by

Frackowiak et al. (2009). In this study, the halation is

suppressed by masking the droplets using an aperture in the

field of view, thus obtaining the vapour field directly from

a fluorescence measurement in the interval between

approximately 1.5 droplets diameters away from the

droplet surfaces up to more than 10 droplet diameters away

from the stream. A comparison of the radial profiles

obtained with and without the mask reveals that the

blooming effect extends as far as 10 droplet diameters

away from the droplet surfaces for 173 and 238 lm

droplets. The present study confirms this observation for

the examined range of up to 10 droplet diameters away

from the liquid–vapour interface. For 173 lm droplets at

22.4 �C, the agreement between measured radial profile by

Frackowiak and co-workers (with the implementation of

the mask) and the present study is good. The numerical

results, however, overestimate the experimental measure-

ments of both studies. Additionally, the phosphorescence

results suggest a much sharper increase in the acetone

vapour concentration in the vicinity of the droplet stream

(0.5 droplet diameters away from the droplet surface) than

the numerical results under consideration. The experiments

by Castanet et al. (2007) also appear to support the former.

In order to demonstrate the applicability of the LIP

technique over a wide range of droplet diameters, further

experiments were conducted for two evaporative droplet

streams of 124 and 226 lm droplet diameters. The vapour

phase quantification procedure for both droplet streams was

identical to the one adopted before, as were all optical set-

up parameters. Once again, no halation effects are detected

and the observed droplet sizes match those of the theoret-

ical calculation. Single-shot images of both droplet streams

are presented in Fig. 13.

Having provided an account of the applicability of LIP

for the investigation of evaporative droplet streams, the

potential application of the technique to different studies

will now be discussed. In internal combustion engines,

spray diagnostics are often conducted by doping a base

fuel, which is ‘‘transparent’’ for the employed excitation

wavelength, with a co-evaporating tracer. Acetone, for

example, has been employed in the design and experi-

mental investigation of a multi-component mixture, com-

prised of three tracers and three base fuels of different

Fig. 12 Average radial profiles of vapour acetone mole fractions (xa)

plotted against the distance from the droplet centre normalized to the

average droplet diameter (Xd/Dd), obtained by both laser induced

fluorescence and phosphorescence imaging

Page 12 of 15 Exp Fluids (2013) 54:1518

123

volatilities, in an attempt to match the volatility range of

commercial automotive fuels (Williams et al. 2010; Ste-

vens et al. 2007). In particular, acetone was the low-vola-

tility fuel tracer, with the low-volatility base fuel

component being isopentane. The applicability of the LIP

technique in this case is subject to the doping concentration

of acetone, which should be high enough to ensure ade-

quate signal levels from both vapour and liquid phases. In

addition, the phosphorescence properties of the acetone/

solvent mixture would have to be calibrated in a similar

manner to this study, in order to account for any collisional

quenching effect by the base fuel. Also, the application of

another excitation wavelength may be useful. Generally,

longer excitation wavelengths result in increased phos-

phorescence efficiency and longer liquid and vapour phase

lifetimes; following intersystem crossing, the excited triplet

state acetone molecules migrate closer to the thermalisation

level, for which the dissociation quantum yield is very low.

Excitation at 320 nm, for example, is highly recommended

when employing the particular technique (Weckenmann

et al. 2011). Despite the fact that liquid acetone phospho-

rescence displays spectral variations depending on the

excitation wavelength (Tran et al. 2005), for the purposes

of the LIP technique, it is preferable to collect the phos-

phorescence signal over the longest spectral range possible,

in order to enhance the signal level.

Another important aspect that must be considered prior

to expanding the LIP technique to more general experi-

ments is the effect of reduced laser fluence, as a result of

absorption across a flow field, on vapour phase concen-

tration measurements. As it has already been noted, in the

case of phosphorescence, a lower fluence will enhance the

phosphorescence efficiency and, hence, the signal drop

cannot be straightforwardly calculated on the basis of the

Beer–Lambert law. Assuming a flow field that is imaged

across 5 cm, the maximum drop in pulse energy would

result for a fully saturated acetone vapour environment;

under these conditions and for a temperature of 297 K and

excitation at 308 nm, the laser energy would drop to

approximately 57 % of its original value. In the case of

fluorescence, this corresponds to a 57 % decrease in the

vapour signal intensity. In order to quantify the corre-

sponding decrease in the vapour phase phosphorescence

signal, the phosphorescence against energy curve can be

used (Fig. 9). In this case, the phosphorescence signal

corresponding to 57 % of the original excitation energy is

only reduced to approximately 74 % of its original value.

This 17 % disparity between the fluorescence and phos-

phorescence measurements represents the maximum error

that could be induced in a vapour tracer concentration

measurement under the particular conditions. If the vapour

concentration varies across the flow field, and hence, the

laser pulse is not so strongly absorbed, this error would be

substantially reduced. The use of longer excitation wave-

lengths would help diminish this inaccuracy due to reduced

absorption and hence reduced triplet–triplet annihilation

effects; for example, at 320 nm, the absorption cross sec-

tion is reduced to 32 % of the equivalent value at 308 nm

(Lozano 1992). Additionally, reducing the laser fluence by

either reducing the excitation energy or increasing the laser

sheet thickness would also assist in the minimization of the

particular error.

7 Conclusions

Two-phase flow visualization using acetone fluorescence

was shown to present two significant challenges: the large

intensity disparity between the two phases that forces the

vapour phase signal to near noise levels and the ensuing

effect of halation around the droplets. In an attempt to

overcome these obstacles, a strategy using the phospho-

rescence, rather than the fluorescence emission of liquid

and vapour acetone, is presented. Phosphorescence lifetime

data are, therefore, collected by examining the emission

decays of liquid and vapour acetone at different boundary

conditions, following excitation at 308 nm. In air, liquid

acetone phosphorescence is shown to decay extremely fast

(73 ns) due to the presence of dissolved oxygen within the

tracer. By purging liquid acetone with nitrogen, the effect

of oxygen quenching is reduced, and the lifetime of

nitrogen-purged liquid acetone in a nitrogen atmosphere

increases to 213 ns. When the case of unpurged acetone in

nitrogen bath gas was examined, it was demonstrated that

due to oxygen diffusion out of the liquid, an increase in

both phosphorescence signal strength and lifetime would

ensue. This effect was present even within the confines of

the imaging field, and thus, it is strongly advised to purge

liquid acetone prior to using it in two-phase flow imaging

experiments by LIP. With regard to acetone vapour, the

Fig. 13 Single-shot LIP images of evaporative 124 and 226 lm

droplet streams injected in an ambient nitrogen environment (297 K,

1 atm)

Exp Fluids (2013) 54:1518 Page 13 of 15

123

phosphorescence emission is completely quenched in air

by molecular oxygen, whereas in nitrogen, the vapour

phosphorescence decays relatively slowly (radiative life-

time of 920 ns). In consequence of the different kinetics

governing the triplet emission of liquid and vapour acetone,

the phosphorescence signals from the two phases are of

similar order, compared to the respective fluorescence

intensities, and thus, a more favourable signal ratio

between the two phases is achieved. In that respect, the two

hindrances associated with two-phase flow laser induced

fluorescence imaging are resolved.

In order to demonstrate the advantages of this strategy,

two cases were examined on a qualitative basis: a non-

evaporating droplet stream in saturated acetone vapour in

nitrogen and an evaporating droplet stream in pure nitro-

gen. These experiments clearly indicate that by employing

LIP, the intense signal disparity between the liquid and

vapour phases, as well as the halation effect, is overcome.

In addition, the location of the liquid–vapour interface is

clearly identified. Following, the qualitative comparison, a

quantitative analysis of an isothermal evaporative droplet

stream was carried out. In response to the potential

emergence of triplet–triplet annihilation as a major deex-

citation channel of acetone triplets, the vapour phase

phosphorescence was calibrated for both excitation energy

and tracer number density. Whereas a direct proportion-

ality is observed between the aforementioned quantities

and laser induced fluorescence, LIP calibrations are better

fitted with second-order polynomial functions. The quan-

titative two-phase flow measurements by LIP indicate the

formation of a diffusion-driven, low concentration vapour

tunnel along the droplet stream. Comparing (amongst the

two techniques) the vapour fields obtained from radial

profiles emanating from droplet surfaces and extending to

10 droplet diameters away from the interface, it is shown

that the halation effect, at least in the particular experi-

ment, is present throughout this range. Literature data also

support this finding. Both the presented fluorescence data

as well as the literature data overestimate the mole frac-

tions observed in the vicinity of liquid droplets obtained

by the phosphorescence measurement. In addition, in the

present experiments as well as in the literature, the

quantification of the vapour field by laser induced fluo-

rescence is impossible to within one droplet diameter

away from the interface. A single comprehensive study

that has successfully performed a halation correction by

means of the Mie scattering signal from ethanol droplets

is shown to closely agree with the LIP results, thus sup-

porting their validity.

Acknowledgments Supply of the droplet generator by Y. Hard-

alupas and A.M.K.P. Taylor is gratefully acknowledged.

References

Ammigan K, Clack HL (2009) Planar laser-induced fluorescence

imaging of the spatial vapor distribution around a monodisperse

droplet stream exposed to asymmetric radiant heating. Proc

Combust Inst 32:2179–2186

Badcock CC, Sidebottom HW, Calvert JG, Blaine RR, Damon EK

(1972) A study of the triplet–triplet annihilation reaction in

biacetyl vapor excited at 4365 A and 25�. J Am Chem Soc

94(1):19–24

Bazile R, Stepowski D (1995) Measurements of vaporized and liquid

fuel concentration fields in a burning spray jet of acetone using

planar laser induced fluorescence. Exp Fluids 20:1–9

Braeuer A, Beyrau F, Leipertz A (2006) Laser-induced fluorescence

of ketones at elevated temperatures for pressures up to 20 bars by

using a 248 nm excitation laser wavelength: experiments and

model improvements. Appl Opt 45(20):4982–4989

Bryant RA, Donbar JM, Driscoll JF (2000) Acetone laser induced

fluorescence for low pressure/low temperature flow visualiza-

tion. Exp Fluids 28(5):471–476

Castanet G, Maqua C, Orain M, Grisch F, Lemoine F (2007)

Investigation of heat and mass transfer between the two phases

of an evaporating droplet stream using laser-induced fluores-

cence techniques: comparison with modeling. Int J Heat Mass

Transfer 50:3670–3683

Connon CS, Dimalanta R, Choi C, Dunn-Rankin D (1997) LIF

measurements of fuel vapor in an acetone droplet stream.

Combust Sci Technol 129:197–216

Frackowiak B, Strzelecki A, Lavergne G (2009) A liquid-vapour

interface positioning method applied to PLIF measurements

around evaporating monodisperse droplet streams. Exp Fluids

46:671–682

Groh HJ, Luckey GW, Noyes WA (1953) The mechanism of acetone

vapor fluorescence. J Chem Phys 21(1):115–118

Halpern AM, Ware WR (1971) Excited singlet state radiative and

nonradiative transition probabilities for acetone, acetone-d6, and

hexafluoroacetone in the gas phase, in solution, and in the neat

liquid. J Chem Phys 54(3):1271–1276

Hardalupas Y, Sahu S, Taylor AMKP (2010) Measurement of vapour

concentration and droplet size and velocity in an evaporative

spray. Paper presented at the 15th international symposium on

applications of laser techniques to fluid mechanics, Lisbon,

Portugal, 05–08 July 2010

Heicklen J (1959) The fluorescence and phosphorescence of biacetyl

vapor and acetone vapor. J Am Chem Soc 81(15):3863–3866

Hu H, Koochesfahani MM (2002) A novel method for instantaneous,

quantitative measurement of molecular mixing in gaseous flows.

Exp Fluids 33:202–209

Kaskan WE, Duncan ABF (1950) Mean lifetime of the fluorescence

of acetone and biacetyl vapors. J Chem Phys 18(4):427–431

Koch JD, Hanson RK, Koban W, Schulz C (2004) Rayleigh-

calibrated fluorescence quantum yield measurements of acetone

and 3-pentanone. Appl Opt 43(31):5901–5910

Loffler M, Beyrau F, Leipertz A (2010) Acetone laser-induced

fluorescence behavior for the simultaneous quantification of

temperature and residual gas distribution in fired spark-ignition

engines. Appl Opt 49(1):37–49

Lozano A (1992) Laser-excited luminescent tracers for planar

concentration measurements in gaseous jets. Dissertation, Stan-

ford University

Orain M, Mercier X, Grisch F (2005) PLIF imaging of fuel-vapor

spatial distribution around a monodisperse stream of acetone

droplets: comparison with modelling. Combust Sci Technol

177:249–278

Page 14 of 15 Exp Fluids (2013) 54:1518

123

Pergamalis H (2002) Droplet impingement onto quiescent and

moving liquid surfaces. Dissertation, Imperial College London

Pfadler S, Beyrau F, Loffler M, Leipertz A (2006) Application of a

beam homogenizer to planar laser diagnostics. Opt Express

14(22):10171–10180

Ritchie BD, Seitzman JM (2001) Quantitative acetone PLIF in two-

phase flows. Paper presented at the 39th aerospace sciences

meeting and exhibit, Reno, Nevada, 08–11 January 2001

Ritchie BD, Seitzman JM (2004) Simultaneous imaging of vapor and

liquid spray concentration using combined acetone fluorescence

and phosphorescence. Paper presented at the 42nd aerospace

science meeting and exhibit, Reno, Nevada, 05–08 January 2004

Rothamer DA, Snyder JA, Hanson RK, Steeper RR, Fitzgerald RP

(2009) Simultaneous imaging of exhaust gas residuals and

temperature during HCCI combustion. Proc Combust Inst

32:2869–2876

Sahu S (2011) Experimental study of isothermal and evaporative

sprays. Dissertation, Imperial College London

Schulz C, Sick V (2005) Tracer-LIF diagnostics: quantitative

measurement of fuel concentration, temperature and fuel/air

ratio in practical combustion systems. Prog Energy Combust Sci

31(1):75–121

Sidebottom HW, Badcock CC, Calvert JG, Rabe BR, Damon EK

(1972) Lifetime studies of the biacetyl excited singlet and triplet

states in the gas phase at 25�. J Am Chem Soc 94(1):13–19

Stevens RE, Ma H, Stone CR, Walmsley HL, Cracknell R (2007) On

planar laser-induced fluorescence with multi-component fuel and

tracer design for quantitative determination of fuel concentration

in internal combustion engines. Proc Inst Mech Eng Pt D

221(6):713–724

Thurber MC, Hanson RK (1999) Pressure and composition depen-

dences of acetone laser-induced fluorescence with excitation at

248, 266, and 308 nm. Appl Phys B 69(3):229–240

Thurber MC, Hanson RK (2001) Simultaneous imaging of temper-

ature and mole fraction using acetone planar laser-induced

fluorescence. Exp Fluids 30(1):93–101

Thurber MC, Grisch F, Kirby BJ, Votsmeier M, Hanson RK (1998)

Measurements and modeling of acetone laser-induced fluores-

cence with implications for temperature-imaging diagnostics.

Appl Opt 37(21):4963–4978

Tran T (2008) Acetone planar laser-induced fluorescence and

phosphorescence for mixing studies of multiphase flows at high

pressure and temperature. Dissertation, Georgia Institute of

Technology

Tran T, Kochar Y, Seitzman J (2005) Measurements of liquid acetone

fluorescence and phosphorescence for two-phase fuel imaging.

Paper presented at the 43rd aerospace sciences meeting and

exhibit, Reno, Nevada, 10–13 January 2005

Tran T, Kochar Y, Seitzman J (2006) Measurements of acetone

fluorescence and phosphorescence at high pressures and tem-

peratures. Paper presented at the 44th aerospace sciences

meeting and exhibit, Reno, Nevada, 09–12 January 2006

Turro NJ, Ramamurthy V, Scaiano JC (2010) Modern molecular

photochemistry of organic molecules, 1st edn. University

Science Books, Sausalito, CA

Weckenmann F, Bork B, Oldenhof E, Lamanna G, Weigand B, Bohm

B, Dreizler A (2011) Single acetone droplets at supercritical

pressure: droplet generation and characterization of PLIFP.

Z Phys Chem 225:1417–1431

Williams B, Ewart P, Wang X, Stone R, Ma H, Walmsley H,

Cracknell R, Stevens R, Richardson D, Fu H (2010) Quantitative

planar laser-induced fluorescence imaging of multi-component

fuel/air mixing in a firing gasoline-direct-injection engine:

effects of residual exhaust gas on quantitative PLIF. Combust

Flame 157(10):1866–1878

Exp Fluids (2013) 54:1518 Page 15 of 15

123