Penn State Chemical Engineering
Undergrad Research
Colloidal Silica Nanoparticles
Alec Myers
Alec Francis Myers
August 2012
Introduction
Nanofluids are making a stride in various engineering applications; they are among the groups of
ever changing materials due to the fact that their properties vary between each situation. This
makes them widely usable in different industrial processes but also makes it difficult from the
sometimes unpredictable characteristics. Nanofluids have begun to show promise as advanced
coolant agents. Early experiments have shown increase in thermal conductivity for well
dispersed, low volume fluids. This is consistent with the Maxwell mean-field theories but more
recent research shows conflicting behavior such as increasing thermal conductivity with
decreasing nanoparticle size, saturation at higher volume fractions, non-correlation to the
intrinsic thermal conductivity of nanoparticles and a relatively large enhancement at low volume
fractions. Some proposals to the unexpected results have included the nanoparticles participating
in Brownian motion, microscale convection, layering of the liquid at solid-fluid interfaces,
differing shapes, cluster effects and combinations of many. [1, 2, 3]
The main idea behind these investigations of nanofluid characteristics is to maximize the data
collection under different constraints and analyze if any pattern or trend is observed; once
concrete evidence is obtained, the predicted results following theory or straying from the
expected will allow us to ask and try to determine why what happened did in fact happen. In
order to understand the effects that particle size has on thermal conductivity enhancement, the
variations seen in aggregate sizes must be explained first; the following experiments deal with
the analyses of sizes based off of different volume fractions and pH levels of nanofluid sample.
Procedures
Each experiment performed was made using Ludox - 40% Silica Solution, deionized water and
simple acid and base. The percentages are all based off of volume fraction in the samples. Each
step was also repeated exactly the same for each sample to minimize the variables changing
between measurements.
1) After mixing the silica and water the sample vials were placed on stir plates.
2) TMP (used as coating) added to the mixtures while stirring occurred; in order to make the
coating process precise and more accurate each drop of TMP was allowed to mix thoroughly
with the sample volume before adding more (approx 15 seconds between drops).
3) The samples were then closed and allowed to stir completely (for 24 hours).
4) The pH of each coated sample was brought to ~9.7 which is the point where they start to turn
into a gel, or aggregate particle size approaches infinity.
5) Each gel was reversed by lowering the pH to the desired measurement.
6) To fully redisperse the particles, the samples were sonicated (for 60 minutes).
7) When the desired samples were obtained, the aggregate particle sizes were measured using a
Light Scatter Machine.
Samples Made

To make 16mL 5% samples, 4mL silica, 12 mL of water and 0.23g TMP were used.

To make 50mL 5% samples, 12.5mL silica, 37.5mL of water and 0.625g TMP were used.

To make 50mL 10% samples, 24.5mL silica, 24.5mL of water and 1.275g TMP were
used.

To make 50mL 20% samples, 50mL silica and 2.55g TMP were used.

To calculate different size samples, the proportion of each material must be the same.
Tests & Results
*Graphical data follows each test summary*
Control Test
To better understand the activity of the Ludox a control test was performed. In this test 2 types of
samples (both of which were not gelled) were ran against each other: 5% samples that were
coated with .23g TMP then brought to pH and 5% samples that skipped the coating/gel process
altogether.
The coated sample shows increase in size at more basic pH while control size is constant (+/-)
5nm. The coating used causes the repulsion to change. The control sample should be expected to
not vary much, which is what was observed.
Control Test
70
65
60
55
Size (nm)
Control
50
Coated
45
40
35
3
5
7
pH
9
Initial Gelation Test (5%, .23g TMP)
The first true test of size vs pH using the gelation process did not yield any significant results.
Each sample was measured from the initial day to 10 days post formation: the growth rates of the
sizes differed greatly and no relationship can be seen between size and pH. This was the last test
made with the year old Ludox however, so the measurements could have reflected breaking
down of the particle integrity overall. (The new Ludox arrived within a week of this experiment,
and it produced more accurate results in general).
[Initial Gelation] 5% Coated Measurements
70
65
60
55
3DAYS
Size (nm)
5DAYS
50
6DAYS
7DAYS
45
10DAYS
40
35
3
4
5
6
7
pH
8
9
10
[Initial Gelation] pH 3.5
40
39
38
Size (nm) 37
36
35
34
3
5
7
Time (Days)
9
11
[Initial Gelation] pH 7.04
58
56
54
52
50
Size (nm)
48
46
44
42
40
3
5
7
Time (Days)
9
11
Stir Test (5%, .23g TMP)
An interesting side note was brought up concerning the coating possibly reacting with itself or
not coating the nanoparticles evenly. One variable that was assumed constant before, the stirring
speed was tested due to that inquiry.
As seen by the graphs, the samples stirred at a lower speed had higher sizes (on average). No
conclusive decision was made either way regarding the stir speed’s importance though; no
further experiments focused on it but each one carried out from here on used the same setting.
[Stir Test] 1Day
175
155
135
Size 115
(nm)
95
1500 Speed
800 Speed
75
55
35
3
5
7
pH
9
[Stir Test] 4 Days
115
105
95
85
Size
(nm) 75
65
1500 SPEED
800 SPEED
55
45
35
3
5
7
9
pH
[Stir Test] 6 Days
49
47
45
Size 43
(nm)
41
1500 SPEED
800 SPEED
39
37
35
3
5
7
pH
9
TMP Gelations
With the new gelation process taking longer, investigation into the easiest samples to gel and
redisperse was started. The differing factors were made from changing values of TMP added.
Samples were tested with .1g, .15g, .18g and .2g of TMP. Each set of samples were gelled,
brought to pH’s of 6, 7 and 8 and sonicated to test reversibility. Samples that could not
redisperse are chemically bonded and no amount of sonication would be able to break them
apart. The lower amounts of TMP gelled very fast but were not redispersable at higher pH. The
.1g and .15g only ‘dissolved’ at pH 6 and below. For the .18g samples the sonication required to
break apart the molecules was too great, so the .2g of TMP was decided to be the most effective.
The .2g is a good amount in terms of enough coating to gel fast enough, but not too much to
oversaturate the system and react/bond with itself.
TMP Test (5%, .15g / .2g TMP)
The next test was taken to see if the relationship between amount of coating and size was
significant enough to be a considered variable. Both .15g and .2g TMP samples were measured
over a period of time and each day the .2g measurements showed a higher aggregate size. These
results describe that more coating in the system leads to bigger particle aggregates; it is not sure
whether this could be attributed to partially coated particles in lower coating amounts or the
TMP reacting with itself in higher coating amounts.
[TMP Test] 1 Day
635
535
435
Size
(nm) 335
.15g
.2g
235
135
35
3.5
4.5
5.5
6.5
7.5
8.5
pH
[TMP Test] 4 Days
635
535
435
Size
(nm) 335
.15g
.2g
235
135
35
3.5
4.5
5.5
6.5
pH
7.5
8.5
Stir Bar Test (5%, .23g TMP)
Stir bar test was made to check random, uncontrollable variable. The importance of coating
process was tested by using the same stir bar for the coating and first 10 minutes of stirring. Each
sample was made separately but the coating stir bar used was the same. All other variables were
controlled as usual.
Sizes obtained were approximately equivalent. It can be deduced that stirring during the coating
process is VERY IMPORTANT part of making the samples.
[Stir Bar Test] 1 Day
100
95
90
Size 85
(nm)
80
75
70
7.4
7.6
7.8
8
pH
8.2
8.4
[Stir Bar Test] 4 Days
115
110
105
Size 100
(nm) 95
90
85
80
7.4
7.6
7.8
8
8.2
8.4
8.2
8.4
pH
[Stir Bar Test] 10 Days
135
130
125
120
Size 115
(nm) 110
105
100
95
90
7.4
7.6
7.8
8
pH
Reproducability Test (5%, .23g TMP)
Large variations in measurements were seen throughout the tests often. To test the
reproducibility of the samples as well as random uncontrollable variables, 3 equivalent samples
were made on the same day the exact same way from initial mixing to pH adjustments.
Not many similarities could be drawn between any of the samples. Sporadic sizes mean the
variable(s) affecting variation is not some superfluous procedural error.
[Reproducible Test] 1 Day
200
180
160
140
120
Size
100
(nm)
80
60
40
20
0
5
6
7
pH
8
9
[Reproducible Test] 3 Days
250
200
Size
(nm)
150
100
50
0
5
6
7
pH
8
9
8
9
[Reproducible Test] 7 Days
300
250
200
Size
150
(nm)
100
50
0
5
6
7
pH
Split Fractions Test (5%, 10% and 20%)
The final test performed analyzed the relationships between volume fraction and size for each
pH. Separate graphs comparing the growth of each basic pH over time also was related to each
volume amount.
Test comparing growth and aggregate size of 5% 10% and 20% volume fractions. Each coated
for 5 minutes with same stir bar; 50mL were jar made of each %. The jars’ contents were then
split into vials for each pH of each volume fraction to maximize the similarities of each mixture
post-coating and while changing the pH’s.
As seen from the size vs pH graphs, the 5% volume fraction gelled fastest, changed more rapidly
at acidic pH’s and had the highest aggregate particle size at each pH. On the other hand the 20%
reflected the opposite for each category. The samples above pH 9.0 eventually maximized to
infinity (gel) but the lower volume fractions reached those points quicker.
The growth graphs also reflect the same observations. The basic pH’s were analyzed more
closely due to those samples changing significantly for each volume fraction. The 5% grew the
fastest and the 20% grew the slowest. Another important note by pH 9.5 is that by day 3 each 5%
and 10% sample had fully solidified; the 20% sample still was increasing and eventually would
have grew to particle size infinity.
[Split Fractions] Right After Coat
535
485
435
385
335
Size
285
(nm)
235
185
135
85
35
5%
10%
20%
3
5
7
9
pH
[Split Fractions] 6 Days Measurements
485
435
385
335
Size 285
(nm) 235
5%
10%
185
20%
135
85
35
3
5
7
pH
9
[Split Fractions] pH 8.5 Growth
250
230
210
190
170
Size
150
(nm)
130
110
90
70
50
5%
10%
20%
1
2
3
4
Time (days)
5
6
[Split Fractions] pH 9 Growth
400
350
300
Size 250
(nm)
200
5%
10%
20%
150
100
50
1
2
3
4
Time (days)
5
6
[Split Fractions] pH 9.5 Growth
600
500
Size
(nm)
400
5%
10%
300
20%
200
100
1
2
3
4
Time (days)
5
6
Conclusions
Many simple relationships and key facts about colloidal nanoparticle function can be interpreted
out of the obtained results.

Control volume samples (without coating) have ~ the same size at each pH. Therefore,
the coating material used must respond to the changes in basicity and orient/bond in
different ways at each side of the pH scale.

There must be some material in the overall Ludox solution that adds an effect to the
coating response or gelation process, because the newer solution reacted different
compared to the samples made from the year old bottle.

Stir speed as well as stir bars have an effect on the size; this is most likely attributed to
the effectiveness of coating process when kept as constant as possible

The amount of TMP used changes the results greatly

When very low amounts of TMP are used the particles are not reversible/dispersable after
gelation. And when large amounts are used the gelation process is very hard to start.

Increased TMP samples (that are still able to gel) result in higher aggregate sizes.

It is very hard to reproduce the same results, even while controlling every possible
variable. This means that some unknown item is affecting the changes in results for each
sample.

The lower % volumes respond more to lower pH’s while the higher % particles only
increase in size at highly basic pH’s.

The lower % volumes increase more linearly in size below the 8-9 pH threshold, while
the higher % volumes’ sizes stay constant till hitting the threshold. When any sample hits
that 8 or 9 pH point however the sizes of all increase exponentially.
All of those remarks may seem unimportant by themselves, but when combined they reveal a
greater idea of the nature behind the nanofluid particles’ functions. From the observations made,
it can be recognized that some extraneous factor is coming into account on the Nanoscale for
each sample; there is probably chemical as well as physical variables at play, but they are
unknown at this point. The many issues and unexpected changes in results observed from only a
handful of experiments convey the unknown characteristics and mystery of nanofluid particles,
but some of the relationships and overall trends seen among each group can begin to tie together
the pieces.
References
[1] Kleinstreuer, Clement and Feng, Yu. Experimental and theoretical studies of nanofluid
thermal conductivity enhancement: a review. 2011.
<http://www.nanoscalereslett.com/content/pdf/1556-276X-6-229.pdf>
[2] Eapen, Jacob, Rusconi, Roberto, Piazza, Roberto and Yip, Sydney. The Classical Nature of
Thermal Conduction in Nanofluids. 2008.
<http://arxiv.org/ftp/arxiv/papers/0901/0901.0058.pdf>
[3] Eapen, Jacob, Li, Ju and Yip, Sydney. Beyond the Maxwell Limit: Thermal Conduction in
Nanofluids With Percolating Fluid Structures. 2007.
<http://arxiv.org/ftp/arxiv/papers/0707/0707.2164.pdf>