HIGH SURFACE AREA AEROGELS FOR ENERGY STORAGE AND EFFICIENCY
by
Ryan Patrick Maloney

A DISSERTATION
Submitted
to Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
MATERIALS SCIENCE AND ENGINEERING
2012

ABSTRACT
HIGH SURFACE AREA AEROGELS FOR ENERGY STORAGE AND EFFICIENCY
by
Ryan Patrick Maloney
It is becoming increasingly self-evident that energy efficiency, generation and storage will be one
of the defining technological challenges for our society in the coming decades. Aerogels, with their
high surface area, tailorable microstructure and ease of sol-gel processing, are uniquely suited to
assist by both increasing the efficiency of thermal systems and improving power- and energydensities of electrochemical energy storage systems. In this spirit, this dissertation describes the
fabrication and characterization of a variety of aerogels for such applications. It is shown that
aerogels perform well in both situations, and that further work is warranted to take full advantage
of their unique properties.
The dissertation is divided into two main chapters, each focused on a different application for
aerogel. The first chapter concerns the development of silica aerogel for thermal insulation. It
begins with initial characterization of a silica aerogel insulation for a next-generation Advanced
Stirling Radioisotope Generator for space vehicles. While the aerogel as made performs well, it is
apparent that further improvements in mechanical strength and durability are necessary.
The chapter then continues with the exploration of chlorotrimethylsilane surface modification,
which somewhat surprisingly provides a drastic increase in mechanical properties, allowing the
inherently brittle silica network to deform plastically to >80% strain. It is hypothesized that the
hydrophobic surface groups reduce capillary forces during drying, lowering the number of microcracks that may form and weaken the gel.
This surface modification scheme is then implemented in a fiber-reinforced, opacified aerogel
insulation for a prototypical thermoelectric generator for automotive waste heat recovery. This is
the first known report of aerogel insulation for thermoelectrics. The aerogel insulation is able to decrease the parasitic heat flux of the thermoelectric generator by 30% compared with a commercial

high-temperature insulating wool. Unfortunately, the supercritical drying process adds significant
cost to the aerogel insulation, limiting its commercial viability.
The chapter then culminates in the development and characterization of an Ambiently Dried
Aerogel Insulation (ADAI) that eliminates the need for expensive supercritical drying. It is believed that this report represents the first aerogel insulation that can be dried without undergoing
a large volume change before "springing back" to near its original volume, which allows it to be
cast into place into complex geometries and around rigid inclusions. This reduces a large barrier
to the commercial viability of aerogel insulation. The advantages of ADAI are demonstrated in a
third-generation prototypical thermoelectric generator for automotive waste heat recovery.
The second chapter then details two different aerogel-based materials for electrochemical energy storage. It begins with lithium titanate aerogel, which takes advantage of the high surface area
of the aerogel morphology to display a batt-cap behavior. This should allow the lithium titanate
aerogel to perform at higher rates than would normally be expected for the bulk oxide material.
Additionally, the flexibility of the sol-gel process is demonstrated through the incorporation of
electrically conductive high-surface area exfoliated graphite nanoplatelets in the oxide.
The last section describes the characterization of a LiMn2 O4 spinel coated carbon nanofoam
in a non-aqueous electrolyte. The short diffusion path, high surface area and intimately wired
architecture of the nanofoam allows the oxide to retain its capacity at significantly higher rates
when compared with literature values for the bulk oxide. Additionally, the nanometric length scale
improves cycle life, and the high surface area dramatically increases the insertion capacity by
providing a higher concentration of surface defects.
Taken together, it is clear that aerogels are an extremely attractive class of material for applications pertaining to energy and efficiency, and further research in this area will provide valuable
solutions for pressing societal needs.

Copyright by
Ryan Patrick Maloney
2012

To my family and friends, thank you.

v

ACKNOWLEDGEMENTS

First and foremost, I must thank my advisor Dr. Jeff Sakamoto for his advice and guidance.
Without his coaching, and the numerous opportunities he has provided, this work would not have
been possible. I would also like to thank Dr. Larry Drzal, Dr. Don Morelli and Dr. Harold Schock
for their time and guidance as my committee members.
I extend a special thanks to all my lab mates over the years: Dan Lynam, Ezhiylmurugan Rangasamy, Apoorv Shaligram, HyunJoong Kim, Long Zhang, Travis Thompson and Dena Shahriari,
as numerous excellent undergraduate coworkers. You have all made this lab a fun, supportive
environment to work in.
Additionally, I must give thanks for their assistance at various points in my work to Mike Rich,
Ed Drown, Brian Rook and Per Askeland at the Composite Materials and Structures Center at
MSU. I’d like to also thank Ed Timm, Trevor Ruckle and Matthew Lyle at the Energy and Automotive Research Center at MSU for their help and collaboration on the thermoelectric projects.
I also thank Carol and Stanley Flegler at the Center for Advanced Microscopy, not only for their
assistance and inexhaustible patience, but for their generosity in making Michigan feel like more
like home for those students whose families are far away.
Finally, I thank my friends and family for their patience and unyielding support. This was a
long process, and it would not have been possible if not for the love given by those closest to me.
Thank you.

vi

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

x

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

CHAPTER 1

1

OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CHAPTER 2 INTRODUCTION . . . . . . . . . . . . . .
2.1 Energy and efficiency needs . . . . . . . . . . . . . .
2.2 Aerogel history . . . . . . . . . . . . . . . . . . . .
2.3 Sol-gel synthesis . . . . . . . . . . . . . . . . . . .
2.3.1 Metal-alkoxide hydrolysis/condensation . . .
2.3.2 Resorcinol-Formaldehyde polymerization . .
2.3.3 Alternative sol-gel techniques . . . . . . . .
2.4 Drying methods . . . . . . . . . . . . . . . . . . . .
2.4.1 Supercritical drying . . . . . . . . . . . . . .
2.4.2 Freeze drying . . . . . . . . . . . . . . . . .
2.4.3 Ambient drying . . . . . . . . . . . . . . . .
2.5 Electrochemical Characterization . . . . . . . . . . .
2.5.1 Charge storage mechanisms . . . . . . . . .
2.5.2 Cyclic voltammetry . . . . . . . . . . . . . .
2.5.3 Galvanostatic cycling (Chronopotentiometry)

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

CHAPTER 3 SILICA AEROGELS FOR THERMAL INSULATION . . . .
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Fiber-Reinforced Aerogel for Advanced Stirling Radioisotope Generator .
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Large Deformation of Silica Aerogels via Surface Modification
Chlorotrimethylsilane . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.2 Experimental Techniques and Methods . . . . . . . . . . . . . . .
3.3.2.1 Sample Preparation . . . . . . . . . . . . . . . . . . . .
3.3.2.2 Sample Characterization . . . . . . . . . . . . . . . . .
3.3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

5
5
6
8
8
10
11
11
12
15
15
16
17
21
25

. . .
. . .
. . .
. . .
. . .
. . .
. . .
with
. . .
. . .
. . .
. . .
. . .
. . .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

27
27
30
30
34
36
39

.
.
.
.
.
.

.
.
.
.
.
.

41
41
42
42
43
44

3.4

3.5

3.3.3.1 Composition and Morphology . . . . . . . . . . . . . . . . .
3.3.3.2 Sintering and Mechanical Properties . . . . . . . . . . . . . .
3.3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Aerogel Insulation for a Prototypical Skutterudite-based Thermoelectric Generator for Automotive Waste Heat Recovery . . . . . . . . . . . . . . . . . . . .
3.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.2.1 Aerogel Encapsulation . . . . . . . . . . . . . . . . . . . . .
3.4.2.2 Characterization . . . . . . . . . . . . . . . . . . . . . . . .
3.4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Ambiently Dried Aerogel Insulation . . . . . . . . . . . . . . . . . . . . . . .
3.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.2.1 Chemicals . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.2.2 Gel preparation and drying . . . . . . . . . . . . . . . . . . .
3.5.2.3 Characterization . . . . . . . . . . . . . . . . . . . . . . . .
3.5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CHAPTER 4 AEROGELS FOR ELECTROCHEMICAL ENERGY STORAGE
4.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Lithium Titanate Spinel Aerogel for Next-Generation Lithium-ion Batteries .
4.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.2.1 Sample Preparation . . . . . . . . . . . . . . . . . . . . . .
4.2.2.2 Characterization . . . . . . . . . . . . . . . . . . . . . . .
4.2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 Lithium Manganese Oxide Spinel Coated Carbon Nanofoams . . . . . . . . .
4.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.

44
45
50
52

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

53
53
54
54
56
59
63
64
64
67
67
67
68
69
80

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

81
81
83
83
84
84
86
86
91
95
95
97
98
103

CHAPTER 5

CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

CHAPTER 6

SUGGESTIONS FOR FUTURE WORK . . . . . . . . . . . . . . . . . 111

viii

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

ix

LIST OF TABLES

3.1

Shrinkage of control and CTMS treated samples due to drying and subsequent heat
treatment. "Dried" shrinkage is measured with respect to the wet gel volume, while
500◦ C and 700◦ C shrinkage values are measured with respect to the dried gel. . . . . . 44

3.2

Physical properties of control and CTMS treated samples. . . . . . . . . . . . . . . . . 45

3.3

Heat flux measurement parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

x

LIST OF FIGURES

2.1

Future liquid fuel demand and supply projections, from ref. [1]. (For interpretation
of the references to color in this and all other figures, the reader is referred to the
electronic version of this dissertation.) . . . . . . . . . . . . . . . . . . . . . . . . . .

6

2.2

A simplified procedure for forming an aerogel. . . . . . . . . . . . . . . . . . . . . . .

8

2.3

Silica aerogel, with a typical "pearl necklace" structure. . . . . . . . . . . . . . . . . .

9

2.4

Various ways a sol-gel can be dried; Supercritically, to form an aerogel (top); Ambiently, without any modification, to form a dense xerogel (second row); Ambiently,
with modification and solvent exchange, to form a slightly less dense porous aerogel
(third row). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.5

Phase diagram detailing the three main ways a gel can be dried: Freeze Drying,
Ambient Drying, and Supercritical Drying. . . . . . . . . . . . . . . . . . . . . . . . . 14

2.6

A simplified diagram of a Graphite/LiCoO2 cell, showing both charge and discharge
processes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.7

Example Cyclic Voltammetry plot for a Faradaic charge storage mechanisms (Data
from Section 4.2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.8

Example Cyclic Voltammetry plot for a Capacitive charge storage mechanisms (Data
simulated) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.9

Example Cyclic Voltammetry plot for a Pseudocapacitive charge storage mechanisms (Data from Section 4.2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.10 A typical chronopotentiometry plot, showing the voltage response of an electrode
cycled at three different current densities. . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1

Modes of thermal transport in an aerogel, and how the aerogel structure selectively
reduces each. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.2

Thermal conductivity of aerogel insulation developed at JPL, as a function of titania
content. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

xi

3.3

Aerogel insulation developed at JPL in a heat-dump test apparatus, before (top) and
after (bottom) being subjected to a simulated runaway thermal event. The excessive
heat caused the aerogel to sinter, opening pathways for thermal energy to vent into
the environment. The interior reached a maximum temperature of 1040◦ C, well
within the required specification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.4

The anisotropic nature of the aerogel insulation can be clearly seen in this photograph of two aerogel samples, as the sample on the right has become oval in crosssection after heating. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3.5

After heating to 950◦ C in vacuum, the aerogel shows a reduction in the number of
pores below 1 nm wide, which is indicative of preferential sintering of the necks
between adjacent silica particles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.6

Silica aerogels reinforced with various loadings of chopped fibers. . . . . . . . . . . . 38

3.7

Effect of chopped fiber loading on sintering of silica aerogel insulation as a function
of temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.8

Effect of chopped fiber loading on mechanical properties of sintered silica aerogels.
Data from aerogel samples provided by JPL is given as a reference. . . . . . . . . . . . 40

3.9

SEM micrographs of CTMS Treated and Control samples, before and after heating
at 500◦ C. All images are at the same magnification. . . . . . . . . . . . . . . . . . . . 46

3.10 Differential Pore Volume Distribution from Nitrogen Adsorption (<5nm pores, HK
method). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.11 FTIR spectra of all samples. CTMS modified samples show residual organic peaks
after heating at 500◦ C for three hours, but after 700◦ C for one hour all organics are
removed and the spectra matches that of the control heated at 500◦ C. . . . . . . . . . . 48
3.12 Stress vs. strain compression curves of all samples. Even after baking at 500◦ C,
CTMS modified samples still exhibit impressive mechanical properties. Heating at
700◦ C removes all organics and the sample exhibits the same strain to failure as the
control heated at 500◦ C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.13 Image of both a CTMS modified sample (left) and control sample (right) after compression testing. Both samples underwent the same strain. Note the extensive damage in the control, which is not present in the modified sample. . . . . . . . . . . . . . 50

xii

3.14 Schematic of CTMS modification. The CTMS creates a conformal coating on the
surface of the silica, filling in and reinforcing any microcracks present. . . . . . . . . . 51
3.15 Challenges in thermal management of terrestrial TEGs. Thermal insulation for a
terrestrial TEG must prevent conductive heat transport (left), withstand high temperatures (middle), and be cast into place around complex geometries (right). . . . . . 54
3.16 Image of the process for integrating the aerogel in the TEG . . . . . . . . . . . . . . . 55
3.17 The custom apparatus used to measure heat flux in thermoelectric unicouples . . . . . 56
3.18 Parasitic heat loss was measured on two types of samples: Aerogel insulated (top),
and an uninsulated control (bottom) . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.19 Parasitic heat loss comparison of uninsulated skutterudite couples and aerogel insulated couples, showing that the aerogel insulation is effective in a test setup that
approximates real working conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.20 Heat flux and thermal efficiency testing: the heated gas is directed via a manifold
to impinge directly upon the couples (a), the assembled TEG is instrumented with
thermocouples in the HCs and in the cooling fluid located before and after each
panel (b), the fully assembled TEG test bed (c). . . . . . . . . . . . . . . . . . . . . . 61
3.21 This figure shows the ultimate efficacy of the aerogel insulation in an actual thermoelectric generator. Heat flux was measured on all four panels that comprised
the TEG. Plates 1 and 3 did not have couples. Plate 2 had 25 skutterudite couples
insulated with aerogel. Plate 4 had 25 couples insulated with commercial insulation. . . 63
3.22 Various ways a sol-gel can be dried; Supercritically, to form an aerogel (top); Ambiently, with modification and solvent exchange, to form a slightly less dense porous
aerogel (second row); Ambiently, without modification, solvent exchange, or springback (bottom row). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.23 The microstructure of ADAI exhibits a wide range of particle sizes, ranging from
1-20 um in diameter, with relatively large pores. . . . . . . . . . . . . . . . . . . . . . 69
3.24 ADAI is thermally stable up to 250◦ C, after which it slowly loses surface methyl
groups. By 600◦ C, all methyl groups have oxidized. . . . . . . . . . . . . . . . . . . . 70
3.25 The mass loss between 300◦ C and 800◦ C is largely due to oxidation of surface
methyl groups, as evidenced by the loss of the Si-C peak. . . . . . . . . . . . . . . . . 71

xiii

3.26 Shrinkage occurs in three stages. Up to 90 minutes, bulk pore fluid is evaporating
from large pores, resulting in minimal shrinkage. From 90min to 4 hours, residual
pore fluid is evaporating from smaller pores where capillary forces are greater, causing the bulk of the shrinkage. After 4 hours, drying is complete. The gels are limited
to 15%vol shrinkage, and do not experience "spring-back." . . . . . . . . . . . . . . . 73
3.27 Diagram detailing how the pore size can affect the gel size at different stages of
ambient drying. A gel with larger pores will experience reduced shrinkage, at the
expense of increased thermal conductivity due to gas convection. . . . . . . . . . . . . 74
3.28 Visual confirmation of the lack of spring-back during drying, as well as a sample
after heating to 500◦ C in air for 1 hour. The gradient in discoloration of the heated
sample is due to presence of oxygen during heating - the white section was closer to
the open end of the quartz tube in which the sample was heated. . . . . . . . . . . . . 75
3.29 Thermal conductivity of ADAI is comparable to other high temperature insulation
materials. Error bars represent one standard deviation. . . . . . . . . . . . . . . . . . . 77
3.30 Third generation skutterudite-based thermoelectric generator for automotive waste
heat recovery, seen during fabrication. The white layer in the cross section is ADAI
insulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.31 Assembly of the ADAI insulation for the thermoelectric generator. Top left: aluminum molds spray-coated with boronitride mold release. Top right: detail picture
of aluminum posts around which the ADAI will be cast. The posts will be removed
when the gel forms, and the actual thermoelectric unicouple legs will be inserted
during final assembly. Middle: The dried ADAI insulation, with a mica sheet to
provide an initial barrier to gas diffusion and improve handleability. Note the clean
holes where the aluminum posts once were. Bottom: side view of the ADAI insulation.

79

4.1

Increasing the lithium:titanium precursor ratio increases the phase purity of the spinel. . 87

4.2

Increasing the lithium:titanium precursor ratio decreases the surface area of the calcined aerogel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

4.3

The calcined powder shows nanoscale crystallites (bottom) in good contact with the
conductive additive (top) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

4.4

Cyclic voltammetry shows a batt-cap behavior for the aerogel. Testing was done at
0.5 mV/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4.5

At a C/3 rate, the aerogel has a capacity comparable to the commercial product. . . . . 93
xiv

4.6

Cyclic voltammetry of LiMn2 O4 coated nanofoams cycled from 3.5 to 4.4 V vs.
Li/Li+ at 0.5 mV s-1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

4.7

Cyclic voltammetry of LiMn2 O4 coated nanofoams cycled from 2.3 to 4.4 V vs.
Li/Li+ at 0.5 mV s-1 . Mechanical degradation of the electrode during sweeping to
low potentials caused a large amount of noise below 2.75 V. . . . . . . . . . . . . . . . 100

4.8

Lithium insertion capacity of the LiMn2 O4 coated nanofoams over multiple cycles
at a symmetric 1C rate. Every 12th cycle was done at slower C/5 symmetric cycling. . 102

4.9

At a slower C/15 cycling rate, bulk LiMn2 O4 (x=0, above) exhibits a much poorer
cycling performance. Figure taken from ref. [84]. . . . . . . . . . . . . . . . . . . . . 103

4.10 Galvanostatic cycling of the LiMn2 O4 coated nanofoam at 150 mA g-1 (approximately 1C). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.11 Galvanostatic cycling of the LiMn2 O4 coated nanofoam at 300 mA g-1 (approximately 2C). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.12 Galvanostatic cycling of the LiMn2 O4 coated nanofoam at 1350 mA g-1 (approximately 9C). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.13 Galvanostatic cycling of bulk LiMn2 O4 (a), from ref. [81]. . . . . . . . . . . . . . . . 107

xv

Chapter 1
OBJECTIVES

It is the goal of this dissertation to develop and characterize aerogels for use in applications relating to energy and efficiency - arguably one of the most pressing of today’s societal needs (or at
least the most pressing of those for which there is a technological solution). The dissertation is
largely divided into two main thrusts: Aerogels for thermal insulation (efficiency), and Aerogels
for electrochemical energy storage (energy). Aerogels are possibly unique, as a class of material,
in that they are well suited for such seemingly disparate applications. It will be shown that the
flexibility of sol-gel synthesis, combined with a high surface area and tailorable pore morphology,
makes aerogels of profound interest for both thermal insulation and electrochemical reactions.
Aerogels for thermal insulation
The first section details the characterization of a silica aerogel as a candidate thermal insulation
material for an Advanced Stirling Radioisotope Generator (ASRG) power system for interplanetary
space vehicles. The aerogel, developed and fabricated at the Jet Propulsion Laboratory (JPL) in
Pasadena, CA, has a number of unique performance metrics it must satisfy if it is to be qualified for
space flight. In addition to needing a low thermal conductivity and low density, the insulation must
be thermally stable at operating temperature, yet crack and/or shrink during a runaway thermal
event. This work details the extent to which exposure to standard operating conditions affects

1

the mechanical properties of the aerogel insulation, with particular attention given to how those
changes will affect flight qualification. Additionally, a mechanical reinforcement scheme utilizing
short chopped fibers is explored.
Finding the mechanical properties of typical silica aerogels lacking, the second section details
the investigation of silylating agent chlorotrimethylsilane (CTMS) and its effect on the mechanical properties of the aerogel. Such surface treatments have been explored by other groups as a
method of making the naturally hydrophilic aerogels hydrophobic, and it is expected that this hydrophobicity will make the silica aerogels more durable in humid atmospheres. Additionally, the
hydrophobic surface groups should reduce surface tension during supercritical extraction, yielding
lower density gels with fewer inherent flaws.
The third section details the development of an aerogel insulation for a prototypical thermoelectric generator for automotive waste heat recovery. Thermoelectric generators, with their complex geometries and high operating temperatures, pose a thermal management challenge for which
aerogels are uniquely suited. While fiber-reinforced, opacified silica aerogel monoliths have been
repeatedly demonstrated in literature as excellent thermal insulators, there is relatively little work
detailing the use of aerogel as a cast-in-place insulation technology. An insulating procedure is
developed wherein fiber reinforcement is first packed into the generator, and then the freespace is
flooded with an opacified sol and allowed to gel. This aerogel insulation should dramatically reduce parasitic heat losses around the thermoelectric unicouple legs, increasing the overall thermal
efficiency of the device.
2

The fourth and final section of the silica aerogel work concerns the need for an ambiently dried
aerogel insulation that can be cast into place around rigid inclusions (e.g. thermoelectric legs).
Current ambiently dried aerogel insulation technologies shrink to <40% of their initial volume before "springing back" to a volume close to their original. While this is acceptable for freestanding
aerogel monoliths, such a drastic shrinkage during drying will cause the insulation to strain against
any rigid inclusions it is intended to insulate. This strain leads to extensive cracking in the aerogel, ultimately reducing or even eliminating its ability to provide thermal insulation. This work
details the development of an aerogel insulation that can be ambiently dried without any spring
back, and demonstrates the efficacy of this aerogel as insulation for a third-generation prototypical
thermoelectric generator.
Aerogels for electrochemical energy storage
The first section on aerogels for energy storage details the creation and characterization of a
lithium titanate aerogel as a high-rate anode for lithium ion batteries. The high surface area of
the aerogel morphology is expected to increase reaction kinetics – and thus power density – of
the lithium titanate oxide. Additionally, the flexibility of sol-gel synthesis should allow facile
incorporation of high-aspect ratio conductive additives – in this case, exfoliated graphite nano
platelets – which provide an electrical conductivity that can keep up with the fast reaction kinetics
at the particle surface. While reports exist in literature of sol-gel or hydrothermal methods of
creating lithium titanate particles, this is the first that attempts to control the drying process in
order to create free-standing, high surface area lithium titanate aerogel.
3

The final section of this dissertation explores the capacity and rate capability of LiMn2 O4 oxide
coatings on high-surface area carbon nanofoams for electrochemical energy storage. Such oxide
coated nanofoams combine several advantages of aerogels – high surface areas for increased kinetics, continuous carbon network for electrical conductivity, tailorable pore sizes and fiber reinforcement to allow ambient drying – to create a low-cost, high-performance electrode. It is expected
that the final electrode material will have improved rate-capability compared with the bulk oxide,
and that the high surface area will increase insertion capacity due to an increased concentration of
surface defects. Additionally, the oxide coated nanofoam is well suited to use as a 3D cathode for
a next-generation lithium-ion battery architecture.

4

Chapter 2
INTRODUCTION

2.1

Energy and efficiency needs

It is becoming increasingly self-evident that energy efficiency, generation and storage will be one of
the defining technological challenges for our society in the coming decades. For example, in 2009,
the Department of Energy projected a net deficit in liquid fuels beginning in 2012 and continuing
steadily into 2030, to be resolved through optimistically named "Unidentified Projects" (Figure
2.1) [1]. Shell International predicts [2] that by 2050, even after taking into account expected gains
in efficiency (20%) and production (50%), there will still be "a gap between business-as-usual
supply and business-as-usual demand of around 400EJ/a – the size of the whole industry in 2000."
Recent widespread power outages in the US highlight a need for improved energy storage for the
electrical grid in the US [3], especially as intermittent renewables contribute a larger percentage
of base load. Solving these challenges will require a diverse portfolio of different technologies
and innovations. This dissertation argues that aerogels, as a class of materials, are well suited to
improve both efficiency and energy technologies to meet the challenges of the 21st century.

5

Figure 2.1: Future liquid fuel demand and supply projections, from ref. [1]. (For interpretation of
the references to color in this and all other figures, the reader is referred to the electronic version
of this dissertation.)

2.2

Aerogel history

Aerogels are a class of material with, at best, a hazy definition. This is somewhat fitting, as
silica aerogels (by far the most studied composition) have been nick-named "frozen smoke" due
to their clouded appearance and seemingly amorphous boundaries. Loosely defined, an aerogel
is the solid remnant of a wet gel after all liquid has been removed in such a way as to cause
little or no shrinkage in the solid network [4]. Attempts to further refine this definition are met
with conflict. There are some who would define an aerogel as a gel that has been dried in such
a way as to maintain a high surface area (100’s of m2 g-1 ) and high porosity (>90%). Others will
6

accept the aerogel name only if such high surface areas are obtained through a supercritical drying
process (described in Section 2.4.1); a high surface-area dried gel obtained through ambient drying
methods, for example, would be termed an "ambigel," even if its properties are identical to those
of the supercritically dried gel.
For the purposes of this dissertation, materials will be defined according to their final properties,
and not the method in which those properties were obtained. Aerogels have high surface areas and
high porosities (as previously defined), while xerogels, which experience large shrinkage during
drying, exhibit low surface areas (< 20 m2 g-1 ) and low porosity (< 20%). Ambigels, designed to
undergo limited shrinkage during ambient drying, fall somewhere in between these two extremes.
The invention of Aerogel is credited to Samuel Kistler [5, 6] who, in the early 1930’s, removed
the pore fluid from a jelly by exchanging the water for a liquid with a low critical temperature (ether
or propane, depending on the composition of the solid network) before supercritically drying the
gel (a process further described in Section 2.4.1). Initial aerogels were largely a scientific curiosity; although several patents were awarded and commercial ventures attempted, none were overtly
successful, and interest in the material waned. In recent decades, aerogels have seen a resurgence
in interest for applications as diverse as thermal insulation, acoustic dampeners, catalysts, supercapacitors, 3D batteries, inertial confinement for cryogenic fluids and optically transparent insulators
for high-energy particle physics [7, 8]. It is their unique combination of high surface area, high
porosity, co-continuous solid and pore networks, tailorable pore sizes and room-temperature solgel processing that makes Aerogels so attractive for many applications, and their development will
7

Figure 2.2: A simplified procedure for forming an aerogel.

continue in the years to come.

2.3

Sol-gel synthesis

The procedure for making an aerogel consists of two main parts. First, a wet gel must be formed
with the composition and microstructure desired. Second, the wet gel must be dried in such a way
as to preserve the microstructure of the wet gel. This section details different methods of forming
the wet gel, using sol-gel chemistry. The next section will detail the ways in which a gel may be
dried.

2.3.1

Metal-alkoxide hydrolysis/condensation

The most common method of forming a sol-gel is the two-step process of hydrolysis and condensation of a metal-alkoxide, as diagrammed in Figure 2.2. In the first step, the metal-alkoxide is
8

Figure 2.3: Silica aerogel, with a typical "pearl necklace" structure.

diluted with alcohol, to which water and an acid are added. Hydrolysis occurs according to reaction
2.1, forming relatively stable monomers of metal-hydroxide in solution. Addition of base initiates
condensation of hydroxy groups according to reaction 2.2. Condensation continues (reaction 2.3),

9

slowly growing larger and larger particles of metal-oxide.

M(OR)n + H2 O

−→
−

M(OH)n + n(ROH)

(2.1)

M(OH)n

−→
−

(HO)n−1 M−O−M(OH)n−1 + H2 O

(2.2)

m[M(OH)n ]

−→
−

m(MO0.5n ) + mn(H2 O)

(2.3)

As the metal-oxide particles get larger, it eventually becomes more favorable for particles to coalesce, and further condensation reactions preferentially widen the neck between adjacent particles.
This gives rise to a "pearl necklace" microstructure (Figure 2.2 (bottom left) and Figure 2.3). The
choice of metal-alkoxide precursor, dilution of the sol and concentration of acid and base all affect the condensation reaction, which in turn determines the size at which silica particles begin to
interconnect. Thus the sol-gel route can be customized according to the microstructure desired.
Almost any metal-oxide can be formed through this method, providing a metal-alkoxide can be
synthesized. Examples include silicon dioxide (for thermal insulation), ruthenium dioxide (catalysis and supercapacitors) [9,10], vanadium pentoxide (lithium ion batteries) [9,11,12], and titanium
dioxide (photocatalysis) [13, 14].

2.3.2

Resorcinol-Formaldehyde polymerization

While the metal-alkoxide synthesis route allows the fabrication of many useful sol-gel compositions, including some electrically conductive metal-oxides [10], a highly conductive aerogel would
be advantageous for catalysis and electrochemical applications. An alternative gel formulation

10

was formulated by Pekala in 1989 based on the polycondensation of resorcinol with formaldehyde
under alkaline conditions [15]. This carbon-rich gel can then be pyrolyzed into an electrically
conductive carbon aerogel. A carbon nanofoam based on the resorcinol-formaldehyde chemistry
is used as a conductive scaffolding for the oxide coating described in Section 4.3.

2.3.3

Alternative sol-gel techniques

An important recent development in sol-gel synthesis is the adaptation of epoxide chemistries to
initiate the condensation of hydrated metal salts [16]. This method enables the formation of multicomponent metal oxides, and is particularly useful for the gelation of species for which no alkoxide
or common solvent exist. Chalcogenide sol-gels are also possible, as shown by Brock et al. [17],
which utilize a thiolysis/condensation reaction to create high surface area metal-sulfide gels.

2.4

Drying methods

The drying of an aerogel is often the most influential step in determining the gel’s final properties.
Supercritical drying is the most often utilized, as it best preserves the pore structure of the wet gel,
but other methods have been developed which attempt to do the same for a lower cost. This section provides a brief description and comparison of the main three methods: supercritical drying,
ambient drying, and freeze drying.

11

2.4.1

Supercritical drying

Supercritical drying was the first method used to form an aerogel, and remains the most common.
If a sol-gel is allowed to dry ambiently through simple evaporation, the retraction of the pore fluid
into the pore exerts a capillary force on the solid network according to equation 2.4:

Pc =

8ΦG (γl γs )0.5 − 4γl
d

(2.4)

Where:
ΦG = Interaction parameter describing the work of adhesion, described in ref. [18]
γl = Surface tension of liquid
γs = Surface tension of solid
d = Pore diameter

Due to the small diameter pores of the gel, the capillary forces are so great that they cause the
network to collapse, resulting in a dense, low surface area "xerogel" (Figure 2.4, second row). Supercritical drying avoids this by utilizing heat and pressure to first convert the liquid pore fluid to a
supercritical fluid before removal. This avoids the formation of a liquid-gas interface, circumventing capillary forces and maintaining the original microstructure of the gel.
Figure 2.4 (top row) outlines the process for supercritical drying. First, the wet gels undergo
solvent exchanges to remove any unreacted precursors and replace the pore fluid with one with
a suitably low critical point – generally acetone, acetonitrile or liquid carbon dioxide. If liquid
carbon dioxide is used as the pore fluid, the final solvent exchange takes place inside a pressurized
autoclave. With the wet gels submerged in excess solvent inside an autoclave, the temperature of

12

Figure 2.4: Various ways a sol-gel can be dried; Supercritically, to form an aerogel (top);
Ambiently, without any modification, to form a dense xerogel (second row); Ambiently, with
modification and solvent exchange, to form a slightly less dense porous aerogel (third row).

the solvent is raised while controlling the pressure to prevent evaporation (Figure 2.5). Once the
temperature and pressure of the pore fluid exceeds the critical point, the pressure is slowly relaxed
to ambient while maintaining a constant temperature. Care must be taken to avoid large changes
in pressure within the autoclave, especially with larger gels, as rapid decompression events will
cause gels to fracture.
Supercritically drying a gel with an organic (such as ethanol) as the pore fluid involves relatively low pressures (6.4 MPa) but high temperatures (243◦ C), which are generally beyond the
13

Figure 2.5: Phase diagram detailing the three main ways a gel can be dried: Freeze Drying,
Ambient Drying, and Supercritical Drying.

autoignition temperature of the fluid. As such, it is critical that careful safety measures are taken
to prevent ignition of the flammable, pressurized liquid. A benefit of using an organic as the pore
fluid, however, is that the high temperature of the supercritical organic solvent results in an alkylation of the solid surface, yielding a gel that is hydrophobic after drying. One can also use liquid
carbon dioxide as the extraction fluid, which is both inert and requires a lower temperature (21◦ C)
but higher pressure (7.4 MPa) to reach supercritical conditions. This is inherently safer, but results
in a hydrophilic gel unless the gel surface has been modified prior to drying.

14

2.4.2

Freeze drying

Commonly used for food preservation, freeze drying is an alternative method of removing the pore
fluid of a gel without allowing a liquid-gas interface to form. First, the pore fluid is rapid frozen into
a solid. A vacuum is then applied while maintaining sub-freezing temperatures, thereby removing
the pore-solid via sublimation. Freeze drying is rarely used to form an aerogel monolith, as the
liquid-to-solid phase change generally causes extensive fracturing of the pore microstructure [4].

2.4.3

Ambient drying

The final option for drying a sol-gel is to allow a liquid-gas boundary to form via normal evaporation, while controlling the gel morphology and solvent composition such that degradation of the
gel microstructure is limited.
The most common approach to ambiently dried aerogel involves a combination of exchanging the pore fluid with a low-tension solvent and treating the surface with a silylating agent to
reduce surface tension (Figure 2.4, third row). Such a route has been systematically explored by
A. Venkateswara Rao and A. Parvathy Rao, who report on the effects of various solvents [19, 20]
and surface modifications [21, 22] on ambiently dried silica gels.
These ambiently dried gels could be further improved if a couple key challenges can be overcome. First, the additional processing steps involved in exchanging the pore fluid and chemically
modifying the surface can add significant time and cost. This has been somewhat alleviated by
Schwertfeger et al., who developed a single-step process for surface treatment and solvent ex-

15

change using a mixture of hexamethyldisiloxane and trimethylchlorosilane [23]. The second challenge is that such drying techniques typically rely on the so-called "spring-back" effect [7, 24],
where the gel compresses by >40% of its initial wet volume before expanding back to near its
original volume. While such a technique is perfectly suitable for aerogel monoliths, reliance on
spring-back precludes the use of ambiently dried silica aerogel as a cast-in-place insulation material in applications with complex geometries or rigid inclusions. In such situations, the gel will
crack and break apart during the initial high-shrinkage phase of drying before spring-back occurs.
Another strategy is to reinforce the aerogel structure itself with short fibers [25], or react a
crosslinked polymer on the silica surface [26, 27]. Unfortunately, both efforts still rely on some
degree of spring-back to achieve low densities, and the polymeric coating is not stable at the high
temperatures (>500◦ C) at which aerogel is most attractive as an insulation material.

2.5

Electrochemical Characterization

Some of the more recent, and exciting, developments in aerogel technologies have involved the
use of high surface area aerogels in electrochemical applications, with energy storage receiving a
lot of attention. This section provides a brief description of the main analytical electrochemical
tools used to characterize these materials, with further discussion of their use and performance in
Chapter 4.
All electrochemical energy storage systems involve three main components: an anode (high
potential energy), cathode (low potential energy), and electrolyte. Ions are stored at or in the two
16

electrodes, and while the electrolyte allows transport of the ions from one electrode to the other, it
prevents the transport of those ions’ associated electrons. These electrons are only allowed passage
from one electrode to the other through an external circuit, in which their potential energy is used
to do work.
For characterization of an electrode material for lithium-ion batteries, the electrode of interest
(the "working" electrode) is generally set up as a cathode, with metallic lithium (the "counter" electrode) as an anode. A third metallic lithium electrode may also be used as a "reference" electrode
if peripheral cell components are expected to convolute measurements on the working electrode.
Measuring the potential of the working electrode vs. the reference (which has no applied current
and is thus not taking part in the reaction) allows one to accurately measure the potential of the
working electrode at high current density.

2.5.1

Charge storage mechanisms

There are three main methods in which a material can store charge electrochemically: Faradaic,
Capacitive, and Pseudocapacitive.
In Faradaic processes, charge is stored as a difference in potential energy between two electrodes. For example, graphite and lithium cobalt oxide (LiCoO2 ) are a common anode/cathode
pair. Figure 2.6 shows a simple cell setup, where lithium ions are extracted from the LiCoO2
cathode (low potential energy, reaction 2.5) and placed inside the graphitic anode (high potential energy, reaction 2.6). Each ion shuttled from the low energy state to a higher energy stores

17

Figure 2.6: A simplified diagram of a Graphite/LiCoO2 cell, showing both charge and discharge
processes.

18

an energy equivalent to the difference in the Gibbs free energy associated with the two reactions
(equation 2.7).

LiCoO2

←→
−

CoO2 + Li + + e −

(2.5)

LiC6

←→
−

C6 + Li + + e −

(2.6)

◦
∆G◦ = ∆G◦
cathode − ∆Ganode

(2.7)

) − (∆G◦ −→C +Li + +e − )
∆G◦ = (∆G◦
LiC6 − 6
LiCoO2 −→CoO2 +Li + +e −
−

(2.8)

∆G◦ = 405 kJ mol −1

(2.9)

Once the cell is assembled, a small number of lithium ions will initially migrate back to the lowenergy cathode, leaving their associated electrons behind. However, as the electrolyte does not
permit transport of the electrons associated with these liberated ions, in the absence of an external
circuit this ion migration results in a build up of electrical potential on the two electrodes. The
electric potential directly opposes the chemical potential, preventing further reaction and giving
rise to the cell voltage. The expected voltage of an electrode can be computed from the activity of
the species, according to the Nernst equation (eqn. 2.10):

E = E◦ −

RT ared
ln
nF
aox

Where:
E = Electrode potential
E ◦ = Standard half-cell reduction potential
R = Gas constant
19

(2.10)

T = Temperature
n = number of electrons transferred in the reaction
F = Faraday’s constant
ared = Activity of the reduced species
aox = Activity of the oxidized species
The standard half-cell reduction potential, E ◦ , is the potential of the cell at unity activity, and is
related to the Gibbs free energy by equation 2.11:

∆G◦ = nFE ◦

(2.11)

Where:
∆G◦ = Gibbs free energy of the chemical reaction
E ◦ = Electrical potential of the electrode

There are various characterization methods whereby the performance of an electrode can be
evaluated, with the most basic being a simple measurement of the Open Circuit Potential (OCP).
If this potential differs from that predicted in equation 2.10, it is an indication that the underlying
chemistry has been altered.
Capacitive storage uses an applied electric potential to form a double-layer at the surface of
the electrode, with the electric charge being balanced by a concentration of solvated ions with
countering charge. No underlying chemical reaction or charge transfer occurs. As such, there is
no expected voltage of a capacitor – the voltage is instead determined by the amount of charge
stored in the electrode. Capacitive energy storage has the benefit of being extremely fast compared
to faradaic processes, as it is not limited by the kinetics of a chemical reaction. As the energy is

20

stored at the surface of the electrode, materials with higher surface areas will be able to store more
charge - this makes aerogels a natural candidate for capacitive electrodes.
Finally, pseudocapacitive energy storage is somewhat of a hybrid between pure faradaic and
pure capacitive processes. Charge is stored as a chemical potential, but the surface area of the
electrode is so high that a significant fraction of the ions involved are stored at or near the surface
of the material. Such a short diffusion length for the ions reduces or eliminates some of the kinetic
limitations associated with a faradaic process, allowing the electrode to perform at rates normally
reserved for pure capacitors. This combines the high capacity of faradaic processes with the high
rate capability of a capacitor. Aerogels have been shown to exhibit a pseudocapacitive behavior
[11, 28], and are the subject of much ongoing research.
Further investigation into the thermodynamics and kinetics of an electrode generally relies on
two additional characterization techniques: Cyclic Voltammetry, and Chronopotentiometry.

2.5.2

Cyclic voltammetry

After OCP, Cyclic Voltammetry (CV) is the most fundamental electrochemical characterization
tool used to evaluate an electrode material. An applied potential is swept back and forth through a
voltage range in which an electrochemical reaction is expected, and the mass-normalized current
resulting from the applied potential is measured. Three different CV plots are given as examples in
Figures 2.7-2.9. The first shows a purely faradaic response – sharp peaks in current are observed,
centered on the expected potential of the reaction. The second plot shows the current response of

21

0.0
−1.0

−0.5

Current (A g−1)

0.5

1.0

Faradaic

1.0

1.2

1.4

1.6

1.8

Potential (V vs Li/Li+)

2.0

Figure 2.7: Example Cyclic Voltammetry plot for a Faradaic charge storage mechanisms (Data
from Section 4.2)

22

0.0
−1.0

−0.5

Current (A g−1)

0.5

1.0

Capacitive

1.0

1.2

1.4

1.6

1.8

Potential (V vs Li/Li+)

2.0

Figure 2.8: Example Cyclic Voltammetry plot for a Capacitive charge storage mechanisms (Data
simulated)

23

0.0
−1.0

−0.5

Current (A g−1)

0.5

1.0

Pseudocapacitive

1.0

1.2

1.4

1.6

1.8

Potential (V vs Li/Li+)

2.0

Figure 2.9: Example Cyclic Voltammetry plot for a Pseudocapacitive charge storage mechanisms
(Data from Section 4.2)

24

a pure capacitor, where the observed current is directly proportional to the scan rate, giving rise
to a box-shaped plot. The final plot shows the CV response of a pseudocapacitor – faradaic peaks
superimposed on a capacitive box.

2.5.3

Galvanostatic cycling (Chronopotentiometry)

Chronopotentiometry involves applying a fixed current and measuring the change in electrode potential over time, with the direction of the current switching once the potential reaches a cut-off
value. A typical plot is shown in Figure 2.10. This allows the ultimate capacity of the electrode
to be measured, and successive cycling at the same rate allows one to measure the cycling performance of an electrode as a degradation of capacity vs. cycling. In addition, the capacity and cycle
life can be evaluated at different current densities, providing a measure of the rate capability of the
electrode. This is often the most important metric, as improving the capacity of electrodes at high
charge/discharge rates is the focus of much research.

25

4.8

Chronopotentiometry

4.4
4.2
4.0
3.6

3.8

Potential (V vs Li/Li+)

4.6

Applied Current (mA g−1)
150
300
1350

0

50

100

Capacity (mA h g−1)

150

200

Figure 2.10: A typical chronopotentiometry plot, showing the voltage response of an electrode
cycled at three different current densities.

26

Chapter 3
SILICA AEROGELS FOR THERMAL INSULATION

3.1

Introduction

The low densities and highly tortuous pore networks that define aerogels make them a natural
candidate as a high-performance thermal insulation material. This chapter details work undertaken
to further develop silica aerogels for thermal insulation, beginning with the characterization of
an existing aerogel formulation and culminating in the invention and characterization of a new,
ambiently dried aerogel insulation technology. This section is an effort to provide the background
information necessary to place the remainder of the chapter in proper context.
Silica aerogel is an insulator because it selectively decreases each of the three modes of thermal
transport: solid conduction, gas convection, and infrared radiation (Figure 3.1). The first, solid
conduction, is greatly reduced due to the low density of the aerogel – there simply is very little
solid material to conduct heat through, and what solid does exist is comprised of silicon dioxide
which has one of the lowest bulk thermal conductivites. The second, gas convection, relies on
transport of hot gasses across the thickness of the insulating material. The aerogel microstructure
is such that, while there is a large volume of open porosity through which the gas can move, it
is finely subdivided into very small interlinked pores. These pores are quite often only tens of
nanometers wide, resulting in a tortuous path for the gas to follow. This tortuosity greatly reduces

27

Figure 3.1: Modes of thermal transport in an aerogel, and how the aerogel structure selectively
reduces each.
gas convection across the aerogel. The final mode, radiative transport, can be problematic for silica
aerogel, especially at high temperatures. Silica is transparent to infrared radiation, and as such does
very little to prevent radiative thermal transport. However, the flexibility of sol-gel processing is
such that opacifying particles can be easily incorporated into the gel network. This strategy was
extensively explored by Fricke et al. in the 1990’s, and involves dispersing into the sol particles
which have either: A) a high extinction coefficient to absorb incident radiation, such as carbon
soot (with a specific extinction coefficient ranging from 0.5 - 1.5 m2 g-1 , refs. [29, 30]); or B) a
high albedo to scatter incident radiation, such as titanium dioxide (with an albedo of 0.9, ref. [31]).
With opacification, silica aerogels function extremely well as thermal insulation. Coupled with
the incorporation of fiber reinforcement, surface treatments and other strategies, aerogel holds
much promise as a durable, cast-in-place high-temperature insulation material for a variety of
28

applications.

29

3.2

3.2.1

Fiber-Reinforced Aerogel for Advanced Stirling Radioisotope Generator
Introduction

Thermal insulation is a key component of spacecraft design, particularly for electrical power systems. The efficiencies of both Radioisotope Thermoelectric Generators (RTGs) and Stirling Engines rely on large temperature differences across short spans. Multi-Layer Insulation (MLI) has
worked well, but lacks any ability to prevent gas diffusion and thus requires the vacuum of space
to function as a thermal insulator. Additionally, the relatively heavy MLI insulation comes at a
premium in an environment where every kilogram at launch matters. As NASA prepares for further exploration of Mars and other planetary surfaces, there is a need for a lightweight thermal
insulation material that can work equally well both in the vacuum of space and in an atmosphere.
Silica aerogel, with its low density and low thermal conductivity, is uniquely suited as a
lightweight thermal insulation material for on-board power generation systems. Indeed it has already seen use in specific space vehicle applications – most notably, as insulation for the Warm
Electronics Box in the Mars Exploration Rovers Spirit and Opportunity, one of which is still in
operation at the time of this writing [32]. Silica aerogel is also used as insulation in the heat
exchangers for the Mars Science Laboratory, scheduled to land later this year.
On-board power generation for space vehicles is typically provided by a Radioisotope Thermoelectric Generator (RTG) for situations where solar panels are insufficient – either beyond the orbit
of Mars, or away from its equator. An alternative design has been proposed, based on a Stirling
30

generator. Both thermoelectric and Stirling generators rely on maintaining a temperature gradient
to generate electrical power, and as such need effective thermal insulation for increased efficiency.
The Advanced Stirling Radioisotope Generator (ASRG) requires monoliths of insulation, while
the RTG design has the added challenge of requiring thermal insulation cast around the individual
unicouple legs.
This work is part of a larger effort to develop and characterize silica aerogel as a launch-ready
technology for insulation of an ASRG for next-generation spacecraft designs. The thermal conductivity as a function of temperature was obtained for various opacifying titania loadings (Figure
3.2), showing that a very low thermal conductivity can be obtained at optimal titania loadings.
In addition to requiring a low thermal conductivity, the ASRG design must also protect against
a thermal runaway event, in which the unit suffers mechanical failure and stops converting the
thermal energy of the heat source to electrical power. Without this energy conversion, the temperature of the hot side climbs rapidly. The specifications detail that the unit must contain some
mechanism for dumping excess heat, for if the cladding of the radioisotope heat source exceeds
1250◦ C it will degrade and cause damage the rest of the spacecraft. Figure 3.3 shows the aerogel
insulation, before and after a simulated thermal runaway event. The aerogel passed the test – the
increased temperatures caused the silica network to sinter, shrinking down and opening up gaps in
the insulation through which excess heat could vent to the surrounding environment.
As part of a collaborative investigation into the use of aerogel for thermal insulation in an
ASRG unit, this section details the mechanical characterization of the silica aerogel as a result of
31

140
120
80

100

150 mg/cc aerogel in vacuum (No TiO2)
200 mg/cc aerogel in vacuum (50 mg/cc TiO2)
240 mg/cc aerogel in vacuum (100 mg/cc TiO2)
310 mg/cc aerogel in vacuum (150 mg/cc TiO2)
350 mg/cc aerogel in vacuum (100 mg/cc TiO2)

60

q

40

q

q

20

q
q
q
q

q

q

0

Effective Thermal Conductivity (mW m−1 K−1)

q

200

400

600

800

Temperature (K)

1000

1200

Figure 3.2: Thermal conductivity of aerogel insulation developed at JPL, as a function of titania
content.

32

Figure 3.3: Aerogel insulation developed at JPL in a heat-dump test apparatus, before (top) and
after (bottom) being subjected to a simulated runaway thermal event. The excessive heat caused
the aerogel to sinter, opening pathways for thermal energy to vent into the environment. The
interior reached a maximum temperature of 1040◦ C, well within the required specification.

33

sintering due to high temperatures. This is important for two reasons. First, the aerogel densifies
under normal operating conditions of 700◦ C in vacuum - thus, the degree of densification needs to
be quantified and accounted for if large enough panels of insulation are to be manufactured such
that they are the correct thickness after heating. Second, this densification will affect the stiffness
of the panel. Launch conditions will subject the insulation to large vibrational strains, thus it is
important to have an accurate value for the stiffness of the aerogel panel so that a proper failure
analysis can be conducted.
In addition to evaluating the effect of sintering on mechanical properties of the gels, a short
study was conducted on the use of chopped fibers as an alternative to the fiber felt used in the
standard formulation. The anisotropic nature of the felt gives rise to variations in the mechanical
properties of the aerogel, which may be avoided by utilizing an isotropic short fiber reinforcing
scheme.

3.2.2

Experimental

For the sintering studies, fiber felt reinforced, opacified silica aerogel blocks approximately 4" by
4" by 1" were provided by JPL. Test cylinders 20mm in diameter and approximately 25mm long
were cut from the block using a diamond grit hole saw. The fiber felt reinforcement was situated
perpendicular to the larger face of the block (i.e. 4" by 4" panels of felt were stacked 1" thick),
and test samples were cut both perpendicular to the felt layers (across the 1" block thickness) and
parallel to the layers in order to gauge the degree of anisotropy in the insulation.

34

To evaluate the effect of short chopped fibers on the mechanical properties of sintered silica
aerogels, bulk quartz wool was obtained from Saint-Gobain. To break the long fibrous wool into
short fibers, 0.2 g of wool, 40 mL of ethanol and 15 1/8" PTFE balls were placed into a 55 mL
methacrylate vial and processed in a vibratory mill for 10 minutes. The chopped fibers were then
separated from the ethanol and allowed to dry.
To make the gels, 35.7 mL of Silbond H5 (Silbond Corp.) was added to 35.7 mL of 200 proof
ethanol (Decon Labs) with stirring. In a separate container, a catalyst solution of 24.9 mL of 200
proof ethanol, 53.55 mL of distilled water and 0.15 mL of aqueous ammonium hydroxide (30%
ammonia in water, reagent grade, Columbus Chemical Industries Inc.) were combined with stirring. The catalyst solution was then added to the diluted Silbond solution drop-wise with stirring.
This sol was then added to a vial containing the appropriate amount of chopped fibers and sonicated for 30 seconds before being allowed to gel. After three days of aging at room temperature,
the samples were extracted from the mold and placed in an ethanol wash bath to remove any remaining reactants or catalysts. The bath water was changed once per day for three washes total,
before supercritically drying with liquid carbon dioxide.
After drying, fiber reinforced samples were placed into an open-ended quartz tube and heated
to various temperatures for 1 hour. Compression testing was done with a United Testing Systems model SFM-20 load frame using a 100 lb load cell at a constant crosshead speed of 0.75
mm/min. The surface area and pore size distribution were obtained by nitrogen adsorption with a
Micromeretics model ASAP 2020.
35

Figure 3.4: The anisotropic nature of the aerogel insulation can be clearly seen in this photograph
of two aerogel samples, as the sample on the right has become oval in cross-section after heating.

3.2.3

Results

Figure 3.4 shows two aerogel samples which were heated to 950◦ C in vacuum. The sample on
the left was cut perpendicular to the fiber felt layers, while the sample on the right was cut in the
parallel direction. The fiber layers are visible in the sample on the right, and the oval cross-section
indicates that sintering primarily densified the aerogel in the dimension perpendicular to the fiber
layers.
Figure 3.5 shows the pore size distribution of the aerogel insulation as-received, after heating

36

Figure 3.5: After heating to 950◦ C in vacuum, the aerogel shows a reduction in the number of
pores below 1 nm wide, which is indicative of preferential sintering of the necks between adjacent
silica particles.
to 500◦ C in air, and after heating to 500◦ C in air followed by heating to 950◦ C in vacuum. The
aerogel is largely thermally stable at 500◦ C, but begins to sinter at higher temperatures. It can be
seen from this plot that the sintering is largely due to a widening of the necks between adjacent
silica spheres, which is manifested as a reduction in the volume of the smallest "pores" (< 1nm).
In an effort to reduce the anisotropy of the insulation, silica aerogels reinforced with short
chopped fibers were investigated (Figure 3.6). Figure 3.7 shows the dimensional shrinkage of

37

Figure 3.6: Silica aerogels reinforced with various loadings of chopped fibers.
these samples after exposure to different temperatures in air. A certain degree of anisotropy is still
evident in these gels, as they preferentially reduced in diameter rather than height. This may be due
to either a preferred orientation of the fibers inside the gel, or sedimentation of the fibers during
gelation. Unsurprisingly, aerogels with higher fiber loadings experienced less densification.
The effect of fiber loadings and sintering on the mechanical properties of silica aerogel is shown
in Figure 3.8. Samples needed to reach 0.35 MPa without failure to be considered for use in the
ASRG. Unreinforced samples are more susceptible to sintering, and thus have a higher modulus
than those with higher fiber loadings. As fiber loading increased beyond 3%wt, the sintered aero38

Figure 3.7: Effect of chopped fiber loading on sintering of silica aerogel insulation as a function
of temperature.
gel would fail plastically before reaching the load required for flight qualification. Stress-strain
response of the felt-reinforced aerogel from JPL is also given as a reference.

3.2.4

Conclusions

While fiber-reinforced, opacified silica aerogel is an impressive material with excellent thermal
conductivity, its lack of mechanical strength is a concern for some applications. Additionally, the
choice of reinforcement has a large effect on the sintering of the aerogel network and its resultant
mechanical properties. Felt reinforcement introduces anisotropy to the silica network, which may
be disadvantageous for some applications. This can be partially alleviated by the use of chopped
39

Figure 3.8: Effect of chopped fiber loading on mechanical properties of sintered silica aerogels.
Data from aerogel samples provided by JPL is given as a reference.

fibers. Increasing the fiber concentration of the gel increases the resistance to sintering of the gel,
which has the additional effect of reducing the elastic modulus of the gel after exposure to heat.
This must be considered when designing an aerogel insulation that must be load bearing. There is
a need for further work improving the mechanical properties of silica aerogel insulations, with the
ultimate goal of a stiff, durable material suitable for a variety of applications.

40

3.3

3.3.1

Large Deformation of Silica Aerogels via Surface Modification with
Chlorotrimethylsilane
Introduction

Aerogels were first developed by Kistler [6] in the early 1930’s. These aerogels are a class of low
density materials derived through sol-gel chemistry, of which silica aerogels are the most extensively characterized [7, 33]. The low solids content (<5%) and tortuous interconnected porosity
(5-50 nm diameter pores) allow silica aerogels to have extremely low thermal conductivity, even at
elevated temperatures. Unfortunately, they are inherently fragile, and silica’s brittle nature limits
aerogel to non-load bearing applications. Common methods of reinforcement include incorporating fibers into the sol before gelation [34] or coating the silica surface with a conformal polymer [35], although the use of the latter method limits the maximum operating temperature to that
of the polymer.
Additionally, silica aerogels must be supercritically dried [33], generally with respect to either alcohol or carbon dioxide, otherwise the capillary forces induced during evaporation of the
pore fluid will lead to extensive shrinkage and possible cracking of the gel monolith. This drying
process adds significant cost. Work has been done using various silanes [36] to modify the silica surface, both reducing the surface tension as the pore fluid evaporates and preventing adjacent
silica surfaces from bonding during drying. In particular, chlorotrimethylsilane (CTMS) has been
shown to drastically reduce the shrinkage of silica aerogels derived from sodium silicate [36] and
tetraethylorthosilicate [21]. To date, no studies have been found that have explored the effect of
41

CTMS surface treatment on the mechanical properties of silica aerogels, except in so far as data
for shrinkage during drying are reported [21, 36, 37]. In this work, we use chlorotrimethylsilane
to modify silica aerogels in order to gain dramatic improvements in strain and stress to failure, as
well as enhanced resistance to sintering.

3.3.2

3.3.2.1

Experimental Techniques and Methods

Sample Preparation

To make the gels, 35.7 mL of Silbond H5 (Silbond Corp.) was added to 35.7 mL of 200 proof
ethanol (Decon Labs) with stirring. In a separate container, a catalyst solution of 24.9 mL of 200
proof ethanol, 53.55 mL of distilled water and 0.15 mL of aqueous ammonium hydroxide (30%
ammonia in water, reagent grade, Columbus Chemical Industries Inc.) were combined with stirring. The catalyst solution was then added to the diluted Silbond solution drop-wise with stirring.
The solution was then poured into polystyrene molds and allowed to gel (approximately 30 minutes). After three days of aging at room temperature, the samples were extracted from the mold
and placed in an ethanol wash bath to remove any remaining reactants or catalysts. The bath water
was changed once per day for three washes total.
Surface modification was accomplished by placing 18 samples in a jar with 360 mL of 200
proof ethanol and adding 40 mL of chlorotrimethylsilane (Alfa Aesar, A13651). Approximately
24 hours later, the CTMS solution was poured out and replaced with fresh ethanol. The ethanol

42

was changed once per day for four washes total. Control samples without surface modification
also went through these four additional ethanol washes.
All samples were supercritically dried with carbon dioxide in a 12 L critical point dryer (Accudyne Systems Inc). To resolve the effect of the surface modification on the silica structure, as well
as evaluate sintering resistance, both control and treated samples were heated in a tube furnace.
Samples were heated at 500◦ C for three hours (4◦ C/min ramp rate) in order to remove organics
while avoiding extensive changes to mechanical properties [38]. To completely remove organics,
separate treatment samples were heated at 700◦ C for one hour (4◦ C/min ramp rate). All heating
was done in air.

3.3.2.2

Sample Characterization

Compression testing was done with a United Testing Systems model SFM-20 load frame using a
100 lb load cell at a constant crosshead speed of 0.75 mm/min. Elastic moduli were extracted from
a linear fit of the data between 2-4% strain. The surface area and pore size distribution were obtained by nitrogen adsorption with a Micromeretics model ASAP 2020. FTIR was performed with
a Perkin-Elmer model Spectrum 2000 spectrometer. Being both monolithic and mostly transparent
to IR, the entire sample was placed into the FTIR beam path without the use of KBr. Neither saturation of the detector nor over-absorption of the beam was observed. SEM images were obtained
on osmium-coated samples using a JEOL model 7500F microscope.

43

3.3.3

Results

3.3.3.1

Composition and Morphology

Table 3.1 shows the average dimensions of the samples as wet gels, after supercritical drying,
and finally after heating at 500◦ C. Although the CTMS modified samples have a slightly higher
Height (mm)
%
Avg. ±stdev
Change
Control
Wet
Dried
500◦ C
Treated
Wet
Dried
500◦ C
700◦ C

Diameter (mm)
%
Avg. ±stdev
Change

Volume (cc)
%
Avg. ±stdev
Change

18.50
16.14 ±0.89
13.35 ±0.23

13%
17%

15.50
14.44 ±0.04
12.13 ±0.31

6.9%
16%

3.49
2.64 ±0.15
1.54 ±0.15

24%
42%

2.7%
6.3%
10%

15.50
15.36 ±0.09
14.51 ±0.10
13.78 ±0.22

0.9%
5.6%
10%

3.49
3.34 ±0.06
2.79 ±0.06
2.42 ±0.12

4.4%
26%
28%

18.50
18.00 ±0.28
16.87 ±0.14
16.18 ±0.36

Table 3.1: Shrinkage of control and CTMS treated samples due to drying and subsequent heat
treatment. "Dried" shrinkage is measured with respect to the wet gel volume, while 500◦ C and
700◦ C shrinkage values are measured with respect to the dried gel.

mass (Table 3.2) due to the surface organic coating, they exhibit significantly less shrinkage during drying and thus have a lower dry bulk density. Furthermore, SEM images (Figure 3.9) of the
samples show that the CTMS treatment causes an increase in the porosity of the as-dried aerogels.
Nitrogen adsorption measurements (Figure 3.10) reveal that the treated samples have fewer micropores (herein defined as pores less than 1 nm wide) than the control. After heating to remove
organics, the treated samples exhibit a larger relative increase in micropores than the control, as
evidenced by the change in peak height.
44

Avg.

Density
(mg/cc)
Avg.

203.42

77.0

0.20 ±0.01
3.29 ±0.75

246.90

74.0

0.11 ±0.01
0.54 ±0.17
1.47 ±0.51

Mass (mg)

Control
Dried
500◦ C
Treated
Dried
500◦ C
700◦ C

Elastic Modulus (MPa)
Avg. ±stdev

Table 3.2: Physical properties of control and CTMS treated samples.

FTIR (Figure 3.11) confirms the presence of organic groups on the surface of both the control
and the CTMS treated samples, due to residual pore fluid adsorbed onto the surface of the dry
gel. The treated aerogel has a pair of additional organic peaks at around 1800 cm-1 due to the
methyl surface groups. The organic groups common to both samples are due to residual pore fluid
adsorbed onto the surface of the dry gel. After heating the samples at 500◦ C, all of the organics on
the control are removed, but the treated samples still show residual organics. Heating the treated
samples to 700◦ C removes all organics.

3.3.3.2

Sintering and Mechanical Properties

Figure 3.12 shows the stress-strain response of both treated and control samples before and after
heating at 500◦ C, as well as treated samples heated at 700◦ C. Typically, silica aerogels have a strain
to failure of approximately 10-20% [34], unless fiber reinforcement [34] or polymer crosslinking
[35, 39, 40] is utilized. Polymer crosslinking in particular can withstand high strain, but relies on

45

Figure 3.9: SEM micrographs of CTMS Treated and Control samples, before and after heating at
500◦ C. All images are at the same magnification.

relatively thick coatings of crosslinked polymer. In contrast, CTMS deposits only a monolayer
of non-crosslinked methyl groups, yet the treated silica aerogels are able to compress to over
80% strain. Indeed, the compression tests of CTMS treated samples were terminated not because
the sample failed but because the load cell capacity was exceeded. A test of a single CTMS
treated sample using a higher capacity load cell showed a strain to failure of 96%, at a stress of 15
MPa. The control samples seem to exhibit a similar behavior as the treatment group, albeit with
greater variability. However, a visual inspection of the samples after testing (Figure 3.13) reveals

46

0.06
0.05
0.04

q

0.01

0.02

0.03

q
q
q
q
qq
q
qq
q
q
q q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
qq
qq
q
qqq
q qq q
q
q
q
q q
q q
q
q
q q
q q
q q
q
qq
qqqq
qqq
q
q
qq q
q q
q
q qq
qqq q
q qq
q qq
q
q q
q
qq
q
q
q
q q
q
q
q q
q
q
q
q

0.00

°
Differential Pore Volume dV/dw (cm3g−1A−1)

q

Control, as−dried
Control, 500°C
Treated, as−dried
Treated, 500°C

q
q
q
q

q
q

q

q

q

q

q

0

10

20

30

°
Pore Width (A)

40

50

Figure 3.10: Differential Pore Volume Distribution from Nitrogen Adsorption (<5nm pores, HK
method).

47

Figure 3.11: FTIR spectra of all samples. CTMS modified samples show residual organic peaks
after heating at 500◦ C for three hours, but after 700◦ C for one hour all organics are removed and
the spectra matches that of the control heated at 500◦ C.

that the treated samples remained monolithic while the control samples have broken into multiple
pieces. The control pieces are aligned parallel to the testing axis, and thus they appear to exhibit
mechanical properties similar to the treated samples even after obvious failure of the sample.
After heating at 500◦ C, the control samples have sintered greatly, with a 42% reduction in
volume (Table 1). This resulted in an increase in the elastic modulus at low strains (Table 1), and
a strain to failure typical of silica aerogels. The treated samples have sintered far less - only a 16%
reduction in volume - and they mostly retain an impressive strain to failure of around 75% (Figure
3.12), albeit with greater variation than non-heat treated samples.
48

2.5
1.5
1.0
0.0

0.5

Stress (MPa)

2.0

Control, as−dried
Control, 500°C
Treated, as−dried
Treated, 500°C
Treated, 700°C

0.0

0.2

0.4

0.6

0.8

Strain (mm/mm)
Figure 3.12: Stress vs. strain compression curves of all samples. Even after baking at 500◦ C,
CTMS modified samples still exhibit impressive mechanical properties. Heating at 700◦ C
removes all organics and the sample exhibits the same strain to failure as the control heated at
500◦ C.

49

Figure 3.13: Image of both a CTMS modified sample (left) and control sample (right) after
compression testing. Both samples underwent the same strain. Note the extensive damage in the
control, which is not present in the modified sample.

3.3.4

Discussion

The reduction in drying shrinkage can be attributed to two possible causes. First, the methyl
groups will change the surface tension between the silica and the pore fluid, as well as between the
silica and the supercritical carbon dioxide. This reduces the capillary forces during drying, which
reduces the overall shrinkage and prevents the formation of drying-induced cracks, which would
form micropores (pores smaller than 1 nm). Second, the organic groups may be reinforcing the
weak points of the wet gel formed during gelation, specifically the micropores where the gap is
short enough for the opposing monolayers of methyl groups to interact. (Figure 3.14).
The nitrogen adsorption measurements in Figure 3.10 confirm these two scenarios. The treated
samples have fewer micropores than the control, indicating a reduction in drying-induced stresses.

50

Figure 3.14: Schematic of CTMS modification. The CTMS creates a conformal coating on the
surface of the silica, filling in and reinforcing any microcracks present.

That the treated samples exhibit a larger proportional increase in micropores, while still having
fewer than the control, indicates that the CTMS treatment did fill in the micropores that formed during gelation. This may locally reinforce the structure, despite the treatment being a non-crosslinked
monolayer coating. As evidenced in Figure 3.13, the CTMS treatment leads to a dramatic improvement in strain-to-failure, compared with literature values of 10% strain and 1 MPa stress [34] for
unreinforced silica.
It is concluded that the residual organics in the micropores are reinforcing the weakest parts
of the gel and allowing the sample to deform so extensively. This is confirmed with the compression tests (Figure 3.12), where the treated samples heated to 500◦ C retain a large improvement
in mechanical properties, as well as a small amount of organics (shown via FTIR, Figure 3.11).
After heating at 700◦ C, the treated samples now show no organic groups and demonstrate a strain
to failure similar to that of untreated samples heated at 500◦ C, indicating that the organic groups
51

played a major role in strengthening the gel.

3.3.5

Conclusion

CTMS modification has a dramatic effect on the mechanical properties of silica gels. The shrinkage
due to supercritical drying is reduced, resistance to sintering is greatly improved, and the plasticity
of the aerogel is greatly increased, allowing for upwards of 80% strain with no macroscopic damage. More notable is the fact that after heating at 500◦ C in air for several hours, the samples still
exhibit a compressive strain to failure greater than 50%. This has dramatic implications for the
future of silica aerogel insulation, as previous work has only been able to achieve such levels of
compression using organic surface coatings that necessarily limit working temperatures to that of
the polymer. Using CTMS or similar surface modifications can increase the robustness and compressibility of silica aerogels even in relatively high-temperature environments, where they are the
most useful.

52

3.4

3.4.1

Aerogel Insulation for a Prototypical Skutterudite-based Thermoelectric Generator for Automotive Waste Heat Recovery
Introduction

With the exception of Radioisotope Thermoelectric Generators (RTG) used successfully on a number of NASA science and exploration space missions, the use of high temperature thermoelectric
power generation technology has been very limited. RTGs use a radioisotope heat source to generate >500◦ C thermal gradients that are sustained for multiple decades [41–43]. The relatively high
hot side temperature (1000◦ C) and operation in space vacuum create a near steady-state thermal
environment that is relatively benign compared to the projected thermal environment for terrestrial
applications [44, 45]. For example, converting waste heat from internal combustion engines (ICE)
will result in hot side thermoelectric converter temperatures in the 400-700◦ C range, with the heat
source flowing hot exhaust gas with wide swings in temperature and flow rates during a typical
duty or driving cycle. Operation in ambient air or exhaust gas presents a unique set of thermal
management and chemical stability challenges compared to operation in space vacuum (Figure
3.15). Additionally, the complex geometry of a thermoelectric generator requires an insulation
that can be cast into place around the individual thermoelectric legs. To this end, this work focuses
on the development and application of fiber-reinforced, opacified silica aerogel as cast-in-place
insulation for prototypical skutterudite thermoelectric generators.

53

Figure 3.15: Challenges in thermal management of terrestrial TEGs. Thermal insulation for a
terrestrial TEG must prevent conductive heat transport (left), withstand high temperatures
(middle), and be cast into place around complex geometries (right).
3.4.2

3.4.2.1

Experimental

Aerogel Encapsulation

Aerogel insulation was cast into place using the protocol described elsewhere [46]. Briefly, Silbond H5 (Silbond Corp.) was added to 200 proof ethanol (Decon Labs) with stirring. In a separate
container, a catalyst solution of 200 proof ethanol, distilled water and aqueous ammonium hydroxide (30% ammonia in water, reagent grade, Columbus Chemical Industries Inc.) were combined
with stirring. The catalyst solution was then added to the diluted Silbond solution drop-wise with
stirring. To improve properties such as mechanical integrity, thermal stability, casting and to reduce radiative heat transport, quartz felt (Saint Gobain), fumed silica (Alfa), quartz powder (Alfa)
and titania (Alfa) were added. The solution was then poured into molds containing the couples
and allowed to gel (approximately 30 minutes). After three days of aging at room temperature, the
samples were extracted from the mold and placed in an ethanol wash bath to remove any remaining
reactants or catalysts. The bath fluid was changed once per day for three washes total. All samples
54

Figure 3.16: Image of the process for integrating the aerogel in the TEG

were supercritically dried with carbon dioxide in a 12 liter critical point dryer (Accudyne Systems
Inc).
The process for integrating the aerogel into the full TEG differs slightly from the uncouples,
and has gone through several iterations; however, to demonstrate the essence of the process, a
typical case is presented here (Figure 3.16). Panels consisting of 25 skutterudite couples were prefilled with quartz fiber felt and assembled into a box shaped prototypical TEG (Figure 3.16a-c). A
space filler (aluminum block) is placed in the center of the box, the bottom of the box is sealed
and the liquid aerogel precursor is injected (Figure 3.16d). The liquid precursor gels in less than
one hour, after which the unit is sealed to prevent evaporation of the solvents during three days of
aging (Figure 3.16e) and placed in the critical point dryer for supercritical fluid extraction (Figure

55

Figure 3.17: The custom apparatus used to measure heat flux in thermoelectric unicouples

3.16f).

3.4.2.2

Characterization

Measuring the parasitic heat losses through the individual couples was conducted using a custom
heat-flux meter (Figure 3.17). The cold sides of the couples were soldered onto a copper heat
conductor 2.54 cm in diameter and 1 cm thick. The copper and couple unit was then encapsulated
with aerogel insulation 5.0 cm in diameter, centered around the couple and level with the hot side
interconnect (approximately 1.8 cm in total height). Casting the aerogel to a far larger diameter
than the copper diameter helps to isolate heat flux to the axial dimension, and also better represents
56

the 15:1 heat collector area ratio needed for optimal power output. Viewed from the side, the
heat-flux meter assembly had the following components from top to bottom: a tungsten heating
element 2.54 cm in diameter; the hot side of the couple; the cold side (which was soldered onto
the copper puck); an aluminum centering disk 5.0 cm in diameter and approximately 1mm thick
at the point of contact; a stainless steel heat flux meter; and finally, a copper chiller plate kept
at approximately 25◦ C with water flowing at 12 SCFM. Slight pressure was applied to ensure
intimate thermal contact between all components. The entire assembly was shrouded with a hightemperature machine-able insulation block, all thermal junctions below the copper puck contained
a thermal paste to minimize thermal resistances, and measurements were conducted in a bell jar
filled with inert gas to protect the tungsten element from oxidation. Couples without insulation
were used as a control.
The stainless steel heat flux meter contained two thermocouples spaced 1 cm apart in the direction of heat flow, and by measuring the temperature gradient across a defined distance in a material
with known thermal conductivity, we were able to calculate the total heat flux through the insulated sample. The tungsten heater was raised from room temperature to 500◦ C at 8◦ C/min, and
thermal equilibrium was reached after 30 minutes at temperature. Heat flux through the sample
was measured as an average difference in thermocouple temperatures over 5 minutes. The heat
flux through the sample was computed using Eqn 3.1:

Q = k·A

57

∆T
d

(3.1)

Where:
Q = total heat flux through the sample
k = thermal conductivity of stainless steel
A = cross-sectional area of stainless steel flux meter
∆T = temperature difference between the two thermocouples
d = spacing between thermocouples
The parasitic contribution was obtained by subtracting the heat flux through the couple itself,
as calculated using a finite element model, from the total heat flux computed in Eqn 3.1. The
thermocouple readings were accurate to ±2.2◦ C (or approximately 6.9%), the cross-sectional area
was accurate to ±0.06%, and the thermocouple spacing was accurate to ±0.04%, resulting in an
overall uncertainty of approximately 7%.
Measuring the parasitic heat losses from prototypical couple subassemblies was conducted with
an experimental test bed that simulated an exhaust system. A 13 kW electrical heater (Airtorch R ,
MHI inc) was used to heat nitrogen gas to 850◦ C. The heated gas was directed onto couple hot
shoes using a manifold consisting of a series of stainless steel tubes. Four panels comprised the
prototypical TEG; one panel contained five subassemblies (each subassembly had multiple skutterudite couples) that were insulated with aerogel insulation, another panel also contained the same
number of subassemblies/couples that were insulated with commercial silica wool and the other
two panels only had commercial silica wool. The commercial wool was high temperature silica
wool (McMaster-Carr part number 9356K11). Each TEG panel was actively cooled with water
58

cooling. The rise in temperature of the water entering and leaving could be measured to estimate
the heat flux through each panel.

3.4.3

Results

The aerogel shows uniform dispersion of white titania powder and intimate contact with the couple, as can be seen in Figure 3.18. Four samples with aerogel insulation were compared with four
control couples without any insulation, and the results are plotted in Figure 3.19 as the average
parasitic heat flux, with error bars representing one standard deviation. The large amount of variability in the control measurements is indicative of increased sensitivity to radiative losses from
the reflective copper and aluminum surfaces. Even with this variability, it is evident that the aerogel insulation dramatically reduces parasitic heat losses compared to the uninsulated control, from
4.3W (±2.6W) down to 0.2W (±0.2W). This lends support to the ability of aerogel to perform
as an effective cast-in-place high temperature insulation for thermoelectrics, but the difficulty of
isolating 1D heat flux means that other parasitic losses are possible that are not being measured in
this system. It is for this reason that testing of the thermal efficacy of aerogel on a complete TEG
was also performed.
A prototypical box-shaped TEG was constructed consisting of a manifold that forced heated
nitrogen gas to impinge upon couples mounted on two opposing panels (Figure 3.20a). The panel
of skutterudite couples on the left was insulated with commercial insulation while the panel of
skutterudite couples on the right was insulated with aerogel (Figure 3.20b). The top and bottom

59

Figure 3.18: Parasitic heat loss was measured on two types of samples: Aerogel insulated (top),
and an uninsulated control (bottom)

60

Figure 3.19: Parasitic heat loss comparison of uninsulated skutterudite couples and aerogel
insulated couples, showing that the aerogel insulation is effective in a test setup that approximates
real working conditions.

Figure 3.20: Heat flux and thermal efficiency testing: the heated gas is directed via a manifold to
impinge directly upon the couples (a), the assembled TEG is instrumented with thermocouples in
the HCs and in the cooling fluid located before and after each panel (b), the fully assembled TEG
test bed (c).

61

panels consisted of commercial insulation only. Nitrogen gas was heated by a 13kW electrical
heater (on the left of the TEG in Figure 3.20c). Thermocouples were inserted directly into the HCs
and directly into the cooling water before and after the panels.
Inlet gas temperature
Average HC temperature
Outlet gas temperature
Heated gas flow rate
Increase in aerogel panel fluid temperature
Increase in commercial insulation panel fluid temperature
Cooling fluid rate
Reduction in parasitic heat loss using aerogel insulation

725◦ C
650◦ C
725◦ C
27.25 m3 hr-1
5◦ C
7.1◦ C
0.11 m3 hr-1
30%

Table 3.3: Heat flux measurement parameters

The hot side of the couples was raised to an average value of 650◦ C to simulate exhaust conditions. The heat fluxes were compared by measuring the flow rate and rise in temperature of
the water cooling fluid before and after each panel (Table 3.3). The heat fluxes through the panel
with the commercial insulation and with the aerogel insulation were approximately 780 and 550
W (Figure 3.21). This results in an estimated 30% difference in heat flux. Furthermore, the electrical power output of both panels with couples was determined to be 25 W. Thus, the system-level
thermal-to-electric conversion efficiencies of the panels insulated with commercial and aerogel
insulation could be estimated at 3.2 and 4.5%, respectively.

62

Figure 3.21: This figure shows the ultimate efficacy of the aerogel insulation in an actual
thermoelectric generator. Heat flux was measured on all four panels that comprised the TEG.
Plates 1 and 3 did not have couples. Plate 2 had 25 skutterudite couples insulated with aerogel.
Plate 4 had 25 couples insulated with commercial insulation.

3.4.4

Conclusions

This work represents the first report in literature where aerogel insulation has been used in a terrestrial thermoelectric generator. An opacified, fiber-reinforced aerogel was successfully cast into
place and used in a prototypical skutterudite generator. The aerogel nearly eliminated parasitic heat
flux compared to an uninsulated unicouple, and reduced the parasitic heat loss of the generator by
30% compared to a commercial high-temperature insulating wool. It is important to note that this
is the first iteration of aerogel and TEG design - future work will focus on improved aerogel performance and ease of fabrication, allowing the fabrication of efficient thermoelectric generators for
waste heat recovery.

63

3.5
3.5.1

Ambiently Dried Aerogel Insulation
Introduction

This work details the characterization of an Ambiently Dried Aerogel Insulation (ADAI) based on
methyltrimethoxysilane precursor. The volumetric shrinkage during drying of unreinforced ADAI
is limited to 15%, and does not rely on the so-called "spring-back" effect common with other
ambiently dried aerogels. This allows it to function as a high-temperature insulation that may
be cast into place around rigid components without cracking. Shrinkage during drying, thermal
stability and thermal conductivity as a function of temperature are herein reported. The use of
ADAI in a prototypical thermoelectric device for waste heat recovery is detailed, showing the
ADAI to be an effective high-temperature thermal insulation.
With their high surface areas (300-1000 m2 g-1 ), high porosity (>95%), low density (as low
as 0.003 g cc-1 ) and bicontinuous solid-pore structure, aerogels are a very attractive "space-age
super-insulator" that has remained just outside commercial viability (with the exception of certain
niche applications) [7]. Supercritical drying plays a large role in the high cost of silica aerogel
insulation, as it requires slow solvent exchanges, expensive high-pressure equipment and batch
processing (Figure 3.22, top). Developing a low-cost aerogel insulation material is the subject of
much research worldwide with many different strategies having been developed, but to date none
have been met with widespread success.
The most common approach to ambiently dried aerogel involves a combination of exchang-

64

Figure 3.22: Various ways a sol-gel can be dried; Supercritically, to form an aerogel (top);
Ambiently, with modification and solvent exchange, to form a slightly less dense porous aerogel
(second row); Ambiently, without modification, solvent exchange, or spring-back (bottom row).

ing the pore fluid with a low-tension solvent and treating the surface with a silylating agent to
reduce surface tension (Figure 3.22, middle). Such a route has been systematically explored by A.
Venkateswara Rao and A. Parvathy Rao, who report on the effects of various solvents [19, 20] and
surface modifications [21, 22] on ambiently dried silica gels.
These ambiently dried gels could be further improved if a couple key challenges can be overcome. First, the additional processing steps involved in exchanging the pore fluid and chemically
modifying the surface can add significant time and cost. This has been somewhat alleviated by

65

Schwertfeger et al., who developed a single-step process for surface treatment and solvent exchange using a mixture of hexamethyldisiloxane and trimethylchlorosilane [23]. The second challenge is that such drying techniques typically rely on the so-called "spring-back" effect [7, 24],
where the gel compresses by >40% of its initial wet volume before expanding back to near its
original volume. While such a technique is perfectly suitable for aerogel monoliths, reliance on
spring-back precludes the use of ambiently dried silica aerogel as a cast-in-place insulation material in applications with complex geometries or rigid inclusions. In such situations, the gel will
crack and break apart during the initial high-shrinkage phase of drying before spring-back occurs.
Another strategy is to reinforce the aerogel structure itself with short fibers [25], or react a
crosslinked polymer on the silica surface [26, 27]. Unfortunately, both efforts still rely on some
degree of spring-back to achieve low densities, and the polymeric coating is not stable at the high
temperatures (>500◦ C) at which aerogel is most attractive as an insulation material.
This work details the fabrication and characterization of an ambiently dried aerogel insulation (ADAI) based on base-catalyzed methyltrimethoxysilane (MTMS) precursor, which achieves
<15% volumetric shrinkage without relying on spring-back (Figure 3.22, bottom). MTMS was first
explored as an alkoxide for the synthesis of supercritically dried aerogel by Hegde et al. [47], but
an ambient drying technique was never reported. Baghat et al. describe the ambient drying of acidbase catalyzed MTMS aerogels, but report that the gels underwent spring-back during drying [48].
The ability of ADAI to dry without spring-back, while maintaining relatively low thermal conductivity and mechanical integrity, is highly advantageous in that it allows cast-in-place insulation
66

of complex geometries in high-temperature applications. The microstructure, shrinkage during
drying and thermal conductivity of ADAI is herein reported, as is the performance of ADAI in a
prototypical thermoelectric generator for automobile waste heat recovery.

3.5.2

3.5.2.1

Experimental

Chemicals

Methyltrimethoxysilane (MTMS, Sigma-Aldrich no. 246174), methanol, ammonium hydroxide
(30%vol aqueous) and water (reverse osmosis) were used as received. For opacified samples,
titanium dioxide (1-2 um, Alfa Aesar no. 43047), fumed silica (90 m2 g-1 , Alfa Aesar no. 42737)
and quartz powder (1-2 um, Alfa Aesar no. 13024) were used as received. Fiber reinforcement
utilized quartz felt (Saint-Gobain) baked at 600◦ C for 1 hour to remove the organic binder, or
quartz wool (2 um diameter, Saint-Gobain) milled in a vibratory mill for 10 minutes (0.2 g of
wool, 40 mL of Ethanol and 15 1/8" PTFE balls in a 55 mL methacrylate vial) to obtain short
chopped fibers.

3.5.2.2

Gel preparation and drying

The molar ratio of MTMS, methanol, water and ammonium hydroxide was 1:3.5:4:1.75, respectively. The the water content of the ammonium hydroxide was subtracted from the total water
added to the sol. The MTMS and half of the methanol were stirred together in one vial, and the
remaining reagents were combined in a separate vial. Both solutions were stirred for 10 minutes,

67

and then the catalyst solution was added to the MTMS solution drop-wise with vigorous stirring
at a rate of 1 drop per second. The sol was then stirred for one minute. If powder opacification
or chopped fiber reinforcement was used, it was added to the mixed sol and then dispersed with a
probe sonicator for 30 seconds before pouring the sol into a glass mold. For samples using fiber
felt reinforcement, the felt was placed into the mold before pouring in the sol. Gelation took about
one hour, and the gels were allowed to age for 3 days at room temperature before being extracted
from the molds and dried under ambient conditions inside a fume hood.

3.5.2.3

Characterization

For evaluating shrinkage during drying and as a result of exposure to heat, gels 15.8 mm in diameter
and 17.2 mm tall were prepared without opacification or fiber reinforcement. Measuring the size
of soft gels is challenging, as variations in caliper pressure can cause significant variations in the
measured dimension. To counter this, a custom apparatus was used in which a high-precision linear
actuator with ±50 nm accuracy applied a force of 0.015 N for each data point, thus minimizing
variations in pressure between measurements.
To measure thermal conductivity, opacified, fiber reinforced samples were coated with graphite
and analyzed in a Netzsch LFA 447 Xenon Flash System. Surface areas were measured by nitrogen adsorption with a Micromeretics ASAP 2020. Scanning electron micrographs were obtained
using a JEOL 7500F SEM with a conductive osmium coating. Thermogravimetric Analysis (TGA)
was performed on a TA Instruments Q500, and Fourier-Transform Infrared (FTIR) spectra were

68

Figure 3.23: The microstructure of ADAI exhibits a wide range of particle sizes, ranging from
1-20 um in diameter, with relatively large pores.
obtained on a Perkin-Elmer model Spectrum 2000 spectrometer in ATR mode.

3.5.3

Results

Figure 3.23 shows the microstructure of an unopacified, unreinforced ADAI sample. ADAI appears
to have a wide distribution of particle sizes, ranging from 1-20 um in diameter. Nitrogen adsorption
indicates that the gel has a nominal surface area of 50 m2 g-1 by BET analysis, which is fairly low
for a sol-gel and is indicative of large particles. Additionally, the pore structure is relatively open,
with large open pores.
69

1.00
0.95
0.90
0.80

0.85

Mass (%)

0

200

400

Temperature (°C)

600

800

Figure 3.24: ADAI is thermally stable up to 250◦ C, after which it slowly loses surface methyl
groups. By 600◦ C, all methyl groups have oxidized.
Figure 3.24 shows the mass loss of dried ADAI as a function of temperature, as revealed by
thermogravimetric analysis. The gel remains largely stable until 250◦ C, when it experiences a
sharp 2% mass loss, followed by a gradual loss of mass until 600◦ C, where it remains fairly constant at 20% loss. FTIR spectra (Figure 3.25) of dried gels after exposure to different temperatures
70

Si−O−Si
Si−O−Si

Si−C
Si−C

As Dried

C−H

Absorbance

C−H

300°C

800°C

3500

3000

2500

2000

−1

λ (cm )

1500

1000

Figure 3.25: The mass loss between 300◦ C and 800◦ C is largely due to oxidation of surface
methyl groups, as evidenced by the loss of the Si-C peak.
in air indicate that the mass loss after 300◦ C is due to oxidation of the methyl surface groups, as
indicated by the disappearance of the Si-C peak at 1271 cm-1 . It is unknown what the cause of the
sharp drop in mass at 250◦ C is, as the FTIR spectra for ADAI heated to 300◦ C is largely identical
to the as-dried gel. A supercritically dried MTMS gel characterized by Rao et al. [49] lacks the
71

sudden mass loss, and exhibits a thermal stability approximately 100◦ C higher than ADAI. There
are two possible explanations for this difference in thermal stability. The first is that the acid-base
catalysis used by Rao may have affected the microstructure of the gel in such a way as to lower the
surface energy of the particles compared with ADAI, thus increasing the temperature at which the
methyl surface groups oxidize. A second possibility is that the supercritical extraction technique
removed residual reactants that remained in the ADAI pore structure. This may also explain the
presence of the sharp drop in mass at 250◦ C. The exact nature of the thermal stability of ADAI,
and the functional groups present on the surface, is the subject of an ongoing investigation, and
will be reported in a future publication.
Figure 3.26 shows both the mass loss and volumetric shrinkage of an unopacified, unreinforced
ADAI sample as a function of drying time. The drying process consists of two distinct phases.
During the first 90 minutes of drying, the gel loses approximately half of its mass as methanol
evaporates from the large pores. The size of these pores are large enough that capillary forces are
comparatively small, and the volumetric shrinkage is limited to roughly 5%. Between 90 minutes
and four hours, the gel loses the remaining pore fluid, accounting for another 30% of its wet mass.
This pore fluid is likely contained in smaller inter-particle pores, which cause a greater degree of
surface tension during drying and compress the gel to approximately 15% volumetric shrinkage
(or 5% linear shrinkage). After 4 hours, there is no further appreciable shrinkage or mass loss,
and the gel can be considered dry. It is believed that the relatively large particles and large pores
(seen in Figure 3.23) have the largest impact on reducing capillary forces, and thus shrinkage,
72

0
−40
−80

−60

% change

−20

% Volume Shrinkage
% Mass Shrinkage

20

50

100

200

500

Time (min)

1000

2000

5000

Figure 3.26: Shrinkage occurs in three stages. Up to 90 minutes, bulk pore fluid is evaporating
from large pores, resulting in minimal shrinkage. From 90min to 4 hours, residual pore fluid is
evaporating from smaller pores where capillary forces are greater, causing the bulk of the
shrinkage. After 4 hours, drying is complete. The gels are limited to 15%vol shrinkage, and do
not experience "spring-back."

73

Figure 3.27: Diagram detailing how the pore size can affect the gel size at different stages of
ambient drying. A gel with larger pores will experience reduced shrinkage, at the expense of
increased thermal conductivity due to gas convection.
in ADAI. Figure 3.27 illustrates how dimensional changes are more pronounced during the final
stages of drying, as pore fluid retreats into the smaller pores. Increasing the pore size of the gel
will reduce shrinkage during drying, although this likely comes at the expense of increased thermal
conductivity due to gas convection.
A time-lapse of photographs taken during the drying process is shown in Figure3.28. After
drying, the sample was placed in a quartz tube and exposed to 700◦ C in air for one hour. The
volumetric shrinkage after heating was 33%, which is comparable to other silica aerogel insulation
[46].
This multi-stage drying process is in good agreement with literature reports on ambiently-dried
74

Figure 3.28: Visual confirmation of the lack of spring-back during drying, as well as a sample
after heating to 500◦ C in air for 1 hour. The gradient in discoloration of the heated sample is due
to presence of oxygen during heating - the white section was closer to the open end of the quartz
tube in which the sample was heated.

75

aerogels [7], with a few key exceptions. First, it is important to note that the ADAI gels underwent
no solvent exchanges, in contrast with ambiently dried aerogels that utilize low surface tension
solvents [19, 20]. Second, the gel as prepared includes a hydrophobic surface group, lowering the
surface energy of the wet gel without the need for additional surface treatments [21, 22]. The third
and most important distinction of ADAI is that, while typical ambiently-dried aerogel can shrink
up to 40%vol during drying before "springing back" to 10%vol shrinkage, the ADAI gels in this
study demonstrate a maximum 15%vol shrinkage without relying on any spring-back phenomena.
This is important, as one of the strengths of sol-gel derived thermal insulations is that they can be
cast into place around complex geometries. If a gel needs to shrink by 40%vol during ambient
drying, it will strain and likely crack when cast around a rigid inclusion. Limiting the maximum
shrinkage during drying allows ADAI to be used in applications where typical ambiently-dried
aerogels will break apart.
While an aerogel that may be ambiently dried without relying on spring-back is interesting in its
own right, for it to perform effectively as an insulator it must have a low thermal conductivity. The
large pores which likely enable ambient drying of ADAI are also likely to increase gas convection
within the gel. The thermal conductivity of an opacified, reinforced sample is shown in Figure
3.29, and compared with the supercritically dried aerogel from Section 3.2. Although the thermal
conductivity is higher than the supercritically dried silica aerogel insulation, it is still low enough to
function as a high-performance thermal insulator. Further work will focus on finding an optimum
formulation, balancing shrinkage with thermal conductivity.
76

80

ADAI
Supercritical
q

60

q
q
q

q
q

20

40

q

0

Effective Thermal Conductivity (mW m−1 K−1)

q

0

50

100

150

200

Temperature (°C)

250

300

Figure 3.29: Thermal conductivity of ADAI is comparable to other high temperature insulation
materials. Error bars represent one standard deviation.

77

Figure 3.30: Third generation skutterudite-based thermoelectric generator for automotive waste
heat recovery, seen during fabrication. The white layer in the cross section is ADAI insulation.

ADAI was successfully used as thermal insulation in a thermoelectric generator for waste heat
recovery, as seen in Figure 3.30. A negative mold was designed (Figure 3.31, top) which utilized
an array of aluminum posts around which ADAI was cast. Once the gel had set and aged, the posts
were removed and the gel allowed to dry. The ambiently dried insulation was crack-free, easily
handled and assembled into the final generator design, where it performed extremely well.

78

Figure 3.31: Assembly of the ADAI insulation for the thermoelectric generator. Top left:
aluminum molds spray-coated with boronitride mold release. Top right: detail picture of
aluminum posts around which the ADAI will be cast. The posts will be removed when the gel
forms, and the actual thermoelectric unicouple legs will be inserted during final assembly.
Middle: The dried ADAI insulation, with a mica sheet to provide an initial barrier to gas diffusion
and improve handleability. Note the clean holes where the aluminum posts once were. Bottom:
side view of the ADAI insulation.

79

3.5.4

Conclusions

Ambiently dried aerogel insulation (ADAI) can be easily prepared through base-catalysis of an
MTMS precursor. In contrast with other ambiently dried aerogels, ADAI does not require costly
surface modifications or solvent exchanges, and does not rely on the spring-back effect to achieve
low densities after drying. This allows it to be cast into place in complex geometries and around
rigid inclusions that would otherwise break apart an ambiently dried aerogel. The thermal conductivity of ADAI is comparable to other high temperature insulation materials, and has been proven
effective in a prototypical thermoelectric generator for automotive waste heat recovery. With further optimization of powder opacification and fiber reinforcement, ADAI holds much promise as a
lower-cost high-temperature thermal insulation material.

80

Chapter 4
AEROGELS FOR ELECTROCHEMICAL ENERGY STORAGE

4.1

Background

The properties which make silica aerogels so useful as thermal insulation are also advantageous for
electrochemical energy storage. Capacitive storage systems obviously benefit from an increase in
charge-storing surface area, but even purely faradaic systems are improved through a combination
of increased reaction kinetics and defect-driven charge storage. Some of the initial work on chargestoring aerogels centered on vanadia for lithium-ion batteries [50–52], where the high surface area
aided energy densities. Sakamoto and Dunn extended this work by incorporating carbon nanotubes
into the vanadia solution, taking advantage of the sol-gel chemistry to fabricate an intimatelywired aerogel electrode with excellent rate capability [12]. In addition, Rolison and Dunn provide
an excellent summary of the characterization of aerogels of molybdenum and manganese oxides
[10], with both materials showing an increase in insertion capacity. Further work shows that for
manganese-oxide in particular, the aerogel morphology resulted in improved performance at higher
charge/discharge rates [53, 54].
This chapter details the characterization of two different charge-storing oxides based on aerogel
morphology. The first is a lithium titanate aerogel to be used as a high-power anode for a lithiumion battery, where the high surface area of the oxide results in pseudocapacitive behavior. The other

81

is a thin manganese oxide spinel coating on a carbon nanofoam with aerogel-like morphology.
This oxide-coated nanofoam combines thin diffusion distances in the oxide with the high electrical
conductivity of the carbon nanofoam to create an electrode with high power density. In addition,
the high concentration of surface defects increases the insertion capacity above the theoretical value
of the bulk oxide. Taken together, these two materials demonstrate the full range of improvements
achievable with aerogels for electrochemical energy storage.

82

4.2

4.2.1

Lithium Titanate Spinel Aerogel for Next-Generation Lithium-ion Batteries
Introduction

As society transitions to a carbon neutral energy future, it is becoming increasingly apparent that
advances in energy storage technology are necessary. For automotive applications in particular,
battery technology needs to be more powerful, store more energy, and cost less to manufacture,
while improving both safety and cycle life [55, 56]. Densification via sintering [57], atomic layer
deposition [58], and nanostructuring [59] of the electrode have all been successful in improving
cathode stability and performance. However, limited progress has been made on the anode. Metal
alloy electrodes are attractive due to an extremely large capacity; unfortunately, the excessive
volume change associated with that capacity limits their cycle life [60, 61].
Lithium titanate spinel (Li4 Ti5 O12 , LTO) is another candidate material that holds promise as
an anode material for advanced lithium ion batteries. Known as a zero-strain insertion oxide, it can
undergo thousands of cycles with little capacity loss [62–64]. Additionally, its insertion potential
(1.55 V vs. Li/Li+ ) remains above the onset of Solid Electrolyte Interphase (SEI) formation, allowing one to take advantage of nanoscale features not normally feasible with carbon electrodes
due to excessive SEI buildup. The most basic method of making LTO is through a solid-state reaction, where suitable precursors (usually lithium carbonate and titanium dioxide) are combined
via ball milling, then calcined at elevated temperatures (800-1000◦ C) in air to form the correct
phase. Many groups have made progress in increasing the active surface area through high-energy
83

ball milling [62] of the final LTO powder, or through sol-gel processing methods [65–68]. However, the sol-gel processing generally utilized ambient drying techniques, which lead to more dense
structures and increased particle sizes after calcination.
Aerogels, formed via the supercritical drying of sol-gels, have shown much promise as advanced energy storage materials [11,28,69]. The high surface area of the aerogel increases reaction
kinetics, decreases solid-state ion diffusion distances, and can even enhance capacity by increasing
the number of surface defects in certain oxides [10]. To date, no lithium titanate aerogel has been
reported in literature. This work details the fabrication of such an aerogel, and characterizes the
electrochemical properties of the high surface area electrode.

4.2.2

4.2.2.1

Experimental

Sample Preparation

Lithium acetate dihydrate (Sigma-Aldrich, L6883), titanium isopropoxide (Alfa Aesar, 77115),
triethanolamine (TEA) (Sigma-Aldrich, T58300), chloroform (Mallinckrodt Chemicals, 4440-08),
methanol (Sigma-Aldrich, 34860) and water (reverse osmosis purified) were used as received.
Styrene-Butadiene rubber (SBR) and carboxymethyl cellulose (CMC), both from MTI Corp., were
used as aqueous binding agents for electrode fabrication. Commercial LTO was obtained from NEI
Inc. (BE-10). Graphite nanoPlatlets (GnP) were used as a high-aspect ratio conductive additive,
for which the fabrication procedure is described elsewhere [70].

84

The molar ratio of titanium isopropoxide : TEA : water : methanol : chloroform was kept constant at 5:1.4:15:60:60, with the amount of lithium acetate varying from 5 to 20 moles. In one vial,
the TEA and chloroform were combined with stirring for 10 minutes, to which the titanium isopropoxide was added and stirred for another 30 minutes. In a second vial, the water and methanol
were combined with stirring for 10 minutes, to which the lithium acetate was added and stirred for
another 30 minutes until dissolved. The lithium acetate solution was then added to the titanium
isopropoxide solution drop-wise with vigorous stirring. After stirring for 5 minutes, the solution
was separated into 14mL glass molds, sealed and allowed to gel. The wet gels were aged for three
days, and then supercritically dried with carbon dioxide in a 12L critical point dryer (Accudyne
Systems Inc.) without any intermediate solvent exchanges. The dried gels were calcined at 800◦ C
for 3 hours under flowing argon.
For samples intended for electrochemical testing, 5wt% GnP (with respect to the final
Li4 Ti5 O12 mass) was added to the bulk solution and bath sonicated for 2 minutes before separating
into the smaller molds. Calcining under flowing argon prevents oxidation of the graphite. After
calcination, an aqueous electrode slurry was prepared by ball-milling a mass ratio of 75:15:5:5:540
LTO:GnP:SBR:CMC:water, taking into account the GnP already included in the LTO gel. This was
spread onto an aluminum current collector with a doctor blade at approximately 150 um thick. The
final electrode was calendared down to 27um, with a total areal loading of 0.8 mg cm-2 . Electrodes
were also made using a commercial LTO product from NEI Inc., with an areal loading of 2.48 mg
cm-2 at the same thickness.
85

4.2.2.2

Characterization

X-Ray Diffraction (XRD) spectra were collected with a Bruker D8 ADVANCE diffractometer on
powder samples roughly ground with an alumina mortar and pestle. Surface areas were measured
by nitrogen adsorption with a Micromeretics ASAP 2020. Scanning electron micrographs were
obtained using a JEOL 7500F SEM without the use of a conductive coating. For electrochemical
testing, 1/4" diameter electrodes were punched out, dried under vacuum at 110◦ C for 4 hours and
assembled in a three-electrode Swagelok cell inside an argon-filled glovebox with 1M LiPF6 in
1:1:1 EC:DEC:DMC electrolyte (MTI Corp.), using metallic lithium as both reference and counter
electrodes. Whatman glass filter paper (GF/F) was used as a porous separator.

4.2.3

Results

Solid-state methods of producing lithium titanate spinel utilize a Li:Ti precursor ratio that is typically 10% higher than stoichiometric (4.4:5) in order to compensate for the loss of lithium during
calcination. Figure 4.1 shows the composition of the LTO aerogel as a function of the Li:Ti precursor ratio. It is immediately apparent that even a 25% increase in the Li:Ti ratio (5:5) is insufficient
to obtain phase purity, and that a mixture of LTO spinel and lithium-poor titanates are instead
formed.
There are two possible mechanisms for such extensive loss of lithium. First, the high surface
area of the aerogel may increase the rate of lithium loss during calcination. Alternatively, it is
possible that the wet gel that forms may only be an amorphous titanium oxide gel, with a high con-

86

Li4Ti5O12
TiO2 rutile
(Li0.3Ti2.94)O6

20:5

Intensity

15:5

10:5

7.5:5

5:5

Commercial

20

40

2θ

60

80

100

Figure 4.1: Increasing the lithium:titanium precursor ratio increases the phase purity of the spinel.

87

centration of lithium in the pore fluid. Solvent exchanges would wash away the lithium-rich pore
fluid, leaving behind an insufficient amount of lithium to form the spinel phase. It was found that
placing a wet gel in a solution of methanol and chloroform caused the gel to dissolve, indicating
that the gel components are at best tenuously bound. By increasing the Li:Ti ratio and drying without preliminary solvent exchanges, it was possible to increase the phase purity of the calcined gel.
Unfortunately, this came at the expense of the calcined gel’s surface area, as shown in Figure 4.2.
This is in contrast to the findings of Mohammadi et al., who found an increase in surface area with
increasing lithium precursor content of thin-film LTO [71]. Although the mechanism for the effect
of lithium content on the surface area is not given, it is likely the difference between the two results
can be attributed to the gelation process. In the acid-catalyzed thin film, an increase in the lithium
precursor content reduced the overall HCl molarity, adjusting the isoelectric point of the forming
particles and changing the degree to which they crosslink to form the gel. In the base-catalyzed gel
of this work, the increase in lithium precursor (lithium acetate dihydrate) carried with it an increase
in water content, increasing the rate of hydrolysis and shortening the gelation time. This shorter
gelation time yields larger particles with fewer crosslinks, reducing the overall surface area. It was
decided that a Li:Ti ratio of 10:5 gave the best balance between phase purity and surface area, and
this ratio was used for all subsequent characterization.
Figure 4.3 shows the morphology of the calcined aerogel clustered around the conductive GnP
additive. The primary particle sizes are around 10-50 nm, forming aggregates up to a few microns
in diameter and showing good dispersion of the GnP. GnP was chosen as conductive additive for
88

50

q

20

30

q

q
q

0

10

Surface area (m2 g−1)

40

q

5:5

7.5:5

10:5

15:5

20:5

Li:Ti molar ratio
Figure 4.2: Increasing the lithium:titanium precursor ratio decreases the surface area of the
calcined aerogel

89

Figure 4.3: The calcined powder shows nanoscale crystallites (bottom) in good contact with the
conductive additive (top)

90

two reasons; to demonstrate that novel high-aspect ratio additives can be easily incorporated into
the sol-gel and maintained during calcination, and to enable future work on intrinsically wired
high-power electrodes that can take advantage of the nanoscale LTO aerogel morphology.
Figure 4.4 shows the cyclic voltammetry curves of an LTO aerogel electrode, with a similar
commercial LTO electrode for comparison. The low surface area commercial LTO (5-7 m2 g-1 )
exhibits a pair of redox peaks centered at 1.55 V, typical of an LTO battery. The high surface
area aerogel (30 m2 g-1 ) also has a redox behavior at 1.55 V, but it is superimposed onto a large
pseudo-capacitive box profile. This "batt-cap" behavior is typical of high surface-area electrodes
[11,28], and is advantageous in that it increases the power density of the electrode material without
sacrificing capacity. Galvanostatic charge-discharge curves (Figure 4.5) show that the aerogel has
a capacity comparable to the commercial product (137 vs. 147 mA h g-1 ). The sloping potential
profile of the aerogel is typical of high surface area aerogel electrodes with a distribution of particle
sizes, as the surface energies of the nanoparticles affect the insertion potential [12, 72].

4.2.4

Conclusions

This work represents the first report of producing a lithium titanate spinel aerogel for lithium ion
batteries. Sol-gel derived LTO gels are more susceptible to lithium loss than solid-state processed
compounds, and a higher lithium:titanium precursor ratio is necessary to maintain phase purity.
This purity comes at the expense of surface area, and a ratio of 10:5 appears to be a sufficient
compromise between the two. The high surface area of the aerogel yields a pseudocapacitive

91

0.0
−1.0

−0.5

Current (A g−1)

0.5

1.0

Aerogel
Commercial (NEI)

1.0

1.2

1.4

1.6

1.8

Potential (V vs Li/Li+)

2.0

Figure 4.4: Cyclic voltammetry shows a batt-cap behavior for the aerogel. Testing was done at
0.5 mV/s.

92

2.5
2.0
1.5
1.0

Potential (V vs Li/Li+)

Aerogel
Commercial (NEI)

0

50

100

150

−1

Capacity (mA h g )

Figure 4.5: At a C/3 rate, the aerogel has a capacity comparable to the commercial product.

93

behavior, which should allow for high pulse-power performance. The capacity of the aerogel
was 137 mA h g-1 , which is comparable to a commercial LTO electrode tested under the same
conditions. Incorporation of conductive additives in the sol-gel, and calcination of the gels in an
argon atmosphere, allow for intrinsic wiring of the oxide. This should enable the use of nanoscale
LTO aerogels at high rates, and will be the subject of future work.

94

4.3
4.3.1

Lithium Manganese Oxide Spinel Coated Carbon Nanofoams
Introduction

A carbon-neutral energy future will require advances in electrochemical energy technologies,
whether they are batteries and supercapacitors for energy storage, fuel cells and photovoltaics for
energy generation, or – more likely – a mixture of multiple complementary approaches. Lithiumion batteries are the current state of the art for energy storage, but demands are ever increasing.
Batteries need to be more powerful, store more energy, and last over more cycles, while costing
less to manufacture. The first two requirements are particularly troublesome, as traditional methods of increasing the specific power of an electrode necessarily reduce the specific energy [53, 73].
One method of decoupling specific power from specific energy is to abandon the planar electrode
design and instead construct the battery in three dimensions [8, 73–75]. Building in 3D allows for
very short solid-state ion-diffusion distances (tens of nanometers) between the electrodes, thereby
increasing the specific power of the device and enabling fast charge/discharge rates. Additionally,
and perhaps more importantly, a 3D electrode design allows one to "build up" capacity by adding
layers to increase electrode thickness, in much the same way as an architect increases the square
footage of a skyscraper by adding more floors. Areal capacity is maximized, and the overall cell
specific energy is increased.
Previous research has shown that 10-nm thick insertion oxides can be "painted" onto carbon
nanofoams to create aqueous electrochemical capacitors (ECs) with the potential to out-perform

95

current commercial technologies [76]. Manganese oxide (MnOx) coatings are of particular interest, as they utilize less toxic and more abundant elements than the cobalt or nickel compounds used
in more common lithium-ion battery electrodes. The particular polymorph of the manganese oxide
used in the 3D EC application has a low capacity to begin with (60 mAh g-1 ) [77], and the ultrathin
coating and large pore volume of the painted nanofoam will affect the capacity balance between
anode (Li or Li-based) and cathode in a 3D design. Rough estimates of a simple cell design show
that while a 10-nm-thick oxide coating would comprise 20% of the mass of the cell, it would have
less than 1% of the capacity of the lithium metal anode. Increasing the capacity of the oxide will
dramatically increase the overall specific energy of the cell.
Rather than turning to more exotic, expensive, and generally toxic oxide compounds with
higher capacities, it is possible to improve the MnOx oxide itself. Previous research has shown
that the presence of cation vacancies increases the insertion capacity of mixed-conducting transition metal oxide significantly [78, 79]. One method of introducing defects in MnOx is to increase
the surface area to take advantage of inherently defective surface sites [10, 78]. It has been shown
that increasing the surface area of sol-gel-derived α-MnO2 through supercritical drying to form
an aerogel will increase the capacity of the oxide to 100mAh g-1 , compared to 60 mAh g-1 for
the low-surface-area xerogel derived from the same original gel [75]. Ongoing work at the Naval
Research Lab (NRL) already shows that the lithiated spinel LiMn2 O4 , when painted on the carbon
nanofoams and electrochemically cycled in an aqueous system, exhibits a capacity higher than theoretical for the bulk crystal [80]. This implies that the high surface area of the oxide-coated foam
96

already provides a higher concentration of surface defects that increase insertion capacity.
In this work, the electrochemical performance of LiMn2 O4 spinel coated high-surface area carbon nanofoams is characterized in a non-aqueous electrolyte, with special detail given to insertion
capacity and rate capability. LiMn2 O4 spinel is an attractive oxide for lithium-ion batteries, due
to its low cost and absence of any toxicity. However, it exhibits poor cycling performance, likely
due to Jahn-Teller distortion and lattice parameter differences between adjacent phases [81]. It
is shown that the thin oxide layer and nanofoam morphology explored in this work dramatically
improves capacity retention during cycling at moderate rates. Furthermore, the high surface area
oxide and integrated carbon current collector allows for impressive capacity at high rates when
compared with literature values for the bulk oxide in a conventional electrode [81]. Finally, the
morphology of the oxide-coated nanofoam allows the spinel to better withstand the conversion
reaction at lower voltages when compared with bulk particles. This opens the door to further research on increasing the capacity of insertion oxides, and utilizing oxide-coated carbon nanofoams
for 3D battery architectures.

4.3.2

Experimental

Coating the carbon nanofoam with the oxide is a well-established process [82, 83]. Briefly, carbon
nanofoams are vacuum infiltrated with aqueous 0.1 M H2 SO4 or 0.1 M Na2 SO4 , then soaked
in 0.1 M NaMnO4 in either acidic or neutral conditions for 24-48 hours. The pH and duration
during this self-limiting deposition process controls the morphology of the deposited Na-birnessite

97

MnOx product, with neutral conditions yielding a more conformal coating through the interior of
the nanofoam. The samples are rinsed with water to remove residual permanganate solution, and
then soaked in 1M LiNO3 for 24 hours at room temperature to exchange the Na+ in the layered
birnessite structure with Li+ . The foams are then rinsed with water and dried thoroughly under
vacuum, before heating in flowing argon at 300◦ C for 4 hours to obtain the LiMn2 O4 structure.
For electrochemical testing, rectangular coupons were cut from the foam and affixed to an
aluminum current collector with a carbon paint. The electrodes were assembled into a CR-2032
coin cell with Celgard separator and lithium metal anode. 1M LiPF6 in 1:1:1 EC:DEC:DMC
was used as electrolyte. Both separator and electrode were vacuum infiltrated with the electrolyte
before cell assembly.

4.3.3

Results

Figure 4.6 shows the current response of the electrodes during cyclic voltammetry at a moderate
sweep rate of 0.5 mV s-1 between 3.5 and 4.4 V vs Li/Li+ . Even after 25 cycles at this sweep rate,
two pairs of redox reactions centered at approximately 4.0 and 4.15 V can be clearly identified,
indicating that the oxide coating is LiMn2 O4 , and is electrochemically active. Widening the voltage range (Figure 4.7) shows a third pair of redox peaks centered near 2.9 V, which is associated
with the high-strain conversion reaction. The oxide coating is able to withstand a number of cycles
over this large potential window, however it eventually loses capacity as the extensive dimensional
strain breaks the oxide apart.

98

0.5
0.0
−0.5

Current (A g−1)

Cycle no.
1
5
10
15
20
25

3.6

3.8

4.0

Potential (V vs Li/Li+)

4.2

4.4

Figure 4.6: Cyclic voltammetry of LiMn2 O4 coated nanofoams cycled from 3.5 to 4.4 V vs.
Li/Li+ at 0.5 mV s-1

99

1.5
0.5
0.0
−0.5
−1.5

−1.0

Current (A g−1)

1.0

Cycle no.
5
10
15
20
25

2.5

3.0

3.5

+

Potential (V vs Li/Li )

4.0

Figure 4.7: Cyclic voltammetry of LiMn2 O4 coated nanofoams cycled from 2.3 to 4.4 V vs.
Li/Li+ at 0.5 mV s-1 . Mechanical degradation of the electrode during sweeping to low potentials
caused a large amount of noise below 2.75 V.

100

Figure 4.8 shows the capacity fade of the oxide coating as a function of repeated cycling at a
1C symmetric rate (1 hour charge/discharge), with every 12th cycle at a slower C/5 rate (5 hours
per charge/discharge) in order to check if the capacity loss is due to degradation of the electrode’s
ability to handle the faster 1C rate. The oxide coated nanofoam loses approximately 12% of its
capacity after 100 cycles. For comparison, cycling data of bulk LiMn2 O4 from Tucker et al. is
also shown in Figure 4.9 [84]. The bulk oxide experiences a capacity loss of nearly 40% after
30 cycles at a C/15 symmetric rate. It is clear that the nanometric oxide coating on the nanofoam
improves the cycling performance of LiMn2 O4 , likely due to the nanometric thickness of the oxide
enabling strain relief of lattice distortions. This is supported by recent reports of nanoparticles and
nanowires of LiMn2 O4 exhibiting improved cycling performance [85, 86].
Figures 4.10-4.12 show the voltage response of oxide coated nanofoams during galvanostatic
cycling at different rates. At the lowest rate of 150 mA gactive -1 (approximately a 1C rate), the
electrode delivers 180 mAh gactive -1 , which is well above the theoretical value of 148 mAh g-1 for
full delithiation of LiMn2 O4 . Even after increasing the rate to 300 mA gactive -1 (2C rate) and 1350
mA gactive -1 (9C rate), the electrodes are still able to deliver 170 and 100 mAh gactive -1 , respectively. This is in sharp contrast to Shin et al. [81], who show in Figure 4.13 that bulk LiMn2 O4
achieves only 80 mAh g-1 at 1C, decaying to 50 mAh g-1 at only a 4C rate. The improved rate
capability of the nanofoam electrodes is likely due to short ion diffusion lengths through the oxide
thickness, in combination with the highly-conductive carbon nanofoam support acting as current
collector. Higher-than-theoretical capacities at "slow" rates of 150 or 300 mA g-1 can be explained
101

200
150
100
0

50

Capacity (mA h g−1)

q
qqq
qqqq
q
qq q
qq
q
qqqqq
qqqqq
q
qq
q
qqq
q
qqqq
qq
q
qqqqqq
qqqqq
qqqq
qqqqq q
qq q
qqq
qqqq
qqq
qq
qq
qqq q
qqq q
qqq
qqq
q
q

20

40

60

Cycle No.

80

100

Figure 4.8: Lithium insertion capacity of the LiMn2 O4 coated nanofoams over multiple cycles at
a symmetric 1C rate. Every 12th cycle was done at slower C/5 symmetric cycling.

102

Figure 4.9: At a slower C/15 cycling rate, bulk LiMn2 O4 (x=0, above) exhibits a much poorer
cycling performance. Figure taken from ref. [84].

LiNix Mn2Ϫx O4 , ͑e͒ LiZnx Mn2Ϫx O4 , ͑f͒ Li1ϩx Mn2Ϫx O4 : ͑᭿͒
es in part a represent data for different cells using the same

by a high concentration of surface defects in the high-surface-area oxide coated nanofoam, as seen
in other works [10, 75, 78, 80].

4.3.4

Conclusions

LiMn2 O4 spinel is an attractive insertion oxide for lithium-ion batteries, due to its low cost, comparable capacity and non-toxic nature, although it suffers from poor cycling performance. Depositing
LiMn2 O4 as a nanometric coating onto a high surface area carbon nanofoam allows the oxide to

103

4.2
4.0
3.8
3.6

Potential (V vs Li/Li+)

4.4

Cycle no.
1
5
10
15

0

50

100

Capacity (mA h g−1)

150

Figure 4.10: Galvanostatic cycling of the LiMn2 O4 coated nanofoam at 150 mA g-1
(approximately 1C).

104

200

4.2
4.0
3.8
3.6

Potential (V vs Li/Li+)

4.4

Cycle no.
1
5
10
15

0

50

100

Capacity (mA h g−1)

150

Figure 4.11: Galvanostatic cycling of the LiMn2 O4 coated nanofoam at 300 mA g-1
(approximately 2C).

105

200

4.2
4.0
3.8
3.6

Potential (V vs Li/Li+)

4.4

Cycle no.
1
5
10
15

0

50

100

Capacity (mA h g−1)

150

Figure 4.12: Galvanostatic cycling of the LiMn2 O4 coated nanofoam at 1350 mA g-1
(approximately 9C).

106

200

A208

Journal of The Electrochemical Society, 15

Ͼ3.58
IRC. T
micro
with
Teller
values
oxides
Th
combi
to LiM
exhibi
the pr
prope
cathod
brid v

Fin
istrati
F-125

Th
costs o

Figure 5.Figure 4.13: Galvanostatic cycling of bulk LiMn2 O4 (a), from ref. [81]. 2 O4 , ͑b͒
Comparison of the discharge profiles of ͑a͒ LiMn
LiMn1.85Ni0.075Li0.075O4 , and ͑c͒ LiCoO2 samples at C/10, C/5, C/2, 1C, 2C,
and 4C rates illustrating the rate capability.
107

Fe, Co, Ni, Cu, and Ga, 0 р y р 0.1, and 0 р z р 0.1) oxides

1. M.
Va
2. D.
3. H.
4. X.
5. M.
que
845
6. H.
7. D.
He
8. J. H
7͑
9. Y.
10. Y.

withstand a hundred cycles at a fast 1C symmetric charge/discharge rate with only a 12% capacity
fade, compared with 40% capacity fade after only 30 cycles at slower rates for the bulk oxide. Additionally, the thin oxide coating and massively parallel carbon nanofoam current collector allows
the electrode to retain impressive capacities at high rates. Finally, the high surface area of the oxide
coating increases the concentration of surface defects, yielding a capacity higher than theoretical
for the defect free bulk crystal. The result is a high capacity, low-cost, non-toxic electrode that can
withstand high rates and multiple cycles. Oxide-coated nanofoams hold promise as a traditional
electrode material for lithium ion batteries, and further improvements in insertion capacity will
enable the fabrication of a high-power 3D battery with appreciable charge storage.

108

Chapter 5
CONCLUSIONS

This dissertation represents a focused series of efforts to utilize aerogels towards pressing societal
needs for improved efficiency and energy storage technologies. The high surface area of aerogels, coupled with their tailorable pore structure and flexible sol-gel processing, allows aerogels of
many compositions to be made with customizable properties. The research herein focused on two
broad classes of aerogels: silica aerogels for thermal insulation, and insertion-oxide aerogels for
electrochemical energy storage.
The research on silica aerogels in Chapter 3 began with the characterization of silica aerogel
insulation for an Advanced Radioisotope Thermoelectric Generator for next-generation spacecraft
designs in Section 3.2. The silica aerogel performed well, and it was demonstrated that further
improvements in fiber reinforcement could be achieved. Work then progressed in Section 3.3
with the development of a surface modification scheme using chlorotrimethylsilane. In addition
to making the aerogel hydrophobic, the silane drastically improved the durability and mechanical
properties of the gel, allowing the gel to be compressed beyond 80% strain without the catastrophic
failure typical of unmodified gels. This surface modification scheme is then used in Section 3.4 in
a fiber-reinforced, opacified aerogel for a prototypical thermoelectric generator designed to recover
waste heat from automobile exhaust. The optimized aerogel performs well, reducing the parasitic
heat flux of the device by 30%. Finally, the knowledge gained from these studies was used in
109

Section 3.5 to create an aerogel insulation that could be ambiently dried in place in a thermoelectric
generator – eliminating the need for costly supercritical drying. The ability to be cast into place
represents a key advancement in ambient aerogel technologies, and holds much promise for the
use of aerogel in future thermal insulation applications.
The very properties that make silica aerogel a good insulator were also utilized in Chapter 4 to
improve the performance of electrode materials for lithium-ion batteries. It was shown in Section
4.2 that lithium titanate spinel could be fabricated with a high surface area aerogel morphology,
which demonstrated a pseudocapacitive behavior indicative of a high-rate electrode. A high-aspect
ratio conductive additive was also incorporated into the oxide structure by dispersing the particles
in the sol before gelation; this demonstrated the ease with which the aerogel can be optimized
through sol-gel processing techniques. Finally, in Section 4.3, a lithium manganese oxide spinel
coating was deposited onto a high surface-area carbon nanofoam and characterized. This oxide
coating exhibited excellent rate capability and cycling performance due to the short oxide diffusion length and intimate contact with the conductive carbon network. Most interestingly, the high
surface area of the oxide resulted in an insertion capacity higher than expected for the bulk oxide,
due to the high concentration of defects at the surface. Such electrode materials, taking advantage
of the high surface area and flexible synthesis of aerogels, hold much promise as next-generation
materials for electrochemical energy storage, and will remain a viable subject for research in the
years to come.

110

Chapter 6
SUGGESTIONS FOR FUTURE WORK

It has been shown that aerogels are attractive materials for efficiency and electrochemical energy
storage, owing mainly to their high surface areas and flexible sol-gel chemistry. Future work
will revolve around further optimization of aerogels for these applications, as well as detailed
investigations into the mechanisms through which the material’s properties are improved.
For the Ambiently Dried Silica Aerogel (ADAI) described in Section 3.5, future work will
focus on establishing the exact mechanism by which the material is able to dry without experiencing spring-back. It is suspected that the gel undergoes a two-phase gelation process, where large
particles first form to reinforce the gel, followed by the nucleation of small particles which block
gas convection. This can be verified by utilizing an "arrested gelation" technique, where the gels
are placed into baths of methanol at different stages of their gelation/aging process. The excess
methanol dilutes any unreacted precursor in the gel pore volume, halting further nucleation and/or
growth of silica particles. Nitrogen-sorption and SEM analysis of these arrested gels will allow a
timeline of the gelation process to be constructed, hopefully providing information as to how the
gel forms its bimodal structure.
The lithium titanate aerogel in Section 4.2 has the potential for use as a high-rate anode material, if sufficiently wired for electrical conductivity. Future work should focus on better integration
of conductive additive in order to take full advantage of the high surface area of the oxide material.
111

In addition, lithium titanate aerogel can be used to make a thick electrode, with porosity engineered
in such a way as to facilitate ion transport through the material. Such Highly-Ordered Hierarchical (HOH) electrodes have the potential to decouple the inverse relationship between power- and
energy-density, and are the subject of ongoing work.

112

BIBLIOGRAPHY

113

BIBLIOGRAPHY

[1] Glen Sweetnam. Meeting the World’s Demand for Liquid Fuels. In EIA 2009 Energy Conference. United States Department of Energy, April 2009.
[2] Shell International BV. Shell Energy Scenarios to 2050: Signals and Signposts. Technical
report, February 2011.
[3] Bruce Dunn, Haresh Kamath, and Jean-Marie Tarascon. Electrical Energy Storage for the
Grid: A Battery of Choices. Science, 334:928–935, 2011.
[4] Alain C Pierre. History of Aerogels. In Michel A. Aegerter, Nicholas Leventis, and
Matthias M Koebel, editors, Aerogels Handbook, pages 3–18. Springer New York, New York,
NY, March 2011.
[5] SS Kistler. Coherent Expanded Aerogels and Jellies. Nature, 127(3211):741, 1931.
[6] SS Kistler. Coherent expanded-aerogels. The Journal of Physical Chemistry, 36(1):52–64,
1932.
[7] A Soleimani Dorcheh and MH Abbasi. Silica aerogel; synthesis, properties and characterization. Journal of materials processing technology, 199(1-3):10–26, 2008.
[8] Jeffrey W Long, Bruce Dunn, Debra R Rolison, and Henry S. White. Three-dimensional
battery architectures. Chemical Reviews, 104:4463–4492, 2004.
[9] KE Swider, CI Merzbacher, PL Hagans, and Debra R Rolison. Aerogels as a tool to study the
electrical properties of ruthenium dioxide. Journal of Non-Crystalline Solids, 225(1):348–
352, 1998.
[10] Debra R Rolison and Bruce Dunn. Electrically conductive oxide aerogels: newmaterials in
electrochemistry. Journal Of Materials Chemistry, 11(4):963–980, 2001.
[11] Winny Dong, Jeffrey S Sakamoto, and Bruce Dunn. Electrochemical properties of vanadium
oxide aerogels. Science and Technology of Advanced Materials, 4:3–11, 2003.
[12] Jeffrey S Sakamoto and Bruce Dunn. Vanadium Oxide-Carbon Nanotube Composite Electrodes for Use in Secondary Lithium Batteries. Journal Of The Electrochemical Society,
149(1):A26, 2002.
[13] John M Watson, Adrienne T Cooper, and Joseph R.V. Flora. Nanoglued titanium dioxide
aerogels for photocatalysis. Environmental Engineering Science, 22(5):666–675, 2005.

114

[14] Hiroshi Hirashima. Preparation of TiO2 Aerogels-Like Materials Under Ambient Pressure. In
Michel A. Aegerter, Nicholas Leventis, and Matthias M Koebel, editors, Aerogels Handbook,
pages 145–153. Springer New York, New York, NY, March 2011.
[15] R W Pekala. Organic aerogels from the polycondensation of resorcinol with formaldehyde.
Journal of Materials Science, 24(9):3221–3227, September 1989.
[16] Theodore F Baumann, Alexander E Gash, and Joe H Satcher. A Robust Approach to Inorganic Aerogels: The Use of Epoxides in Sol-Gel Synthesis. In Michel A. Aegerter, Nicholas
Leventis, and Matthias M Koebel, editors, Aerogels Hansbook, pages 155–170. Springer New
York, New York, NY, March 2011.
[17] Stephanie L Brock and Hongtao Yu. Chalcogenide Aerogels. In Michel A. Aegerter, Nicholas
Leventis, and Matthias M Koebel, editors, Aerogels Handbook, pages 367–384. Springer New
York, New York, NY, March 2011.
[18] John H Harreld, Winny Dong, and Bruce Dunn. Ambient Pressure Synthesis of AerogelLike Vanadium Oxide and Molybdenum Oxide. Materials Research Bulletin, 33(4):561–567,
April 1998.
[19] A. Parvathy Rao, A. Venkateswara Rao, and JL Gurav. Effect of protic solvents on the
physical properties of the ambient pressure dried hydrophobic silica aerogels using sodium
silicate precursor. Journal of Porous Materials, 15(5):507–512, 2008.
[20] A. Parvathy Rao, GM Pajonk, and A. Venkateswara Rao. Effect of preparation conditions on
the physical and hydrophobic properties of two step processed ambient pressure dried silica
aerogels. Journal of Materials Science, 40(13):3481–3489, 2005.
[21] A. Venkateswara Rao, GM Pajonk, SD Bhagat, and Philippe Barboux. Comparative studies
on the surface chemical modification of silica aerogels based on various organosilane compounds of the type Rn Si4-n . Journal of Non-Crystalline Solids, 350:216–223, 2004.
[22] A. Parvathy Rao, A. Venkateswara Rao, and G.M. Pajonk. Hydrophobic and physical properties of the ambient pressure dried silica aerogels with sodium silicate precursor using various
surface modification agents. Applied Surface Science, 253(14):6032–6040, 2007.
[23] F Schwertfeger, D Frank, and M Schmidt. Hydrophobic waterglass based aerogels without
solvent exchange or supercritical drying. Journal of Non-Crystalline Solids, 225(1):24–29,
1998.
[24] Douglas M Smith, David Stein, Julie M Anderson, and William Ackerman. Preparation of
low-density xerogels at ambient pressure. Journal of Non-Crystalline Solids, 186:104–112,
1995.

115

[25] Jeff S Sakamoto, Harold Schock, T Caillat, J P Fleurial, Ryan Maloney, Matt Lyle, Trevor
Ruckle, Ed Timm, and Long Zhang. Skutterudite-Based Thermoelectric Technology for
Waste Heat Recovery: Progress Towards a 1 kW Generator. Science of Advanced Materials, 3(4):621–632, August 2011.
[26] Lynn A Capadona, Mary Ann B Meador, Antonella Alunni, Eve F. Fabrizio, Plousia Vassilaras, and Nicholas Leventis. Flexible, low-density polymer crosslinked silica aerogels.
Polymer, 47(16):5754–5761, 2006.
[27] Nicholas Leventis, Chariklia Sotiriou-Leventis, Guohui Zhang, and Abdel-Monem M.
Rawashdeh. Nanoengineering strong silica aerogels. Nano Letters, 2(9):957–960, January
2002.
[28] Paul E Tang, Jeffrey S Sakamoto, Emmanuel Baudrin, and Bruce Dunn. V2 O5 aerogel as a
versatile host for metal ions. Journal of Non-Crystalline Solids, 350:67–72, 2004.
[29] J F Widmann, J C Yang, T J Smith, S L Manzello, and G W Mulholland. Measurement of
the optical extinction coefficients of post-flame soot in the infrared. Combustion and Flame,
134(1-2):119–129, July 2003.
[30] J Fricke, X Lu, P Wang, D Buttner, and U Heinemann. Optimization of monolithic silica
aerogel insulants. International Journal of Heat and Mass Transfer, 35(9):2305–2309, January 1992.
[31] J Kuhn, T Gleissner, MC Arduini-Schuster, S Korder, and J Fricke. Integration of mineral
powders into SiO2 aerogels. Journal of Non-Crystalline Solids, 186:291–295, 1995.
[32] Steven M Jones and Jeffrey Sakamoto. Applications of Aerogels in Space Exploration. In
Michel A. Aegerter, Nicholas Leventis, and Matthias M Koebel, editors, Aerogels Handbook,
pages 721–746. Springer New York, New York, NY, March 2011.
[33] C J Brinker and G W Scherer. Sol-Gel Science. The Physics and Chemistry of Sol-Gel
Processing. Academic Press, San Diego, CA, January 1990.
[34] Kelly E Parmenter and F Milstein. Mechanical properties of silica aerogels. Journal of
Non-Crystalline Solids, 223(3):179–189, January 1998.
[35] Te-Yu Wei, Shih-Yuan Lu, and Yu-Cheng Chang. Transparent, Hydrophobic Composite
Aerogels with High Mechanical Strength and Low High-Temperature Thermal Conductivities. Journal of Physical Chemistry B, 112(38):11881–11886, September 2008.
[36] A. Parvathy Rao, A. Venkateswara Rao, and Uzma K.H. Bangi. Low thermalconductive,
transparent and hydrophobic ambient pressure dried silica aerogels with various preparation conditions using sodium silicate solutions. Journal of Sol-Gel Science and Technology,
47(1):85–94, 2008.
116

[37] A. Venkateswara Rao and Ravindra R Kalesh. Organic surface modification of TEOS based
silica aerogels synthesized by co-precursor and derivatization methods. Journal of Sol-Gel
Science and Technology, 30(3):141–147, 2004.
[38] WL Vasconcelos and LL Hench. Mechanical properties evolution during sintering of optical
sol-gel silica. Journal of Non-Crystalline Solids, 121(1-3):132–135, 1990.
[39] Guohui Zhang, Amala Dass, Abdel-Monem M. Rawashdeh, Jeffrey Thomas, Joseph A.
Counsil, Chariklia Sotiriou-Leventis, Eve F. Fabrizio, Faysal Ilhan, Plousia Vassilaras,
Daniel A Scheiman, Linda Mccorkle, Anna Palczer, J. Chris Johnston, Mary Ann Meador,
and Nicholas Leventis. Isocyanate-crosslinked silica aerogel monoliths: preparation and
characterization. Journal of Non-Crystalline Solids, 350:152–164, 2004.
[40] Mary Ann B Meador, Eve F. Fabrizio, Faysal Ilhan, Amala Dass, Guohui Zhang, Plousia
Vassilaras, J. Chris Johnston, and Nicholas Leventis. Cross-linking amine-modified silica
aerogels with epoxies: Mechanically strong lightweight porous materials. Chemistry Of Materials, 17(5):1085–1098, January 2005.
[41] G. L. Bennet, J. J. Lombardo, R. J. Hemler, G. Silverman, C. W. Whitmore, W. R. Amos,
E. W. Johnson, A. Schock, R. W. Zoecher, T. K. Keenan, J. C. Hagan, and R. W. Englehart.
Mission of Daring: The General-Purpose Heat Source Radioisotope Thermoelectric Generator. In 4th International Energy Conversion Engineering Conference and Exhibit (IECEC),
pages 26–29, San Diego, CA, June 2006.
[42] C. B. Vining. page 329. CRC Handbook on Thermoelectrics. CRC Press, Boca Raton, FL,
1995.
[43] E. Skrabek. page 267. CRC Handbook on Thermoelectrics. CRC Press, Boca Raton, FL,
1995.
[44] F. Stabler. Automotive Applications of High Efficiency Thermoelectrics. In DARPA/ONR
Program Review and DOE High Efficiency Thermoelectric Workshop, San Diego, CA, pages
24–27, 2002.
[45] J Yang and F.R. Stabler. Automotive applications of thermoelectric materials. Journal of
electronic materials, 38(7):1245–1251, 2009.
[46] Ryan Maloney and Jeffrey Sakamoto. Large deformation of chlorotrimethylsilane treated
silica aerogels. Journal of Non-Crystalline Solids, 357(10):2059–2062, May 2011.
[47] Nagaraja D Hegde and A. Venkateswara Rao. Physical properties of methyltrimethoxysilane
based elastic silica aerogels prepared by the two-stage sol–gel process. Journal of Materials
Science, 42(16):6965–6971, 2007.

117

[48] Sharad D Bhagat, Chang-Sup Oh, Yong-Ha Kim, Young-Soo Ahn, and Jeong-Gu Yeo.
Methyltrimethoxysilane based monolithic silica aerogels via ambient pressure drying. Microporous and Mesoporous Materials, 100(1-3):350–355, 2007.
[49] A. Venkateswara Rao, MM Kulkarni, DP Amalnerkar, and T Seth. Superhydrophobic silica aerogels based on methyltrimethoxysilane precursor. Journal of Non-Crystalline Solids,
330(1-3):187–195, 2003.
[50] Hiroshi Hirashima and Kazumi Sudoh. Preparation of V2 O5 aerogels and aerogel coatings.
Journal of Non-Crystalline Solids, 145:51–54, 1992.
[51] F Chaput, B Dunn, P Fuqua, and K Salloux. Synthesis and characterization of vanadium
oxide aerogels. Journal of Non-Crystalline Solids, 188(1-2):11–18, July 1995.
[52] D. B. Le, A. L. Tipton, B. B. Owens, and W. H. Smyrl. Aerogels and Xerogels of V2 O5 as
Intercalation Hosts. Journal Of The Electrochemical Society, 142(6):L102, 1995.
[53] Jeffrey W Long, Rhonda M Stroud, and Debra R Rolison. Controlling the pore-solid architecture of mesoporous, high surface area manganese oxides with the birnessite structure.
Journal of Non-Crystalline Solids, 285(1-3):288–294, June 2001.
[54] Christopher P Rhodes, Jeffrey W Long, Michael S Doescher, Brett M Dening, and Debra R Rolison. Charge insertion into hybrid nanoarchitectures: mesoporous manganese oxide
coated with ultrathin poly (phenylene oxide). Journal of Non-Crystalline Solids, 350:73–79,
2004.
[55] A.A. Pesaran, T. Markel, H.S. Tataria, and D Howell. Battery Requirements for Plug-In
Hybrid Electric Vehicles - Analysis and Rationale. 23rd International Electric Vehicle Symposium (EVS-23), July 2009.
[56] Bruno Scrosati and Jürgen Garche. Lithium batteries: Status, prospects and future. Journal
of Power Sources, 195(9):2419–2430, May 2010.
[57] Wei Lai, Can K Erdonmez, Thomas F Marinis, Caroline K Bjune, Nancy J Dudney, Fan Xu,
Ryan Wartena, and Yet-Ming Chiang. Ultrahigh-Energy-Density Microbatteries Enabled by
New Electrode Architecture and Micropackaging Design. Advanced Materials, 22(20):E139–
E144, May 2010.
[58] YS Jung, AS Cavanagh, Anne C Dillon, MD Groner, SM George, and SH Lee. Enhanced
Stability of LiCoO Cathodes in Lithium-Ion Batteries Using Surface Modification by Atomic
Layer Deposition. Journal Of The Electrochemical Society, 157:A75, 2010.
[59] Antonino Salvatore Arico, Peter Bruce, Bruno Scrosati, Jean-Marie Tarascon, and Walter van
Schalkwijk. Nanostructured materials for advanced energy conversion and storage devices.
Nature Materials, 4(5):366–377, January 2005.
118

[60] Uday Kasavajjula, Chunsheng Wang, and A John Appleby. Nano- and bulk-silicon-based
insertion anodes for lithium-ion secondary cells. Journal of Power Sources, 163(2):1003–
1039, January 2007.
[61] Robert A Huggins. Negative Electrodes in Lithium Cells. In Aerogels Handbook, pages
123–149. Springer Science+Business Media, LLC, New York, NY, 2009.
[62] W Iwaniak, J Fritzsche, M Zukalova, R Winter, M Wilkening, and P Heitjans. Li conductivity
of nanocrystalline Li4 Ti5 O12 prepared by a sol-gel method and high-energy ball milling.
Defect and Diffusion Forum, 289-292:565–570, 2009.
[63] A S Prakash, P Manikandan, K Ramesha, M Sathiya, J-M Tarascon, and A K Shukla.
Solution-Combustion Synthesized Nanocrystalline Li4 Ti5 O12 As High-Rate Performance LiIon Battery Anode. Chemistry Of Materials, 22(9):2857–2863, May 2010.
[64] E Ferg, RJ Gummow, A de Kock, and MM Thackeray. Spinel Anodes for Lithium-Ion Batteries. Journal Of The Electrochemical Society, 141(11):L147–L150, January 1994.
[65] Yan-Jing Hao, Qiong-Yu Lai, Ji-Zheng Lu, Dong-Qiang Liu, and Xiao-Yang Ji. Influence of
various complex agents on electrochemical property of Li4 Ti5 O12 anode material. Journal
Of Alloys And Compounds, 439(1-2):330–336, 2007.
[66] C Shen, X Zhang, Y Zhou, and H Li. Preparation and characterization of nanocrystalline
Li4 Ti5 O12 by sol-gel method. Materials Chemistry and Physics, 78(2):437–441, 2003.
[67] Yan-Jing Hao, Qiong-Yu Lai, Ji-Zheng Lu, Hong-Li Wang, Yuan-Duan Chen, and Xiao-Yang
Ji. Synthesis and characterization of spinel Li4 Ti5 O12 anode material by oxalic acid-assisted
sol-gel method. Journal of Power Sources, 158(2):1358–1364, 2006.
[68] Yan-Jing Hao, Qiong-Yu Lai, Dong-Qiang Liu, Zhi-ui Xu, and Xiao-Yang Ji. Synthesis by
citric acid sol-gel method and electrochemical properties of Li4 Ti5 O12 anode material for
lithium-ion battery. Materials Chemistry and Physics, 94(2-3):382–387, 2005.
[69] Jeffrey W Long, Michael S Logan, Christopher P Rhodes, Everett E Carpenter, Rhonda M
Stroud, and Debra R Rolison. Nanocrystalline Iron Oxide Aerogels as Mesoporous Magnetic
Architectures. Journal of the American Chemical Society, 126(51):16879–16889, December
2004.
[70] H. Fukushima. Graphite nanoreinforcements in polymer nanocomposites. PhD thesis, Michigan State University, East Lansing, MI, 2003.
[71] M R Mohammadi and D J Fray. Low temperature nanostructured lithium titanates: controlling the phase composition, crystal structure and surface area. Journal of Sol-Gel Science
and Technology, 55(1):19–35, July 2010.
119

[72] W J H Borghols, M Wagemaker, U Lafont, E M Kelder, and F M Mulder. Size Effects in
the Li4+x Ti5 O12 Spinel. Journal of the American Chemical Society, 131(49):17786–17792,
December 2009.
[73] Debra R Rolison, Jeffrey W Long, Justin C Lytle, Anne E Fischer, Christopher P Rhodes,
Todd M McEvoy, Megan E Bourg, and Alia M Lubers. Multifunctional 3D nanoarchitectures
for energy storage and conversion. Chemical Society Reviews, 38(1):226–252, January 2009.
[74] Timothy S Arthur, Daniel J Bates, Nicolas Cirigliano, Derek C Johnson, Peter Malati,
James M Mosby, Emilie Perre, Matthew T Rawls, Amy L Prieto, and Bruce Dunn. Threedimensional electrodes and battery architectures. MRS bulletin, 36(07):523–531, 2011.
[75] Christopher P Rhodes, Jeffrey W Long, Katherine A Pettigrew, Rhonda M Stroud, and Debra R Rolison. Architectural integration of the components necessary for electrical energy
storage on the nanoscale and in 3D. Nanoscale, 3:1731, 2011.
[76] Jeffrey W Long, Daniel Bélanger, Thierry Brousse, Wataru Sugimoto, Megan B Sassin, and
Olivier Crosnier. Asymmetric electrochemical capacitors—Stretching the limits of aqueous
electrolytes. MRS bulletin, 36(07):513–522, 2011.
[77] Jeffrey W Long, Karen E Swider-Lyons, Rhonda M Stroud, and Debra R Rolison. Design of
Pore and Matter Architectures in Manganese Oxide Charge-Storage Materials. Electrochemical and Solid State Letters, 3(10):453–456, 2000.
[78] Benjamin P Hahn, Jeffrey W Long, and Debra R Rolison. Something from nothing: Enhancing electrochemical charge storage with cation vacancies. Accounts of Chemical Research,
In Press:1–29, 2012.
[79] Paul Ruetschi and R Giovanoli. Cation Vacancies in MnO2 and Their Influence on Electrochemical Reactivity. Journal Of The Electrochemical Society, 135(11):2663–2669, November 1988.
[80] Megan B Sassin and Jeffrey W Long. Unpublished. Naval Research Laboratory, 2011.
[81] Youngjoon Shin and Arumugam Manthiram. Factors Influencing the Capacity Fade of Spinel
Lithium Manganese Oxides. Journal Of The Electrochemical Society, 151(2):A204, 2004.
[82] Anne E Fischer, Katherine A Pettigrew, Debra R Rolison, Rhonda M Stroud, and Jeffrey W
Long. Incorporation of Homogeneous, Nanoscale MnO2 within Ultraporous Carbon Structures via Self-Limiting Electroless Deposition: Implications for Electrochemical Capacitors.
Nano Letters, 7(2):281–286, February 2007.
[83] Megan B Sassin, Azzam N Mansour, Katherine A Pettigrew, Debra R Rolison, and Jeffrey W
Long. Electroless Deposition of Conformal Nanoscale Iron Oxide on Carbon Nanoarchitectures for Electrochemical Charge Storage. ACS nano, 4(8):4505–4514, 2010.
120

[84] Michael C Tucker, Jeffrey A Reimer, and Elton J Cairns. A 7 Li NMR Study of Capacity Fade
in Metal-Substituted Lithium Manganese Oxide Spinels. Journal Of The Electrochemical
Society, 149(5):A574, 2002.
[85] Do Kyung Kim, P Muralidharan, Hyun-Wook Lee, Riccardo Ruffo, Yuan Yang, Candace K
Chan, Hailin Peng, Robert A Huggins, and Yi Cui. Spinel LiMn2 O4 Nanorods as Lithium
Ion Battery Cathodes. Nano Letters, 8(11):3948–3952, November 2008.
[86] C H Jiang, S X Dou, H K Liu, M Ichihara, and H S Zhou. Synthesis of spinel
LiMn2 O4 nanoparticles through one-step hydrothermal reaction. Journal of Power Sources,
172(1):410–415, October 2007.

121