- Email: [email protected]
Characterisation of liquid film in a microstructured falling film reactor using laser scanning confocal microscopy
Experimental Thermal and Fluid Science 30 (2006) 463–472 www.elsevier.com/locate/etfs
Characterisation of liquid film in a microstructured falling film reactor using laser scanning confocal microscopy Kay Kin Yeong
a,1
, Asterios Gavriilidis a,*, Ralf Zapf b, Hans-Joachim Kost b, Volker Hessel b,c, Alan Boyde d
a
Department of Chemical Engineering, University College London, London WC1E 7JE, UK Institut fu¨r Mikrotechnik Mainz GmbH, Carl-Zeiss-Str. 18-20, D-55129 Mainz, Germany c Department of Chemical Engineering and Chemistry, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands d Biophysics OGD, Dental Institute, Barts & The London School of Medicine & Dentistry, Queen Mary, University of London, London E1 1BB, UK b
Received 15 October 2004; accepted 23 September 2005
Abstract Reflection confocal microscopy was used to measure the thicknesses of falling thin films generated using microstructured plates. Each plate contained straight parallel channels through which liquid fell vertically under gravity. Three different channel dimensions were used: 300 · 100 lm, 600 · 200 lm and 1200 · 400 lm (width · depth). Measurements were performed for acetone, ethanol and isopropanol and gave reasonable agreement with theoretical predictions and correlations from the literature. Discrepancies were attributed to the fact that these equations do not consider parameters that affect three-phase contact lines such as surface tension and contact angle. No trend of film thickness on channel width was observed and was probably due to different surface heterogeneity and roughness characteristics amongst the plates. Measurement difficulties were encountered at low liquid film thicknesses and where the angle of the liquid meniscus was too steep. 2005 Elsevier Inc. All rights reserved. Keywords: Microstructured reactor; Falling film; Liquid film thickness measurement
1. Introduction Falling film reactors generate thin liquid films (typically in contact with either a flat wall or stack of pipes) that fall under gravitational pull. The thinness of these films results in rapid heat and mass transfer. As such, these systems find application in extraction, evaporation and highly exothermic processes [1–8]. Conventional falling film systems on tubes or flat surfaces generate films with thicknesses in the order of 0.5–3 mm [9,10]. A microstructured falling film reactor (l-FFR) developed by the Institut fu¨r Mikrotechnik Mainz can generate stable films less than 100 lm thick. This reactor offers excellent heat removal capabilities and *
Corresponding author. Tel.: +44 20 76793811; fax: +44 20 73832348. E-mail address: [email protected] (A. Gavriilidis). 1 Present address: Xennia Technology, Lumen House, Lumen Road, Royston, Hertfordshire SG8 7AG, UK. 0894-1777/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.expthermflusci.2005.09.006
has been used safely in the direct fluorination of aromatics [11,12]. Our research aims to evaluate the performance of the l-FFR in solid catalysed gas–liquid hydrogenations. The development of suitable catalyst incorporation methods (e.g. via alumina coatings) and reaction studies have been reported elsewhere [13,14]. This paper presents investigations of liquid film thickness and profile measurements. A variety of methods have been employed to measure the thickness of liquid films. Examples include conductivity probes [15–17], ultrasound detectors [18], fibre-optic sensors [19], microwave techniques [20,21], laser interferometry [22,23], fluorescence [24], laser-induced fluorescence [25,26], optical measurement methods [27–29], rainbow refractometry [30] and infrared thermography [31]. The use of conductivity probes was not considered for this investigation because they require physical contact with the liquid film which would greatly affect flow characteristics due to the small flow channels and liquid film
464
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
Nomenclature g KF Re d C h
gravitational acceleration l4 g qr3
Reynolds number ¼ Cl liquid film thickness mass flowrate per wetted perimeter inclination angle from the horizontal
thicknesses encountered. Some of the optical methods listed above were also unsuited for the system being investigated (e.g. required translucent substrates, circular surfaces). To add to the arsenal of optical-based methods, reflection confocal laser scanning microscopy (CLSM) was employed to study this system. CLSM is a popular imaging and analysis technique that has been widely used in many areas, e.g. biology (mostly in fluorescent mode [32–35]), semiconductor materials science (mostly in reflection mode [33]) and polymer science [36–39]. The key feature of CLSM is that it has a very short depth-of-field. In conventional optical microscopes, the entire object within the lensÕ field of view is seen; areas at the focal point of the lens appear sharp while those towards or beyond the focal point are blurry. In CLSM, however, due to its limited depth-offield, regions which are removed from the focal point are not seen at all due to the confocal pinhole which restricts light from levels other than the focal region from reaching the detector. The effect is similar to taking a slice through an object. By varying the depth at which the microscope is focused and scanning the beam across each layer, a succession of slices can be taken, which can then be summed to give a sharp, high resolution image of the object in question. Because CLSM collects data point-by-point on each slice, a very detailed map of a three-dimensional object can be obtained, meaning that depth discrimination and optical tomography can be performed to high accuracy [33,34]. 2. Experimental apparatus and procedure 2.1. Microstructured falling film reactor The main part of the reactor is the structured stainless steel plate for the generation of the falling film. Each plate contained straight parallel channels 78 mm long fabricated using isotropic wet etching. More details about the reactor can be found elsewhere [11–14]. Three different channel configurations were used, as shown in Table 1. In addition, a c-alumina coated version of each plate was used as well (details of the coating procedure are given in [13] while pictures of cross sections of the coated channels are given in [40]). Slots were cut through the plates at both ends of the channels to form the liquid inlet and exit ports. Each
l m m0 q qc r
dynamic viscosity kinematic viscosity m 0:6106
liquid density density of adjoining phase surface tension
Table 1 Channel dimensions of the microstructured plates Plate
Channel depth (lm)
Channel width (lm)
Number of channels
Type I Type II Type III
100 200 400
300 600 1200
64 32 16
plate measured 89 · 46 mm and was housed in a stainless steel enclosure. Liquid was fed to and removed from the back of the plate. A stainless steel gasket was placed on top of the plate. Its geometry was such that the whole length of the channels were exposed with the exception of the first and last 5 mm, which helped to distribute the liquid across the plate. The channels were exposed to ambient air. 2.2. Confocal microscope The experiments were performed using a Noran Odyssey video-rate laser point scanning confocal microscope, in reflection mode with a 633 nm laser. A Nikon Mplan extra large working distance (20/0.4 lens, i.e. magnification = 20·, numerical aperture = 0.4) was used. To allow the examination of the l-FFR in a vertical orientation, the objective had to be attached to the microscope via a 90 bend (incorporating a 45 mirror) such that this lens projected outward horizontally. Because of this, the focus axis of the microscope was changed to the horizontal. As such, a microscope stage had to be constructed to allow focussing movement in the horizontal plane. It was especially important that the movement of the stage in the direction parallel to the projection of the lens could be measured accurately, as this now represented the depth at which the microscope was focused. Two screw micrometers were used to provide movement and measurement in the horizontal plane (see Fig. 1). Because of the manual setup of this system, a restricted number of data points could be collected compared to an automated system. This led to the need for some amount of extrapolation in order to interpret the data collected. The manual data collection employed in this work resulted in experimental error of ±4 lm while confocal microscopy with high numerical aperture objectives may have sub-micron depth and lateral resolution [33].
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
465
the channels, this made the surface of the liquid film identifiable, except when: (a) the liquid film was very thin. When the film surface was very close to the bottom of the channel, reflection from the channel surface was brighter than the gas/ liquid interface and it became very difficult to identify the location of the film surface, (b) the angle of the film/channel surface was too steep. Only light reflected back along the incoming path is detected. This does not present a problem when
Fig. 1. Microscope stage with focusing mechanism (to and away from the laser beam) controlled using a screw micrometer (for depth measurements), and a secondary horizontal motion mechanism perpendicular to the laser beam.
2.3. Experimental procedure A syringe pump (Razel A-99) was used to feed the test liquid to the reactor. Three liquids of differing viscosity were employed: ethanol, isopropanol and acetone. Measurements were taken at three channels located at the left, central and right parts of the reactor, 2.1 cm below the liquid inlet slot, and then averaged. This height was chosen as a compromise between minimising evaporation and entrance effects. Exploratory experiments showed that evaporation was negligible at this distance from the entrance. Some entrance effects were found, e.g. the film was a few microns thicker 0.5 cm below this height, but there were difficulties in performing measurements at a lower location due to physical obstruction by parts of the microscope. 3. Preliminary results The difference between the refractive indices of air and liquid results in reflection occurring at the gas–liquid interface [41]. In addition to light reflected from the bottom of
Fig. 2. Confocal microscopy images of a microchannel filled with ethanol at different focal depths (a)–(c) are images of a stainless steel channel, (d) is an image of a c-alumina coated channel: (a) top of channel walls, (b) on fluid at the centre of the channel (i.e. the lowest point of the liquid surface), (c) bottom of channel, (d) bottom of c-alumina coated channel (300 · 100 lm channel).
466
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
observing surfaces which are located normal to the incidence path. However, when the surface being observed is at an angle, the returning light will fail to enter the working aperture of the objective lens. Hence, image intensity would be significantly reduced, if anything can be seen at all [42]. Fig. 2 shows a series of images taken using the CLSM. Fig. 2a shows the microscope focussed on the top of the stainless steel channel walls, making that area much brighter than all surrounding regions which are at a different depth. The surface of the liquid film can also be dimly seen in the middle of the figure. In true confocal mode, the liquid surface should not be visible because it is at a different depth, but here we have a rather low numerical aperture objective and the focal-level discrimination is consequently limited. Focusing deeper into the channel locates the deepest point of the liquid surface, which corresponds to the centre of the channel width-wise (see Fig. 2b). Moving below the liquid surface eventually focuses the microscope on the channel bottom, shown in Fig. 2c. Fig. 2d shows a c-alumina coated channel for comparison. The surface of the liquid was less bright than the channel surface (with the stainless steel channels brighter than the
alumina-coated channels), which therefore required greater image contrast in order to be seen. Fringes were observed on the liquid surface due to interference with light reflected from the channel surface. These fringes were in constant motion, which suggests that the liquid film was perturbed. These disturbances may be an indication of the presence of hydrodynamic instabilities [43,44]. 4. Channel profile measurements Comparisons were made between the manuallyobtained data from the Noran Odyssey CLSM and measurements made using an automated laser-scanning surface profilometer (NanoFocus). This is an autofocusing laser system using light reflection to measure surface profiles. The laser source is moved to find the maximum laser intensity and in this way images the surface profile. The latter instrument took measurements every 0.5 lm across the width of the channels. The results are shown in Fig. 3. The measurements were in good agreement for the 300 · 100 and 600 · 100 lm channels, but the confocal system depicted a shallower, more rounded surface than the profilometer for the 1200 · 400 lm channel. This was most likely due to channel size variations during manufacturing.
120 Laser-scanning surface profilometry Confocal microscopy
Y (μm)
80
40
0 0
100
200
(a)
300
X (μm) 250 Laser-scanning surface profilometry Confocal microscopy
Y (μm)
200 150 100 50 0 0
200
400
(b)
600
800
X (μm)
Y (μm)
400 Laser-scanning surface profilometry Confocal microscopy
300 200 100 0 0
(c)
500
1000
1500
X (μm)
Fig. 3. Comparison of channel cross-sectional profiles using confocal microscopy and laser-scanning surface profilometry: (a) 300 · 100 lm channel, (b) 600 · 200 lm channel, (c) 1200 · 400 lm channel.
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
5. Liquid film surface profile measurements
Channel height/liquid film thickness (μm)
Liquid flowing down a flat plate can take several forms. At large flowrates, it takes the form of continuous sheets (which may demonstrate waviness). At either the front end of those sheets or at low flowrates, flow breaks up into rivulets or fingers [45,46]. The purpose of the microchannels in the l-FFR is to prevent this break-up of the liquid film at very low flowrates. Due to the combination of capillary forces and small channel widths, liquid is pulled up along the sides of the channels and takes up a significant portion of the channel width. The surface of the liquid film takes on the form of a flowing meniscus. The profile of the film surface was measured at various flowrates in each of the l-FFR plates detailed earlier, and a selection is shown in Figs. 4 and 5. It should be noted that the measurements by which the plates are referred to (e.g. 300 · 100 lm) are
Channel height/liquid film thickness (μm)
merely convenient descriptors for those plates. The actual plate dimensions and geometry can be found in Fig. 3. For the majority of the measurements, the film profiles are shown as being incomplete. As mentioned in the previous section, this was due to measurement difficulties caused by the curvature of the liquid surface. Also, in Fig. 4a, the surface profile for the lowest flowrate is depicted as flat. This was due to errors in determining the true liquid profile for very thin films over the highly reflective metal surface of the channel. In general, however, the surface of the film curves upwards moving from the centre of the channel to the walls. As the flowrate increases, the film thickness also increases and the surface profile becomes flatter. When the channel is completely filled the profile becomes flat (Fig. 4a, flowrate 84.9 ml/h). Further increase in the flowrate causes the liquid to bulge out of the channel (Fig. 4a, flowrate 113 ml/h). When the hydrodynamic pressure exceeds
9.90 ml/h
120 15.6 ml/h
100
24.0 ml/h
80
33.2 ml/h 42.4 ml/h
60 48.1 ml/h
40
62.9 ml/h
20
84.9 ml/h 113 ml/h
0 0
50
100
(a)
Channel height/liquid film thickness (μm)
467
150 Channel width (μm)
200
250
300
200
channel
9.92 ml/h 13.9 ml/h
150 19.4 ml/h 25.0 ml/h
100
36.5 ml/h
50
51.6 ml/h channel
0
0
100
200
(b)
300
400
500
600
Channel width (μm)
350 300 30.4 ml/h
250
44.6 ml/h
200
60.1 ml/h
150
97.0 ml/h
77.8 ml/h 111 ml/h
100
channel
50 0
(c)
0
200
400
600
800
1000
1200
Channel width (μm)
Fig. 4. Cross-sectional profile of the liquid film surface at various flowrates in different microchannels: (a) 300 · 100 lm channel, (b) 600 · 200 lm channel, (c) 1200 · 400 lm channel (liquid ethanol). Fig. 4a reprinted with permission from Ind. Eng. Chem. Res. Copyright 2005 American Chemical Society.
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472 Channel height/liquid film thickness (μm)
468
100
5.95 ml/h 7.54 ml/h
80
9.12 ml/h
60
16.3 ml/h
40
22.2 ml/h
20
32.1 ml/h 43.6 ml/h
0 0
50
100
150
200
250
300
alumina
-20
channel
Channel width (μm)
Fig. 5. Cross-sectional profile of the liquid film surface at various flowrates in the 300 · 100 lm alumina coated plate. ‘‘alumina’’ indicates the liquid– alumina interface, while ‘‘channel’’ indicates the alumina–stainless steel interface (liquid ethanol).
Channel height/liquid film thickness (μm)
surface tension forces the liquid bursts out of its channel wall confines. Such behaviour has been suggested by Adam [47] in the context of liquid penetration in pores with constrictions. Namely, as the liquid penetrates a pore with complex shape, the meniscus gets pinned when it reaches a constriction, resisting increasing pressure difference by changing its curvature. This is equivalent to a change of the apparent contact angle from its intrinsic value x, to x + /Wall, where /Wall is the solid wall inclination change at the pore mouth. In our case, the pore mouth corresponds to the microchannel edges. This phenomenon has been recently examined theoretically by minimisation of liquid free energies for open microchannels [48]. Comparing the film profiles of coated and uncoated channels (see Fig. 6), it can be seen that the liquid surface is located higher up in the coated channels at a given flowrate. Note that in Fig. 6 the baseline reference is the top of the channel. This upward displacement of the liquid surface is noticeably greater in the middle of the channel, and has an average value of 17 lm. In comparison, the thickness of the alumina layer was estimated to be 10 lm. This displacement can be rationalised as follows. The curvature and location of the interface is determined (a) by the balance between the Laplace pressure and the interfacial tension according to the well known Laplace equation [49], where the key parameter is the contact angle, (b) by the location of the three-phase contact line and (c) by the ability of meniscus to contort if the three-phase contact line
is pinned at the channel edges [47]. Depending on the contact angle, the (cross sectional) shape of the channel and the liquid volumetric flowrate, the three-phase contact lines can reside at the sides or at the edges of the channel. In Fig. 6, where the three-phase contact lines seem to be pinned at the channel edges, the difference between the two interface profiles can be attributed to different channel depth/shape, different contact angles at stainless steel and alumina and meniscus contortion. The fact that the contact angle also depends on surface roughness and heterogeneity [50] can play a role, particularly at lower flowrates where the three-phase contact line may be located at the channel sides and not edges. It had been expected that the profiles of the film surface would have been circular arcs as the capillary (Ca) and Weber (We) numbers are very small (<103) for this system at the range of flowrates studied, indicating that the liquid surface should not be deformed due to flow [51]. This is true for the lower-to-mid flowrates in the 300 · 100 lm channels (both coated and uncoated). However, at larger flowrates in those channels and at all flowrates in the larger channels, the profiles were found to be distorted and best approximated as superellipses. This finding could be the result of experimental error (due to difficulty in accurately detecting the curved liquid surface and disturbances in the liquid flowrate) or contortion of the three-phase contact line at its pinning location due to surface roughness and heterogeneity [52].
100 80 43.6 ml/h (coated)
60 42.4 ml/h (uncoated)
40
alumina
20
channel
0
0
50
100
150
200
250
300
-20 Channel width (μm)
Fig. 6. Comparison of cross-sectional liquid film surface profiles in the 300 · 100 lm coated and uncoated channels. ‘‘alumina’’ indicates the surface of the alumina-coated channel, while ‘‘channel’’ indicates the surface of the uncoated channel (liquid ethanol).
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
469
2
1.8
Feind correlation Nusselt equation Kapitza equation 300x100 plate
1.6 log δ
300x100, Al2O3 plate 600x200 plate 600x200, Al2O3 plate
1.4
1200x400 plate 1200x400, Al2O3 plate
1.2
1 -1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
log Re
Fig. 7. Relationship between liquid film thickness, d, and Reynolds number, Re, obtained experimentally (points) and theoretically (lines) (liquid ethanol).
6. Liquid film thickness Due to the fact that neither channel bottom nor liquid surface were flat (and also possessing different curvatures), the average liquid film thickness was defined based on the cross-sectional area of the liquid film profile divided by the wetted channel perimeter. The experimental film thickness as a function of Reynolds number is shown in Fig. 7. The figure also shows predictions of film thickness using the following equations and correlations [53]: 2 1=3 1=3 3m Re q Nusselt : d ¼ g sin h q qc 1=3 1=3 2:4m2 Re q Kapitza : d ¼ g sin h q qc 1=3 1=3 2 3m q 00:11 00:025 Feind : d ¼ Re0:333m m q qc g sin h for 0:72K F0:1 < Re < 1:35K F0:1 Since the system was a vertical falling film, h = 90. The denq sity factor qq was ignored as q qc (note: the Re for the c flowrates studied were 0.04–0.86, which was much smaller than the range of the Feind equation criterion as K F0:1 ¼ 8:2). Since the calculation of d and Re is based on the wetted perimeter, it is sensitive to the location of the three-phase contact line. It must be emphasised that these equations do not account for surface tension and contact angle, as they have been developed for two-dimensional films. The results follow the trend of the predictions, but are scattered over a fairly broad range. The Kapitza equation appears to give the closest overall prediction. In general, the liquid film thicknesses measured for the alumina-coated channels were greater than those for the uncoated channels. This behaviour may be due to the fact that the porous nature of the alumina coating may have drawn the liquid film higher up the channel wall as well as to the different contact angle for the two materials. A common feature of the data is that each set exhibits a sudden drop of the film thickness at lower Re. This is most likely due to experimen-
tal error when measuring very thin films (below ca. 30 lm) and identifying the points of contact between the liquid and the channel walls, as detailed previously. In general, the 1200 · 400 lm plate results in the smallest film thicknesses and the 600 · 200 lm plate in the thickest films for the uncoated plates, while the 300 · 100 lm plate results in the smallest film thicknesses and the 600 · 200 lm plate in the thickest films for the alumina-coated plates. The disparity of the experimental results amongst different plates and with the theoretical predictions is not surprising, if one considers that the equations do not account for three-phase contact and that the hydrodynamics near the point of three-phase contact are not well understood. The classical Young equation which relates contact angle to the various interfacial tensions is applicable to smooth, homogeneous, isotropic solid surfaces. However, apparent contact angle depends further on the so-called line tension, whose effect is magnified by surface roughness and hetereogeneity by additional corrugation of the three-phase contact line. Surface roughness and hetereogeneity affect further the apparent contact angle by increasing the real solid surface and hence its total interfacial energy, according to the Wenzel and Cassie equations [54]. Moreover, the effect of roughness can be even stronger under dynamic conditions [55]. Hence, the absence of a specific trend of film thickness vs. channel size in Fig. 7 may be attributed to surface heterogeneity and roughness. 7. Comparison of different liquids In order to determine the behaviour of different fluids, experiments were also performed using isopropanol and Table 2 Physical properties of acetone, ethanol and isopropanol
Acetone Ethanol Isopropanol
Viscosity (mPa s)
Surface tension (mN/m)
Density (kg/m3)
0.34 1.2 2.3
14.4 21.6 22.7
792 790 790
470
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472 2.2
2 Feind correlation Nusselt equation Kapitza equation
1.8 log δ
300x100 plate 600x200 plate
1.6
1.4
1.2
1 -1.7
-1.5
-1.3
(a)
-1.1
-0.9
-0.7
-0.5
log Re 1.8
Feind correlation
1.6 log δ
Nusselt equation Kapitza equation 300x100 plate
1.4
1.2 -0.2
(b)
-0.1
0
0.1
0.2
0.3
0.4
log Re
Fig. 8. Relationship between liquid film thickness, d, and Reynolds number, Re, obtained experimentally (points) and theoretically (lines): (a) isopropanol, (b) acetone.
acetone (see Table 2 for the physical properties). The relationship between liquid film thickness and Re for isopropanol and acetone can be seen in Fig. 8. The Kapitza equation gives the closest overall prediction for isopropanol while the Nusselt equation or Feind correlation seem to give better predictions for acetone (though the number of data points is not sufficient for conclusive evidence). Comparing the results of the three liquids, the liquid with the highest viscosity (i.e. isopropanol) resulted in the greatest liquid film thickness at a given Re, which is consistent with the theoretical predictions.
the equations do not account for surface tension and contact angles. The greatest discrepancy between the predictions and experiments were found at the lowest flowrates and for the 600 · 200 alumina-coated plate. The lowerthan-expected film thicknesses measured experimentally may be due to difficulties in identifying and locating the surface of the liquid film at film thicknesses less than 30 lm. The cause of the nonmonotonic behaviour of film thickness vs. channel size may be attributed to different surface heterogeneity and roughness characteristics of the various plates. Acknowledgments
8. Conclusions Thin liquid films were generated using microstructured falling film plates with different channel sizes (300 · 100 lm, 600 · 200 lm and 1200 · 400 lm) and surfaces (stainless steel and c-alumina on stainless steel). Film thicknesses were measured using reflection confocal microscopy at a range of flowrates for acetone, ethanol and isopropanol. The Kapitza equation was found to give the best overall predictions for ethanol and isopropanol, while the Nusselt equation and Feind correlation appeared to give better predictions for acetone. The agreement between the equations and the experimental data is good, considering the fact that
The authors would like to thank the European Commission (G1RD-CT2000-00469) for partial financial support. Help and advice provided by Enhong Cao, Carlos Amador, Maureen Arora, Roy Radcliffe and Peter Howell were invaluable to this work. The Odyssey CSLM was funded by the Wellcome Trust and its conversion to red laser light operation by a Royal Society grant. References [1] A.M. Yih, C.C. Kuo, Design and testing of a new type of falling film gas–liquid contacting device, AIChE Journal 34 (1988) 499–501.
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472 [2] E.A. Knaggs, Development of continuous falling film SO3 sulfonation technology, Chemtech (July) (1992) 436–445. [3] S.M. Walas, Chemical reactors, in: R.H. Perry, D.W. Green (Eds.), Chemical EngineerÕs Handbook, seventh ed., McGraw-Hill, 1997 (Chapter 23). [4] M.H.H. Van Dam, J.-P. Corriou, N. Midoux, A.-S. Lamine, C. Roizard, Modelling and measurement of sulfur dioxide absorption rate in a laminar falling film reactor, Chemical Engineering Science 54 (1999) 5311–5318. [5] S. Bessenet, V. Renaudin, J.-P. Leclerc, J.-M. Hornut, RTD studies in a falling film graphite evaporator with internal spiral fins: influence of the geometry, Canadian Journal of Chemical Engineering 78 (2000) 486–494. [6] W. Chen, R.N. Christensen, Inlet subcooling effect on heat and mass transfer characteristics in a laminar film flow, International Journal of Heat and Mass Transfer 43 (2000) 167–177. [7] S. Kulankara, K.E. Herold, Theory of heat/mass transfer additives in absorption chillers, HVAC&R Research 6 (2000) 369–380. [8] A.S. Rahmah, M.M. Elsayed, N.M. Al-Najem, A numerical investigation for the heat and mass transfer between parallel flow of air and desiccant falling film in a fin-tube arrangement, HVAC&R Research 6 (2000) 307–323. [9] B. Dabir, M.R. Riazi, H.R. Davoudirad, Modelling of falling film reactors, Chemical Engineering Science 51 (1996) 2553–2558. [10] G. Karimi, M. Kawaji, An experimental study of freely falling films in a vertical tube, Chemical Engineering Science 53 (1998) 3501– 3512. [11] V. Hessel, W. Ehrfeld, K. Golbig, V. Haverkamp, H. Lo¨we, M. Storz, Ch. Wille, A. Guber, K. Ja¨hnisch, M. Baerns, Gas/liquid microreactors for direct fluorination of aromatic compounds using elemental fluorine, in: Proceedings of the 3rd International Conference on Microreaction Technology, Frankfurt, 1999, pp. 526–540. [12] K. Ja¨hnisch, M. Baerns, V. Hessel, W. Ehrfeld, V. Haverkamp, H. Lo¨we, Ch. Wille, A. Guber, Direct fluorination of toluene using elemental fluorine in gas/liquid microreactors, Journal of Fluorine Chemistry 105 (2000) 117–128. [13] K.K. Yeong, A. Gavriilidis, R. Zapf, V. Hessel, Catalyst preparation and deactivation issues for nitrobenzene hydrogenation in a microstructured falling film reactor, Catalysis Today 81 (2003) 641–651. [14] K.K. Yeong, A. Gavriilidis, R. Zapf, V. Hessel, Experimental studies of nitrobenzene hydrogenation in a microstructured falling film reactor, Chemical Engineering Science 59 (2004) 3491–3494. [15] M.W.E. Coney, The theory and application of conductance probes for the measurement of liquid film thickness in two-phase flow, Journal of Physics E: Scientific Instruments 6 (1973) 903–910. [16] S. Saha, G.V. Tomarov, O.A. Povarov, Experimental investigation on the effect of vibration on flow of liquid film in a horizontal duct, Experiments in Fluids 19 (1995) 383–387. [17] R. Kumar, M. Gottmann, K.R. Sridhar, Film thickness and wave velocity measurements in a vertical duct, Transactions of the ASME I: Journal of Fluids Engineering 124 (2002) 634–642. [18] R.S. Dwyer-Joyce, B.W. Drinkwater, C.J. Donohoe, The measurement of lubricant-film thickness using ultrasound, Proceedings of the Royal Society of London A 459 (2003) 957–976. [19] A.J. Addlesee, K. Cornwell, Liquid film thickness above a bubble rising under an inclined plate, Transactions of the IChemE 75A (1997) 663–667. [20] R.P. Roy, J. Ku, I. Kaufman, J. Shukla, Microwave method for measurement of liquid film thickness in gas–liquid flow, Review of Scientific Instruments 57 (1986) 952–956. [21] R.B. Hurley, I. Kaufman, R.P. Roy, Noncontacting microstrip monitor for liquid film thickness, Review of Scientific Instruments 61 (1990) 2462–2465. [22] L.M. Peurrung, D.B. Graves, Film thickness profiles over topography in spin coating, Journal of the Electrochemical Society 138 (1991) 2115–2124. [23] W.M. Nozhat, Measurement of liquid-film thickness by laser-interferometry, Applied Optics 36 (1997) 7864–7869.
471
[24] M.F.G. Johnson, R.A. Schluter, S.G. Bankoff, Fluorescent imaging system for global measurement of liquid film thickness and dynamic contact angle in free surface flows, Review of Scientific Instruments 68 (1997) 4097–4102. [25] C.H. Hidrovo, D.P. Hart, Emission reabsorption laser induced fluorescence (ERLIF) film thickness measurement, Measurement Science and Technology 12 (2001) 467–477. [26] H. Cho, K. Min, Measurement of liquid fuel film distribution on the cylinder liner of a spark ignition engine using the laser-induced fluorescence technique, Measurement Science and Technology 14 (2003) 975–982. [27] T.A. Shedd, T.A. Newell, Automated optical liquid film thickness measurement method, Review of Scientific Instruments 69 (1998) 4205–4213. [28] J.T. Zhang, B.X. Wang, X.F. Peng, Falling liquid film thickness measurement by an optical-electronic method, Review of Scientific Instruments 71 (2000) 1883–1886. [29] A.A. Mouza, S.V. Paras, A.J. Karabelas, The influence of small tube diameter on falling film and flooding phenomena, International Journal of Multiphase Flow 28 (2002) 1311–1331. [30] C.L. Adler, J.A. Lock, I.A. Rafferty, W. Hickok, Twin-rainbow metrology. 1. Measurement of the thickness of a thin liquid film draining under gravity, Applied Optics 42 (2003) 6584–6594. [31] P.B. Hoke, J.J. McGrath, Measurement of condensate film thickness on reflective surfaces with infrared thermography, Measurement Science and Technology 13 (2002) 1467–1474. [32] A. Boyde, Confocal optical microscopy, in: P.J. Duke, A.G. Michette (Eds.), Modern Microscopies: Techniques and Applications, Plenum Press, New York, 1990, pp. 185–204. [33] T. Wilson (Ed.), Confocal Microscopy, Academic Press, London, 1990. [34] J.B. Pawley (Ed.), Handbook of Biological Confocal Microscopy, second ed., Plenum Press, New York, 1995. [35] A. Boyde, C. Gray, S.J. Jones, Real-time confocal microscopy and cell biology, in: J.E. Cells (Ed.), second ed., Cell Biology: A Laboratory Handbook, vol. II, Academic Press, San Diego, 1998, pp. 179–186. [36] A.E. Ribbe, T. Hashimoto, H. Jinnai, Complex image generation in the laser scanning confocal microscope of a polymer blend system, Journal of Material Science 31 (1996) 5837–5847. [37] A.E. Ribbe, M. Hayashi, M. Weber, T. Hashimoto, Identification of the morphology of a thin film by fluorescence laser-scanning confocal microscopy, Polymer 39 (1998) 7149–7151. [38] J.P.S. Farinha, M.A. Winnik, K.G. Hahn, Characterization of oil droplets under a polymer film by laser scanning confocal fluorescence microscopy, Langmuir 16 (2000) 3391–3400. [39] M. Moffit, Y. Rharbi, W. Chen, J-D. Tong, M.A. Winnik, D.W. Thurman, J.P. Oberhauser, J.A. Kornfield, R.A. Ryntz, Stratified morphology of a polypropylene/elastomer blend following channel flow, Journal of Polymer Science B: Polymer Physics 40 (2002) 2842– 2859. [40] R. Zapf, C. Becker-Willinger, K. Berresheim, H. Bolz, H. Gnaser, V. Hessel, G. Kolb, P. Lob, A.K. Pannwitt, A. Ziogas, Detailed characterization of various porous alumina-based catalyst coatings within microchannels and their testing for methanol steam reforming, Chemical Engineering Research and Design 81 (2003) 721–729. [41] C.J. Cogswell, C.J.R. Sheppard, Confocal brightfield imaging techniques using an on-axis scanning optical microscope, in: T. Wilson (Ed.), Confocal Microscopy, Academic Press, London, 1990. [42] A. Boyde, S.J. Jones, Mapping and measuring surfaces using reflection confocal microscopy, in: J.B. Pawley (Ed.), Handbook of Biological Confocal Microscopy, second ed., Plenum Press, New York, 1995, pp. 255–266. [43] S.V. Alekseenko, V.E. Nakoryakov, B.G. Pokusaev, Wave formation on vertical falling films, International Journal of Multiphase Flow 11 (1985) 607–627. [44] R.V. Birikh, V.A. Briskman, M.G. Velarde, J.C. Legros, Liquid Interfacial Systems. Oscillations and Instability, Marcel Dekker, New York, 2003.
472
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
[45] A. Oron, S.H. Davis, S.G. Bankoff, Long-scale evolution of thin liquid films, Reviews of Modern Physics 69 (1997) 931–980. [46] D.T. Hughes, T.R. Bott, Minimum thickness of liquid film flowing down a vertical tube, International Journal of Heat and Mass Transfer 41 (1998) 253–260. [47] N.K. Adam, Principles of penetration of liquids into solids, Discussion of the Faraday Society 3 (1948) 5–11. [48] R. Seemann, M. Brinkmann, E.J. Kramer, F.F. Lange, R. Lipowsky, Wetting morphologies at microstructured surfaces, Proceedings of the National Academy of Sciences 102 (2005) 1848–1852. [49] A.W. Adamson, A.P. Gast, Physical Chemistry of Surfaces, Chapter II—Capillarity, sixth ed., Wiley-Interscience, New York, 1997. [50] D. Li, Drop size dependence of contact angles and line tensions of solid–liquid systems, Colloids and Surfaces A: Physicochemical and Engineering Aspects 116 (1996) 1–23.
[51] S. Hardt, F. Doffing, H. Pennemann, Simulation of hydrodynamic dispersion in gas/liquid microreactors, in: Proceedings of the International Conference on Modelling and Simulation of Microsystems, Puerto Rico, 2002. [52] T. Cubaud, M. Ferigier, Advancing contact lines on chemically patterned surfaces, Journal of Colloids and Interfacial Science 269 (2004) 171–177. [53] L. Horoba, R.X. Tudose, Gravitational film flow in liquid–liquid systems, Revue Roumaine de Chimie 45 (2000) 955–961. [54] P.S. Swain, R. Lipowsky, Contact angle on heterogeneous surfaces: a new look at CassieÕs and WenzelÕs laws, Langmuir 14 (1998) 6772– 6780. [55] C.H. Lin, R.M. Ybarra, P. Neogi, Three and two-dimensional effects in wetting kinetics, Advances in Colloid and Interface Science 67 (1996) 185–204.
Characterisation of liquid film in a microstructured falling film reactor using laser scanning confocal microscopy Kay Kin Yeong
a,1
, Asterios Gavriilidis a,*, Ralf Zapf b, Hans-Joachim Kost b, Volker Hessel b,c, Alan Boyde d
a
Department of Chemical Engineering, University College London, London WC1E 7JE, UK Institut fu¨r Mikrotechnik Mainz GmbH, Carl-Zeiss-Str. 18-20, D-55129 Mainz, Germany c Department of Chemical Engineering and Chemistry, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands d Biophysics OGD, Dental Institute, Barts & The London School of Medicine & Dentistry, Queen Mary, University of London, London E1 1BB, UK b
Received 15 October 2004; accepted 23 September 2005
Abstract Reflection confocal microscopy was used to measure the thicknesses of falling thin films generated using microstructured plates. Each plate contained straight parallel channels through which liquid fell vertically under gravity. Three different channel dimensions were used: 300 · 100 lm, 600 · 200 lm and 1200 · 400 lm (width · depth). Measurements were performed for acetone, ethanol and isopropanol and gave reasonable agreement with theoretical predictions and correlations from the literature. Discrepancies were attributed to the fact that these equations do not consider parameters that affect three-phase contact lines such as surface tension and contact angle. No trend of film thickness on channel width was observed and was probably due to different surface heterogeneity and roughness characteristics amongst the plates. Measurement difficulties were encountered at low liquid film thicknesses and where the angle of the liquid meniscus was too steep. 2005 Elsevier Inc. All rights reserved. Keywords: Microstructured reactor; Falling film; Liquid film thickness measurement
1. Introduction Falling film reactors generate thin liquid films (typically in contact with either a flat wall or stack of pipes) that fall under gravitational pull. The thinness of these films results in rapid heat and mass transfer. As such, these systems find application in extraction, evaporation and highly exothermic processes [1–8]. Conventional falling film systems on tubes or flat surfaces generate films with thicknesses in the order of 0.5–3 mm [9,10]. A microstructured falling film reactor (l-FFR) developed by the Institut fu¨r Mikrotechnik Mainz can generate stable films less than 100 lm thick. This reactor offers excellent heat removal capabilities and *
Corresponding author. Tel.: +44 20 76793811; fax: +44 20 73832348. E-mail address: [email protected] (A. Gavriilidis). 1 Present address: Xennia Technology, Lumen House, Lumen Road, Royston, Hertfordshire SG8 7AG, UK. 0894-1777/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.expthermflusci.2005.09.006
has been used safely in the direct fluorination of aromatics [11,12]. Our research aims to evaluate the performance of the l-FFR in solid catalysed gas–liquid hydrogenations. The development of suitable catalyst incorporation methods (e.g. via alumina coatings) and reaction studies have been reported elsewhere [13,14]. This paper presents investigations of liquid film thickness and profile measurements. A variety of methods have been employed to measure the thickness of liquid films. Examples include conductivity probes [15–17], ultrasound detectors [18], fibre-optic sensors [19], microwave techniques [20,21], laser interferometry [22,23], fluorescence [24], laser-induced fluorescence [25,26], optical measurement methods [27–29], rainbow refractometry [30] and infrared thermography [31]. The use of conductivity probes was not considered for this investigation because they require physical contact with the liquid film which would greatly affect flow characteristics due to the small flow channels and liquid film
464
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
Nomenclature g KF Re d C h
gravitational acceleration l4 g qr3
Reynolds number ¼ Cl liquid film thickness mass flowrate per wetted perimeter inclination angle from the horizontal
thicknesses encountered. Some of the optical methods listed above were also unsuited for the system being investigated (e.g. required translucent substrates, circular surfaces). To add to the arsenal of optical-based methods, reflection confocal laser scanning microscopy (CLSM) was employed to study this system. CLSM is a popular imaging and analysis technique that has been widely used in many areas, e.g. biology (mostly in fluorescent mode [32–35]), semiconductor materials science (mostly in reflection mode [33]) and polymer science [36–39]. The key feature of CLSM is that it has a very short depth-of-field. In conventional optical microscopes, the entire object within the lensÕ field of view is seen; areas at the focal point of the lens appear sharp while those towards or beyond the focal point are blurry. In CLSM, however, due to its limited depth-offield, regions which are removed from the focal point are not seen at all due to the confocal pinhole which restricts light from levels other than the focal region from reaching the detector. The effect is similar to taking a slice through an object. By varying the depth at which the microscope is focused and scanning the beam across each layer, a succession of slices can be taken, which can then be summed to give a sharp, high resolution image of the object in question. Because CLSM collects data point-by-point on each slice, a very detailed map of a three-dimensional object can be obtained, meaning that depth discrimination and optical tomography can be performed to high accuracy [33,34]. 2. Experimental apparatus and procedure 2.1. Microstructured falling film reactor The main part of the reactor is the structured stainless steel plate for the generation of the falling film. Each plate contained straight parallel channels 78 mm long fabricated using isotropic wet etching. More details about the reactor can be found elsewhere [11–14]. Three different channel configurations were used, as shown in Table 1. In addition, a c-alumina coated version of each plate was used as well (details of the coating procedure are given in [13] while pictures of cross sections of the coated channels are given in [40]). Slots were cut through the plates at both ends of the channels to form the liquid inlet and exit ports. Each
l m m0 q qc r
dynamic viscosity kinematic viscosity m 0:6106
liquid density density of adjoining phase surface tension
Table 1 Channel dimensions of the microstructured plates Plate
Channel depth (lm)
Channel width (lm)
Number of channels
Type I Type II Type III
100 200 400
300 600 1200
64 32 16
plate measured 89 · 46 mm and was housed in a stainless steel enclosure. Liquid was fed to and removed from the back of the plate. A stainless steel gasket was placed on top of the plate. Its geometry was such that the whole length of the channels were exposed with the exception of the first and last 5 mm, which helped to distribute the liquid across the plate. The channels were exposed to ambient air. 2.2. Confocal microscope The experiments were performed using a Noran Odyssey video-rate laser point scanning confocal microscope, in reflection mode with a 633 nm laser. A Nikon Mplan extra large working distance (20/0.4 lens, i.e. magnification = 20·, numerical aperture = 0.4) was used. To allow the examination of the l-FFR in a vertical orientation, the objective had to be attached to the microscope via a 90 bend (incorporating a 45 mirror) such that this lens projected outward horizontally. Because of this, the focus axis of the microscope was changed to the horizontal. As such, a microscope stage had to be constructed to allow focussing movement in the horizontal plane. It was especially important that the movement of the stage in the direction parallel to the projection of the lens could be measured accurately, as this now represented the depth at which the microscope was focused. Two screw micrometers were used to provide movement and measurement in the horizontal plane (see Fig. 1). Because of the manual setup of this system, a restricted number of data points could be collected compared to an automated system. This led to the need for some amount of extrapolation in order to interpret the data collected. The manual data collection employed in this work resulted in experimental error of ±4 lm while confocal microscopy with high numerical aperture objectives may have sub-micron depth and lateral resolution [33].
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
465
the channels, this made the surface of the liquid film identifiable, except when: (a) the liquid film was very thin. When the film surface was very close to the bottom of the channel, reflection from the channel surface was brighter than the gas/ liquid interface and it became very difficult to identify the location of the film surface, (b) the angle of the film/channel surface was too steep. Only light reflected back along the incoming path is detected. This does not present a problem when
Fig. 1. Microscope stage with focusing mechanism (to and away from the laser beam) controlled using a screw micrometer (for depth measurements), and a secondary horizontal motion mechanism perpendicular to the laser beam.
2.3. Experimental procedure A syringe pump (Razel A-99) was used to feed the test liquid to the reactor. Three liquids of differing viscosity were employed: ethanol, isopropanol and acetone. Measurements were taken at three channels located at the left, central and right parts of the reactor, 2.1 cm below the liquid inlet slot, and then averaged. This height was chosen as a compromise between minimising evaporation and entrance effects. Exploratory experiments showed that evaporation was negligible at this distance from the entrance. Some entrance effects were found, e.g. the film was a few microns thicker 0.5 cm below this height, but there were difficulties in performing measurements at a lower location due to physical obstruction by parts of the microscope. 3. Preliminary results The difference between the refractive indices of air and liquid results in reflection occurring at the gas–liquid interface [41]. In addition to light reflected from the bottom of
Fig. 2. Confocal microscopy images of a microchannel filled with ethanol at different focal depths (a)–(c) are images of a stainless steel channel, (d) is an image of a c-alumina coated channel: (a) top of channel walls, (b) on fluid at the centre of the channel (i.e. the lowest point of the liquid surface), (c) bottom of channel, (d) bottom of c-alumina coated channel (300 · 100 lm channel).
466
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
observing surfaces which are located normal to the incidence path. However, when the surface being observed is at an angle, the returning light will fail to enter the working aperture of the objective lens. Hence, image intensity would be significantly reduced, if anything can be seen at all [42]. Fig. 2 shows a series of images taken using the CLSM. Fig. 2a shows the microscope focussed on the top of the stainless steel channel walls, making that area much brighter than all surrounding regions which are at a different depth. The surface of the liquid film can also be dimly seen in the middle of the figure. In true confocal mode, the liquid surface should not be visible because it is at a different depth, but here we have a rather low numerical aperture objective and the focal-level discrimination is consequently limited. Focusing deeper into the channel locates the deepest point of the liquid surface, which corresponds to the centre of the channel width-wise (see Fig. 2b). Moving below the liquid surface eventually focuses the microscope on the channel bottom, shown in Fig. 2c. Fig. 2d shows a c-alumina coated channel for comparison. The surface of the liquid was less bright than the channel surface (with the stainless steel channels brighter than the
alumina-coated channels), which therefore required greater image contrast in order to be seen. Fringes were observed on the liquid surface due to interference with light reflected from the channel surface. These fringes were in constant motion, which suggests that the liquid film was perturbed. These disturbances may be an indication of the presence of hydrodynamic instabilities [43,44]. 4. Channel profile measurements Comparisons were made between the manuallyobtained data from the Noran Odyssey CLSM and measurements made using an automated laser-scanning surface profilometer (NanoFocus). This is an autofocusing laser system using light reflection to measure surface profiles. The laser source is moved to find the maximum laser intensity and in this way images the surface profile. The latter instrument took measurements every 0.5 lm across the width of the channels. The results are shown in Fig. 3. The measurements were in good agreement for the 300 · 100 and 600 · 100 lm channels, but the confocal system depicted a shallower, more rounded surface than the profilometer for the 1200 · 400 lm channel. This was most likely due to channel size variations during manufacturing.
120 Laser-scanning surface profilometry Confocal microscopy
Y (μm)
80
40
0 0
100
200
(a)
300
X (μm) 250 Laser-scanning surface profilometry Confocal microscopy
Y (μm)
200 150 100 50 0 0
200
400
(b)
600
800
X (μm)
Y (μm)
400 Laser-scanning surface profilometry Confocal microscopy
300 200 100 0 0
(c)
500
1000
1500
X (μm)
Fig. 3. Comparison of channel cross-sectional profiles using confocal microscopy and laser-scanning surface profilometry: (a) 300 · 100 lm channel, (b) 600 · 200 lm channel, (c) 1200 · 400 lm channel.
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
5. Liquid film surface profile measurements
Channel height/liquid film thickness (μm)
Liquid flowing down a flat plate can take several forms. At large flowrates, it takes the form of continuous sheets (which may demonstrate waviness). At either the front end of those sheets or at low flowrates, flow breaks up into rivulets or fingers [45,46]. The purpose of the microchannels in the l-FFR is to prevent this break-up of the liquid film at very low flowrates. Due to the combination of capillary forces and small channel widths, liquid is pulled up along the sides of the channels and takes up a significant portion of the channel width. The surface of the liquid film takes on the form of a flowing meniscus. The profile of the film surface was measured at various flowrates in each of the l-FFR plates detailed earlier, and a selection is shown in Figs. 4 and 5. It should be noted that the measurements by which the plates are referred to (e.g. 300 · 100 lm) are
Channel height/liquid film thickness (μm)
merely convenient descriptors for those plates. The actual plate dimensions and geometry can be found in Fig. 3. For the majority of the measurements, the film profiles are shown as being incomplete. As mentioned in the previous section, this was due to measurement difficulties caused by the curvature of the liquid surface. Also, in Fig. 4a, the surface profile for the lowest flowrate is depicted as flat. This was due to errors in determining the true liquid profile for very thin films over the highly reflective metal surface of the channel. In general, however, the surface of the film curves upwards moving from the centre of the channel to the walls. As the flowrate increases, the film thickness also increases and the surface profile becomes flatter. When the channel is completely filled the profile becomes flat (Fig. 4a, flowrate 84.9 ml/h). Further increase in the flowrate causes the liquid to bulge out of the channel (Fig. 4a, flowrate 113 ml/h). When the hydrodynamic pressure exceeds
9.90 ml/h
120 15.6 ml/h
100
24.0 ml/h
80
33.2 ml/h 42.4 ml/h
60 48.1 ml/h
40
62.9 ml/h
20
84.9 ml/h 113 ml/h
0 0
50
100
(a)
Channel height/liquid film thickness (μm)
467
150 Channel width (μm)
200
250
300
200
channel
9.92 ml/h 13.9 ml/h
150 19.4 ml/h 25.0 ml/h
100
36.5 ml/h
50
51.6 ml/h channel
0
0
100
200
(b)
300
400
500
600
Channel width (μm)
350 300 30.4 ml/h
250
44.6 ml/h
200
60.1 ml/h
150
97.0 ml/h
77.8 ml/h 111 ml/h
100
channel
50 0
(c)
0
200
400
600
800
1000
1200
Channel width (μm)
Fig. 4. Cross-sectional profile of the liquid film surface at various flowrates in different microchannels: (a) 300 · 100 lm channel, (b) 600 · 200 lm channel, (c) 1200 · 400 lm channel (liquid ethanol). Fig. 4a reprinted with permission from Ind. Eng. Chem. Res. Copyright 2005 American Chemical Society.
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472 Channel height/liquid film thickness (μm)
468
100
5.95 ml/h 7.54 ml/h
80
9.12 ml/h
60
16.3 ml/h
40
22.2 ml/h
20
32.1 ml/h 43.6 ml/h
0 0
50
100
150
200
250
300
alumina
-20
channel
Channel width (μm)
Fig. 5. Cross-sectional profile of the liquid film surface at various flowrates in the 300 · 100 lm alumina coated plate. ‘‘alumina’’ indicates the liquid– alumina interface, while ‘‘channel’’ indicates the alumina–stainless steel interface (liquid ethanol).
Channel height/liquid film thickness (μm)
surface tension forces the liquid bursts out of its channel wall confines. Such behaviour has been suggested by Adam [47] in the context of liquid penetration in pores with constrictions. Namely, as the liquid penetrates a pore with complex shape, the meniscus gets pinned when it reaches a constriction, resisting increasing pressure difference by changing its curvature. This is equivalent to a change of the apparent contact angle from its intrinsic value x, to x + /Wall, where /Wall is the solid wall inclination change at the pore mouth. In our case, the pore mouth corresponds to the microchannel edges. This phenomenon has been recently examined theoretically by minimisation of liquid free energies for open microchannels [48]. Comparing the film profiles of coated and uncoated channels (see Fig. 6), it can be seen that the liquid surface is located higher up in the coated channels at a given flowrate. Note that in Fig. 6 the baseline reference is the top of the channel. This upward displacement of the liquid surface is noticeably greater in the middle of the channel, and has an average value of 17 lm. In comparison, the thickness of the alumina layer was estimated to be 10 lm. This displacement can be rationalised as follows. The curvature and location of the interface is determined (a) by the balance between the Laplace pressure and the interfacial tension according to the well known Laplace equation [49], where the key parameter is the contact angle, (b) by the location of the three-phase contact line and (c) by the ability of meniscus to contort if the three-phase contact line
is pinned at the channel edges [47]. Depending on the contact angle, the (cross sectional) shape of the channel and the liquid volumetric flowrate, the three-phase contact lines can reside at the sides or at the edges of the channel. In Fig. 6, where the three-phase contact lines seem to be pinned at the channel edges, the difference between the two interface profiles can be attributed to different channel depth/shape, different contact angles at stainless steel and alumina and meniscus contortion. The fact that the contact angle also depends on surface roughness and heterogeneity [50] can play a role, particularly at lower flowrates where the three-phase contact line may be located at the channel sides and not edges. It had been expected that the profiles of the film surface would have been circular arcs as the capillary (Ca) and Weber (We) numbers are very small (<103) for this system at the range of flowrates studied, indicating that the liquid surface should not be deformed due to flow [51]. This is true for the lower-to-mid flowrates in the 300 · 100 lm channels (both coated and uncoated). However, at larger flowrates in those channels and at all flowrates in the larger channels, the profiles were found to be distorted and best approximated as superellipses. This finding could be the result of experimental error (due to difficulty in accurately detecting the curved liquid surface and disturbances in the liquid flowrate) or contortion of the three-phase contact line at its pinning location due to surface roughness and heterogeneity [52].
100 80 43.6 ml/h (coated)
60 42.4 ml/h (uncoated)
40
alumina
20
channel
0
0
50
100
150
200
250
300
-20 Channel width (μm)
Fig. 6. Comparison of cross-sectional liquid film surface profiles in the 300 · 100 lm coated and uncoated channels. ‘‘alumina’’ indicates the surface of the alumina-coated channel, while ‘‘channel’’ indicates the surface of the uncoated channel (liquid ethanol).
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
469
2
1.8
Feind correlation Nusselt equation Kapitza equation 300x100 plate
1.6 log δ
300x100, Al2O3 plate 600x200 plate 600x200, Al2O3 plate
1.4
1200x400 plate 1200x400, Al2O3 plate
1.2
1 -1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
log Re
Fig. 7. Relationship between liquid film thickness, d, and Reynolds number, Re, obtained experimentally (points) and theoretically (lines) (liquid ethanol).
6. Liquid film thickness Due to the fact that neither channel bottom nor liquid surface were flat (and also possessing different curvatures), the average liquid film thickness was defined based on the cross-sectional area of the liquid film profile divided by the wetted channel perimeter. The experimental film thickness as a function of Reynolds number is shown in Fig. 7. The figure also shows predictions of film thickness using the following equations and correlations [53]: 2 1=3 1=3 3m Re q Nusselt : d ¼ g sin h q qc 1=3 1=3 2:4m2 Re q Kapitza : d ¼ g sin h q qc 1=3 1=3 2 3m q 00:11 00:025 Feind : d ¼ Re0:333m m q qc g sin h for 0:72K F0:1 < Re < 1:35K F0:1 Since the system was a vertical falling film, h = 90. The denq sity factor qq was ignored as q qc (note: the Re for the c flowrates studied were 0.04–0.86, which was much smaller than the range of the Feind equation criterion as K F0:1 ¼ 8:2). Since the calculation of d and Re is based on the wetted perimeter, it is sensitive to the location of the three-phase contact line. It must be emphasised that these equations do not account for surface tension and contact angle, as they have been developed for two-dimensional films. The results follow the trend of the predictions, but are scattered over a fairly broad range. The Kapitza equation appears to give the closest overall prediction. In general, the liquid film thicknesses measured for the alumina-coated channels were greater than those for the uncoated channels. This behaviour may be due to the fact that the porous nature of the alumina coating may have drawn the liquid film higher up the channel wall as well as to the different contact angle for the two materials. A common feature of the data is that each set exhibits a sudden drop of the film thickness at lower Re. This is most likely due to experimen-
tal error when measuring very thin films (below ca. 30 lm) and identifying the points of contact between the liquid and the channel walls, as detailed previously. In general, the 1200 · 400 lm plate results in the smallest film thicknesses and the 600 · 200 lm plate in the thickest films for the uncoated plates, while the 300 · 100 lm plate results in the smallest film thicknesses and the 600 · 200 lm plate in the thickest films for the alumina-coated plates. The disparity of the experimental results amongst different plates and with the theoretical predictions is not surprising, if one considers that the equations do not account for three-phase contact and that the hydrodynamics near the point of three-phase contact are not well understood. The classical Young equation which relates contact angle to the various interfacial tensions is applicable to smooth, homogeneous, isotropic solid surfaces. However, apparent contact angle depends further on the so-called line tension, whose effect is magnified by surface roughness and hetereogeneity by additional corrugation of the three-phase contact line. Surface roughness and hetereogeneity affect further the apparent contact angle by increasing the real solid surface and hence its total interfacial energy, according to the Wenzel and Cassie equations [54]. Moreover, the effect of roughness can be even stronger under dynamic conditions [55]. Hence, the absence of a specific trend of film thickness vs. channel size in Fig. 7 may be attributed to surface heterogeneity and roughness. 7. Comparison of different liquids In order to determine the behaviour of different fluids, experiments were also performed using isopropanol and Table 2 Physical properties of acetone, ethanol and isopropanol
Acetone Ethanol Isopropanol
Viscosity (mPa s)
Surface tension (mN/m)
Density (kg/m3)
0.34 1.2 2.3
14.4 21.6 22.7
792 790 790
470
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472 2.2
2 Feind correlation Nusselt equation Kapitza equation
1.8 log δ
300x100 plate 600x200 plate
1.6
1.4
1.2
1 -1.7
-1.5
-1.3
(a)
-1.1
-0.9
-0.7
-0.5
log Re 1.8
Feind correlation
1.6 log δ
Nusselt equation Kapitza equation 300x100 plate
1.4
1.2 -0.2
(b)
-0.1
0
0.1
0.2
0.3
0.4
log Re
Fig. 8. Relationship between liquid film thickness, d, and Reynolds number, Re, obtained experimentally (points) and theoretically (lines): (a) isopropanol, (b) acetone.
acetone (see Table 2 for the physical properties). The relationship between liquid film thickness and Re for isopropanol and acetone can be seen in Fig. 8. The Kapitza equation gives the closest overall prediction for isopropanol while the Nusselt equation or Feind correlation seem to give better predictions for acetone (though the number of data points is not sufficient for conclusive evidence). Comparing the results of the three liquids, the liquid with the highest viscosity (i.e. isopropanol) resulted in the greatest liquid film thickness at a given Re, which is consistent with the theoretical predictions.
the equations do not account for surface tension and contact angles. The greatest discrepancy between the predictions and experiments were found at the lowest flowrates and for the 600 · 200 alumina-coated plate. The lowerthan-expected film thicknesses measured experimentally may be due to difficulties in identifying and locating the surface of the liquid film at film thicknesses less than 30 lm. The cause of the nonmonotonic behaviour of film thickness vs. channel size may be attributed to different surface heterogeneity and roughness characteristics of the various plates. Acknowledgments
8. Conclusions Thin liquid films were generated using microstructured falling film plates with different channel sizes (300 · 100 lm, 600 · 200 lm and 1200 · 400 lm) and surfaces (stainless steel and c-alumina on stainless steel). Film thicknesses were measured using reflection confocal microscopy at a range of flowrates for acetone, ethanol and isopropanol. The Kapitza equation was found to give the best overall predictions for ethanol and isopropanol, while the Nusselt equation and Feind correlation appeared to give better predictions for acetone. The agreement between the equations and the experimental data is good, considering the fact that
The authors would like to thank the European Commission (G1RD-CT2000-00469) for partial financial support. Help and advice provided by Enhong Cao, Carlos Amador, Maureen Arora, Roy Radcliffe and Peter Howell were invaluable to this work. The Odyssey CSLM was funded by the Wellcome Trust and its conversion to red laser light operation by a Royal Society grant. References [1] A.M. Yih, C.C. Kuo, Design and testing of a new type of falling film gas–liquid contacting device, AIChE Journal 34 (1988) 499–501.
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472 [2] E.A. Knaggs, Development of continuous falling film SO3 sulfonation technology, Chemtech (July) (1992) 436–445. [3] S.M. Walas, Chemical reactors, in: R.H. Perry, D.W. Green (Eds.), Chemical EngineerÕs Handbook, seventh ed., McGraw-Hill, 1997 (Chapter 23). [4] M.H.H. Van Dam, J.-P. Corriou, N. Midoux, A.-S. Lamine, C. Roizard, Modelling and measurement of sulfur dioxide absorption rate in a laminar falling film reactor, Chemical Engineering Science 54 (1999) 5311–5318. [5] S. Bessenet, V. Renaudin, J.-P. Leclerc, J.-M. Hornut, RTD studies in a falling film graphite evaporator with internal spiral fins: influence of the geometry, Canadian Journal of Chemical Engineering 78 (2000) 486–494. [6] W. Chen, R.N. Christensen, Inlet subcooling effect on heat and mass transfer characteristics in a laminar film flow, International Journal of Heat and Mass Transfer 43 (2000) 167–177. [7] S. Kulankara, K.E. Herold, Theory of heat/mass transfer additives in absorption chillers, HVAC&R Research 6 (2000) 369–380. [8] A.S. Rahmah, M.M. Elsayed, N.M. Al-Najem, A numerical investigation for the heat and mass transfer between parallel flow of air and desiccant falling film in a fin-tube arrangement, HVAC&R Research 6 (2000) 307–323. [9] B. Dabir, M.R. Riazi, H.R. Davoudirad, Modelling of falling film reactors, Chemical Engineering Science 51 (1996) 2553–2558. [10] G. Karimi, M. Kawaji, An experimental study of freely falling films in a vertical tube, Chemical Engineering Science 53 (1998) 3501– 3512. [11] V. Hessel, W. Ehrfeld, K. Golbig, V. Haverkamp, H. Lo¨we, M. Storz, Ch. Wille, A. Guber, K. Ja¨hnisch, M. Baerns, Gas/liquid microreactors for direct fluorination of aromatic compounds using elemental fluorine, in: Proceedings of the 3rd International Conference on Microreaction Technology, Frankfurt, 1999, pp. 526–540. [12] K. Ja¨hnisch, M. Baerns, V. Hessel, W. Ehrfeld, V. Haverkamp, H. Lo¨we, Ch. Wille, A. Guber, Direct fluorination of toluene using elemental fluorine in gas/liquid microreactors, Journal of Fluorine Chemistry 105 (2000) 117–128. [13] K.K. Yeong, A. Gavriilidis, R. Zapf, V. Hessel, Catalyst preparation and deactivation issues for nitrobenzene hydrogenation in a microstructured falling film reactor, Catalysis Today 81 (2003) 641–651. [14] K.K. Yeong, A. Gavriilidis, R. Zapf, V. Hessel, Experimental studies of nitrobenzene hydrogenation in a microstructured falling film reactor, Chemical Engineering Science 59 (2004) 3491–3494. [15] M.W.E. Coney, The theory and application of conductance probes for the measurement of liquid film thickness in two-phase flow, Journal of Physics E: Scientific Instruments 6 (1973) 903–910. [16] S. Saha, G.V. Tomarov, O.A. Povarov, Experimental investigation on the effect of vibration on flow of liquid film in a horizontal duct, Experiments in Fluids 19 (1995) 383–387. [17] R. Kumar, M. Gottmann, K.R. Sridhar, Film thickness and wave velocity measurements in a vertical duct, Transactions of the ASME I: Journal of Fluids Engineering 124 (2002) 634–642. [18] R.S. Dwyer-Joyce, B.W. Drinkwater, C.J. Donohoe, The measurement of lubricant-film thickness using ultrasound, Proceedings of the Royal Society of London A 459 (2003) 957–976. [19] A.J. Addlesee, K. Cornwell, Liquid film thickness above a bubble rising under an inclined plate, Transactions of the IChemE 75A (1997) 663–667. [20] R.P. Roy, J. Ku, I. Kaufman, J. Shukla, Microwave method for measurement of liquid film thickness in gas–liquid flow, Review of Scientific Instruments 57 (1986) 952–956. [21] R.B. Hurley, I. Kaufman, R.P. Roy, Noncontacting microstrip monitor for liquid film thickness, Review of Scientific Instruments 61 (1990) 2462–2465. [22] L.M. Peurrung, D.B. Graves, Film thickness profiles over topography in spin coating, Journal of the Electrochemical Society 138 (1991) 2115–2124. [23] W.M. Nozhat, Measurement of liquid-film thickness by laser-interferometry, Applied Optics 36 (1997) 7864–7869.
471
[24] M.F.G. Johnson, R.A. Schluter, S.G. Bankoff, Fluorescent imaging system for global measurement of liquid film thickness and dynamic contact angle in free surface flows, Review of Scientific Instruments 68 (1997) 4097–4102. [25] C.H. Hidrovo, D.P. Hart, Emission reabsorption laser induced fluorescence (ERLIF) film thickness measurement, Measurement Science and Technology 12 (2001) 467–477. [26] H. Cho, K. Min, Measurement of liquid fuel film distribution on the cylinder liner of a spark ignition engine using the laser-induced fluorescence technique, Measurement Science and Technology 14 (2003) 975–982. [27] T.A. Shedd, T.A. Newell, Automated optical liquid film thickness measurement method, Review of Scientific Instruments 69 (1998) 4205–4213. [28] J.T. Zhang, B.X. Wang, X.F. Peng, Falling liquid film thickness measurement by an optical-electronic method, Review of Scientific Instruments 71 (2000) 1883–1886. [29] A.A. Mouza, S.V. Paras, A.J. Karabelas, The influence of small tube diameter on falling film and flooding phenomena, International Journal of Multiphase Flow 28 (2002) 1311–1331. [30] C.L. Adler, J.A. Lock, I.A. Rafferty, W. Hickok, Twin-rainbow metrology. 1. Measurement of the thickness of a thin liquid film draining under gravity, Applied Optics 42 (2003) 6584–6594. [31] P.B. Hoke, J.J. McGrath, Measurement of condensate film thickness on reflective surfaces with infrared thermography, Measurement Science and Technology 13 (2002) 1467–1474. [32] A. Boyde, Confocal optical microscopy, in: P.J. Duke, A.G. Michette (Eds.), Modern Microscopies: Techniques and Applications, Plenum Press, New York, 1990, pp. 185–204. [33] T. Wilson (Ed.), Confocal Microscopy, Academic Press, London, 1990. [34] J.B. Pawley (Ed.), Handbook of Biological Confocal Microscopy, second ed., Plenum Press, New York, 1995. [35] A. Boyde, C. Gray, S.J. Jones, Real-time confocal microscopy and cell biology, in: J.E. Cells (Ed.), second ed., Cell Biology: A Laboratory Handbook, vol. II, Academic Press, San Diego, 1998, pp. 179–186. [36] A.E. Ribbe, T. Hashimoto, H. Jinnai, Complex image generation in the laser scanning confocal microscope of a polymer blend system, Journal of Material Science 31 (1996) 5837–5847. [37] A.E. Ribbe, M. Hayashi, M. Weber, T. Hashimoto, Identification of the morphology of a thin film by fluorescence laser-scanning confocal microscopy, Polymer 39 (1998) 7149–7151. [38] J.P.S. Farinha, M.A. Winnik, K.G. Hahn, Characterization of oil droplets under a polymer film by laser scanning confocal fluorescence microscopy, Langmuir 16 (2000) 3391–3400. [39] M. Moffit, Y. Rharbi, W. Chen, J-D. Tong, M.A. Winnik, D.W. Thurman, J.P. Oberhauser, J.A. Kornfield, R.A. Ryntz, Stratified morphology of a polypropylene/elastomer blend following channel flow, Journal of Polymer Science B: Polymer Physics 40 (2002) 2842– 2859. [40] R. Zapf, C. Becker-Willinger, K. Berresheim, H. Bolz, H. Gnaser, V. Hessel, G. Kolb, P. Lob, A.K. Pannwitt, A. Ziogas, Detailed characterization of various porous alumina-based catalyst coatings within microchannels and their testing for methanol steam reforming, Chemical Engineering Research and Design 81 (2003) 721–729. [41] C.J. Cogswell, C.J.R. Sheppard, Confocal brightfield imaging techniques using an on-axis scanning optical microscope, in: T. Wilson (Ed.), Confocal Microscopy, Academic Press, London, 1990. [42] A. Boyde, S.J. Jones, Mapping and measuring surfaces using reflection confocal microscopy, in: J.B. Pawley (Ed.), Handbook of Biological Confocal Microscopy, second ed., Plenum Press, New York, 1995, pp. 255–266. [43] S.V. Alekseenko, V.E. Nakoryakov, B.G. Pokusaev, Wave formation on vertical falling films, International Journal of Multiphase Flow 11 (1985) 607–627. [44] R.V. Birikh, V.A. Briskman, M.G. Velarde, J.C. Legros, Liquid Interfacial Systems. Oscillations and Instability, Marcel Dekker, New York, 2003.
472
K.K. Yeong et al. / Experimental Thermal and Fluid Science 30 (2006) 463–472
[45] A. Oron, S.H. Davis, S.G. Bankoff, Long-scale evolution of thin liquid films, Reviews of Modern Physics 69 (1997) 931–980. [46] D.T. Hughes, T.R. Bott, Minimum thickness of liquid film flowing down a vertical tube, International Journal of Heat and Mass Transfer 41 (1998) 253–260. [47] N.K. Adam, Principles of penetration of liquids into solids, Discussion of the Faraday Society 3 (1948) 5–11. [48] R. Seemann, M. Brinkmann, E.J. Kramer, F.F. Lange, R. Lipowsky, Wetting morphologies at microstructured surfaces, Proceedings of the National Academy of Sciences 102 (2005) 1848–1852. [49] A.W. Adamson, A.P. Gast, Physical Chemistry of Surfaces, Chapter II—Capillarity, sixth ed., Wiley-Interscience, New York, 1997. [50] D. Li, Drop size dependence of contact angles and line tensions of solid–liquid systems, Colloids and Surfaces A: Physicochemical and Engineering Aspects 116 (1996) 1–23.
[51] S. Hardt, F. Doffing, H. Pennemann, Simulation of hydrodynamic dispersion in gas/liquid microreactors, in: Proceedings of the International Conference on Modelling and Simulation of Microsystems, Puerto Rico, 2002. [52] T. Cubaud, M. Ferigier, Advancing contact lines on chemically patterned surfaces, Journal of Colloids and Interfacial Science 269 (2004) 171–177. [53] L. Horoba, R.X. Tudose, Gravitational film flow in liquid–liquid systems, Revue Roumaine de Chimie 45 (2000) 955–961. [54] P.S. Swain, R. Lipowsky, Contact angle on heterogeneous surfaces: a new look at CassieÕs and WenzelÕs laws, Langmuir 14 (1998) 6772– 6780. [55] C.H. Lin, R.M. Ybarra, P. Neogi, Three and two-dimensional effects in wetting kinetics, Advances in Colloid and Interface Science 67 (1996) 185–204.