Delayed appearance of the liquidcondensed phase in 1octadecanol films onlevitated waterdropsA. E. Frost, Mark Seaver, and Glenn O. Rubel Citation: The Journal of Chemical Physics 100, 3268 (1994); doi: 10.1063/1.466417 View online: http://dx.doi.org/10.1063/1.466417 View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/100/4?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Surface and bulk phase behavior of dry and hydrated tetradecanol:octadecanol alcohol mixtures J. Chem. Phys. 116, 8056 (2002); 10.1063/1.1465401 Magnetic levitation of condensed hydrogen Rev. Sci. Instrum. 62, 3022 (1991); 10.1063/1.1142148 A reversal of lateralization, with images appearing on the side of the delay J. Acoust. Soc. Am. 68, S16 (1980); 10.1121/1.2004603 Delayed Appearance of OH in Acetylene—Oxygen Reaction J. Chem. Phys. 42, 2636 (1965); 10.1063/1.1696365 Delayed Appearance of OH in the Acetylene—Oxygen Reaction J. Chem. Phys. 40, 1771 (1964); 10.1063/1.1725396

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

Delayed appearance of the liquid-condensed phase in 1-octadecanol films, on levitated waterdrops

A. E. Frosta) and Mark Seaver Code 5610, Naval ResearchLaboratory, Washington D.C. 20375-5338

Glenn O. Rubel u.s. Army Armament Munitions and Chemical Command, Chemical Research Development and Engineering Center, Aberdeen Proving Ground, Maryland 21010-5423

(Received 7 July 1993; accepted 5 November 1993)

We have observed dramatic delays in the formation of the liquid-condensed phase of l-octadecanol thin films on the surface of levitated water drops as the drops evaporate into a flowing gas stream. When the aqueous substrate starts with more than 1.0 X 10-4 M EuH .

(EDTA), the liquid-condensed phase of l-octadecanol appears, as expected, at a surface con­centration equivalent to one monolayer. As we reduce the initial EuH (EDTA) content, higher and higher surface concentrations of l-octadecanol are reached prior to the appearance of the liquid-condensed phase. At initial concentrations of EuH (EDTA) below -4.0X 10-7 M, the . average surface concentration at the liquid-condensed phase change exceeds ten monolayers .. Light-scattering data attest to the presence of small, <400 nm radius, l-~ctadecanol particles initially dispersed throughout the substrate. We demonstrate that these partIcles are collected by the shrinking drop surface with minimal diffusion of the particles to the surface.

INTRODUCTION

Studies of monolayer phases and their effects on mass transport are usually carried out with a Langmuir trough. I- 3 Standard procedure in such studies is to clean the water surface and deposit the monolayer forming ma­terial on the surface using a volatile solvent that is not retained in the monolayer. An alternative procedure which works with many surfactants, including l-octadecanol, re­lies on the spontaneous spreading of the monolayer from the solid material. When a monolayer is spread from the solid, the two surfactant phases reach equilibrium at the equilibrium spreading pressure (ESP). Other monolayer properties often reported include surface pressure-area (1T­

A) isotherms, monolayer resistivities (R), and spreading rates from the solid.

For monolayers of l-octadecanol, the 1T-A isotherms at 20·C show two phase transitions.4

,5 The low-pressure gas to liquid-condensed (G/LC) transition occurs at surface pressures below 1 mN/m and molecular areas between 0.210 and 0.225 nm2/molecule.4,5 The high-pressure tran­sition, referred to as the liquid-condensed to solid (LC/S) transition, occurs near 1T=13 mN/m and 0.195-0.205 nm2/molecule.4

,5 The resistivity measurements of Costin and Barnes indicate that for a pure l-octadecanol mono­layer the large increase in resistivity correlates with the G/LC phase change.6 The spreading rate for l-octadecanol at 20·C is 10-3 cm2/s per cm of perimeter of the solid7 and the ESP at 20 ·C is 35 mN/m.8

During the course of an investigation in which we used the fluorescence from EuH (EDTA) to remotely monitor the temperature oflevitated water drops,9-11 we discovered that when the drop contained dissolved EuH (EDTA) the liquid-condensed l-octadecanol monolayer appeared sooner than when the drop contained no dissolved EuH

')NRL/NRC Resident Research Associate.

3268 J. Chern. Phys. 100 (4), 15 February 1994 .

(EDTA). This observation led us to investigate the dy­namics of the G/LC phase change in our levitated drops. In this report, we present detailed experimental evidence showing that the timing of the G/LC phase change in levitated water drops coated with l-octadecanol depends strongly on the amount of dissolved Eu3+ (EDTA) in the drop.

EXPERIMENT

The apparatus, which uses an acoustic standing wave to levitate 0.4-3.0 mm diameter drops in the jet of a wind tunnel, has been described in detail. 12 For these experi­ments, the gas jet is dry nitrogen to which water vapor has been added to produce 20%-30% relative humidity. The gas stream temperature is 20.4 ·C. Production and handling of the solutions have also been described previ­ously. Stock solutions are 3.7x 10-4 M l-octadecanol in 100% ethanol and 2.0X 10-4 M EuH (EDTA), as EuCl3 and H4 EDTA, in lS.2 MO cm resistivity water. The pH of the EuH (EDTA) stock solution is 3.4. Most working mixtures are made by adding 25 ml of ultrapure water or EuH (EDTA) solution to 1.0 ml of the l-octadecanol solution. This mixture produces working solutions that are 0.67 M in ethanol and l.4X 10-5 Min l-octadecanol. Oc­casionally we add the EuH (EDTA) solution to 2.0 ml of alcohol mixture to produce solutions that are 1.29 M in ethanol and 2.7X 10-5 M in l-octadecanol. The EuH

(EDTA) concentration is varied by diluting the stock so­lution with ultrapure water prior to its addition to the alcohol mixture. All working solutions for a variety of EuH (EDTA) concentrations are mixed at the beginning of each day. A syringe is used to draw 50-100 pJ of the working mixture from its flask and place a single (-SILl) drop in the levitator. For some of the working solutions, the monolayers show a reduced final resistivity which we interpret as indicating contamination. In such cases, we

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

Frost, Seaver, and Rubel: Octadecanol films on levitated waterdrops 3269

7.0 0.0 •• l~:"".~Y"~ . -1.0

6.0 -2.0

Oi 5.0 ~ -3.0 .§. .=!o . on .

-4.0 :g .. E e 4.0 'C

-5.0

3.0 .'. -6.0 ~ . .. . " -

o 200 400 600 800 1000 1200 1400

time (8)

FIG.!. Mass and evaporationt:ate (dm/dt) data for a water drop ini­tially containing 0.67 Methanol, 2X 10-5 M Eu3+ (EDTA), and 2.7 X 10-5 M l-octadecanol. The l-octadecanol eventually coats the drop surface and we see the gas to liquid-condensed phase change near 800 s.

discard the solution, clean the glassware, and remix the working solution until the median value of the final resis­tivityexceeds 1.5.

Data -are taken by imaging the backlit drop onto a video camera with a long working distance microscope. The drop images are digitized and analyzed in real time for perimeter and area. This information is converted to sur., face area and volume information by assuming rotational symmetry about the vertical axis through the drop. Mea­surements of known size objects sets the 95% confidence limits for the precision of our volume measurements (aVI V) at or b.elow 1.5X 10-4 when 15 measurements are av­eraged. -

Light-scattering experiments are performed on solu­tions in a glass cell using a 4 mW helium-neon laser. Scat­tered light is-detected 'with a photomultiplier mounted at 90' with respect to the incident laser beam. The incident beam is chopped _at -190 Hz allowing lock-in detection. Solutions tested are (a) pure water; (b) 2.1XIO-s M l-octadecanoL in ib6% ethanol; (c) 4.2X 10-5 M l-octadecanol in 100% ethanol; (d) 6.3 X 10-5 M l-octadecanol in' 100% ethanol; (e) 2.7X 10-5 M l-octadecanof in ultrapure water; (f) identical to (e) ex­cept using 1.2 X 10-'-5 M Eu3+' (EDTA) solution instead of ultrapure water; and (g) same as (e) except using 2.0 X 10-4 M Eu3+ (EDTA) solution.

DROP EVAPORATION

Figure 1 displays the mass (circles) and evaporation rate (squares) data fora drop starting at 2.0X 10-5 M Eu3+ (EDTA). The dmldt data represent the point-by­point differences between the mass data and have been smoothed with a moving 15 point average. At 0.67 M eth­anol, our drop density exceeds 0.993 g/mlY Thus, the mass, which is displayed, equals the volume, which is mea­sured. The evaporation behavior is best illustrated by the dmldt data. These data show a rapid decline in evapora­tion rate during the first 200 s as most of the ethanol evap­orates. The evaporation rate continues to decline but at a

-2.0

~ -4.0 .;:!:

~ ... E -6.0 'C

-8.0

-1 0.0 L-..-,-~-,-........ ....I-....... -,-L....-,-o.....1.""-....... ....I-,-,--,-L...o...-,-.....J

o 200 400 600 800 1000 1200 1400 time (s)

FIG. 2. Evaporation rate comparison between the data in Fig. 1 and our model for the evaporation of a waterdrop containing 0.67 Methanol. Both sets of data indicate that the ethanol effectively disappears within the first 200 s.

slower rate between 200 and 700 s. _ This decline occurs because dmldt is proportional to the drop'radius. Between 700 and 850 s, we see an order of magnitude reduction in evaporation rate associated with the formation of a liquid­condensed phase l-octadecanol monolayer. After 850 s the evaporation rate remains constant because the drop radius decreases only slightly. As we change the concentration of Eu3+ (EDTA) the behavior illustrated in Fig. 1 remains the same. The only difference is the length of time that passes before the liquid-condensed phase appears.

In describing Fig. 1, we said that ethanol evaporation occurs in the first 200 s. Previously, we have seen good agreement between a nonideal solution model for two com­ponent systems and evaporation experiments using I-propanol/wate'r mixtures. 14 We have modified this model to treat ethanol/water mixtures. The results appear in Fig. 2 where we compare our model's prediction for the evap­oration rate of a drop of water initially containing 0.67 M ethanol with the experimental data from Fig. 1. At 200 s, the model predicts an ethanol concentration of 0.024 M and shows that concentrations below this value do not affect our evaporation rate measurements. We do not' be­lieve that any residual ethanol is' affecting Qur other mea­surements because we routinely measure resistivities for the LC monolayer in excess of 2, indicating insignificant ethanol in the monolayer. Also, as the ethanol content vs time will be nearly' constant for all drops, any residual ethanol will not affect the drop to drop comparisons.

To ease comparisons both between drops in our exper­iments and with the results from Langmuir trough work, we follow the lead of previous researchers and define the term critical coverage. 15,16 As a minor distinction, we de­fine critical coverage as occurring at the end of the sudden change in evaporation rate rather than the beginning. In Fig. 1, e.g., critical coverage occurs near 825 s and the mass or volume at critical coverage is 2.65 mg (fLl). Based on the results of Costin and Barnes, we believe critical coverage corresponds to the G/LC phase change. 6

There are two parameters that describe monolayer be-

J. Chern. Phys., Vol. 100, No.4, 15 February 1994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

3270 Frost, Seaver, and Rubel: Octadecanol films on levitated waterdrops

8.0

6.0

e- 4.0

~ a: 2.0

0.0

-2.0 0 400 800

. .

o with Eu(EDTA)

• without Eu(EDTA)

1200

time (s)

1600 2000

FIG. 3. Resistivity vs time data for l-octadecanol on a drop containing 2X 10-4 M Eu3+ (EDTA) and on a drop with no added Eu3+ (EDTA). In both cases the median resistivity exceeds 3.0. The noise in the resistiv­ity values after the phase change is discussed in the text.

havior in terms which depend only on the condition of the liquid/air interface. These parameters are the resistivityl,lo and the accommodation coefficient. ls

,17 We use the resis­tivity description because, for large drops like ours, a sin­gularity occurs in calculating the accommodation coeffi­cient. lo For a water drop surrounded by moving air, the expression describing the evaporation, including mono­layer resistivity, is

(1)

In this equation, m is the mass (g), a is the drop radius (cm), M is the molecular weight, Rg is the gas constant (erg mol- I K - I), P a and P 00 are, respectively, the partial pressure of water at the drop surface and in the flowing gas (dyn/cm2), Ta and Too are the drop temperature and the gas temperature (IC), respectively, Du is the vapor diffu­sivity (cm2/s), lv is the mass ventilation factor, and R is the surface resistivity (s/cm). The ventilation factor takes into account convective enhancement of the vapor phase transport due to the flowing gas surrounding the drop. 18

This factor has been determined to follow the empirical correlation 19

lv=0.78 +0.302(Re) 1/2(SC) 1/3, Re>2.5. (2)

In this equation, Re is the Reynolds number and Sc is the Schmidt number. Previously, we have measured the drop temperature during the monolayer phase change and found that internal thermal adjustments in the drop are rapid relative to the changing evaporation rate throughout the phase change.9 This means that we can calculate T a and therefore Pa from the measured evaporation rate dm/dt. Knowledge of these two parameters allows us to invert Eq. ( 1) and determine the interfacial resistivity R.

We use the median final resistivity of the monolayer to determine which working solutions are acceptably pure. Literature reports show R> 1.5 for l-octadecanol at 25°C and 1T> 2 mN/m.6 Thus, we have required a median final R value in excess of 1.5, with most of our results showing median final R values between 2.2 and 4.0. Typical R val­ues are shown in Fig. 3 for the data in Fig. 1, circles, and for a substrate without Eu3+ (EDTA), diamonds. In both cases the median value of R after the phase change reaches ~ 3.0. The resistivities shown in Fig. 3 are inversely related to dm/ dt which is the difference between subsequent mass measurements. The scatter seen in this figure results from the fact that, after the phase change, dm/ dt is similar in magnitude to our measurement precision, 1.5 X 10-4

• The inverse relationship between Rand dm/ dt skews R toward larger values which is why we use the median rather than the mean to describe the distribution of final R values.

AQUEOUS PHASE CHEMISTRY AND TRANSPORT OF 1-0CTADECANOL

A key difference between our drop experiments and Langmuir trough experiments is that we disperse the sur­factant throughout the substrate and transport it to the surface from the drop's interior. Thus, in order to properly interpret our results, we must understand the chemistry of the l-octadecanol in the aqueous solution and the trans­port mechanismCs) that carry it to the drop surface.

The solubility of l-octadecanol in water at 298 K is reported as 4.0X 10-9 M.20 This is over four orders of magnitude less than our working solution concentrations. The addition of ethanol may improve the solubility some­what, but l-octadecanol particle formation seems likely. To test this hypothesis, we have made light-scattering mea-

TABLE 1. Light-scattering results from l-octadecanol solutions and quantities derived from those results. The derived quantities are for qualitative comparison only. See the text for details.

Solution composition Signal (mY) Radius CA) D(cm2/s) LV (particles/C)

2.1 X 10-5 M l-octadecanol in ethanol 13 5.1 3.5X 10-6 1.3X 1019

4.2X 10-5 M l-octadecanol in ethanol 16 5.1 3.5XlO-6 2.6X1019

6.3 X 10-5 M l-octadecanol in ethanol 18 5.1 3.5XlO-6 3.9X1019

Ultrapure water 3.6 2.7 X 10-5 M l-octadecanoI,

no Eu3+ (EDTA) 60 90 2.0XlO-7 3.1 X lOIS 2.7X 10-5 M l-octadecanol,

1.2 X 10-5 M Eu3+ (EDTA) 90 138 1.3 X 10-7 8.0x1014

2.7X 10-5 M l-octadecanoI, 240 377 4.7X 10-8 4.2x 1013 2.0X 10-4 M Eu3+ (EDTA)

J. Chern. Phys., Vol. 100, No.4, 15 February 1994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

Frost, Seaver, and Rubel: Octadecanol films on levitated waterdrops 3271

surements on a series of bulk solutions similar in compo­sition to our working solutions. Table I shows the results and derivations made from these measurements.

The first three rows in Table I measure the scattering from three concentrations of l-octadecanol dissolved in pure ethanol. In all three cases the l-octadecanol is com­pletely soluble and should be present only as the monomer. We can use these three points to calibrate the light­scattering intensity for l-octadecanol monomers in etha­nol. In the absence of strong solvent-solute interactions, the slope of this calibration line should predict the scatter­ing signal from l-octadecanol monomers in any solvent. A plot of signal strength vs l-octadecanol concentration gives a calibration line whose slope is 1.19 X 105 m V 1M and in­tercept is 10.7 mY. Thus, the expected scattering signal from a 2.7x 10-5 M solution Qf l-octadecanol monomer should be 3.2 m V plus the solvent scattering contribution. The calibration procedure predicts a scattering signal of 6.8 mV from an aqueous solution of l-octadecanol mono­mer at 2.7x 10-5 M rather than the 60 mV observed.

The large increase in signal seen upon addition of l-octadecanol stock solution to ultrapure water demon­strates the existence of suspended l-octadecanol particles. Analysis with respect to head group size and chain length suggests that the l-octadecanol will not form micelles.2I

Thus, we expect the size distribution of l-octadecanol par­ticles to be polydisperse. We do not have sufficient light­scattering data to describe such a size distribution. To fa­cilitate a qualitative analysis of our evaporation data, we will treat these size distributions as if they were monodis­perse and the particles spherical. This treatment enables us to use the light-scattering data to estimate the particle ra­dii, diffusivities, and number densities, given in columns 3-5 for the l-octadecanol in water suspensions.

Rayleigh scattering depends on the number of scatter­ers N and their radius to the fourth power, i.e., scattering signal <xNr4. Therefore, the values derived by assuming a monodisperse particle distribution represent a weighted mean that is strongly biased toward larger radii. The par­ticle radii can best be thought of as an upper limit, while the diffusivities and number densities suggest lower limits. Keep tn mind that the values given in "columns 3-5 of Table I are used for qualitative comparisons only.

We estimate a l-octadecanol molecular radius of 5.1 A. from its molar volume, 333 cm3/mol, assuming a spherical shape. Our calibration procedure predicts 3.2 m V. as the solvent-independent scattering signal from 2.7X 10-5 M l-octadecanol monomer. Given estimates for the monomer radius and scattering signal, we deduce the radius and number density for sphericall-octadecanol particles in ul­trapure water based on the signal strength ratio (60-3.6)/3.2, where the solvent scattering contribution is subtracted from the observed signal. Assuming that the total volume of l-octadecanol is conserved, i.e., NI VI =N2V2 , the signal strength ratio

. (N )1/3 :~::=~= N:

predicts the particle radius and number density. For the

mixtures that include Eu3+ (EDTA) the assumption that the total particle volume is conserved may not be correct. However, the insensitivity of the signal ratio with respect to particle number density means that changes in signal strength will be dominated by changes in radius.

We estimate the diffusivity ( D) of l-octadecanol monomer in water using the Wilke-Chang method as de­scribed in Reid et al.22 From the monomer diffusivity, we can estimate the particle diffusivities using the Stokes­Einstein equation

RT DAB 61M1BrA

and the particle radii given in Table I. Consider the two limiting descriptions for the transfer

of l-octadecanol to a drop's surface from its interior. These descriptions are diffusion of the particles to the surface or "collection" of the particles by the surface as the drop evaporates. Prior to the monolayer phase change, values for the surface area's rate of change, d ( 41T?) I dt, range from 0.8-1.3 X 10-4 cm2 Is. Removing the factor. of 41T gives d?ldt-z 1 X 10-5 cm2/s. This compares with a diffu­sivity of 2.0X 10-7 cm2/s for the 90 A. particles in water. Thus, we can treat the l-octadecanol particles as stationary with respect to changes in the position of the surface due to evaporation. Hence the average concentration of l-octadecanol in the drop interior remains constant throughout the experiment and l-octadecanol particles are scavenged by the moving surface. In this stationary particle description, the number of l-octadecanol molecules on the surface at a given time, Nit), will be determined by the volume evaporated and the initial l-octadecanol concen­tration, Co,

(3)

where NA is Avogadro's number and Vo is the initial drop volume.

To compare our drop data with the Langmuir trough results, we must relate Nit) to the number of molecules needed to form· a LC monolayer, N L. For a surface area

. 2 S(t), we have NL=S(t)IA L , where A L =0.22 nm I molecule is the Langmuir trough result for l-octadecanol at the G/LC phase change. The ratio NsCt)IN L then com­pares the number of molecules on our drop surface with the number of molecules needed to cover that surface with a LC monolayer. We define this ratio as the quantity Ps, the average surface density in units of LC monolayers, although the actual molecular arrangement at the surface is unknown. We obtain

Ps(t) [VO-V(t)]CoN~L

Set)

At critical coverage, Eq. (4) becomes

(4)

(5)

When a drop reaches critical coverage at the surface den­sity of l-octadecanol as predicted by the Langmuir trough

J. Chem. Phys., Vol. 100, No.4, 15 February 1994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

3272 Frost. Seaver. and Rubel: Octadecanol films on levitated waterdrops

3.00

I'!'! 4x1(f6M Eu3

+(EDTA)

2.50 • - 2 x 1 (f 4M EU3 +(EDTA)

• ~ • • .,

2.00 • >-.. 0 • • c 0 ,§, 1.50

" . Co

• • • • • • • 1.00 •

0.50 1.-."'---'--'-....1..--'--''--''--..1-....... ___ ....... ---1_ ....... ''-:--'-.....

o 4 12 16

FIG. 4. Average surface density at critiCal coverage Ps vs initial drop volume for 2.0XIO-F (circles) and 4.0X1O-6 M -(squares) Eu3+ (EDTA) substrates. The 2X1O-4 M solutions show a lack of correlation between the two variables which indicates negligible diffusion of sus­pended l-octadecanol particles in that substrate. The 4.0X 10-6 M solu­Hons show a relationship which results in a 30% increase in Ps as the initial volume rises from 2.0 to 15 p,1.

results, we infer coverage of the surface with a LC mon07 layer, and expect to measure Ps= 1.0. Because the surfac­tant reaches the surface as particles, we might expect to measure Ps> 1 due to the finite spreading rate of l-octadecanol from the solid. This point will be discussed in detail in a later section.

Two complications might invalidate the use of Eqs. ( 4) and (5). These are diffusion of surfactant partiCles to the surface and evaporation of l-octadecanol from the sur­face. We can test for particle diffusion by plotting Ps vs initial drop volume Vi' When diffusion is not significant, Ps will be independent of the initial volume. Figure 4 plots Ps vs initial volume for drops containing either 2 X 10-4 or 4 X 10-6 M Eu3+ (EDT A) . In this figure, we see no cor­relation between Ps and Vi for the high Eu3+ (EDTA) concentration, which demonstrates that diffusion is negli­gible. In contrast, at 4X 10-6 M Eu:l+ (EDTA), ciiffusion is noticeable and we see a 30% increase in Ps across the range of initial volumes. To minimize variations in Ps due to diffusion, we limit the range of our fnitial drop volumes to 7-10 J.L1. _

We can asses the l-octadecanol evaporation rate from a monolayer using the continuum description for evapora­tion from a sphere into a gas containing no l-octadecanol vapor:

dm 41TaD,],)1 Pa dt - Rg Til'

(6)

Solution of Eq. (6f requires knowledge -of the vapor phase diffusivity and partial pressure of l-octadecanol. Ki­netic theory of gases predicts Duo:. ClIm) 112. We estimate Dv by plotting the measured diffusivities for the C1-Cs alcohols23 vs (mass)-I12. This gives Dv=0.02 cm2/s. (Other extrapolation procedures give smaller values for Dv .) We obtain a value for the sublimation vapor pressure of l-octadecanol at 20·C, Pa=3X1O- 8 Torr, byextrapo-

16

I ~.- ' +

12

I '" >-., o. c 0 8.0 ,§,~ "'-

4.0 ,~ I

• .. • .. • ••

0.0 1 o· 7

[Eu(EDTA)] '(moleslliter)

FIG.,S. Average surface density at critical coverage Ps of l-octadecanol vs initial Eu3+ (EDTA) concentration. Based on Langmuir trough re­sults, we expect to see the phase change at ps= l.0. _.

lating the 47-56 ·C vapor pressure measurements of Davies and Kybett.24 With DvandPa estimated, we set a=O,JO em and fu=S and, calculate dm/dt= 1.2 X 108 molecules/so For a,liquid-condensed monolayer, each molecule occupies 0.22 nm2 which gives an area loss rate of2.7XlO-7 cm:/s. By comparison, the smallest Sc seen' in this work is 3XJO-2 crn2

• Thus, errors due to l-octadecanol :evapora­tion during a 3000 s experiment should not exceed 3%.

MONOLAYER FORMATION

Figure Sdisplays the results of it serieS of es "measure­ments on drops starting at different initial Eu + (E])TA) concentrations. In this figure, the solid'line at ps'::::13.8 shows the average surface density when ultrapure water is the substrate. When Eu3+ (EDT A) is added to the sub­strate, we see that Ps> 1 and varies inversely with ion1c concentration. AtEi:L3+ (EDTA) concentrations above -1 X 10-4 M, Ps is unity which means that the l-octadecanol film is behaving as expected based on the Langmuir trough results: At lower Eu3+ (EDTA)conceJi-' trations, the appearance of the liquid-condertsed phase is delayed relative to Langmuir trough predictions. We also see that within experimental error; the result for 4X 10-7

M Eu3+ (EDTA) is indistinguishable from the value mea­sured without added Eu3+ (EDTA). , ,<

As a check on the' purity and behavior of our l-oCtadecanol and substrates, we have measured'1T=,,4 iso­themis Iii a Langmuir trough for l-octadecanol on both the ultrapure water and SX1O-5 M Eu3+ (EDTA) substrates.­The results, bbtained by spreading· pure l-oCtadecanol from a chloroform solution, show the G/LC phase change at 0.227±0.001 nm2/molecUleand 0.S2±0.03 mN/m and the LC to solid phase change at 0.207±0.001 nm2j molecule and 13.3 ±0.2 mN/m for both substrates. These results duplicate the values reported by Harkins and Copeland5 and demonstrate that the effects seen in Fig. S are inherent in the experimental method and not due to the chemicals used.

J. Chern. Phys .• Vol. 100. No, 4. 15 February 1994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

Frost, Seaver, and Rubel: Octadecanol films ori levitated waterdrops 3273

We elm think of several other explanations for the re­sults seen in Fig. 5. The first of these is that in the working solution flask the bulk concentration of l-octadecanol is reduced by diffusion to ihe air/water and glass/water in­terfaces. (Note that we load "our syringe from the interior of the working ~ solution.) This diffusive loss mechanism would be maximized for the smallest l-octadecanol parti­cles;"the pure water mixtu'res.""The addition of EuH

(EDTA),which results in larger particles, will reduce the particle diffusion rate and wouid slow the loss to the inter­faces'restilHng in higher conceritrations of l-octadecanol in the drops. To test this hypothesis, we mixed l-octadecanol stock solution with ultrapure water and with 4 X 10-6 M Eu3+ (EDTA). We measured p~= 11 for the mixture with ultrapure water andps=1.6 for the 4X1O-6 M EuH

(EDTA) mixture. 'Then, we used a clean syringe to re­move 5 ml of mixture from the working solution contain­ing no EuH (EDTA)~'This 5'inlportion is placed in a new clean flask 'arirlEuH (EDTA) stock added to make a second 4X 10-6 M mixture. The measured Ps from this second mixture is 1.5 which agrees well with Ps= 1.6 mea­sured for the first 4 X 10-6 M mixture and is much smaller than p'};= 11 for the mixture without Eu3 +. (EDT A). This result demonstrates that diffusive' losses to the vanous in­te,Jfaces ciuring mixing. and storing of working solutions arid drop injection ar€? pegligible. ~ ,

A s~nd possible explanation for the concentration dependence of Ps is that more than 90% of the l-octadecanol is lost to the glassware and that the EuH

(EDTA) gets incorporated in the monolayer. In this case, the result froma mixture with pure 'Y~terwould represent the equilibrium amount of l-octadecanol left. in the bulk wQrking ~olution .. The apfarent increase in surfactant seen with thea.dg~tion of Eu + (EDTA) would arise from a mixed monolayer containing both l-octadecanol and Eu3+ (EDTA). Tp test for, surface activity of the EuH

(EDTA), we have measured the surface tension of the ultrapure water (72 dyn/cm2) and a 2X 10-4 M EuH

(EDTA) solution (71 dyn/cm2). The small change in sur­face tension shows that the EuH (EDTA) is not intrinsi­cally surface active. Measurements on mixed monolayers, where the second component has a low resistivity, show 10 X reductions in R with the addition of < 5 mol % cho­lesteroes" and a 5 X reduction in R ,with the addition of 10 ' mol % sodium docosyl sulfate.26 We do not expect EuH

(EDTA) to exhibit significant resistivity since measure­ments on the fatty alcohols show that a saturated 16 car­bon backbone is lleeded to produce resistivities that exceed unity.27 Therefore, the fact that we routinely measure R > 2 for all concentrations of EuH (EDTA) indicates that our condensedmonolayers contain little if any EuH (EDTA).

,There are two possible explanations for Ps> 1 that we cannot exclude. These are (1) that the l-octadecanol par­ticles formed in our working mixtures do not spread as rapidly upon reaching the drop surface as the literature prediCts and that the EuH . (EDTA) ions somehow in­crease the spreading rate or (2) on a levitated drop, where there are no walls, Eu3+ (EDTA) in the substrate provides nucleation sites for the change in molecular orientation

16

14

~ 12

!II ... CI) 10 » ~ 0 8

'" 0 E 6 . c.

4

2

0 0

...... ~~.": ................ .. --2.0x10'4M Eu3 +(EDTA) ;'

••••••• no Eu3 +(EDTA) !

....................•... .....

400 800

.' .. , ....

.'

/

1200

time (5)

i I

I

1600 2000

FIG. 6. Average surfaCe density Ps of l-octadecanol vs time for a drop without ,added Eu3+ (EDTA), solid line, and for a drop starting at 2.0 X.1O- 4 M Eu3+ (EDTA).

associated with the G/LC phase transition.28 Since we know that we are forming particles, the first of these two hypotheses seems most likely. However, the second hy­pothesis cannot be ruled out with our current experiments.

The l-octadecanol spreading rate at 20·C has been measured as 10-3 cm2/s per cm perimeter of pure solid at 1T=3 mN/m.7 We can use this value, the l-octadecanol particle size Jnd number density from. the mixture with pure water given in Table I, and our volume and surface measurements to estimate when we should see the liquid­condensed monolayer on a drop. In Fig. 6, we plot Ps vs time for a drop starting with 2.0X 10-4 M EuH (EDTA), solid line, and one without any EuH (EDTA), dotted line. For the drop without EuH(EDTA), we see that one monolayer's worth of l-octadecanol molecules have'been collected by the surface near [=480 s.At this time, the drop volume' has been reduced from· 7.87 to 4.27 ,ul, a change of 3.60 f-Ll. If the surfactant remains in particle form, we can refer to Table I and estimate that 9.6 X 109

particles of l-octadecanol are on the surface. At neutral buoyancy, the surface area occupied by each spherical par­ticle will be 1T? Thus, the particles will occupy 0.027 out of the 0.128 cm2 bf surface area and have a total perimeter of 5.6 X 104 cm. Multiplying the total perimeter times 10-3

gives a spreading rate of 56 cm2/s. Since we only have 0.101 crn2 of surface to 'cover, the liquid-condensed phase should appear instantly (1.8 ms).

For the drop starting at'2.0X 10-4 M EuH (EDTA), we see that one monolayer's worth of l-octadecanol has been collected by the surface near 570 s. At this time 3.53 f-Llhas evaporated. This predicts 1.45X 108 particles on the surface .. These particles cover 6.5 X 1O-3cm2 out of 0.138 cm2 total surface area. The total perimeter of these parti­cles is 3.5 X 103 cm for a spreading rate of 3.5 cm2/s. As in the previous case, this would result in the instantaneous appearance of the LC phase monolayer. Unlike the drop without added Eu3+. (EDTA), we in fact see the phase change at this time.

These calculations rely on assuming ·-that the

J. Chern. Phys., Vol. 100, No.4, 15 February 1994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

3274 Frost, Seaver, and Rubel: Octadecanol films on levitated waterdrops

8

~ 6

~ .2 4 g 0-2 2 c

o

_2~~~~~-L-L-L~~~~L-~~~

o 500 1000 1500

time (8)

22

18 ~~ E

14 co ~ co

10 ., Q

~ 6 " !II

0-2

2 c

-2

FIG. 7. Plot of drop volume (filled circles) and surface area (crosses) vs time. For clarity, every tenth surface area point is shown. The addition of l-octadecanoVethanol solution is apparent as the break in the curves near the upper left. The two "missed" data points are seen on the volume curve at lower left. The change in slope after 1000 s corresponds to formation of the solid phase, which greatly reduces evaporation. This drop yielded a p~ value of 1.14 and a final resistivity to evaporation of 3.5.

l-octadecanol particles are neutrally buoyant and mono­disperse. Neither assumption is likely to be true. However, any reduction in total perimeter for particles that are not neutrally buoyant will be more than compensated for by the increased total perimeter associated with the more nu­merous but smaller radius particles expected in our poly­disperse distribution.

Although qualitative, the preceding analysis clearly il­lustrates that if the l-octadecanol on the drop surface re­mains as particles, the spreading rate in the absence of Eu3+ (EDTA) is nowhere near the rate reported in the literature while the literature rate may apply to particles formed in 2x 10-4 M Eu3+ (EDTA) solutions. In this scenario, the explanation for the data shown in Fig. S is that when l-octadecanol particles are formed in ultrapure water, they spread much more slowly than when they are formed in water containing dissolved Eu3+ (EDTA). In addition, the spreading rate increases with Eu3+ (EDTA) concentration.

An alternate explanation is that thel-octadecanol par­ticles spread as rapidly as expected regardless of the Eu3+ (EDTA) concentration but the LC phase change is de­layed. In this picture, when Ps exceeds unity we expect the surface to be composed of l-octadecanol particles sur­rounded by a monolayer of unspecified phase. If the parti­cles and the monolayer are in equilibrium the surface pres­sure is 3S mN/m. This pressure is well above the transition pressure for either the GILC ( < 1 mN/m) or LC/S (13 mN/m) phase change. Thus, even though our surface film, composed of particles and monolayer, is unlikely to be at equilibrium, the monolayer portion of this film should be above the density at which the G/LC phase change is seen in the Langmuir trough. To explain the observed delays in the resistivity increase, we suggest that in the absence of walls or barriers, ions in the substrate are needed to nucle­ate the change in molecular orientation seen between the gas and liquid-condensed phases of the monolayer.

Nucleation theory describes a phase change in terms of the flux of "germs" making the transition from critical to above critical germ size, with critical size determined by thermodynamic considerations. In this picture, we account for the low concentration threshold behavior by assuming that the ultrapure water contains approximately 10-7 M dissolved ions or other nucleation sites. At high ion con­centrations, there is an optimum number of nucleation sites (>10-7 ) for germ formation such that the phase change appears as soon as the surfactant density reaches 22 A 2/moleeule. When the nucleation sites are fewer than this optimum number, the phase change will be delayed as time is required for additional germs to exceed"cdtical size and increase the number of growing nuclei. Thus, the length of time needed to see the phase change will increase as the number of available nucleation sites decreases. Experimtm­tally, such a delay would be seen as an increase in the average l-octadecanol surface density at critical coverage. The data in Fig. 6 indicate that in our experiments these delays would range from 0-1000 s.

CONCLUSIONS

We have monitored the drop evaporation rate to de­termine - when the liquid-condensed phase of a l-octadecanol thin film appears on levitated wat~r drops. Our experimental technique produces small particles of l-octadecanol, ,dispersed throughout the drops. These par­ticles are collated by the shrinking surface as the drop evaporates into a moving stream of air. We find that for pure water or low concentration Eu3+ (EDTA) substrates the G/LC phase change is delayed with respect to predic­tions based on Langmuir trough results. At Eu3+ (EDTA) concentrations above 1.0 X 10-4 M our results agree with the Langmuir trough results. In terms of the average sur­face density at critical coverage, p~, of the l-octadecanol, we find values' of p~ that range from 14 monolayers on ultrapure water to 1.0 monolayer when the Eu3+ (EDTA) concentration exceeds ~ I X 10-4 M. These average sur­face densities correspond to delays in the G/LC phase change of 0-1000 s.

We propose two possible explanations for our observa­tions. The first explanation is that the suspended 1 :octadecanol particles created by our experimental proce­dure have a spreading rate much reduced relative to the literature value and that the presence of Eu3+ (EDTA) in the water when these particles are formed increases the spreading rate. The second explanation assumes that the particles spread as described in the literature, but that in the absence of walls, the change in molecular orientation associated with the GILC phase change requires nucle­ation by dissolved ions in the substrate. Currently, we are testing the effects of different ions and pH on this behavior.

Note added in proof

We have succeeded in further clarifying both when the delayed phase change occurs and the underlying mecha­nism. Experiments that vary the pH and the species of dissolved ion showed that delays in the phase change are

J. Chern. Phys., Vol. 100, No.4, 15 February 1994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18

Frost, Seaver, and Rubel: Octadecanol films on levitated waterdrops 3275

most strongly influenced by the pH of the substrate, with basic solutions showing long delays at all ion concentra­tions.29 As part of the pH studies, we also calibrated the surface tension after the phase change and determined that the high resistivities coincide with surface tensions appro­priate for solid phase l-octadecanol monolayers. We have learned how to add surfactant dissolved in 100% ethanol to levitated drops. This change in technique allows us to duplicate the spreading procedure used in a Langmuir trough. Application of this new technique to ultrapure wa­ter, pH=5.6, and to 2X 10-4 M EuH (EDTA), pH=3.4, provides results which allow us to distinguish between the two hypotheses proposed in the results section of this re­port.

Figure 7 shows the data from one experiment where we added l-octadecanol dissolved in ethanol to a levitated drop of ultrapure water. In this figure, the crosses show the drop surface area (only every tenth data point for clarity) and the solid circles the drop volume (all points). What we see is a drop of water evaporating for about 100 s at which point we add the dissolved surfactant and "miss" two data points. The 2-4 size measurements missed during the ad­dition procedure compel us to extrapolate to find the added volume of l-octadecanol. We estimate the added volume by fitting the volume data for the -100 s preceding and following the addition with third-order polynomials and extrapolating both to the time of the addition. The differ­ence is the volume of ethanol solution added, from which the number of surfactant molecules (N) on the drop is calculated. At critical coverage we obtain Nc' by dividing the measured surface area by the area per surfactant mol­ecule (0.20 nm2/molecule for Sphase l-octadecanol). The ratio N / Nc is equal to Ps·

In six measurements with drops having an ultrapure­water substrate, we measured p~= 1.07 ±0.23, and in six measurements of drops with substrates of 2 X 10-4 M Eu (EDTA), we obtained ps=0.90±O.15. The uncertainties given are the standard deviations in the groups of six mea­surements. We attribute this variation to the extrapolation procedqre used to estimate the added volume.

The p~ values for pure water and the 2.0X 10-4 M EuH (EDTA) substrates are indistinguishable within ex­perimental error. This result is in striking contrast of the order-of-magnitude difference seen in Fig. 5. This change in procedure should not affect the nucleation of one surface phase from another. Thus this new result demonstrates that a switch from heterogeneous to homogeneous nucle­ation does not account for the delayed appearance of re-

sistive monolayers on a pure water substrate. Therefore, the kinetic differences observed in Fig. 5 and in Ref. 29 must result from the effect of ions and pH on the spreading rate of l-octadecanol particles formed in aqueous solution.

ACKNOWLEDGMENT

We wish to thank W. R. Barger for making the 'IT-A isotherm measurements.

I A. W. Adamson, Physical Chemistry of Surfaces, 5th ed. (Wiley­Interscience, New York, 1990).

2M. J. Jaycock and G. D. Parfitt, Chemistry of Interfaces (Ellis Hor­wood, Chichester, 1981).

3G. L. Gaines, Insoluble Monolayers at Liquid-Gas Interfaces (WiIey­Interscience, New York, 1966).

4G. T. Barnes and D. S. Hunter, J. Colloid Interface Sci. 136, 198 (1990).

5W. D. Harkins and L. E. Copeland, J. Chem. Phys. 10,272 (1942). 61. S. Costin and G. T. Barnes, J. Colloid Interface Sci. 25, 584 (1967). 7 J. H. Brooks, in Retardation of Evaporation by Monolayers, edited by V. K. LaMer (Academic, New York, 1962), p. 251.

8 J. H. Brooks, in Retardation of Evaporation by Monolayers, edited by V. K. LaMer (Academic, New York, 1962), p. 259.

9 EDT A is ethylenediaminetetraacetic acid. 10M. Seaver, J. R. Peele, T. J. Manuccia, G. O. Rubel, and G. Ritchie, J.

Phys. Chem. 96, 6389 (1992). 11M. Seaver and J. R. Peele, Appl. Opt. 29, 4956 (1990). 12M. Seaver, A. Galloway, and T. J. Manuccia, Rev. Sci. Instrum. 60,

3452 (1989). 13 J. Timmermans, The Physico-Chemical Constants of Binary Systems in

Concentrated Solutions (Wiley-Interscience, New York, 1960), Vol. 4. 14M. Seaver and A. Barrett, J. Appl. Meterol. (in press). ISG. O. Rubel and J. W. Gentry, J. Phys. Chem. 88, 3142 (1984). 16B. V. DeIjaguin, V. A. Fedoseyev, and L. A. Rosenzweig, J. Colloid

Interface Sci. 22, 45 (1966). I1N. A. Fuchs, Evaporation and Droplet Growth in Gaseous Media (Per­

gamon, New York, 1959). ISH. R. Pruppacher and J. D. Klett, Microphysics o/Clouds and Precip-

itation (Reidel, Dordrecht, 1980). 19K. V. Beard and H. R. Pruppacher, J. Atmos. Sci. 28, 1455 (1971). 20M. H. Abraham, J. Chem. Soc. Faraday Trans. 180, 153 (1984). 21 J. N. IsraelachviIi, in Physics of Amphiphiles: Micelles, . Vesicles, and

Microemulsions, edited by V. De Giorgio and M. Corti (North­Holland, New York, 1985), p. 24.

2lR. C. Reid, J. M. Prausnitz, and B. E. Poling, The Properties of Gases and Liquids, 4th ed. (McGraw-HilI, New York, 1987), p. 598.

23G. A. Lugg, Anal. Chem. 7, 1072 (1968). 24M. Davies and B. Kybett, Trans. Faraday Soc. 61, 1608 (165). 2SG. T. Barnes, K. J. Bacon, and J. M. Ash, J. Colloid Interface Sci. 76,

263 (1980). 261. S. Costin and G. T. Barnes, J. Colloid Interface Sci. 51, 122 (1975). 27V. K. LaMer, T. W. Healy, and L. A. G. Aylmore, J. Colloid Interface.

Sci. 19, 673 (1964). 28C. M. Knobler and R. C. Desai, Annu. Rev. Phys. Chem. 43,

207 (1992). 29 A. E. Frost and Mark Seaver, Thin Solid Films (in press).

J. Chem. Phys., Vol. 1 ~O, No.4, 15 February 1994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

129.105.215.146 On: Thu, 18 Dec 2014 19:53:18