Você está na página 1de 40

MEASURING THE TEMPERATURE OF STATIONARY FLUID IN A

MICROCHANNEL USING THERMOCHROMIC LIQUID CRYSTALS

by

Meir Basson
_____________________________________________________________

A Thesis Presented to the


FACULTY OF THE USC VITERBI SCHOOL OF ENGINEERING
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(AEROSPACE ENGINEERING)

May 2010

Copyright 2010 Meir Basson


Table of Contents
________________________________________________________________________

List of Figures iii

Abstract v

Introduction 1

Materials & Methods 8

Results 23

Discussion 32

Conclusion 34

Bibliography 35

ii
List of Figures

Figure 1: TLC's in their solid, liquid, and intermediate states 2

Figure 2: The relation between the skew angle and the pitch length of the structure 3

Figure 3: Typical reflected wavelength vs. temperature response of a temperature 4


Sensitive TLC formulation at a constant angle of reflection

Figure 4: TLC circular dichroism 5

Figure 5: HS vector of a typical pixel 7

Figure 6: Square flexible silica capillary tube 8

Figure 7: Block used to control the temperature 10

Figure 8: Microscope and Plexiglas box 11

Figure 9: types of light in the illumination path 13

Figure 10: The effect of the polarizer and ¼ wave plate in the illumination path 15

Figure 11: Block with the microchannel inserted 16

Figure 12: The distance in pixels that the field of view of the camera has gone 18

Figure 13: Flowchart of the algorithm used to determine x, y and average hue 19
For TLC microcapsules in the images

Figure 14: Each particle’s location in the axial direction of the microchannel 22

Figure 15: Histogram of set temperature 37.15 oC 23

Figure 16: Calibration curve between set temperatures and average 24


hue with σT(1) for uncertainty

Figure 17: Calibration curve between set temperatures and average 25


hue with σT(ave) for uncertainty

Figure 18: Temperature uncertainty vs. Hue 26

Figure 19: Lower expect linear temperature curve with the TLC temperature data 27

iii
Figure 20: Higher expect linear temperature curve with the TLC temperature data 28

Figure 21: TLC Temperature data plotted with the combined expected curve 30

Figure 22: Expected curve with combined curve fit 31

iv
Abstract

Thermochromic liquid crystals (TLC’s) have a unique chemical composition that makes

them reflect different wavelengths of the visible light spectrum at different temperatures,

causing them to change their color. This quality makes them a good candidate for

measuring temperatures of fluids in a microchannel. Images of TLC’s in a microchannel

were acquired at different controlled temperatures and their color was quantified based on

average hue values. Their use was tested by comparing the temperature values they

measured with predicted temperature values in a simple configuration. The results

confirmed TLC’s as a useful temperature measurement technique in the microscale.

TLC’s could also be used to simultaneously measure the velocity and temperature in

various microfluidic applications.

v
Introduction

Thermochromic liquid crystals (TLC’s) have been used in the past as a temperature

measurement technique of fluids in the macro scale [4,7]. They can also be used in the

microscale as a temperature measurement technique of stationary fluids in a

microchannel. If this measurement technique proves to be practical, it could be used to

learn more about the temperature and how it changes in various micro-fluidic

applications. For example, many microfluidic devises use thermocapillary flow to throw

a switch in a MEMs relay [6] or to eject liquids through microtubes [1,2,3].

Thermocapillary flow is caused due to a temperature gradient along the microtube where

both air and liquid are present. This causes a surface tension gradient on the air-liquid

interface. The exact details on the air-liquid interface are not fully understood. For

example, the temperature at a particular location on the interface is not equivalent to the

local temperature predicted by heat conduction along the microtube. This is due to the

fact that convective heat transfer exists on the interface. If more knowledge could be

acquired on the temperature gradients that drive the flow, designs for more efficient

microfluidic devices could be made.

Thermochromic Liquid Crystals (TLC’s) are optically active mixtures of organic

chemicals that have chiral structures. While matter conventionally exists in three forms,

solids, liquids, and gases, TLC’s have an intermediary state between the solid and liquid

phases called the liquid crystalline state. When TLC’s are in this state, they possess the

mechanical properties of liquids and the optical properties of crystalline solids. Optically

1
active, chiral neumatic TLC’s in this state are comprised of helical aggregates of

molecules. The aggregates can be thought of as being comprised of ‘sheets’ of molecules.

The molecules in each ‘sheet’ point in a particular direction as shown in figure 1.

Figure 1: TLC's in their solid, liquid, and intermediate states (based on figure 11 of the Hallcrest
Handbook of TLC Technology) [5]

Each molecule is skewed at an angle θ with respect to its neighbors in the adjacent sheets,

separated by a distance λ, directly above and below them. The pitch length is the length

along the axis of the helix through which the direction vector of the molecules has

revolved 360 degrees. Thus, the skew angle θ and the sheet distance λ determine the pitch

length P (see figure 2).

 360 
P= λ .
 θ 

With increasing temperature of the liquid crystals, the skew angle θ increases and the

distance between each sheet λ increases. Depending on the particular TLC formulation,

either the increase in θ dominates over the increase in λ or vice versa. Usually the

2
increase in θ dominates. Increasing θ contributes to decreasing pitch length P while

increasing λ contributes to increasing pitch length P.

The structure of TLC’s forms a volume grating that interacts with incident light. Thus,

they cause Bragg scattering. Depending on the pitch length and angle between the

incident light and reflected light φ, a certain wavelength of light λo will be reflected.

λo = 2P sin φ

Figure 2: The relation between the skew angle and the pitch length of the structure (based on figure 9
of the Hallcrest Handbook of TLC technology)[5]

Thus, TLC’s will have different colors at different angles of reflection and different

temperatures. The TLC colors range through the colors of the visible light spectrum.

Figure 3 shows the relationship between reflected wavelength and temperature of a

temperature sensitive TLC formulation whose pitch length P increases with temperature.

3
Since the TLC’s color plays are reversible, upon cooling TLC’s, the color change

sequence is reversed. Different TLC formulations cause TLC mixtures to have different

color plays. Thus, they can have different starting temperatures and different color play

lengths.

Figure 3: Typical reflected wavelength vs. temperature response of a temperature sensitive TLC
formulation at a constant angle of reflection.

Another property of TLC is circular dichroism. Light that illuminates TLC is resolved by

the unique TLC structure into its two circularly polarized components, one right-handed

and the other left handed. One component is completely reflected while the other is

completely transmitted (see figure 4).

Most TLC mixtures are very susceptible to degradation from UV light, and are also very

sensitive to certain organic chemicals. Thus, TLC mixtures that serve as temperature

indicators are typically microencapsulated. Tiny droplets of TLC are surrounded with a

continuous layer of polymer coating, forming a microcapsule. Microcapsule sizes range

from a few microns to a few milliliters. The microencapsulation process stabilizes and

protects TLC’s.
4
Figure 4: TLC circular dichroism (Based on figure 12(c) of the Hallcrest Handbook of TLC
Technology)[5]

Color Measurement

The pixels in an image of TLC particles would have unique Red, Green, and Blue (RGB)

values that represent the TLC’s apparent colors. These values can be used to determine

the hue, saturation, and intensity (HSI) values for each pixel. Intensity, I, is the sum of

the three components R, G and B.

5
Saturation is a measure of how far a pixel is from white and hue is a measure of the color

(see figure 5). The quantities v1 and v2 are related to the RGB values by

and

After they have been determined, they can be used to determine the values for hue, H,

and saturation, S.

With the hue values for each pixel, an average hue can be calculated for each individual

TLC particle.

Due to the sensitivity of TLC’s to the optical arrangement and aging of the sample,

TLC’s must be calibrated with controlled ambient temperatures.

The hue value obtained from pictures of TLC’s is expected to have two main sources of

uncertainty. The first source of uncertainty would be due to particle-to-particle color

variations. Since no one TLC particle is exactly the same as the other, the calculated hue

values are expected to vary by a determinable amount. The second source of uncertainty

6
would be due to variations in reflected wavelength due to the location of the TLC’s in the

field of view of the camera.

Figure 5: HS vector of a typical pixel. Here the hue value is closest to the red axis, which is at 0
degrees. Thus, this pixel is expected to look red.

After quantifying uncertainties, the usefulness of the measurement technique can be

tested by comparing measured temperature values with predicted temperature values in a

simple configuration.

7
Materials & Methods

A microencapsulated TLC sample (NSL33R35C5W, Hallcrest), with a nominal color

play range of 35 oC to 45 oC was used. The average TLC microcapsule diameter in the

sample was approximately 10 to 15 microns, but the sample had a full distribution of

microcapsule diameters ranging from 1 micron to 30 microns.

The microchannel used in the experiment was a fused silica capillary tube (see figure 6).

The center of the tube was void of the Polyamide coating, making it transparent so that

the particles in that section of the tube are visible to the microscope.

Figure 6: Square flexible silica capillary tube

An Aluminum block was used to control the temperature around the microchannel. It had

five temperature sensors along its axial direction, and had a slot through which the

capillary tube was designed to be inserted (see figure 7). The three temperature sensors in

the center of the block were thermistor probes from U.S Sensor Corp with an uncertainty

of ±0.02 Co. The middle sensor was 1.5 mm away from the other two sensors along the

axial direction of the block, and was placed in the center of the block. Three alignment

marks were made on the


8
other side of block, where the axial slot is located, to indicate the axial location of the

three middle sensors (figure 7). The shop provided no uncertainty estimate of how well

the alignment marks line up with the sensors, but a conservative estimate for this

uncertainty is ±1 mm. The two temperature sensors on the outer edges of the block were

FerroTech Corp part numbers 201347 with an uncertainty of ±0.1 oC.

To control the temperatures, six Peltier devices wired in series were mounted on the two

ends of the block. They heated or cooled the surface of the block, depending on the

direction of the current through them. Each set was controlled by a bipolar PID

temperature controller (FTC100, FerroTec Corp) allowing the temperature to be

increased and decreased. The feedback input for the temperature control was the

thermistors at the outer edges of the block.

9
Figure 7: Block used to control the temperature

10
The microscope used was an Olympus IX81 with a halogen lamp. The microscope was

used at its 40X magnification. The microscope stage was enclosed by a Plexiglas

incubator. The stage could be moved manually in the two horizontal directions as shown

in figure 8. A 3CCD, Progressive scan, color camera (Toshiba, IK-TF7C ) was connected

to the microscope’s left side port. The camera acquired a smaller field of view than that

of the eye ports, but centered at the same location (see figure 8).

The images were captured at a computer using a frame grabber (X64-CL_Express_1,

Dalsa Coreco), and a CamExpert imaging software.

As mentioned in the introduction, TLC particles reflect left-handed circularly polarized

(LHCP) light without changing the sense of polarization and transmit right-handed

circularly polarized (RHCP) light, but other surfaces such as the edges of the

microchannel and the encapsulation of the TLC’s reverse the handedness upon reflection.

11
Since the TLC color is the only part of the image that should be interpreted as the data,

the microscope was configured to illuminate with only LHCP light and to allow only

LHCP light to reach the camera. With this configuration, ideally only the TLC’s would

be visible in the images.

The Halogen lamp produces unpolarized light. Unpolarized light contains a uniform

distribution of random linear polarizations. A polarizer (47216, Edmund Optics) and a ¼

wave plate were placed in the illumination path. The light that comes from the lamp

passes through the polarizer and, due to the orientation of the polarizer, one linear

polarization (P1) is transmitted, while the orthogonal sense of polarization (P2) is blocked.

The light then passes through the ¼ wave plate. The ¼ wave plate is oriented with its fast

axis rotated 45o away from the orientation of the polarizer. This causes equal parts of the

light to travel at the fast and slow speeds of the crystal. When the light exits the ¼ wave

plate, the part of the light that travelled slower is a ¼ of a wave length behind the part

that has travelled faster. As a result, the linearly polarized light (P1) that passes through

the 1/4 wave plate becomes left-handed circularly polarized. Right-handed circularly

polarized light would be obtained by rotating the ¼ plate 90o about the optical axis. The

light then reaches the TLC mixture in the microchannel.

The reflected light off of the TLC’s remains LHCP while the reflected light off of the

other surfaces is reversed and becomes RHCP. The reflected light passes back through

the ¼ plate and is resolved back into its linearly polarized components, with LHCP light

becoming P1 and RHCP light becoming P2. The light then passes through the polarizer,

12
causing only the P1 component of the light to reach the camera (see figure 9). Each letter

in figure 9 labels the type of light in the illumination path.

Figure 9: (a) Unpolarized light. (b) One sense of linearly polarized light, P1 (c) LHCP light. (d) Both
LHCP and RHCP light. (e) P1 and P2 linearly polarized light (f) Only P1 linearly polarized light.

This configuration ideally increases the signal to noise ratio to infinity. However, in

reality there are three imperfections in the setup that cause the encapsulations of the TLC

and the sides of the microchannel to still be somewhat visible in the field of view of the

camera. First, the polarizer has a finite extinction ratio. Second, the fast axis of the ¼

wave plate cannot be oriented exactly 45o away from the orientation of the polarizer.

Third, the ¼ wave plate separates equal parts of the light by exactly ¼ of a wave length at

only one wavelength of the visible light spectrum. These imperfections cause some

RHCP light to reach the microchannel. As a result, a small amount of the reflected light

13
off surfaces other than TLC’s comes back through the ¼ wave plate and the polarizer and

reaches the camera as P1.

The effect of placing the polarizer and ¼ wave plate in the illumination path is shown in

figure 10.

With all the equipment configured to take images, TLC mixture was inserted into the

microchannel. A syringe was filled with the TLC mixture. A fitting assembly (P-662,

Upchurch Scientific) was then used to connect the capillary tube to the syringe. Next, a

Harvard PHD Ultra syringe pump was used to steadily infuse the TLC mixture into the

microchannel.

The syringe was disconnected and the fitting and the capillary tube were placed inside the

aluminum block. The microchannel was inserted into the slot. The transparent part of the

microchannel was centered in the block. Two pieces of double sided tape were then used

to attach the microtube to the block (see figure 11).

All sides of the block were covered with insulating tape except for the heater sides and

the side which is shown face up in figure 11. This side was insulated using a sheet of

Mica Tape. It was taped to the block with the double sided tape once the microchannel

was in place. Finally, the block and microchannel were placed on the microscope stage.

The microchannel was placed so that the x axis illustrated in figure 7 would make the

smallest angle possible with the axis of the direction of the stage movement.

14
Figure 10: (1) Image of TLC microcapsules near the edge of the microchannel, (2) same setup but
with the polarizer and ¼ wave plate in the illumination path. Notice the edges of the microchannel
and encapsulations are nearly invisible in (2)

15
Figure 11: Block with the microchannel inserted

For the calibration, both sides of the block were set to the same temperature and images

of the TLC mixture along the microchannel were taken. The stage was moved after each

image was taken so that the field of view of the camera would cover a new section of the

microchannel. This process was carried out at temperatures from 35 oC to 45 oC in

approximate increments of 0.4 or 0.6 degrees. For each temperature, 75 to 100 images

were taken, yielding 300 to 700 TLC particles.

To test the measurement technique, two linear temperature gradients were imposed in the

middle of the block, one higher and one lower. To impose the lower temperature

gradient, one side of the block was set to 31.2 oC and the other side was set to 44.4 oC. To

impose the higher temperature gradient, one side of the block was set to 33.9 oC and the

16
other side was set to 46.6 oC. From the principle of heat conduction through the shape

and material of the block, a linear temperature gradient was predicted to be present at

least 0.5 mm away from the center of the block in each direction. Thus, for each

temperature gradient imposed, an expected linear temperature curve was plotted. The

temperature sensors in the center gave two points through which the expected linear

curves were fit on an x vs. T graph. For the lower temperature curve, the temperature

range was 35.91 ± 0.02 oC to 40.08 ± 0.02 oC. For the higher temperature curve, the

temperature range was 38.62 ± 0.02 oC to 42.76 ± 0.02 oC.

Next, for each temperature gradient, images of the TLC mixture were taken along the

microchannel. First, an image was taken in the center of the block as indicated by the

center alignment mark. Following this, images were taken by moving the stage towards

one side of the microchannel. One overlapping TLC microcapsule was included in every

two images in order to keep track of the distance in pixels that the field of view of the
r
camera has gone, S (See figure 12).

Images were taken until the camera’s FOV has moved 0.5 mm away from the center. The

same procedure was done on the other side of the microchannel. In order to know when

the camera had travelled 0.5 mm away from the center, the pixel to length conversion

factor was determined. A dual axis linear scale was imaged and the conversion facto was

found to be 0.116 ± 0.002 µm/pix.

17
Figure 12: By keeping one TLC microcapsule in two images taken along the microchannel , the
distance in pixels that the field of view of the camera has moved from the fist image can be
r r r
determined: S = I 1 − I 2

After the images for both the calibration and the experiment were acquired, an algorithm

was implemented in Matlab to process the images and determine for each TLC

microcapsule found in the images the distance in pixels from the left edge of the image,

X, the distance in pixels from the top edge of the image, Y, and average hue value H.

Refer to figure 13 for a flowchart outlining the main operations of the algorithm.

The algorithm receives an image of TLC particles, converts the RGB values for every

pixel into HSI values and saves it as a 3D array. From the HSI data for every pixel, it

filters out the pixels which have intensity and saturation lower than a given threshold.

Next, it averages the saturation around each remaining pixel over a square centered at the

pixel. The square will sit on a certain amount of pixels and its size is determined based on

an input integer ‘rad’. ‘rad’ is half the length of the square minus one pixel.

18
Figure 13: Flowchart of the algorithm used to determine x, y and average hue for TLC microcapsules
in the images.
19
The algorithm then further filters out the pixels of which the average saturation of the

pixels around it was lower than a given threshold. With this step completed, the algorithm

has the X and Y values of the pixels of interest. However, if two retained pixels are too

close to each other, they most likely sit on the same TLC particle. As a result, the

algorithm then groups different pixels if they are close to each other by a given distance

in pixels. Finally, after it has them in groups, it looks for the pixel in each group which

has the highest intensity and drops the rest of the pixels in each group. If two pixels in a

group both have the highest intensity, the algorithm chooses the one which has a higher

saturation. The pixels that are left each represent a TLC particle whose hue is important.

The next step is to average the hue based on all of the pixels in a square around that

location. If the saturation of a pixel in the square is lower than a given threshold, than the

algorithm does not include it in the averaging process.

Finally, the algorithm reports the number of TLC particles found, and displays each one

representing pixel’s X, Y and average hue values.

As part of the calibration, after the X, Y and H data for every TLC microcapsule at each

set temperature from 35 C to 45 C was acquired, the uncertainty was quantified. First, the

uncertainty due to particle-to-particle variations was quantified. For each set temperature

value, an average hue, H_ave, was calculated and a histogram was made from the H data.

In addition, the uncertainty of each individual particle hue values, σH(1), and the

uncertainty of the averaged hue value, σH(ave), were determined using two standard

deviations, 2σ.

20
σH (1) = 2σ

σH (ave) =
n particles

Following this, the uncertainty due to variations in reflected wavelength due to the

location of each given TLC in the field of view was quantified. For each set temperature

value, the hue data was divided into 12 groups. Each group corresponded to a different

section in the field of view. For each group, an average hue was determined and was

given the corresponding uncertainty using two standard deviations. Since no noticeable

trend between the particles’ group and their average hue was found, TLC location in the

field of view was eliminated as a source of uncertainty.

To complete the calibration, a T vs. H_ave graph was made with linear interpolations

between the calibration points. The uncertainty in hue, both for an individual particle’s

hue and average hue was used to determine the uncertainty in temperature,

dT
σT (1) = σH (1)
dH
dT .
σT (ave) = σH (ave)
dH

To complete the experiment, the images that were taken when the temperature gradients

were imposed on the block were processed to acquire the X, Y and H data. The X and Y

data were used to determine each particle’s location in the axial direction of the

microchannel. As described above, the first image taken was in the center of the block,

and one of the TLC particles in that image was called the origin O, (see figure 14).

21
r r r r r
Figure 14: O , I and S are in image coordinates. O has known X and Y values, O = ( X o , Yo ) .

This figure yields


r r r r
P = I + S −O .

r
P is in spatial coordinates. The x component of the P vector is the axial distance of the

TLC particles from the origin. Next, the H data was converted to temperature data for

each TLC particle using the calibration curve.

22
Results

Calibration Results

Most of the histograms made resulted in a similar shape. Figure 15 shows a typical

histogram.

Figure 15: Histogram of set temperature 37.15 oC.

A typical histogram resembles a Gausian. As the number of data points increased, the

histogram looked more and more like a Gausian. For the particular histogram shown,

H_ave = 131.7o.
The fact that the peak is not at H_ave in figure 18 is atypical. This particular histogram

was selected to emphasize that the hue distribution for the set temperatures is not exactly

a Gausian.

23
The following two figures display the calibration curve with linear interpolations between

calibration points and the corresponding uncertainty. Figure 16 shows the uncertainty in

temperature based on one particle’s hue, σT (1) , and figure 17 shows the uncertainty in

temperature based on an average of particle hues, σT (ave) . The second temperature in

the calibration, 35.4 Co, resulted in a very wide hue distribution. Since its uncertainty in

temperature was very high, the first two set temperatures were excluded from further

analysis and their error bars were not displayed.

Figure 16: Calibration curve between set temperatures and average hue with σT(1) for uncertainty

24
Figure 17: Calibration curve between set temperatures and average hue with σT(ave) for uncertainty

1
As expected, since σT (ave) α , the uncertainty in figure 17 is much smaller
n particles

than the uncertainty in figure 16. nparticles was different for each calibration point. The

reduction in uncertainty from σT(1) to σT(ave) will always depend on the value of

nparticles.

Based upon the measured uncertainties, the T range for the linear temp gradient was

restricted to (35.99 oC - 43.69 oC) to avoid the very high uncertainties at higher

temperatures.

25
The following figure compares the uncertainty in temperature measurement of one

individual particle with the uncertainty in temperature measurement of an average of the

particles. With an increased number of particles, the uncertainty decreases considerably.

Figure 18: Temperature uncertainty vs. Hue

Experiment Results

The temperature data and its uncertainty were plotted with the two expected linear

temperature curves on a T vs. x graph (see figures 19 & 20).

26
Figure 19: Lower expect linear temperature curve along with the TLC temperature data. The blue
line is the linear fit to the data

The slope of the expected curve is 0.416 Co/mm and the slope of the linear fit is 0.393
o
C/mm.

27
Figure 20: Higher expected linear temperature curve along with the TLC temperature data. The blue
line is the linear fit to the data

The slope of the expected curve is 0.414 oC/mm and the slope of the linear fit is 0.474
o
C/mm. The data on the positive x side of figure 25 has a significant amount of outliers.

The reason for this is that at temperatures of around 42 oC or higher, small changes in hue

cause the temperature readings from the calibration to vary significantly. Therefore, that

data was excluded from further analysis.

The offset between the curve fits and the expected curves in both graphs is due to the

uncertainty of how well the alignments marks line up with the three thermistors in the

block.

28
To correct this offset, first the two expected curves were combined into one. The slop of

the combined curve is the weighted average of the curve fit of figure 19 and the curve fit

of the negative x side of figure 20,

2mlow + mhigh
mcomb =
3

Second, to show the full useful range of the TLC’s on one plot, all the TLC temperature

data except for the data on the positive x side of figure 25 was plotted on one T vs. x

graph. The data in figure 20 required a shift in its x-coordinates which was determined

using the starting temperature of the higher expected curve,

Tstart = ( −5 mm) * mcomb + bhigher


Tstart − blower
xshift = + 5 mm
mcomb
The combined TLC temperature data was given a linear curve fit. The temperature value

in the center of the curve fit was placed at x=0. Finally, the y-intercept of the expected

curve was corrected so that it intercepts the center of the curve fit at x=0 (see figure 21).

29
Figure 21: TLC Temperature data plotted with the combined expected curve

The standard error for the combined curve fit was calculated,

n _ particles

∑ [T fit ( xi ) − Ti ]
2

S xT = i =1
= 0.474 o C .
n _ paticles − 2

Where T fit ( xi ) is the temperature value of the curve fit at each data point’s axial

location xi , and Ti is the temperature value of each data point. For a 95% confidence

interval for the curve fit,

σT fit = 2 S xT = 0.948 oC

The following figure compares the expected curve with the combined curve fit with its

uncertainty limits plotted. The expected curve falls well within the uncertainty limits of

the curve fit to the TLC data.


30
Figure 22: Expected curve with combined curve fit. The uncertainty limits are indicated with the thin
green lines

31
Discussion

Using thermochromic liquid crystal mixtures to measure the temperature of stationary

fluid along a microchannel produces results that are consistent with prescribed

temperatures to within measured uncertainties that depend on the number of TLC

particles sampled. Thus, for the TLC formulation used in this study, this measurement

technique is practical between 36.30 oC and 41.30 oC. The calibration curve showed that

this measurement technique produces small uncertainties starting at 120 degrees of hue to

nearly 240 degrees of hue. Thus, the practical color play range for this TLC formulation

is from pure green to nearly pure blue. Other TLC formulations are expected to have a

similar practical color play range but a calibration would be necessary to test that.

This measurement technique has room for further improvement. The calibration and the

experiment have related uncertainties. These uncertainties can be minimized by focusing

on three main aspects:

1) Filtering the TLC mixture to result in a more uniform distribution of particle sizes

2) Further improving the method of illumination of the particles in the microchannel

to filter out everything but the relevant, color changing, part of the TLC microcapsules

3) Optimizing the image processing algorithm written to quantify the apparent

particle colors

TLC mixtures could also be used as a velocity measurement technique of fluids flowing

in a microchannel. Flow could be generated in the microchannel and the TLC

32
microcapsules could be used as tracer particles in a micro particle image velocimetry

(µPIV) application. Ultimately, this measurement technique could be used to

simultaneously measure the velocity and temperature in various microfluidic applications

such as thermocapillary flow.

33
Conclusion

Using TLC mixtures to measure the temperature in microchannels has been proven to be

practical. After making improvements to the measurement technique, it can be used for

simultaneous velocity and temperature measurements to acquire data relevant to the

design of various microfluidic devices.

Thermocapillary flow could be generated by injecting an air bubble into the microchannel

and imposing a temperature gradient on the microchannel using the aluminum block. The

bubble will move towards the hotter side of the microchannel. The velocity and

temperature of the air/water interface could be measured during this motion and a model

relating the temperature gradients and the forces they cause could be found.

34
Bibliography

[1] Burns, M.A., C.H. Mastrangelo, T.S. Sammarco, F.P. Man, J.R. Webster, B.N.
Johnson, B. Foerster, D. Jones, Y. Fields, A.R. Kaiser, and D.T. Burke, "Microfabricated
structures for integrated DNA analysis." Proceedings of the National Academy of
Sciences of the United States of America 93(11): p. 5556-5561, 1996.

[2] Burns, M.A., B.N. Johnson, S.N. Brahmasandra, K. Handique, J.R. Webster, M.
Krishnan, T.S. Sammarco, P.M. Man, D. Jones, D. Heldsinger, C.H. Mastrangelo, and
D.T. Burke, "An integrated nanoliter DNA analysis device." Science 282(5388): p. 484-
487, 1998.

[3] Meinhart, C.D. and H.S. Zhang, "The flow structure inside a microfabricated inkjet
printhead." Journal of Microelectromechanical Systems 9(1): p. 67-75, 2000.

[4] Park, H.G. and M. Gharib, "Experimental study of heat convection from stationary
and oscillating circular cylinder in cross flow." Journal of Heat Transfer-Transactions of
the Asme 123(1): p. 51-62, 2001.

[5] Parsley, M., The Hallcrest Handbook of Thermochromic Liquid Crystal Technology.
Glenview, IL: Hallcrest, 1991.

[6] Simon, J., S. Saffer, and C.J. Kim, "A liquid-filled microrelay with a moving mercury
microdrop." Journal of Microelectromechanical Systems 6(3): p. 208-216, 1997.

[7] Valentino, J.P., S.M. Troian, and S. Wagner, "Microfluidic detection and analysis by
integration of thermocapillary actuation with a thin-film optical waveguide." Applied
Physics Letters 86(18), 2005.

35

Você também pode gostar