07-04-2012, 11:30 AM
NANOLITHOGRAPHIC CONTROL OF CARBON NANOTUBE SYNTHESIS
[attachment=19675]
ABSTRACT
Nanolithographic Control of Carbon Nanotube Synthesis. (December 2007)
David Huitink, B.S. Texas A&M University
Chair of Advisory Committee: Dr. Debjyoti Banerjee
A method offering precise control over the synthesis conditions to obtain carbon
nanotube (CNT) samples of a single chirality (metallic or semi-conducting) is presented.
Using this nanolithographic method of catalyst deposition, the location of CNT growth is
also precisely defined.
This technique obviates three significant hurdles that are preventing the exploitation
of CNT in micro- and nano-devices. Microelectronic applications (e.g., interconnects,
CNT gates, etc.) require precisely defined locations and spatial density, as well as
precisely defined chirality for the synthesized CNT. Conventional CVD synthesis
techniques typically yield a mixture of CNT (semi-conducting and metallic types) that
grow at random locations on a substrate in high number density, which leads to extreme
difficulty in application integration.
Dip Pen Nanolithography (DPN) techniques were used to deposit the catalysts at
precisely defined locations on a substrate and to precisely control the catalyst
composition as well as the size of the patterned catalyst. After deposition of catalysts, a
low temperature Chemical Vapor Deposition (CVD) process at atmospheric pressure
was used to synthesize CNT. Various types of catalysts (Ni, Co, Fe, Pd, Pt, and Rh) were
deposited in the form of metal salt solutions or nano-particle solutions. Various
iv
characterization studies before and after CVD synthesis of CNT at the location of the
deposited catalysts showed that the CNT were of a single chirality (metallic or semiconducting)
as well as a single diameter (with a very narrow range of variability).
Additionally, X-ray photoelectron spectroscopy (XPS) was used to characterize the
deposited samples before and after the CVD, as was lateral force microscopy (LFM) for
determination of the successful deposition of the catalyst material immediately after
DPN as well as following the CVD synthesis of the samples. The diameter of the CNT
determines the chirality. The diameter of the CNT measured by TEM was found to be
consistent with the chirality measurements obtained from Raman Spectroscopy for the
different samples. Hence, the results showed that CNT samples of a single chirality can
be obtained by this technique. The results show that the chirality of the synthesized CNT
can be controlled by changing the synthesis conditions (e.g., size of the catalyst patterns,
composition of the catalysts, temperature of CVD, gas flow rates, etc.).
ACKNOWLEDGEMENTS
First of all, I would like to thank Dr. Banerjee for providing me with an incredible
opportunity to conduct meaningful research in such an interesting field. I can definitely
say that I have learned more in the process of researching over the past year than I would
have ever expected. Also, for his guidance and support as my research advisor and
committee chair, I am grateful for his efforts to aide my pursuit of this degree. I am also
thankful for the opportunity for which he allowed me to be a co-participant in the
Summer Research Fellowship Program for the Air Force Research Laboratory at Wright-
Patterson Air Force Base in Dayton, Ohio. It was a valuable and memorable experience.
Additionally, I extend my thanks to Dr. Cagin and Dr. Heffington for their participation
in my oversight committee, and to Dr. Heffington, once more, for his support and
mentoring throughout my academic career.
To all who helped in the process of completing the research – thank you! Dr. Sinha
played a vital role in the actual CVD synthesis of the samples. Rohit Gargate took on an
immense task of trying to fill in for me during my absence to Dayton. Thanks also to
Dr. Raj Ganguli of the Materials Directorate at the Air Force Research Laboratory in
Dayton, OH for help with SEM measurements during my stay there. And to the staff of
the Texas A&M Materials Characterization Facility and the Microscopy and Imaging
Center, thanks for your help in troubleshooting the equipment, and giving the occasional
needed “pointers.”
vii
Furthermore, I would like to thank my wonderfully supportive wife, Linsay, and our
son, Davis, for putting up with the odd hours, late night studying, and travel which
separated us for several weeks. I love you both. And to my family and friends who
encouraged me, thanks for always being there.
INTRODUCTION
1.1 Carbon Nanotube Synthesis
At the present state, production of carbon nanotubes (CNTs) has been accomplished*
via several techniques, both natural1,2 and artificial; however, the primary means of
obtaining “high purity” nanotubes is fairly limited. Two popular methods of
synthesizing large quantities of CNTs include laser ablation of graphite rods and arc
discharge which require a plasma source to induce the growth of nanotubes in the
presence of catalysts (e.g. Pd, Ni, Co, etc.).3 In 1991, carbon soot formed between two
graphitic electrodes operating at 100 amps was found to contain nanotubes;4 and later,
this method of arc discharge was refined by Ebbesen, et al.5 The high temperatures
involved in the electric arc causes the carbon in the negative graphitic electrode to
sublimate and form hollow rods. The high yield for this method (up to 30%) allows for
the macro-generation of tubes in measures of grams, which are comprised of a mixture
of single-walled and multi-walled tubes up to 50 microns in length.6 Guo, et al., first
demonstrated laser ablated CNT formations by creating a target composed of graphite
powder, carbon cement, and catalytic metal.7 In the presence of argon gas flow, the
resultant baked rod produces CNTs under incident laser irradiation when heated to
approximately 1200°C. Carbon Nanotechnologies, Inc (CNI) in Houston alternatively
utilizes a high-pressure carbon monoxide (HiPco) flow method in the presence of an
This thesis follows the style of Nano Letters.
2
electric arc - a procedure developed in 1999 - to generate larger amounts of single wall
nanotubes (SWNTs) than the original plasma discharge methods.8 Furthermore,
chemical vapor deposition (CVD) provides another useful means of producing large
quantities of nanotubes as first demonstrated by M. José-Yacamán et al.,9 which has
gained popularity due to the fact that CVD allows for synthesis of CNTs in the absence
of vacuum, unlike the other popular methods. Using this method, a substrate coated with
metal catalyst particles is heated near 700°C over which a carbonaceous gas (such as
ethanol or acetylene) is introduced alongside a process gas (like nitrogen). The carbon
in the reactor chamber collects on the metal catalyst from which tubes are formed as the
atoms migrate along the surface of the metal particles. It has also been shown that
application of a strong electric field which results in a plasma inside the reactor (known
as plasma enhanced chemical vapor deposition) will generate CNT growth in the
direction of the electric field.10 Using this method, it is possible to obtain a sample of
vertically aligned CNT by arranging the CVD reactor perpendicular to the substrate, a
feat not reproducible using the other methods.6
Irrespective of the synthesis process (with the exception of arc discharge), in-situ
TEM observations have shown that carbide precursors gather on the surface of catalyst
particles (transition metals, typically Fe, Ni, Co, and intermingled combinations, among
others) at which point rapid rod-formations occur on the catalyst metals, which are metastable
carbide particles.11 Following the appearance of these rods, graphitic structures
form on the rod slowly, resulting in the final, hollow nanotube.11 Studies12 have
suggested that this rapid rod-forming process is the result of the carbon extruding from
3
the catalyst particle labeled "base (or root) growth" and "tip growth" depending on
whether the metal particle remains at the base or moves at the end of the tube proceeding
forth from the substrate, respectively. These models suggest, which has been
validated,13 that the size of the catalyst particles play a direct role in the diameter of the
synthesized CNTs, for which it appears that the chirality (orientation of hexagonal bonds
comprising tube wall) of the formed nanotubes are somewhat dependent on the tube
diameter. This is of particular interest, since the chiral behavior of CNT determines the
electrical properties of the tube, where it may be either conducting or semi-conducting;
and, furthermore, theory has indicated that a conducting CNT can handle 1,000 times the
current density of conventional conductive materials like copper or silver.6 Additionally,
a nuclear magnetic resonance study of SWNT formations by Wu, et al., showed that
changing the catalyst composition in a laser ablation target and by increasing the
exposure to oxygen resulted in varying proportions of metallic nanotubes in the
synthesized CNT sample.14 According to the authors, these results "indicate that the
chirality distribution in SWNT samples is not always random and might be controllable
by synthesis conditions."
1.2 Carbon Nanotube Characterization
Among many of the various techniques used for the evaluation of CNTs and their
properties, Raman Spectroscopy has emerged as a prominent method for verification of
tube presence when imaging is difficult to obtain, as well as for the determination of
structural characteristics.15,16 In the presence of the laser excitation, the CNTs can
4
exhibit three very distinct resonance peaks, each representing a different tube feature.
The first observed peak (or set of peaks) is called the "radial-breathing mode" or RBM
(sometimes, RB mode), whose Raman shift is observed in the range of roughly 50 - 400
cm-1. The reason for this nomenclature is explained by the fact that this peak is the
result of the radial pulsing of the tube cylinder which perhaps mimics the expanding and
contracting of a lung during breathing. Because the relationship seen between the shift
location of this peak and the diameter of the tube, Dresselhaus developed the following
empirical correlation for the determination of SWNT diameters from Raman
information.17
RBM
SWNT D
ω
= 248 (1)
Here, DSWNT is the diameter of the SWNT in nm, and ωRBM is the Raman shift location
of the observed peak in cm-1. The two other typically observed peaks are the D and G
bands, which occur near 1300 and 1600 cm-1 respectively, and which refer to the
disorder (D) and graphitic (G) structure of the CNT sample. Prior to Dresselhaus, a
study by Eklund, et al., calculated the numerator in this equation to be 223.8 nm/cm
when predicting the theoretical behavior of the rolled graphitic structure.18 Other
authors have noted the use of 234 nm/cm,19 as well; however, the value of 248 nm/cm as
found by Dresselhaus, et al., has been exclusively used in a variety of recent studies.
This formula is only valid for tube diameters ranging from 1 to 2 nm. Since the size of
these peaks are a statistical representation of the concentration of these features in the
sample, a Raman response with a D-band larger than the G-band typically reflects that
5
the sample is predominantly multi-walled. In other words, an entirely single-walled tube
sample will show little or no evidence of this peak. The G-band alternatively provides
insight into the chiral nature of the tube. This "chirality" refers to the orientation of the
hexagonal carbon-carbon bonding patterns that are seen in CNTs. Depending of the
angle described by the (n,m) indices, the CNTs may be either conducting or semiconducting.
Chiral angles resulting in an "armchair" pattern are "metallic" (where n – m
equals a multiple of 3); those exhibiting the "zig-zag" pattern can either be conducting
(1/3 of n, m combinations) or semi-conducting (2/3 of (n,m) combinations). The
remaining index combinations in between armchair and zig-zag are referred to as
"chiral" and are semi-conducting. Figure 1 is a schematic of these two CNT types,
where the origins of the names for these types are seen in the shapes made at the end of
the open tubes. When the chiral bond angles of the tube are aligned in such a way that
Figure 1. Armchair and Zig-Zag Type CNT Chiralities
Armcahir
(Conducting)
Zig-Zag
(Conducting or
Semi-conducting)
6
results in a "metallic" tube sample, the G-band will shift to a lower Raman shift
frequency than seen in semi-conducting samples.20 Essentially all CNT samples exist in
a mixture of all chirality types with diameters varying within a certain range, and
therefore, Raman is most often used for classifying the statistical distribution of the
presence of these tube types.
[attachment=19675]
ABSTRACT
Nanolithographic Control of Carbon Nanotube Synthesis. (December 2007)
David Huitink, B.S. Texas A&M University
Chair of Advisory Committee: Dr. Debjyoti Banerjee
A method offering precise control over the synthesis conditions to obtain carbon
nanotube (CNT) samples of a single chirality (metallic or semi-conducting) is presented.
Using this nanolithographic method of catalyst deposition, the location of CNT growth is
also precisely defined.
This technique obviates three significant hurdles that are preventing the exploitation
of CNT in micro- and nano-devices. Microelectronic applications (e.g., interconnects,
CNT gates, etc.) require precisely defined locations and spatial density, as well as
precisely defined chirality for the synthesized CNT. Conventional CVD synthesis
techniques typically yield a mixture of CNT (semi-conducting and metallic types) that
grow at random locations on a substrate in high number density, which leads to extreme
difficulty in application integration.
Dip Pen Nanolithography (DPN) techniques were used to deposit the catalysts at
precisely defined locations on a substrate and to precisely control the catalyst
composition as well as the size of the patterned catalyst. After deposition of catalysts, a
low temperature Chemical Vapor Deposition (CVD) process at atmospheric pressure
was used to synthesize CNT. Various types of catalysts (Ni, Co, Fe, Pd, Pt, and Rh) were
deposited in the form of metal salt solutions or nano-particle solutions. Various
iv
characterization studies before and after CVD synthesis of CNT at the location of the
deposited catalysts showed that the CNT were of a single chirality (metallic or semiconducting)
as well as a single diameter (with a very narrow range of variability).
Additionally, X-ray photoelectron spectroscopy (XPS) was used to characterize the
deposited samples before and after the CVD, as was lateral force microscopy (LFM) for
determination of the successful deposition of the catalyst material immediately after
DPN as well as following the CVD synthesis of the samples. The diameter of the CNT
determines the chirality. The diameter of the CNT measured by TEM was found to be
consistent with the chirality measurements obtained from Raman Spectroscopy for the
different samples. Hence, the results showed that CNT samples of a single chirality can
be obtained by this technique. The results show that the chirality of the synthesized CNT
can be controlled by changing the synthesis conditions (e.g., size of the catalyst patterns,
composition of the catalysts, temperature of CVD, gas flow rates, etc.).
ACKNOWLEDGEMENTS
First of all, I would like to thank Dr. Banerjee for providing me with an incredible
opportunity to conduct meaningful research in such an interesting field. I can definitely
say that I have learned more in the process of researching over the past year than I would
have ever expected. Also, for his guidance and support as my research advisor and
committee chair, I am grateful for his efforts to aide my pursuit of this degree. I am also
thankful for the opportunity for which he allowed me to be a co-participant in the
Summer Research Fellowship Program for the Air Force Research Laboratory at Wright-
Patterson Air Force Base in Dayton, Ohio. It was a valuable and memorable experience.
Additionally, I extend my thanks to Dr. Cagin and Dr. Heffington for their participation
in my oversight committee, and to Dr. Heffington, once more, for his support and
mentoring throughout my academic career.
To all who helped in the process of completing the research – thank you! Dr. Sinha
played a vital role in the actual CVD synthesis of the samples. Rohit Gargate took on an
immense task of trying to fill in for me during my absence to Dayton. Thanks also to
Dr. Raj Ganguli of the Materials Directorate at the Air Force Research Laboratory in
Dayton, OH for help with SEM measurements during my stay there. And to the staff of
the Texas A&M Materials Characterization Facility and the Microscopy and Imaging
Center, thanks for your help in troubleshooting the equipment, and giving the occasional
needed “pointers.”
vii
Furthermore, I would like to thank my wonderfully supportive wife, Linsay, and our
son, Davis, for putting up with the odd hours, late night studying, and travel which
separated us for several weeks. I love you both. And to my family and friends who
encouraged me, thanks for always being there.
INTRODUCTION
1.1 Carbon Nanotube Synthesis
At the present state, production of carbon nanotubes (CNTs) has been accomplished*
via several techniques, both natural1,2 and artificial; however, the primary means of
obtaining “high purity” nanotubes is fairly limited. Two popular methods of
synthesizing large quantities of CNTs include laser ablation of graphite rods and arc
discharge which require a plasma source to induce the growth of nanotubes in the
presence of catalysts (e.g. Pd, Ni, Co, etc.).3 In 1991, carbon soot formed between two
graphitic electrodes operating at 100 amps was found to contain nanotubes;4 and later,
this method of arc discharge was refined by Ebbesen, et al.5 The high temperatures
involved in the electric arc causes the carbon in the negative graphitic electrode to
sublimate and form hollow rods. The high yield for this method (up to 30%) allows for
the macro-generation of tubes in measures of grams, which are comprised of a mixture
of single-walled and multi-walled tubes up to 50 microns in length.6 Guo, et al., first
demonstrated laser ablated CNT formations by creating a target composed of graphite
powder, carbon cement, and catalytic metal.7 In the presence of argon gas flow, the
resultant baked rod produces CNTs under incident laser irradiation when heated to
approximately 1200°C. Carbon Nanotechnologies, Inc (CNI) in Houston alternatively
utilizes a high-pressure carbon monoxide (HiPco) flow method in the presence of an
This thesis follows the style of Nano Letters.
2
electric arc - a procedure developed in 1999 - to generate larger amounts of single wall
nanotubes (SWNTs) than the original plasma discharge methods.8 Furthermore,
chemical vapor deposition (CVD) provides another useful means of producing large
quantities of nanotubes as first demonstrated by M. José-Yacamán et al.,9 which has
gained popularity due to the fact that CVD allows for synthesis of CNTs in the absence
of vacuum, unlike the other popular methods. Using this method, a substrate coated with
metal catalyst particles is heated near 700°C over which a carbonaceous gas (such as
ethanol or acetylene) is introduced alongside a process gas (like nitrogen). The carbon
in the reactor chamber collects on the metal catalyst from which tubes are formed as the
atoms migrate along the surface of the metal particles. It has also been shown that
application of a strong electric field which results in a plasma inside the reactor (known
as plasma enhanced chemical vapor deposition) will generate CNT growth in the
direction of the electric field.10 Using this method, it is possible to obtain a sample of
vertically aligned CNT by arranging the CVD reactor perpendicular to the substrate, a
feat not reproducible using the other methods.6
Irrespective of the synthesis process (with the exception of arc discharge), in-situ
TEM observations have shown that carbide precursors gather on the surface of catalyst
particles (transition metals, typically Fe, Ni, Co, and intermingled combinations, among
others) at which point rapid rod-formations occur on the catalyst metals, which are metastable
carbide particles.11 Following the appearance of these rods, graphitic structures
form on the rod slowly, resulting in the final, hollow nanotube.11 Studies12 have
suggested that this rapid rod-forming process is the result of the carbon extruding from
3
the catalyst particle labeled "base (or root) growth" and "tip growth" depending on
whether the metal particle remains at the base or moves at the end of the tube proceeding
forth from the substrate, respectively. These models suggest, which has been
validated,13 that the size of the catalyst particles play a direct role in the diameter of the
synthesized CNTs, for which it appears that the chirality (orientation of hexagonal bonds
comprising tube wall) of the formed nanotubes are somewhat dependent on the tube
diameter. This is of particular interest, since the chiral behavior of CNT determines the
electrical properties of the tube, where it may be either conducting or semi-conducting;
and, furthermore, theory has indicated that a conducting CNT can handle 1,000 times the
current density of conventional conductive materials like copper or silver.6 Additionally,
a nuclear magnetic resonance study of SWNT formations by Wu, et al., showed that
changing the catalyst composition in a laser ablation target and by increasing the
exposure to oxygen resulted in varying proportions of metallic nanotubes in the
synthesized CNT sample.14 According to the authors, these results "indicate that the
chirality distribution in SWNT samples is not always random and might be controllable
by synthesis conditions."
1.2 Carbon Nanotube Characterization
Among many of the various techniques used for the evaluation of CNTs and their
properties, Raman Spectroscopy has emerged as a prominent method for verification of
tube presence when imaging is difficult to obtain, as well as for the determination of
structural characteristics.15,16 In the presence of the laser excitation, the CNTs can
4
exhibit three very distinct resonance peaks, each representing a different tube feature.
The first observed peak (or set of peaks) is called the "radial-breathing mode" or RBM
(sometimes, RB mode), whose Raman shift is observed in the range of roughly 50 - 400
cm-1. The reason for this nomenclature is explained by the fact that this peak is the
result of the radial pulsing of the tube cylinder which perhaps mimics the expanding and
contracting of a lung during breathing. Because the relationship seen between the shift
location of this peak and the diameter of the tube, Dresselhaus developed the following
empirical correlation for the determination of SWNT diameters from Raman
information.17
RBM
SWNT D
ω
= 248 (1)
Here, DSWNT is the diameter of the SWNT in nm, and ωRBM is the Raman shift location
of the observed peak in cm-1. The two other typically observed peaks are the D and G
bands, which occur near 1300 and 1600 cm-1 respectively, and which refer to the
disorder (D) and graphitic (G) structure of the CNT sample. Prior to Dresselhaus, a
study by Eklund, et al., calculated the numerator in this equation to be 223.8 nm/cm
when predicting the theoretical behavior of the rolled graphitic structure.18 Other
authors have noted the use of 234 nm/cm,19 as well; however, the value of 248 nm/cm as
found by Dresselhaus, et al., has been exclusively used in a variety of recent studies.
This formula is only valid for tube diameters ranging from 1 to 2 nm. Since the size of
these peaks are a statistical representation of the concentration of these features in the
sample, a Raman response with a D-band larger than the G-band typically reflects that
5
the sample is predominantly multi-walled. In other words, an entirely single-walled tube
sample will show little or no evidence of this peak. The G-band alternatively provides
insight into the chiral nature of the tube. This "chirality" refers to the orientation of the
hexagonal carbon-carbon bonding patterns that are seen in CNTs. Depending of the
angle described by the (n,m) indices, the CNTs may be either conducting or semiconducting.
Chiral angles resulting in an "armchair" pattern are "metallic" (where n – m
equals a multiple of 3); those exhibiting the "zig-zag" pattern can either be conducting
(1/3 of n, m combinations) or semi-conducting (2/3 of (n,m) combinations). The
remaining index combinations in between armchair and zig-zag are referred to as
"chiral" and are semi-conducting. Figure 1 is a schematic of these two CNT types,
where the origins of the names for these types are seen in the shapes made at the end of
the open tubes. When the chiral bond angles of the tube are aligned in such a way that
Figure 1. Armchair and Zig-Zag Type CNT Chiralities
Armcahir
(Conducting)
Zig-Zag
(Conducting or
Semi-conducting)
6
results in a "metallic" tube sample, the G-band will shift to a lower Raman shift
frequency than seen in semi-conducting samples.20 Essentially all CNT samples exist in
a mixture of all chirality types with diameters varying within a certain range, and
therefore, Raman is most often used for classifying the statistical distribution of the
presence of these tube types.