Interaction of hydrogen with novel carbo

Interaction of hydrogen with novel carbon materials - CaltechTHESIS
CaltechTHESIS
A Caltech Library Service
About
Browse
Deposit an Item
Instructions for Students
Interaction of hydrogen with novel carbon materials
Citation
Ye, Yun
(2001)
Interaction of hydrogen with novel carbon materials.
Dissertation (Ph.D.), California Institute of Technology.
doi:10.7907/z8wn-gs34.
Abstract
The hydrogen storage properties of sonic carbon materials were studied. Graphite nanofibers (GNF) were synthesized by catalytic decomposion of ethylene and hydrogen. Catalyst supported carbon materials were prepared by impregnation process. Hydrogen desorption and adsorption properties of graphite nanofibers, single-walled carbon nanotubes (SWNT), fullerene materials and catalysts supported carbon materials were measured volumetrically using a Sievert's apparatus. The hydrogen desorption capacity of GNF was typically less than 0.2 wt.%. A phase transition between crystal SWNT and a new hydride phase was found at high pressures at 80K. The phase transition was of first order, and involved the separation of the individual tubes within a rope, exposing a high surface area for hydrogen adsorption. From the change in chemical potential of the hydrogen gas upon adsorption, we were able to calculate the cohesive van der Waals energy between the tubes as 5 mcV/C atom. This is much smaller than expected from previous theoretical work, and shows that defects in the crystal structure cause large suppressions of the cohesive energy. We were able to alter this cohesive energy by changing the state of the material. Over several cycles of isotherm measurements at 77 K, the hydrogen storage capacities of one of the fullerite samples increased from an initial value of 0.4 wt% for the first cycle to a capacity of 4.2 wt% for the fourth cycle. Correspondingly, the surface area increased from 0.9 m^2/gm to 11 m^2/gm and showed a phase transformation, characterized by X-ray powder diffraction. By adding Ni particles onto the sample, the hydrogen storage capacity of fullerite and activated carbon sample was increased. The adsorption of hydrogen on Ni particle can not account for the total increased capacity even by assuming complete coverage of hydrogen molecules on the Ni particle surface.
Item Type:
Thesis (Dissertation (Ph.D.))
Subject Keywords:
Materials Science
Degree Grantor:
California Institute of Technology
Division:
Engineering and Applied Science
Major Option:
Materials Science
Thesis Availability:
Public (worldwide access)
Research Advisor(s):
Fultz, Brent T. (advisor)
Ahn, Channing C. (co-advisor)
Thesis Committee:
Unknown, Unknown
Defense Date:
8 August 2000
Funders:
Funding Agency
Grant Number
DOE Office of Science
DE-FG03-94ER 14493
Record Number:
CaltechTHESIS:10122010-083542958
Persistent URL:
DOI:
10.7907/z8wn-gs34
Default Usage Policy:
No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:
6129
Collection:
CaltechTHESIS
Deposited By:
Benjamin Perez
Deposited On:
12 Oct 2010 15:54
Last Modified:
16 Apr 2021 22:58
Thesis Files
Preview
PDF
- Final Version
See Usage Policy.
7MB
Repository Staff Only:
item control page
CaltechTHESIS is powered by
EPrints 3.3
which is developed by the
School of Electronics and Computer Science
at the University of Southampton.
More information and software credits
Interaction of Hydrogen with Novel Carbon
Materials

Thesis bv

YUH Y('

III Partial Fulfilllllf'lIt of tll(' Hequin'llH'llts
for 11)(' Degree of
Doctor of Philosophy

r--

California Institute of T('cllllology

Pasadella, (:aliforIlia

2001

(SlIhmit.t.ed August ~th, :WOO)

2001

'fUI1 'Y"e

All Bights H('scrvt'd

III

Acknow ledgements
My 1II0st sincer<' thallks must go to my advisor. Proressor Brellt Fult.;.>; ror guiding
my scient.ific research, for being a caring llH'ut.or alld a sourcc of illspirat.ioll, support

alld CllCOli ragcllwllt..
I am far more grateful thall a r(,\\! words can express. to Dr. ('hallnillg AIm, who
has lwell collahorat.illg \vit.h Ill<' on this work. He was always willillg t.o hplp, to talk
and listcl\.
Dr.

('harks 1\. Witham pat.ielltly trained ll1e 011 t.he Sicvcrt's apparat.1ls and

sharf'c\ with me lots useful kllovvledge, hoth in and out of the lah and officc.
Dr. Hobert. C. Bowman, .Jr., and Dr. Hatnakulllar V. Bugga of the .let Propulsion
Lahorator.V provided a lot. of uSf'ful ad vise on Illy research. I am al so than krul t.o Dr.
John Vajo at Hughs Hest'arch Laboratory. for his sharp and inspiring discussion on

t.\w work.
I would likf' to thank Professor. George H. Gavalas and his graduat.e student Be
Lai for helping Ill£' wit.h the BET llleaSUJ'('nH'llts.
Professor G('org(' n. Hosslllall and Dr.

Elizah<'1h ArredolIdo in Geological and

Pla.n<'1ary Sciellce Division kindly provided F011l'ier TrallSfol'lll Infrared eq1lipllIClIt.
and guided Ill<' t.o operat.(' t.he f'(jl1ipllJ('nt amI interpret. the data.
I am grateful t.o Professor Glf'n Cass ror his kindllf'ss. rrif'll(lship and gell('\'osity.
Tlw work was support.ed in part by the grallt. lI11111her DE-FGO:~-9·lEH 144~n rrom
thc DOE Office of Scicuc<" amI hy a gift from IIH L Laboratory.
Finally I would like to thank my family and all tlJ(' peopk at (~alt('ch who lllad('
my liff' ill the past flv(, :vears elljoyable.

Abstract
TIlt' hydrogf'll st.orage propert.ies of sOllle carbon Inaterials were studi('<\. (;rapltite
lIallofil)prs (GNF) were syntltf'sized h~' catalytic decolllposioll of dlt\'I('lle alld It.vdrogell.

('at.al~'st

supported carholl materials W('1"e prepared by illlpregnatioll proccss.

Hydrogcll dcsorptioll and adsorption propnties of graphite nallofilwrs, sillgle-walled
carboll lIallotulws (S\VNT), fulkrell(, mat.erials alld catal~'s1s supported carl)()n 1I1crystal S\VNT ami a new h.vdridc pha.'-;e was foulld at high pressures at t-IOI\:. The
phase transitioll was of first ordcl", and involved the separation of the individual
tubes withill a rope, exposing a high surfacc area for hydrogcl] adsorpt.ioll. From the
chang<' ill cl](,lllicaI potential of til<' h.nlrogcll gas IIpOIl adsorpt.ioll, wc \\'('1"e able to
calculate t.he colwsive van dn \Vaals Cll<'rgy Iwtw('ell the tubes as G lllCV/C atom.
This is Illuch smalln than expectcd frolll pn'vious t1]('orctical work, alld shows that
defects ill 11](' crystal structure cause large supprcssiolls of the cohesive ('I]('rg,\·. \Vc
we're able to alt.er this cohesive el)('rg~' hy challging the state of the lIlat(')'ia1. Over
several cycks of isothnJl1 IllCaSI\l'('IlH'llts at II I\:, thc hydrogcn st.orage capacitics of
01]('

of the fullerik samples ill

cycle to a capacity of 4.2 wlarea increased from 0.9 lJ12/ gm to 11 1ll:2 /gllJ, alld showed a phase t.rallsforlllation,
characterized by X-ray powder diffraction.

B.v adding Ni part icks onto the sa1l1pk,

til(' h.vdrogf'll st.orage capacity of fulkrit(, and activat.ed carbon sample' was increased.

The adsorption of h.vdrogc'll all N i part ide call 1101 aCcollllt for 1he 10tal i n(Teased capacity evell by assuming complete coverage of hydrogen molecules 011 tllf' Ni particle
SlJ rface.

Contents
Acknowledgements

III

Abstract

Introduction
l.1

1.2

I.:~

Hydrogell
1.1.1

General Aspect.s .

1.1.2

Productioll of Hydrogell

l.1.:~

Storage Methods

Fuel Cells . . . . . . . .

1.2.1

Fuel Cel\ Principles

1.2.2

Classification of Fuel Cells

1.2.:3

Fuel Choice for Fuel (~dl Vehicles

(~a.rbOll

l\lat.erials . . . . . . . . . . . . .

I!)

1.:3.\

Graphite aucl Graphite Materials

\.S

1.:~.2

Act.ivated ('arbon

I!)

1.:3.:3

Fullerenes

1.:3A

Nanotu\ws

2·1

l.:LS

Graphite Nanofibcrs

Ph.vsisorptioll and (']H'lllisorpt.ioll

1.4.1

Langmuir Isot.herm

:l\

1.4.2

BET Isotlwrlll . . .

Bibliography

Experimental Techniques, Equipment, and Data Analysis

2.1

Sample Preparation auel Modification . . . . . . . . . . . . .

\,1

2.2

2.:3

2.1.1

Na1lofilH'r Synt.lwsis ..

2.1.2

I~ 6it.sc 1111If'l"- II 11 fflll a1l Met hod

2.1.:3

Sillgk-Walled Nanotulws Prodl1ctio1l

2.1.-1

Sonicatio1l and Filtrat.ioll .

1:3

2.1.!"i

Cat.alyst. I1llpregnat.ioll

10
SYllthesis of Flllkr(,lH's

Sa1llplc' (~haract(')'i7,at.ion
X-ray Diffraction

2.2.2

Translllission Electron Microscopy.

-\9

2.2.:3

BET . . . . . . . . . . . . . . . . .

!"i0

2.2.4

Fourier Transform Infrared Spectroscopy

Sa1llpk l\lcasUH'llH'nt. and Data Alia lysis

2.:3.1

Sieverfs Apparatus . . . . . . . .

2.:3.2

Hydrogcll (~olllpressihlit..\' and Idc'al Gas ('orredion

Hydrogen Desorption and Adsorption Measurements on Graphite
Nanofibers
:3.1

Hist.ory and ('oll1llH'rciallll1erest.s

:3.2

(~at.alyst.s and Sample' Pn'paratioll .

:3.:3

Sample lVIorphology . . . . . . . . .

:3.4

Isot.herm l\kasurellH'llt. a1l(1 Data Analysis

:3.(i

Discussion

:3.7

(~oncll1sions

Bibliography

2.2.1

Bibliography

.t I

Hydrogen Adsorption and Cohesive Energy of Single-Walled Carbon
Nanotubes
4. j

Background

4.2

ExperimeIlts

77
~~

I I

VII

L:~

Isoll1<'1'111 H<'sults amI ('olwsivc Ellngy ('akulalioll

,\"1

Discussioll.

,U'i

(\mdusioll.

Bibliography

Hydrogen Adsorption and Phase Transitions in Fullerites
:>.1

Backgroulld .

~)l

!'i. 2

Experilllcllt.s.

~)2

[).:3

X-Ray Diffractioll Pattern amI Phase Trallsformatioll of Fllilcritc.

!) .1

Discussion

100

Bibliography

103

Carbon Supported Catalysts

105

6.1

IO!)

6.2

6.:~

Ni Cat.alyst Particl(' SizE' Control allcl (:akl1latioll
G.I. L

Nil Activated Carbon

107

6.1.2

Ni/FullNitc.

IO~

Hcsl1it.s aud Analysis

10!)

G.2.1

Isotherm l\kasUrellH'lIt

IW)

c ,) ,)
t).~. ~

Isotherm Hesuits

. . .

110

G.2.:~

Fourier Transform Infrared S pect.rol1lf'try

112

COlle! USiOll .

Bibliography

116

Conclusions and Outlook for Further Work

118

7.1

Conclusions .

II~

Furtiwr Work

120

VIII

List of Figures
1.1

Compressed hydrogen densit.y (daja arc ohtailled frolll [I]).

1.2

PEM fuel cell (Ballard InC'.) . . . . . . . . . . . . . . . . . .

I.:~

The cryst.al st.ructur<' of graphite showillg AIL\B stackillg.

1(i

1.4

Proposed Illodel of the tal\gled structure of vitreolls carboll [IOJ.

l.!)

(a) The icosalwdral
III 0

C(iU

llIolecule.

(b) The rugby hall-sllaped

('70

Iec III c.. . . . . . . . .

1.6

The fcc cr~'stal structurc of (\>0'

1.7

Classificatioll of carholl lIa.not.ulH's: (il) ilrtllc1!air, (h) /,ig/,ag, chiralnauotul)('s.

1.~

. ..

20
21

2-1

The tlllroJ\f'd hOllf'ycolllh lattice of il nallotulw (frolll [10]). Whell we
COlllH'd sites () aud

.\, and lJ and HI, il nallolul)(' call 1)(' cOllstructed.

1.9

The IlTPAV(' classificatioll of adsorption isotl 1('1"1 liS.

2.1

The configuratiou in til{' 2" syst.em with ahout 1 gjda.v productioll rate
(frolll [~j) . . . . . . . . . . . . . . . .

2.2

Vacculll t.rap for filtration (from [1 OJ).

2.:3

Physical adsorptioll isoth('J'llls: Langmuir (l\=:H)) and BET (c=:H))
isotllf'rIn . . . . . . .

2.4

Microlllf'ritics ASAP 2010 BET surfacf' aualysis apparat.us.

2.!)

Sdwlllatic drawillg of Sievert's apparatus.

2.G

(~oIllprf'ssibility factor for hydrogell . . . . .

:3.1

High lllaglli f\catioll TEM shot of graphi te uallofi IJf'r sam pIc #2 showi ng

!)I

herrillgbolle structure.

(i0

:3.2

Low magnificatiou TE1\1 shot of graphite llanofilH'r salllpk #!).

3.:3

SEM micrograph of graphite n(lnofihers iu sample #!) ..... .

fiG

:~.4

High resolution illlagf' frol11 ('ud of graphite llallofihcr sl)()\\'illg herrillgbOlle morphology. Iuset at. 10w('1" left shows lattic(' plallcs from boxed
reglOll. . . . . . .

:~.:)

TEM micrograph of carholl llallot.u1ws ill GNF sample #.1.

:tG

Haw data of desorptioll st.eps at roOlll t('llllwratu]'('. . . . .

:3.7

Ca.lculated ~x (roolll tf'lIllwrat.ure) for sarall carholl salllplc alld Ihc

(i7

f'lll ply rf'ador. . . . .
:u~

lksorpt.ioll isot.herllls [or saran carholl at rOOlll 1cmperalun' and liqllid
llitrogell telllperature (77 I~).

:~.9

A set o[ rUllS o[ desorption isot.herm for sarall carholl sampk at liquid
II

itrogen t.f'ml){'rat Ure.. . . .

3.10 A set of ruus of desorptioll isotherm for GNF sampk #·1 at. room
tf'llll){'rat u re ..

:3.11 Desorption isotherllls for GNF sample #:). #7. and #'(1, at roOlll 1('Illperat II rc.
:~.12

Log-log plot o[ 17 and :WO J\ isotherm data sho,,\,illg pressure. \\'hen 11111ltiplf' rllllS were I akell. eITor hars arc showlI. with
only tIl(' top half o[ t.he nror bars drawn [or clarity. Traces frolll saran
carho]) arc also shm",,]) for comparison.

4.1

7:)

A low resolution trallslllissioll f'1f'ctroll micrograph of til(' S\\'NT ll1at.erials, showing tIl(' rope st.ructure.

4.2

High rf'solutioll trallsmissioll electron lllicrographos of the S\\"NT mat.f']"ials. (a) as-pn'pared. showillg cross sectiolls of t.uhes t.owards 10w('1"
ccntrr, and (h) afler sonicatioll in dill)('lhyl fOl"lllalllide. . . . . . . ..

4.:3

X-ray diffract.ioll patt.erll of single-walled materials (a) as-prepar('(1 a])(l
(b) after sonication ill dimethyl forll1all1ide ..

4.4

'(1,0

(:}If'lllical potential of hydrogcll gas at 71 h:.

'(1,1

,1.!)

Isot.herms of compositioll versllS pressure at. ~o

\\ for salllpies of as-

prepared S\\'NT material. t.he S\NNT Illaterial aner sOllicatioll ill clillH't h~'l
forlll a III ide, and a high surface area saran carboll, Adjacellt, pairs of
cUr\'ps (Iaheled "'S\VNT") \\'('1"e sequelltial rUlIs 011 the salll(' salllple.
Also silowll is the curve of tile sarall carbon scaled to lo',,\,er 11/(' r;-dio
by til(' surfac(, area ratio of 28!)/lGOO . . . . . . . . . . .

!).I

Desorpt.ion isotllf'\'II1S of cOll1positioll versus pressure at III"': for two
differellt batches of fullerite salllple # I of (~H() (',0 fullerite Illaterials.
The upper sf'! illcludes a trace for Saran carl)oll. The lo\\,('(' set shows
identical isotherm hehavior as a fUllctioll of adsorptioll/desorpt.ioll cy-

ck llumlwr. . . . . . . . . . . . .
Transmission electron micrograph of fulkrik #1, ('.;u/(',o.

(a) low

IlIagllificatioll bright field, and (h) higher magnification dark field of
inset area showillg reco\l{icllscd fullerite nallocrystais. . . . . . . . . .
!).:~

9()

Phase diagram showillg t.ile various pilases of (~,(), the possil)k phase
t.ra.nsitiolls, aud t.he stacking of the molecules [1 '~l· ...

!),4

X-ray diffractioll pat1<'rtl of (b) P1l1'(' ('I)() UJ9.9+(;{) alld (a) silJlulated
with ('rystallograpilica . . . . . . . . . . .

101

;).!)

X-ray diffraction of all fu\lncllf' samples.

102

G.I

Dark field TE1\l images of Ni catalyst particles sllpport.ed Oil tl](' Darco
activated carh01I. . . . . . . . . . . . . . . . . . . . . . . .

10,

G.2

X-ray diffractioll patterns of Ni/adivated carholl samples.

108

().:~

X-ra.v diffraction pat.t.el'lls of Ni/fllileritesalllpies. (a) fllllerite frolll Aesea, (b) fulkritcfrolll MEl{, and (c) ful1<'rit<, frolll MEH with sOllication
used for catalyst prpparation pro('('ss.

6.4

Dpsorptioll isot.herms of composit.ioll VPrsus pn'ssur<' at iiI"': for fulkritf'
and Ni/fullerit.e samples . . . . . . . . . . . . . . . . . . . . . . . . . ,

6.!)

1m)

110

Desorptioll isot.herms of COlll posi tion versus 1)J'cssurp at. :WOI\ and ,'1.10 I\:
rpspectivel.v, for Ni/fullcrit.e sampks.

11 I

G.G

Adsorpt.ion isot.hf'rlllS of compositioll Vf'rsus presslll'C al :WO!\ alld FiO!\
respectively. for Darco activated carboll. NijDA( '. fllilerilc and Nijfllilerit.c
samples. . . . . . . . . . . . . . . . . .

G.7

. . . . . . . . . .

Fourier transform infrared spe1rulll of Darco ac1.ival('d carboll alld
NijDAC samples.

G.S

112

Fourier 1. ransforlll infrared Spf't.rlll1l of flllleri t.p sam pie and N i j fullcri 1c
SCUllplcs . . . . . . . . . . . . . . . . . . . . . . . . . . . .

III

XII

List of Tables
1.1

(;Pllcral propprt.ies of hydrogell. . . . . . . . . . .

1.2

Physical prop(>rties of ('()(J molecules alld crystals.

l.:J

Physical propprt.ies of (:70 molecules alld crystals.

2.1

Sekctioll of metals a1ld llH'tal compoullds.

·1 !i

2.2

Ahsorbance wavellulllhers of SOllle functional groups alld lIlolecules.

!i4

:~.I

Hallge of catalyst compositiolls and reactallt gasps used to produce
graphit.e lIallofihers. . . . . . . . . . . . . . . . . . .

:3.2

. . . ..

Comparison of surface arf'a as measurf'(j by' BET. desorlwd at.omic
rat.io of II t.o carb01l. ami t.otal lh coverage assuming diameter of solid
molecular 11:2 of 0.:351 nm[ll]. . . . . . . . . . . . . . . . . . . . . . ,

!i. I

Ib st.orage capacities aile! BET surfacr arras of fullcrel1es. .

G.I

Preparation and characterizat.ion of carhon support.ed Ni cat.aly·st particles.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

lOG

G.2

Hydrogcll st.orage capacitif's of carbol1 samples with or without cat.alysts.lll

i.1

SUllllllary of hydrogf'n storage capacities alld sllrface areas of carholl
samples . . . . . . . . . . . . .

I l,s

Chapter 1
1.1

Introduction

Hydrogen

III this sectioll, w(' shall discuss the properties of hydrogclI, it.s cconOIllIC and
techllological importance alld its storage Illethods.

1.1.1

General Aspects

Hydrogen has be('1l proposed as a clean fuel for the fllt.lI1"e, hot.h in t.ransportation
and stationary power applicat.iolls.
The fut.ure use of hydrogell to gelleratc electricit.y, I)('at hOllles and busill('sses,
and fUf'1 vehidp.s ,viII require til<' creatioll of a distribut.ioll infrastructll1"e of gas alld
cost.-effective trallsport. alld storage'. Pres('nt st.orage llIdhods arc t.oo pxpellsiv(' and
willllot llIed thc I><'rforlllance requir<'lllellts of future applications.
Poteutial UseS of hydrogen as a fuel and ell erg.\' carrier include powerillg y('hicles,
running turbines or fuel cells t.o produce ('kdric-it..\', alld cogcllcrating heat and electricity for buildings. Research ill hydrogcn utilizat.ion is focllsed 011 jcchllologies that.
will most diredly facilitat.c tht' progressioll to a hydrogf'll en('!"gy ('C()I\OIUY.
Transport technologies will need t.o 1)(' dewloppd based 011 the produdioll and
storage systems that. come into use as the hydrogell ('l)('rgy ('conomy ('volv('s.

General Properties of Hydrogen
Hydrogell is a colorless, t.ast.eless, ami odorkss gas. It. liqucfies at. at1l10spllCric
preSSll1"e at. 20.27 J\. SOIllC of its physical properties arc listed in Tahle I.! [1].

Qualltit y
atomic w('ight
llwltillg point.
boiling point.
d('lIsity at STP
d(,lIsity of liquid
(lPnsit ..v of solid
spf'cific I}('at at cOllstant pressure
lalent heat of vaporizat.ioll

Valuc
1.00191
-2})9.1·\0(, (I~ 1\)
_.)r:.)
~I..,o('.
~.)~., ,,,

(.)()
.)~ I;)
~.~ I

(J.OK99 kg/1l1: l
IO.I~ kg/111

IO.G kg/lIl:J
I·\.}) .J / g.1\
·14})}) .J / g

Tahle 1.1: Gf'lleral properties of hyd rogen.

Advantages and Disadvantages of Hydrogen as a Fuel

Advantage8
Hydrogen call he produccd from almost allY f'1lf'rgy sourc(', including rf'lI('wahle ('n('rg~'
sourcf'S and most of all the futul'f' primary encrgy sourcf's. Hydrogell can 1)(' produccd
from f'lectricity and can be convntecl into f'!edricity at n'lativf'ly high efTiciencics.
\Vhf'n cOlllhustC'd with air the only products art' wa1cr vapor and oxides of Ilitrogcll (the NO.r can bc controlled and suppresscd by selection of comhustion conditions). Thus hydrogcn is ellvironmcntally compatible. IIydrogcn can 1)(' cOlllbillecl
with gasolinc, ethanol. llwthanoL or lIatural gas. It is proposed to add hydrogcll into
gasoline to boost pf'rformancf' and reduce pollution. Adding just })(I<. hydrogcn can
the gasolilH'/air mixture in all iniemal combustion cllgillc cOllld rcducc nitrogcn oxide
cmissions hy :lO% toW%.
Energy df'llsity based 011 per unit wf'ight for hydrogell is larger t.hall for any other
practical fuel. Gasoline is 14 hJIW /kg, while t.Iw hydrogcn is :l~ !,IIW/kg.
Hydrogf'n is an excellent gelleral purpose fucl. probahly bd1cr than any other
pract.ical fucl. This is partly bccause in air mixt1\l'cs the range of flalllllwhilit.v is vny
hroad (4.1 % to 14(/i'.) [2]. Spills of hydrogen dissipatf' rapidly under conditions of good
v('ntilation alld it is less toxic than other fuels.

Di8advantage8
Hydrogen is gas at room temperature. Compact storage requires high pressures,

liquefactioll, or chelllical combinatioll. The tf'Ill)('ratlll'<' of liqll<'faclioll is ('Xl1"<'111('1\'
low (20.:31\ or -2!):~ CO) and the liquid has relatively low dellsit.\· (0.07 ) g/cm: l . Liquefaction of hydrogell COllSUlllf'S f'llf'rgy. The heal of combustion per unit VOIIlIl1<' al

STP is relatively small dlW to the low Ckllsity of the gas.
I1ydrog<'11 is explosive. III air mixt.ures, t.he rallgc of flalll111ability is V('J"Y broad
(-1 (X, to 7·1%) and t.1l<' lllaxi lllU III flame veloci Iy is qui t.e high. The igll i t.ioll ell<'rg.\· is
v<'ry small Cllld easily g('lwrau,d from a spark of static ('lectricil.Y. This cOllld callse

safety COllcern.
Hydrogell can 1)(' transport.ed by pipelines similar to hose 1\sed t.0 transport nal111'al
gas. Hovv'<'ver, hydrogPll tellds to leak more readil.\· than mosl gas<'s, and call el11briUle
SOllle Ill<'tals used for pipelillE's.

The major drawbacks t.o Ilw use of h.\·drogen are ils cost alld lhe problcl11S of
storage.

However, as hydrocarboll flwls lwn)llw relatively lIIore ex)wnsivf' as lo\\,-

cost rescrw's diminish, alld as til<' at lllosphnic ('0 2 level rises, increasing tl1<' IllCall
t.<'I1I)wrat1ll'f's as predicted by til<' gr<'f'llhouse df(,ct, hydrogen willlwconl<' illcrcasinglY
more attractive as a lIlliv<'rsal (,lwrg.\' carrin and fuel.

1.1.2

Production of Hydrogen

The choic(' of productioll lllethods will vary, depending Oil t.he q1\ant.ity and d<'sired
purity of hydrogen. The future visioll is that it will be cost-effectivel.\' produced frolll
n'nf'wable <'Iwrgy sources.

Steam Reforming of Natural Gas
Hydrogell is now producf'cl prima.rily by this md.hod, \'\'hich Itas beell the mosl
efficif'nt., ecollomical and pract.ical t.echniqup available for sevcral df'cades.

It is a

t.wo-step proc('ss. The natural gas is C'xposed to high 1('lll)H'rature st.('alll to produCf'
hydrogen, carbon monoxide, and carbo II dioxide. Next t.Iw gas is furtlwr refOrlll('d
with steam t.o convert. carbOll mOlloxide t.o produce additional hydrogcll. The yi('ld
of hydrogen by this process is approximately 10% t.o 90% Hydrogell can 1)(' cOllverled

from hydrocarbons ranglllg frolll l1}('thaup to heavy lIapht ila IIsillg steam rdonni1lg
11](,( hod

[:{j.

Electrolysis
For applications rf'quirillg f'xt.n'lllely purc hydrog<'n, f'lectrol)'sis is IIsed to prod lice
hydrogell. It is a rdatively (,X)H'IlSivc process (hat uses ('lectricit.y to dissociate water
illto hydrogcll and oxygell gas. The f'tficif'lIc(' of this process is about (i7('!{,. Iklle\\'ahk
energy sources of f'kctricity such as solar, wind, and hydropov\'er call 1)(' used ill
f'If'ct.rolysis.

Photoelectrolysis
PhotoeiE-ctrochell1ieal (PEe) productioll lIses semiconductor t.echnology ill a 01]('stf'P procf'SS of splittillg water din'ctly upon sunlight illumination. A PE(' systl'm
cOllli>ilH's a photovolt.aic cellillalnial that produces f'lcct.ric current wht'll exposed (0
light) and an electrolyzl'r in a si ngle devicc. Thl' proc<'ss has shown C'xcl'l kn( po1.ellt ia I
for producing low-cost H'IW\vablc hydrogcll.

Biomass Gasification and Pyrolysis
The prod uction of hydrogell call resul t frolll high-kill ])erat.lI H' gasi fyi ng and lo\\,temperature pyrolysis of biomass (feedst.ocks include wood chips and forest alld agricult.llral residues). III pyrolysis, hiomass is hrokell dowll ill to high Iy react i V<' vapors
and a carbonaceous residue, or char. The vapors, w 1H'1l condl'nsed ill t.o p,vrolysis oils,
call Iw steam refonm,d to producf' hydrogell. A t.ypical biomass fecdst.ock proahollt 6!)(f{, oils and 8% char by weight, with the rcmaiuder consisting of \\'at<'l' and
gas. The citar is burned t.o provide the reqllirf'd procf'SS heat for t.he pyrolysis reaction. F'or hydrogf'll production, a fast-pyrolysis reactor is dircctly linked t.o a st.cam
reformer. Such proc('sses have yields from 12% to 1i% hydrogen hy weight of dry
hiomass.

Photobiological Production
l\Iost. phot.obiological S.VSt.CIl1S usc the natural activity of bacteria and gn'('ll algac t.o producE' hydrogen. ('hlorophyl ahsorbs sunlight. alld ell7,YIlH'S usc t.hc cnergy
t.o dissociatc hydrogf'll from wat.er. Pbot.ohiological t.echnology holds great promise
for long-term sustainahle> hydroge>Jl product.ioll.

Elllplo~'ing catalysts and cllgilleered

systems, Hz productioIl efficiency could reach 24%.

1.1.3

Storage Methods

Thf' potf'n tial uses of hydrogcll as a fuel and cllergy carner inc ludcs powcnng
vehiclcs, rUllnillg turbines or ft\f'l cells to producc f,]ect.ricit.y, and cogcllcrat ing hcat.
alld f'1f'd.ricit.y for buildings.

Different applications will require diff(')"cllt typcs of

st.oragc t.f'chllologies. lTt.ilit..y f'kd.ricity gf'llf'rat.ioll [or huildings will havc st.orag<' fixcd
ill onc locat.ion (st.ationary st.orage), thus t. he sizc and wcigh t \\' i II 1)(' less illl portant.
t.hall cnergy dficicllcy and costs of tllf' syst.elll.

Povvcrillg a vehicle. hOWf'Vcr, will

rf'quire hydrogell st.oragc ill all oil-hoard syst.em (mohile st.orage) vvit.h w(,jgilt. and
size simi lar to t.1lf' gasoline t.ank in today's vehicle.

Compressed Gas Storage Tanks
H~·drogen used t.oday ill illdustrial applicat.ions is gC'lInally in

til(' [01"111 o[ COlll-

pressed gas or cr~'ogenic liquid, referrf'd t.o as "physical storagc." Due to its rdat.ivc
simplicit.y, rapid refueling capability, exccllcnt dorlllallcy charact.nist.ics, alld low illfrastruct.urf' impact, oil-board high-pressure h'\'drogcll st.orage provides a possibly hest.
llear-t.el"ln solutioll for the comlllcrcialization o[ [uel-cell-powcr(,c\ lllot.or v('hidcs [4].
Vehick rangc, storage syst.em weight., cost., durabilit.y, alJd compat.ihilit.y o[ COlllpollC'nt.
Illatf'rials are all key issucs.
N f'W lllatE'rials, such as high-strength carboll fi ber COll1 posi te, hav(' perm i tied storage tanks to be fabricated to hold hydrogen at f'xtreIllely high pressures. Sd('ct.ed
plastic liner materials with low hydrogell permeability and IOllg-krm durability give
bett.er tankage performance. Thiokol Propulsion is working wit.h DOE 011 t.he devel-

Oplllf'nt. of conformahle t.anks wit.h a minimullI burst pl'<'sslll'e of up t.o 12,()()O psi U"i].
Safdy and perforlllallCf' st.andards for high-pressure v('hicle st.orag<' wen' applied to
pnSl1J"(' safe operatioll ow'r the tank Iwing F)-20 year design lif<,.

Liquid Hydrogen
COlldpnsing hydrogpn gas into thp more dpnsp liquid form ('nablcs a larg<'r quantity of hydrogen to Iw stored and transported. The density of liquid hydrog<'ll is 70.t-I
g/lij,f'L which is thrf'E' tillles higher than roOlll tplll]>erature cOlllpressed hydrog<'11 at.
:~,L!)

1\1Pa UU)OOpsi), or equivalent t.o the dCllsit~· of cOlllplTssed gas at 110.2 MPa

(t!),9~O

psi). Figure 1.1 shows the dP11sity for compressed h'\'drogcn a1\(1 liquid hy-

drogcn. How(-'vpr, cOllverting hydrogen gas to liquid hydrogcn is cost.b' and requires
a large input of energy. TI}{' boil-off of liquid hydrog(-,Il is all issuc t.hat. t.akps special
con s ideratioJl.

Metal Hydrides
Various purp or a.\loypd metals can combine with hydrogPll. producillg stahle metal
hydrides. Bpcatlsp of their high volumetric de11sity, safety, and t.i\(' ahility t.o deliver
pure hydrog<'lI at a C011st.ant. pressure, metal hydrides hav<, a great. pot.cllt.ial for h.vdrogell storage. TI}{'re arc 1110re than ~O mclals ill f'xist.<'llc(' t.hat. fOrlll 1lH't.al hydrid<>s by
absorbing alld ret.a.ining hydrogpn under cprtaill t'<'1l1pnatllr<' <1nd pressllre cOllditiolls.
Hydrogen will he released by changing the conditioll, stich as heat.illg.
Duf' to t1wir Imv gravimetric densit.y, metal hydrides Illay be of limited tlse III

vehicles. New mdal alloys havp 1lf'PIl developed that offer better hydrogell st.orage
charactnistics thall single llletal hydridcs. Hydrogpn storage capacit.ies of the allo~'s
var~' I)(-'tw<'cll 2.:)

wex) alHI G.2 wt

magllf'si lllll-i rOll-al umi11ul1l-11ickd- ti talliulll have cxhi bited iIll provpd gra\'i l11d ric all d
volullletric energy dpllsities. Efforts are beillg made t.o scale tip production of tllf'SP
allo~'s.

A llPW dass of maleria.ls, kllown as Ilollclassical polyhydride I1ldal complexes
(PlYlCs), has also 1)('ell dt>vploped. Classical Pl\H's arc kllown to have relatively' high

100

500 K

100

Pressure [MPa]
Figure 1.1: Compressed hydrogell dellsi t.y (data are ohtailwc\ frotH [1]).

gravillld.ric dCllsiti<'s, hut t.]wy g<'I1f'rally ulldergo irreversihle dihyIn the nOll classical PMCs, hydrogen honds t.o the llwt,al while still rC'\.aillillg sigllificant hydrogell-to-hydrogen bonding, allowing a complet.e rf'lcase of h.vdrogt'll under
mild conditions and without. high vacuum. This unique activit.y illustrat.es t.he high
pot.elltial of t.his tYJ)f' of PMC as a hyclrogpn st.orage material.

Gas-on-Solids Adsorption
The abilit.y of high-surfacf'-area carbons, when chemically activatccl, t.o ret.aiu hydrogen on their surfaces is well known.

This process is call (Jr/sorpi iO/l (1 A) and

usuall~' takes place wit.h these mat.nials at. relatively high pressures alld low t.elllper-

atures.

Microspheres
Very small glass spheres, with diamet.ers from 2.1 to .100 Illll amI wall t.hickllcsscs of
approximat.f'ly 1 11111, call hold hydrogcn at high prf'SSl1res. At kllllH'rat.l1res of 200°('
t.o itO 0°( ~, the in(Teast'd permeahility of t.ht' glass pf'rmits the spheres 10 1)(' filled hy
hydrog<'n under pressure by iml1wrsioll in high-pressure hydrogcll gas. \\'Il<'11 coolt'd
to ambieut. tCl1l»erat.urf', the glass is no 10llgpr pernlf'ahk and t.he hydrogell is locked
inside the spheres. SUbSf'qUC'llt raising of t.he temperature will release thc hydrogell.
Rf'sparch has dd.enuinE'd t.hat a slllall bed of such microspherC's call contain h.vdrogcll
at about 62 MPa with a hydrogen Illass fraction of 1O(,1r. [6J.

1.2

Fuel Cells

The fuel cell is a nineteenth-century invention.

Howcvpr, it t.ook more than a

hUlJdred years until t.he National Aeronautics alld Space Admiuistratioll (NASA)
demonst.rated SOlll<' of its potential a.pplications as a powcr source for the space flights
in early 1~60s. Today, resf'arch on fuel cell is drivell hy technical economic, and social
forces such as high performancf' charactf'ristics, reliability, durabilit.y, low cost, and
environmenta.l benefits.

1.2.1

Fuel Cell Principles

A fl1el cell is all electrocllf'lllical device t.hat. COlltilll1ously converts t.he cll('l11ical
f'llerg~' of a fl1el directly int.o electrical ('IH'rgy.

It is two to three t.illles 1110]"(' efficiellt

than all interIlal combustion E'ngille. \Vat.er is the only (,lIIission whell Ilydrogell is
used as fllf'l.
A single cell consists of an electrolyt.e that is sandvviclwd lwtw('('l1 two porous
gas diffusioll electrodes, an allode ami a cathode.

As Ilydrogen flows illto the fuel

cell on the anode side, a platinum catalyst facilitates the separatio1l of the h~'drogen
molecules illto electrolls alld prot.ons (hydrogell iOlls). The hydrogcn ions migrat.('
t. h rough til(' electroly1.f' mf:'m bralle aile! again wi th tile hel p of a plat i1111 11\ catalyst,
combine with oxygen alld electrons Oll the cathode side to form water. The electrolls.
which call not pass through t.he insulatillg lIwllIhralw. flow from the allode to til('
cathode through an extf'rnal circuit containing a motor or other electric load, which
COllSIIIlWS the power g('l1f'rated

hy the cell. The voltage fro11\ Olle single cell is ahollt

0.7 volts [7]. \Vhell thc cells an' COlli hi lied into sta.cks, the opcrati ng vol t.ag(' increases
to til(' vallie of e\ectical power required.
The electrochemical reactioll inside a fllel c('11 is:
Oxidation Heaction 011 Anocle:
2Ih--t4J/+ + .~(Redllction Reaction on Cathode:

O 2 + ·tJJ+ + 4(- - - t IhO

Overall Reaction in Fuel ('f'II:

1.2.2

Classification of Fuel Cells

There are several types of classifications of fud cells. based on diffeJ'{'lIt criteria

til: fuel used, opf'ratillg temperaturt" electrolyte, direct and indirect systems, pri1l1ar~'
and regeucrativf' systems, alld the t.'."(>es of eh,ctrolytes.

Polymer Electrolyte Membrane Fuel Cell (PEMFC)
In the PEMFC, the polymer electrolyt.e membrane is a solid, orgaJllc polYlllcr,
usually poly-perfluorosulfollic acid. A t.ypical llWllIbraJlc materiaL such as the Illost

10
PfM (Proton Fxc;h...,9tt
Mefnbr8n9)

rue! now r",," PIa..
ExIIauoI
Wall!lrVapor
(NoPoiMion)

tow Temperalura
~_rocI>omIcBl

" " " - (90'C)

'uoI(~)

Figure 1.2: PEl\1 fuel cell (Ballard Inc.).
prevalent memhrane, Nafion produced hy DuPont, consists of thr(,(, regions:

1. thc Tefloll-like, fiuorocarholl backbolle', hundreds of repeating
-CF'2 - CF2 - CF2 - CF2 - ullits in length;
2. the side chains, -O-CF2-(,F2-0-('}2-Cf~-, which conn<'ct thelllolecular
backbOlle to tllP third region:
:~. the iOIl clusters consisting of sulfonic acid lOllS, S'()l /1+.

The IH'gat.ivf' ions, S'O~, arc permalleut.ly attached to tlw side chaill ami call not Il\ove.
Howevf'r, whell the membranp IWCOIllf'S hydrated hy absorbing water, til(' hydrogcn
ions lWCOlllf' mobik. lOll mOVPlllclIt occurs by prot.ons, honded t.o wat.er Illokcules,
hopping frolll S'()l sit.e to S'Ol sit.e v"ithin t.he lllelllbrane. Becausf' of this Illf'chanislll.
the solid hydrat.ed ekctro\yt<' is all excellent cOl\ductor for hydrogcn iOIlS, or protons.
Thus PEM fuel cells are somdinlPs called prot.oll f'xchallge lllcl1lhralW fuel cells.
The operating temperature of tile polYlller elect.rolyte lll<'lllhralle fuel cells is limited hy tlw range over which wat.er is liquid sillce t.he presell('e of water is ('ssellt.ial
for the conductivit.y of f'lectrolytcs. The prot.ons are carried from th(' anod(' to the
cathode by llleans of hydronium (Il:l()+) ion within the lllelJ1hra.Il(" which is the characteristic of acid Iud all. Operating PEl\I fucl cells at 1<'llllwratun's exceeding 100

is possible under pr('ssurizecl condit.ions, but. sllOrtells t.he life of the ('('II.
The only stable electrocatalysts in acidic electrolyt.es are platillum awl its alloys.
The best way to lower the Pt cat.alyst. )('ve]s is to const.rllct the cat.alyst laYf'r with

11
the highest possible surface area. Fillf' PI, part.icles arc dispersed 011 high-surfcIlIIpllrities oftell present ill tlw Il2 fueL sllch as ('0. lllay adsorb Ollto t.he sllrface
of tlH' Pl catalyst ill the alloTIlliS th(' allode is Vf'ry sellsitive to CO poisolling.

CO tolerallce level for Pl-HIl

cat.alyst is rv 50 ppm.
Vf'ry t.hill memhranes (10-100 11m) are llsed ill t.he fucl cell to Illinilllize olllllic
r('sistanCf' losses alld hellcf' attaiu high effici(,Ilcies aud power clf'llsit.ies.

Alkaline Fuel Cells CAFC)
Alkalille fuel cells are characterized by hydrox)'l (011- ) ions. The alkalilw e]N'trolylf> is :3.1)% potassium hydroxide (hOB). immobilized ill a J'('collstitut.cd asbest.os
matrix. The clectrochelll ical reactions ill A Fe arc:
'2Hz + 40 H- ---+4II 2 0 + 4(Oxiclatioll Heaction 011 Anode:
Hedllctioll Headion 011 CatllOdC':

O 2 + '2f1 2 0 + el( - ---+'20 11-

Overall Heactioll ill Fuel Cell:

'2H2 + Oz---+'2H zO

The operat.ion temperature of APe's is GO°, The advantages of AFCs (1re:
• reliability and high efficiE'Il<'Y (2: 60(Yr, );
• cathode performance is Illuch Iwtter thall for acid fucl cclls;
• materials of construction t.end to be low in cost.
An at,tractive f('aturc of AFes is: uulike ill acid fuel cells. a wide range of elf,ctrocatalysts can be used including nickel. silver. mdal oxides. spinels, as well as Bob Ie
metals, A major problem with APe is that tIlf'

CO 2 tolerallcf' level is vC'ry low.

('0 2 forms solid carbollate depositing ill the porous gas diffusioll e/<>ctrode and ('(11'-

bOllat.es the electrolyte. which call block flow challllcls, restrict reactant. transport to
the active site on the electrode, and reduce the ionic conductivity of the electrolyte.

Phosphoric Acid FUel Cells (PAFC)
TIl<' PAF(' is the only tyPt' of fllrl cell Illal has r<'aclH'd I II<' COIllIIl<'ITia\ization
stage in the twentieth century. Phosphoric acid (ll:)(), ) is all excellent e\ectroh·te
for fuel c('l1s usillg hydrogell produced by stealll-refol'lllillgjshift cOllv(']'sioll of organic
hydrocarbolls alld alcohols (lll{,thanol or et.hanol). Thr electroclwlllical rcactiolls ill
PAF(' arc:
Oxidatioll Hf'act.ioll 011 Allode:

"2lf.r --tll/+ + . ' j ( -

Reductioll Reactioll 011 (~athode:

0:2 + 4//+ + :k---'tflJ)

Overall Reactioll ill Fuel C('II:

211:2 + 02--'t2// 20

Molten-Carbonate FUel Cells (MCFC)
l\lolten carbollate fllPl cells employ all a.lkali lllet.al (Li. I~, Na) carbonate as the
elect.ro\yte. Sillce these sa.lt.s call functioll as elcctrol.vtc onl~· wl1('n ill the liquid phase,
the cell operates at 600 to 700°(" which is above the melting poillts of the respective
carbonaks. In an MeFe', the cathodic reaction COIISl1llH'S oX~'gell alld carboll dioxide
to produce carbonate ions, which are transported through til<' electrolyte to the allode,
where hydrogen reacts with the carbonat<' to produc(' water and carbon dioxide. The

('0 2 is recycled back to the cathodic stream. The eiedroc\wlllical reactions ill 1\1(,F(,
are:
Oxidat.ioll Hmdioll 011 Anocif.:

2112 + 2C():~- --'tU/ 20 + 2( '0:2 + .\(-

Hedl1ctioll H<'action Oil (~athod(':

O 2 + 2( '0:2 + "i( - --'t2( 'od-

Overall Reactioll in Fuel Cell:

2//2 + O 2 + CO 2 --'t2H2 0 + C(h

The advallta.ges of 1\H '1"(' illc\ucle:
• due to the high operating temperature" ))ohle metal c\cctrocatal.vsts aI'<' 1I0t
IH'("('ssary;
• CO 2 is a fllPI rather than a poison;
• the waste heat from UIP fuel (f.1I is higll enough for additional electric ))()Wf'r
gf'llerat.ioll in a bottoming-cycle st.eam or gas turbine awl/or heal for chemical

processes resulting ill high efficiency;
• it is possible to carr." out t1lf' furl-processing of natural gas ill the fuel ('ell.

Solid Oxide Fuel Cells (SOFC)
Solid oxide flwl cell uses a solid, nonporous metal oxide as electrolyt.e. The o]>el'i-lting t<'llljwrature is (lOO to lOOOoe. Therefore, it also has high ff/llj)(/"a/llf'( (fd"(fllt(fY(S,
which include higher (,{licicncy, and the flexibility to mw 11l0H' l.v]>es of fuels alld illexpellsive catalysts as t.llf' reactiolls involvillg breaking of carboll to carboll bonds ill
larger llydrocarhon fuels occlIr much fast.er as the tf'llll)('rall!J"(' i:-; incr<'Clsed. However,
high tf'lllperature also enhances breakdown of cdl COl1lpOIH'nt.s.

Direct Methanol Fuel Cells (DMFC)
Direct methallol ftwl cell is similar to PEl\l fuel ccll ill that bot.h IISf' the saul('
type of electrolyte - a po'-"-perfluorosulfollic acid protoll exchange IllClllhralle. The
maill difference is DMF( ~s tlse liquid. Stich as l 1\1 llldhanol ill W(I.t.{'1". At the (\lIode,
t.he I1wthanol is oxidized to

CO 2 alld protons and elf'drolls. The clectroc\wllIi('al

("eactions in DMFC are:
Oxidation Heactioll 011 Anode:

H 2 0 + CI{lJIl--+C0 2 + (il/+ + G,-

Reduction Reactioll 011 (~at.hod(':

102 + (iH- + (i,- --+:llhO

OVf'rall Heaction in Fud Cell:

CH30H + H 2 0 + 102--+:Hf.20 + CO 2

Tlw tecllllical a1l(1 economical advantages of DMF(' over tlw PEl\IF(' arc
• it eliminat.es tI](' need for a fuel processor to produce hydrogen on-board. thus
l'f'c\UCf'S

the wf'ight. volullIE', cOlllplexity and t.llC cost of the SYS/.CIll;

• methanol is a liquid with a high encrgy df'llsity and IS cOllsidcrcarryon board t.he vehicle thall hydrogf'n;
• by usillg abollt 1 M methanol directly ill t.he fud cell, t.he problel1ls of wat.er
and thermal management are gn'atly llliniIllized.

11
The cballenges ill developillg DM F( 's arc:
• finding electrocatal.\'sts to cnhallce the kinetics of electro-oxidation of llwfhallol
at the anode by '" :3 or• llJinimizing crosSOVf'r of nwthanol and hellce the loss of faradic efficicllcy:
• red !lei ng tile performallce degradat ion caused hy poi SOli i ng of the elert. roratalyst
by adsorbed reaction intf'rtllecliates. espccially ('0.

1.2.3

Fuel Choice for Fuel Cell Vehicles

Eight major automakers haw' set t.he tinlf'frame to commercialize fllf'l cf'll vehiclf's
to 2004 - 2005. Everyon(' faces the key issuf' of how to provide hydrogcll to tbe fuel
Cf'lIs. Tlw DOE Hydrogen Technical Advisory Panel gave all overview 011 this issuc
ill May 1999 [8J. Tlwre are two fuel cboice options: eit.her produce 111(' hydrogell
OIl

the ground and then stor<' it on board the vehicle (the clirf'c\ hydroW'1l option):

or produce the hydrogen on the vehicle by means of a mini onboard hydrogell plant
(t,hc ollboard fud processor option). Each optioll has Illultiple feedstock choices to
produc(, 1.11f' hydrogen. Onhoard fupl processors call produC(' h~·drogen from llafllral
gas, lllethanol, ethanol, gasolillc, or diesel storcd Oil hoard tbe vchiclc. Grolllld-hase
hydrogcn processors can also use hydrogen from all t.hese feedstocks, or froll! the
f'lf'ctrolysis of water with electricity from many potent.ial rf'll('wablr ellf'rgy sources.
Considering tIw cfficiency. tcchnical difficulty. cost, and f'xisting ('nergy supply
infrastructUlTs, prf'fclTed ft'f'dstock choices for eacb [uri cboice optioll have lllost.iv
narrowed to gasoline or lllf't.hanol for fUf'1 Cf'1I vehicles with Oil hoard fucl processors.
and nat.ural gas for direct hydrogen. At. t.his poillt, llH't,hc\llol appears to havc all
advant.age over gasoline for onboard fuel processors due t.o better Vf'hici<' ]wrformallC('
and If'sS technical difficult.y, despit.e 1ll0rf' costly challgf'S llef'ded to lllakf' methanol
widcly available at flwling stat.iolls. For direct hydrogen vphicles, thc hydrogen call
be st.ored onboard either as a compressed gas at. about 5,000 psi, or as a cryogcuic
liquid.

If)
Eacll option has ach'antages and disadvantages associelted with vehicle dfici('l)('.V,
t.echnical difficulty, cost., fuel infrastructure requirements, saff'ty, and long-term societ.al benefit.s. PerformallCf', t.echnical difficult.v, alJd societal IH'l1efit.s favor direct
hydrogen. Costs on a per vehicle basis are comparable for both opt.ions. But duc to
concems over fud infrastructure requin-'llwllt.s amI, to a lesser ('xt.<'llt safpt.~·, indust.ry
is st.rongly favoring the 011 board processillg opt.ion.

1.3

Carbon Materials

Structural and dWlllical versat.ilit.y t.ogetiwr wit.h a low alOlllic weight make carbollbased materials illt.erest.ing and pot.ent.ially useful as hydrogen st.orage 1lledia. The
carbon atom has an at.omic nllllll)pr of () and is in a 18'22$'22]1'2 electronic ground st.at.e
configurat.ion. The at.omic mass of car bOil at.olll is 12.01 11 f) ellllll. TIl(' at.olllic radills
of carbon is O.ii A, which is measured as half the C'quilihriulll distancc betweell two
carbon atoms of t.he planar graphite st.ructurf'. CarhOll has two major types of b01lding, tf't.ralwdral 8Jl:1 bonding and trigonal 8p2, ill c1ialllOl)(\ and graphitf', J"('sp('cti\'('I~·.
III t.he 8})2 hyhridization state, one' delocalized lloll-hybridized 7) elect-roll is perpelldicular to the plane of the other tlnee hybrid Sp2 orbitals and hecomes available 10
form tIl(' subsidiary 7l" bOllc1. The existeJlce of 7l"-flrctmll bonding is respollsible for
carbon's versatile talents.

Those materials having extended 7l"-df-'ctroll clouds arf'

callf-'d 7l"-fifrlro/l II/(/ifrials, or sometimes as ."i1l1)(rcnrboll [9], \wcause of their fahu101ls llIultiformity and vf-'rsatile propert.i('s. They include graphite, carbon nallotu\wS,
fullercl1f's and various carbonaceous materials.

1.3.1

Graphite and Graphite Materials

lTllder ambiellt cOlldit.ions allcl ill bulk form, graphite is tile stable' phase of carbol1.
Tlw crysta.i slruct.me of graphite consists of s('ries of stacked parallf-'l layers with 8])2
bondillg, as shOVlin in Figure 1.:3. \Vithin each layer plane known as a graphf II( plal/f,
t.he Ilf-'ar('st.-neighbor distance is 1.42J A. The inierplallaI" disiallCf' is relat.ivply larg(',
which is :3.3f) A. This gives an ill-plane lattice cOllslant a of 2.46:L A alld a c-axis

IG
lattice constant (' of G.iO~ A. TIl(' structure IS consistcnt with the J);:" (fJ(h/III1I1C)
space grou p [11].

I~e.--Ir

.I
L.O-~l!11

1;0.--.

Figurc l.:~: The crystal stru\1\ll'f' of graphitc showing ABA 13 stacking.
Since the in-plane C-C bonding is very st.rong compared t.o the weak illt.f'rpla.Hal' bonding, impurity species tcnd to OCCIlP." interst.itial positioll l)('t.\\'e('11 the la.\'('r
platH'S. The w('ak illt.f'rplanar bOllding also allows ent.irc platH'S of dopant atotlls or
molecules to 1)(' int.ercalat.cd lwtv\'eCll t.he carbon layers, forming graphit.c int('('calat ion
lOlllpolIlJds ((;]('s) [12J,
Faults ill the A13AB stacking give rise to a slllall increase in the interiayer distancc,
This contillues ulltil all inj,n-Iay<'r spacing of about :~A40 A is reached, wh('('(' the
stacking of the individual carboll layers l)('conH's ullcorrelatcd, This st.ruct.u('(' with
no thn'e-Othn t.han natural single-crystal graphite flakes, tiwre arc !llan," sYllt.iH't.ic graphit.c
materials, such as pyrolyt.ic graphit.e, carbon fibns, vit.reolls carhoJ\, a])(1 carholl black,
which are actually aggregat.f's of graphite crysta.llites, or polycrystalliue graphites.
Gel\erally, graphite is black, opaque, and met.allic ill lust.f'l'. It has a df'llsity of

2.09 to 2.2.<1/('/1/:1.

Graphite Whisker is a graphit.ic lllat.erial fonncd by rolling a graplwllc sll<'d int.o a
scroll. Graphite whiskers are formed ill a de discharge bet.weell ('a1'bo11 electrodes
using i!)-~O \' aucl 70-iG A [IOJ. The whiskcrs arf' t.o be found up t.o :~ 011 IOllg

17
and 1-:') pm in diameter. The growt.h of graphite witisk('rs has luall.\' siluilariti('s
to tlw growth of carbon nallotu ]ws.
Vitreous carbon or glassy carboll. is a COli III lOll carholJ Ilia/erial pro("('s.sf'd by slow.
cOlltrolled degradat.ion of certain organic precursor al t('lllpcratllres typicall.\'
from GOOO(' to 1000°(' [1:3). The selections of precursors illcludes ]>ol.vfurfur.vl alcohol. phenolics. polymicle. polyacr.\'lonitrile. cellulose. amI polyvillylidell<' c1doride (PVDC), etc [10]. TIl(' microst.ructuH' of vit.reous carbon consist.s of all
extensive and tangled network of graphit.e-like ribbolls or lllicrofihriis (as shown
ill Figure 1.4), about 100 A long alld :30 A in cross sect.ion. alld r<'s('llIbles 1,11('
precursor's chain confignrat.ioll.

Figure 1.4: Proposed model of tlw tangled st.ructure of vitreous carholl [10].
Saran carbon falls int.o this catagory. It is t.he nallW applied t.o carbonaceous
products derived from the pyrolysis of PVJ)(' or Saran resins. T1H' st.ructnrf'
of this material is an assembly of randomly oriented grapititf' microcr.\'stals in
an amorphous carbon matrix. It. contains very small pores. which giv(' t.hem
molecular siev(' propertif's. Both porosity and porf' size call IH' sckcted to SOIllf'
('xt.ent. hy varying til(' t.echniques used to make t.he ma/erial. Tlwr(' are t.wo
basic techniques described ill t.he literature for carbonizing PVD(' t.lwrlllally
[14J. One is a slow solid-state degradatioll \vhich gives more control over tJl('
surface area.s and the pore size of the resultant carboll. It prodllces carboll with
a microporf' structurf' and sUlwrior a.dsorptive propertif's. The other mdhod
involves heatiug at a fairly high rate sud) that the PVD(' melts beforc an~'

sigllificant df'gradation occurs. In the rapid h('atillg IllCthod. tIl(' 111<'11 f'oaills
during til<' (khydrochlorillat iOIl stq). crosslillks alld hardells yieldillg Cl carboll
rich ill lllacropores.

Pyrolytic Graphite or highly orif.lItu/ jJyJ'o/ytir graphitr (HOpe;) is the 1lI0st COIIImOllly used high-quality graphit.ic l1lateriaL It is prepared by the pyrolysis 0('
h.vdrocarholls at tnnperatUlTs above 2000° amI subseqtwutl.v heat tl'<'aled to
higher t.elll]>eraturf' [l!)].

\\'hell colull1nar or laminar pyrolytic graphites are

annea.led abovf' 2700 0 ( " usually ullc1f'l' a prf'SSlll'f' of several atlllosphcres, furtl1{'r ordering amI stress relievillg or the structure' occur with ill each plallc amI
]wtvveell planes [10].

Carbon Fibers consist of a stack of turhostatic layers with little graphitic charactn
alld may in SOIllf' cases illcluck $pl bouds. Examplf's arc polyacrylonitrile (PAN)
alld Hayoll. The proc('ss illvolvcs st'veral steps: st.ahilizatioll, carbollizatioll and
high tf'11llwrature heat trf'atnwnt. Act.ivCl1.t'd carholl fibers an' highly porous
carbon fihers with specific surrac(' arc as of 700 t.o :H)OO 1112/g and pore VOllllllCS
of 0.:3!) to 0.80 cm1jg or 1ll0rf' [Hi]. Fil)('r diallwtt'rS are t.vpically 10 lilli, whilc'
tllf' structure consists almost. ('ntirely or l1licropores wit.h haIr widths 011 til<'
ordcr of I Il1l!.

Carbon Black represents lllallY types of fillely-divided carbo1\ part.icles that an'
produced by hydrocarbon dehydrogenation. The microcrystalline structures of
t.ilf' carbon blacks are found to be small graphit.e-like layers. In each individual
part.icle, t1lf'Sf' graplWIH' layers are couc(,lltric orgauized.

Porous Carbons arf' a lltll111)('1' of carb01l Illaterials with V<'1'." high surface' areas
and P01'('S of nanollletn diulf'nsiolls similar to tlH' dimcllsions of fullf'n'IJ('s, i1lc1uding activakd carbons, exfoliated graphit.e, and car bOil aerogds. Exfoliaj,ed
graphite is prepared by hf'ating a graphite intercalat.ion compound above SOlll('
critical temperat.ur(', t.hus a gigant.ic irreversible ('-axis expansion occurs, with
sample elongat.ions of as llluch as a factor of' :300 [17]. This elongation is called

exfoliatioll. which gIves nsf' to Sp01\g)'. rOi1.Jll~'. low dellsit~,. high surface area
llli1tcrial. (:arholl aerogels are a disordered forlll of '~Jl2-boll(jpd (,(11,1)011 wit h
an especially low bulk density and arc made by i1 supercoolillg process. They
consist of ill!.erCol1l1cded car bOil particlf's with diallH'jers typically Ill'ar J 2 11111.

1.3.2

Activated Carbon

Due to tIwir porous structure. activat.ed carboll has Iwell lIsed ill plII'ilication and
gas adsorption for centuries. ActiV(' carbons have I)('ell comlllercially produced frolll
a tremendous vari('tv of carbonaceous starting 11li1\erials. i nc1ud i ng co('011l11 shell s.
I)pach pits. sawdust wood char. fish. lignin. petroleum cokf'. bOll(, char. anthracite
coal. coffee grounds. molasses. rice hulls. carbon hlack. peat. kelp. and sugar [18).
Aftcr carbo1\izing the source material under i1ppropriate conditions. the result.ing char
is often subjected to oxidation in a controlled E'IlVirOlllllclI1. Carhollizatioll is usually
conducted in t.he ahsf'nce of air at telllperatut'(' of GOOO(' to 900 0 ( ' . III activatioll h~'
oxidat.ion. tIlt' oxidant selectively erodes tlw surface so as to illcrease the surface area.
develop greater porosity. and Jpi1Vf' til(' remaining atoms arranged iII configuratioBs
that hi1ve specific affinities [19). TIlf'sf' characteristics cOIltribute to the remarkahle
adsorption propert.ies of cativated carhons. Activated carho\1s hav(' surface areas tha1
range from 400 to o\'('r 2000 111 1/ g.

1.3.3

Fullerelles

Fullerf'IWS were first discovered iII 19~!) by 1\ roto aud Smalley [:W J by tilllf'-offlight lllass spect.roscopy in the hot carholl plasma gellf'ralc,d during lascr ahlatioll of
graphite. Fullf'rt'llf's are a major extpllsiOll of tlw scope of carho11 Illolecules knowll to
pxist, and opell all entirely llew chapter 01\ the physics and dWlIIistry of carbol1. TIlt'
annOllllcelllellt by I\ratsduller alld Huffman [21. 22J ill 1990 of a sYIJt.1Jesis llldhod
capable of producing gram Cjuant.ities of ful1ereu('s makes it possible for hundreds of
r<'seardlf'l's t.o study these fascinating new materials.

20
Structures of Fullerelles

III (~HO. tlw sixty carbon atoms are located at tlIP vf'rtices of a trullcated icosalwc!ron where all carhol1 sites are e((uiva.leut.. This is consisteut with the observat iOIl
of a singk sharp line in the nuclf'ar maguet.ic l'f-'sonanc<' (l\:MH) spectrUlll [:t~l. As
defilled hy IUPAC, fullerenes are poi)rlwdral closed cagf'S made up ('utirely of n tllrf'(,coordinate carbon atoms anc! containillg onl~' hexagonal alld lH'lltagollal faces. Therf'
must be exactly 12 pentagonal fac(,s alld all arhitra.ry Ilullllwr II of 1}('xagollal fan's.
which follows Euler's t.iwor<'l1l for polyhedra

f +1' = C +:2
where f.

I'.

( 1.1 )

ane! f a1'f' tIw llumhers of facf's. verticf's. alld edge'S of th(' polyhedra.

rf'Slwctively.

Figure 1.0: (a) The icosahedral ('flO molecule. (b) The rugby ball-shap('d C,o molecll Ie.
TIlt' smallest possible fulkrenc is (':20 with 12 pentagonal fau's aile! no hexagollal

faces. However, it. is t.lwrlllodynamically unstable sillc(' it is ('I}('rgeticaJly unfavorable
for two pentagons to be adjacent. to each other, owing t.o high local curvature of t.!)('
fullerene halL anc! hellce morf' strain energy. The resultillg tendf'llcy for pelltagolls
not t.o be adjacent to one another is called the isolated pelltagon ruk [2/1, :20J. The
smallest fullen'llf' to satisfy the isolated pcntagon rule is (\;0, as shown in Figure 1.0
(a). The next possible molecules are C/O with a rugby ball shape. a.s shown ill Figure
1.5 (b). Since the addition of a single hexagoll adds two carbOll atoms. all fullerelles
must have an even nU11l1wr of carbon atOll\s.

21
The' sLnldurf' of ('70 call 1)f' envisioned as adding a ring of 10 carbon atolllS 10

t.Iw ( :!iO lllokcule. III cOlltrast. to C tlU , wllich has only one' ulliqll(' car bOll site and t.wo
ullique bond lengths, the em Illolecule has fiw ine'qllivalcut sit.es alld eight dist inct
hOlld lengths.

Figure 1.6: The fcc crystal structure of eGo.

In t.he solid st.at.e, t.lw CliO st.ructure is facf'-cf'nt.erf'c1-cubic (fcc), as shown in Figure
loG. with a lattice cOllst.ant of 14.17 A. The cryst.al st.rllct.urf' of C,o is Illurh more
complf'x t.han t.hat. of

e60 . TIl(' gelleral agrf'f'lllent. is t.hat. t.here are t.hree regillles

wllf're t.he structure is isot.ropic above trallsit.ion t.clllperat.urf' TUI alld l)(,COlllf'S lllore
anisotropic as T is decreased. This will be discussed ill detail ill (:hapt.er S.
SOIlW physical propertif's of CiU and (',0 Illolf'cul('s and cryst.als are sUIllI1lf'rizf'cl
ill Table 1.2 and 1.:3.

Synthesis, Extraction, and Purification of Fullerenes

III 1990, l\ra.tschnlf'r f't. al. discovered a methocl t.o sYllthesize' fulleJ'('lIes in macroscopic quantit.if's. There arf' many more procf'SSf'S which have bf'f'll report.ed, illcludmg:

Qualltit~,

avcragf' ('-C' distance
('-( ~ bond Iellgth Oll a pcntagon
CliO lIlean ball dianwtcr
('tiU ball outer diallwter
binding energy Iwr atom
heat of formation (P<'r g (' atolll)
colwsivp pncrgy pel' (' atom
fcc lattice const.ant
C60 -C m distance
(~liU-( 'liO cohesi vc energy
jetralwdral interstitial sit,e radius
octa.lwdral intf'rstitial site radius
lllass dellsity
t.ransition t.emp
Jllelting temperature
su blimation tern perattup
Iwat of sublimation

Value
1.'1:1 A
1.46 A
1.1 A
10.:~4 A
1.4 cV
]O.H) kcal
I A (' \' / alolll
14.11 A
10.02 A
I.G eV /molecuk
1.12 A

Hefcrellcc
[26]
[27]

2.DI A
1.72
2G 1 I":
] I~Do('
4:34°('
'10.1 Kcal/mol

[2~]

[29]
[:30]
[:~ I]
[:~2]

[:n]
[:34]

[:n]
[:~:~]

[:n]
[:Fi]
[:~G]

[:{7]
[:3'1]

Table 1.2: Physical properties of e lilJ molecules and crystals.

Quantity
average C-C ~ distance
(~70('-axis dial1wjer
C,O (( - b axis diameter
lIUlllher of distinct (' sit.es
number of distinct (' silt's
heat of format.ion (per g C' atom)
fcc lattice constant
C 70 -('70 distance
he\> latticC' constant
('70-C,O cohesi ve ellerg.Y
Trallsitioll tcmperaturt' T 01 • 1'02
sublimat.ion temperatu!'('
Iwat of su hI i mation

Value
1.1:3 A

RefeJ'f'ncp

7.96 A
7.12 A

[28]

[2!)]
[29]

9.G!) kcal

1.1).01 A
10.G] A
a=1>=10 ..I)6A, c=17.18A
1.6 f'V / llloieCll Ie

[26]

[:W]
[:l2]

:{G1 I":. 282 I":

[:n]
[:~2. :n]
[:{4]
[:~!) ]

4G6°('
,n.D I":cal/mol

[:{I]

[:~I]

Table 1.:3: Physical properties of C/o molecules and crystals.

:2:3
• ac or dc plasma discharge ]Wj,WC'PIl carboll electrodes ill lIe gas
• inductive heating of carbon rods ill vaCHum
• laser ablatiol] of carboll electrodes ill He gas
• sputtering or e\('ctrol1 bealll C'vaporat.iol1 of graphite in all illert atlllospilCre
• oxidative combustion of 1)(,117,(,llf'/argol1 gas mixtures
Among thest' metbods, tlw carboll arcillg process and tlw flaIl\(' process arc the
ollly processes vyhich have lw('n cOI1lIlH'rcialized due t.o silllplicit~" low start lip illvestlllCllt, )'f'latively low operat.iolJal

complexity ami readily available raw materials.

III til(-' process of producing fullerelJ<'s, most lIlCtlJOds sillllzltallcously W'lH'rate a
mixture of stable full('rel1f's (( 'm), C T(), ••• ). impurity molecules such as polyaWllllJlatic
hydrocarbons, and carboll-rich soot. Therefore, the syutiwsis of fullt')'('J]('S must 1)('
followed by procedllH's to extract aue! separat.e fu]]erelles from impurities according
to Blass.
The r~ratschmer-HlIffl1laIl1ll('thod is t.he most efficient way t.o produCf' l(lrge quantities of fullerel1es (Iud is thus widely ill usc. The process is quite st.raightforward.
Graphite rods (Ire evaporated by At' or D(, arcilJg in a quellching atmosphere of ill('rj
gas (typically 100-:200 tOIT of heliulll). The vapori7,('d carboll cOlJdenses and deposits
on the reactor walls as a light, fluffy condeJJsatc. or soot. A substant.ial fraction of
this material is composed of fullnt'nes. The fullc)"('IIf' soot is then collected alld t.lw
fllllert'lH's are extract.f'd, eitiwl' by sublimation from tllP soot via heating ill a vaCUlllll
or in all inert gas, or more cOllltllonly by solvcnt extraction. Sf'v(']'al parallJ('j<'rs a.]'c
known to aJfect t.llf' conversion of the electrodf's to fullerf'lIcs. High-purity 1)('lilllll
is usually used instead of other inert gas such as argon and nitrogen.

Most soo/-

gellcratiJlg chamhns usc a quenching gas pressure ill th(' 100-200 torr range. It is
r('ported tlw optimum pressure is highly' sensitive to design specifications.
Two dist.inct methods arc employed to extract t/j(' fulleH'lJ('s frolll /he soot produced. The solvf'nt l1lf'thod uses toluene to dissolve the fullf'renes, vvhile the soot alld

otiter insolubles are easily sf'paratcd from til<' solut.ioll h.v fill ralioll. Soxhkt exl raction technique comhined witit chromat.ographic llH'titods produces fulkrclH's of high
plII'ity. III t.he sublimation lIlel.hod. t.ite raw soot is l]('(l(ed ill n qllarl7, tllbe jJi lIe g(lS.
or in V(lCUUI1l. to sublillH' thc fu]]ercll<'s. which tll<'1l cOlld('IIS(' ill a cool('l" section or
tite tube. TIl(' sublimation t.em]H'ratures of crystallinc ('HO alld (',0 are re]al ivcly low
(Ts.C'c.o =:3.1)0°(', 1~,c7o =460°(').

1.3.4

Nanotubes

Discovered by Iijillla of the NEC Lahoratories in Japan in 19~) I [:~Sl. carholl llallotubes are usually considered as tubular fulkrPlws.

('arhon llano!ulH's W('J'(' firs!

synthesized as a by-product in arc-discitarge nwtllOd ill s.vlltI]('sis or fullcrcllcs.

('111'-

rently, many different lTwthods have been developed to pn'pare nanotul)('s. includillg arc-discharge. laser ablation and cntalytic decolllPosition of hydrocarbon. ('arbon naJlotuJws call J)(' classified int.o two types: Mlllti- \\'alled ('arbon NallotllJws
(l\1WNT) and Single-Walled Nallolul)('s (SWNTs). TIl(' l\IWNT is COllJpos('d of :2
to :30 concent.ric graphitic layers. diameters ranging from 10 to .1)0 lIlll and Iengt.h or
more than 10 /1m. SWNT is much thillner with the diameters frolll 1.0 t.o 1.8 11Ill.

Figure 1. i: Classification of carbon nanotu bes: (a) armchai r. (b) zIgzag. and (c)
chiral nanotubes.

Structures of Nanotubes
The important. alld essential fact about tile structure of a carboll Ilallotlll)(' is 11)('
orieIltatioIl of t.hE' carhon \wxagollal riug ill tlw honeycomb lattice relative to th(' axis
of the llanotulH'.

Three examples of single-walled carbOll nauotulws (SvVNT) arc

ShQ\l\,1l ill Figure 1.7. The 'zig-zag' alld 'armchair' lJalJotubcs an' t"vo possible high
sYlllllwtry structures for nallotuhes. The armchair tubules are obtailled by bisf'ctillg
a (',m lllolecuk at tlw equator normal to a fivefold axis, and joillillg t.he (."vo resultillg
hemisplwres wit.h a cylindric tube Olle mOllolayer t.hick, as shown in Figure 1.7 (a).

If t.he ('til) molecule is bis('ctpd normal to a t.hreefold axis, the zigzag tuhule showll in
Figurf' 1.7 (a) is fOrIlwd. HoweVf'r, it is believed t.hat 11I0St. nallot.ulws do lIot have
these highly sYlllmetric forllls hut, have structures ill whicll the hexagolls are arranged
helically arouud tlw the tube axis, as ill Figure 1.7 (c). These structures are gennally
know as chiral. sillce the~' can exist. ill two mirror-relatf'd forllls.

Figure 1.8: The unrolled honcycomb lattice of a Ilanot.ulw (frolll [40]). \VIWll we
COlllWCt sit.es 0 and A, and B alJ(1 BI, a llallo/ubc call 1)(' COllst ruct('(1. Here, e h =
(4,2).

An indexing method was developed by Hamada d al. [41]. TIH' llanot.ulw cylinder
is producf'd by rolling up a shE'et of graphelle lattief'. The structure of all individual

Ilgrapll('lle laUicC'. C ll is called chiral vector, alld call be expressed as

( l.:n
Il,l1l arf' int.egers, 0::; 11111 ::;11.

All armchair nanotulH' corresponds t.o the case of 11 = 11/, amI a zigzag lIanotul)('
correspollds t.o the case of 171 = O. All othf'r (/I, 7lI) cbiral vectors corr('s]>01)(1 to cbiraJ
nanot.uhes. Because of t.he hexagonal sYlllmetry of the hOllcycomb lat.ticT, we 1]('<'<\
to consider only 0::;11111::;17.
The diameter of tIlt' carbon nanot.ube, elf, is given by IChl/IT, as

(l.:q

The c!liral angle () is defilled as tJ\f' angle between 11](' vectors C ll and aI, as shown
III Figure 1.8.

Due t.o the llf'xagonal SYllllllet. ry of t IIf' hOlleycollI blat! icc. the val lies

of () is in the range 0::; If) I::;:30. () is gi vell hy

211 + 111
cos () = --;:======
2V1l2 + 1/1'2 + 11171

( 1..1)

Synthesis of Nanotubes
There are currently fOllr main llIethods used to sYIlt.hesize car bOil llallottll>f's:
car bOil arc synthesis, cllE'lllical vapor depo,sitiolL ion bOllJbardmC'nt.. alld laser vaporisation. Among t.hese, the lllOSt. successful has been the laser vaporisation ll]('thod for
the sYllt.ilesis of single-walkd nanotuhes which was introduced hy Smalley's group in
1996 [42].

Production of m.ulti-walled nanotubcs (MWNT)
The original lllethod used hy lijillla t.o prepare llallotulws differcd slightl~· from til<'
l~ratschlllf'r-Huffl1lall U-'cilllique for ('tiU productioll in t.hat t.he graphit.e electrodes

w('r<' l]('ld a short distallc<' apart duriug arcillg. rathf'f tlialJ heillg kq>t. ill COllt.l-lCt. III
this d('ctrie arc dischargf' lllct.hod. the quant.it)' of l\IWNT ohtained depends 011 t.he
PJ'('ssu]'(' of Hf' at.mosphere ill the H'actor. which is til(' most illlportallt pCl1'all]('t('1'. The
highest quantity of MWNTs is ohtaill('d WIWIl the pressure of He is around ;)()O torr.
while thf' best condition for fullerene production is about 100 torr[4:3]. 'I'll(' l\IWNTs
producpd hy lasPI' ablation arp milch shorter than those hy arc-discharge lIl('thod [.I!)].
('atal.Vlie d('col1lposit.ioll of hydrocarholl is suitable for lIIass production of l\1\VNTs
lwcausf' tIl(' M\VNTs synthesized hy t.his method do not cont.ain Ilanopart.ick or
amorphous carhon [4:3].

Production of 8ingie-'loalled nanotube8 (SWNT)
\VIH'reas lIlulti-walled nanotuhf's require no catalyst for their growt.h. cat.alysts arc
necessary for the growth of the single-v,alled nallotulws (S\VNTs). Clnd lllore than
one catalytic species seelll to be ueccssary to gro,,, ropes of S\VNTs.
In the early reports of tll(' las('r synthesis t('chllique. high yields vvith ahout iO t/{,
-90% conversion of graphite to S\VNT were reported in the condensing Vl-lpor of t.ite
heated flow tulw. operating at 1:WO°C. A Co-Nijgraphite composite laser vaporisation
target ,vas used. consisting of 1.2 atom% (:o-Ni alloy with equal l-lIllOunts of Co and
Ni added t.o tltf' graphite (98.8 atolll/(', )[t2]. Two sequellced las('r pulses weI'(' lIsed
to evaporate a target. containing car bOll mixed wit.h a small alllollnt of transitioll
mdal from the t.arget. Flowing argon gas swef'ps the entrained nallotulH's frolll the
high temperature zone to the wat.er-cooled ('11 collector downstream. just outside the
furnace.

rrhe lllatprial thus l)rociu('ed app('ars III a scall1ling pipctroll l11icroscopp illlage as
a lllat of '"ropes" 10-20 lIlll ill dianwtf'l'. and up to 100 Ifill or more in kllgth. By
transmissioll electron microscopy (T El'vI ). each rope is fOll II d t.o consist pri Illari ly of a
bundle of S\\,NT aligned along a common axis. The diameters of the SWNTs have
a strongly Iwaked dist.rihut.ion at 1.:38±0.02 11m. very dose to t.he diameter of all
ideal (10,10) nanotube. The SWNTs are h('ld (ogdlwr by weak van df'r vVaals illternanotube bonds to form a two-dimensional triangular lattice with a latt.ice COil stant

28
of 1.( 11m, and all inter-tuhc separat.ioll of O.:Jj!) lllll Cit closcs1 Clpproacll \\·i1 hill ;-1
rope. By varying the growth t.emj)f'\"at.nrp, the cat.alyst. cOlllpositioll and o1.II<'r growth
paraIllct.ers, tlw average llanotulH' ciiallwt.<'r and diallwter distributioll CClII 1)(' \'(1ri('d

[44J.
Otlwr Illethods of nanotube synthesis include carbon arc elcctrode, vapor grovvth,
and carbon ion bombardment.
Pur~f£c(J,f'ion of nanotube8

In most. synthesis procedures, carboll nanotubf's are produced along with otl}('r lllat.<'rials, such as amorphous carbon and carbon llanoparticles. Therefore, purificat.ion is
necessar~' to isolate the carhon 11anotul)('s from other entit.ies.

Threp basic Ilwthods

have be('n used: gas phase, liquid phase, alld intercalation methods.

1.3.5

Graphite Nanofibers

Carbon nanofibers (aJso known as carbon filamcnt.s) call b(' grown from thc cat.alytic decomposition of certain hydrocarbolls over small metal particks, such as iron,
cobalt., nickeL and SOJllC of their alloys. The diallwter of thc nanostructur(' is COlltrolled by the size of the catalyst partick and can vary Iwt.weell 2 alld 100 nlll, and
lengths ranging from .f) to 100 I'lll.
A model of the mechanism of carbon nanofilwr formation has 1)('('11 proposed
by Baker and coworkers using controlled atmosplwr(' dectroll III icroscopy [1GJ.

III

their proposed mcchanism, the key st.eps are the adsorption and decomposition of a
hydrocarbon Oll a metal surface to produce carbon species which dissolve alld diffuse
through the bulk and ultimately precipitate at the rear of t.he particle t.o form tll('
nanofilwr. As a result of this process the cat.alyst. particle is df't.adwd from its original
support and remains at the growing end of the carbon structure.

11 has been suggested t.hat there is subtle relationship between the cryst.alline quality of the gra.piwne layers in t.he deposited carbon nanofilwrs aIle! the ability of tlw
Illetal cat.alyst particle to undergo a strong illteractioll with graphite ill til(' presence
of hydrogen. It is observed that cobalt forms highly graphit.ic carbon lIallostructur('s,

wlwrcas other llwtals ill the sallle group, iroll and nickel, exhihit (\ sOIll<'what weaker
interaction with graphitf' and product' llauofilH'r structures \vith a large fraction of
amorphous carbon. On the ot.her hand, when either copper or iron is lllixecl with
llickf'L there is a significant increase in ill(' wetting characteristics with resp('ct to
graphite and such catalyst particles teud to gClwrate high ly ordered carholl llaIlO.Structures. TI1('1'f'forf', it. is possiblf' to control the lllorphology of the nallofilH'rs hy
11]('1'('1.1'

altering til(' ratio of t.il(' allo.v cOllst.ituC'llts. Particles rich ill Ilickel were found

t.o product' much bf'tter crystallinity ill the filamellt st.ructures thall t.hose growll frol])
eithf'r pure nickel or copper-rich particles [4ei].
In generaL t.he structure of nanofi\wrs consist of an inll('J' core of alllorphous carholl
StllTOllllc\f'd hy a skin of graphitic platelets. It is also fOllnd that the intf'rnal st.ructural
arrangf'lllent call he significantly different from that described above, alld is extl'<'llw\y
sensitivf' to a number of factors, particularly t.lw dwmical nature of the catalyst alld
tlw composition of the gaseous reactant.

1.4

Physisorption and Chemisorption

Gas adsorption 1 on solid surfaces and in pore spaces is acorn plex phenomenon
involving Illass and energy interaction aud phase changes. In gcneraL two types of
adsorpt.ion arf' distinguished, physical adsorpt.ion (abbrf'viated as ph ysisol'ptioll) and
chemical adsorption (abbrf'viated as rlwTllisol'ptirm).
Adsorption at a surface or interface is largely the result of billding forcf's betweell
the individual atoms, ions, or molf'Cldf's of an adsorbate and the surfacE', all of thesf'
forcf's having t.heir origin ill f'lectromagndir interactions.
In physisorption, there is a van (leI' \Vaals intera.ction (sonwtinJes consid('rec\ a.s all
1Tlw krm ar/s01pli01I was introduced by haYi'er [47] ill lKK 1 to COlll1otf' t.hp cOlldelH;ation of ga~ps
011 frpp surfacpi', in cont.radict.ion 1.0 gaseOllS al!sorpl um WhNP thp 1Il0lecuks of ga~ [wnd-rate into
thp maHS of tllf' absorbing solid. Adsorption (strictly, physisorpt.ion) has now Iwen intefllatiomdly
dpfined as t.he enricillllPllt of onf' or 11101'1' compolIPnts in an interfacial layer. Anot.hN terll1 so/plum,
which was proposf'd by M,Bain [4K] ill 19(m to el1liJracp adsorption on thf' surfacp, a[li;orption by
penetratioll illto t.he latticp of the solid, and capillary condensation wit.hin t.he porps, has IIPVer
enjoyed really wide usage. TllUfi the clesignation adsorption is frequently f'lllployec! to denote upt.ake
whether by capillary condensation or by surface adsorption.

:w
induced dipole-dipole int.pract.ion) betwe('n the adsorbat.e and t.he substrate. V;-111 dcr
\Vaals interact.ions are weak, and are responsible for the (,llergy release wilen a part.icle
is ph:·{sisorhed. Such small amounts of energy call 1)(' trallsfelTcd int.o vibrat.ions of
tile lattice and dissipated by t.hermal trallsporL alld a 11I0lrCUle bouncing across til('
smface will gradually los(' it.s enngy and finally adsorb to it in the procf'SS calle<\
accommodation. The ('nthalpy of pilysisorpt.ion call be ll1easuH'd hy lllollitorillg til('
ris(' ill telllperaturf' of a sample of known l)('at capacity. and typical val1\es are ill tl)(,
range of 20 k.J / mol or less [S 1]. This slllall enthaJpy change is insufficient to cause bond
breaking, so a physisorb('d molecule retains its idplltity, altho1\gh it might 1)(' distorted
by tllf' i)!'ps('nce of the surface. In chemisorptiou. the particles st.ick to the smface by
forming a c1lf'micaJ (usually cova1ent.) bowl, aud tend to find sites that maximize their
coordination llUmber with the s1\bstrate. The ('nthalp~' of chemisorptioll is VNy 1lluch
greater than that for physisorption, and typicaJ values are in the region of 200 k.J / lllol
[S ']. A clwmisor1)f'd lllolecule may he torn apart at the df'llland of the ullsatisfied
valences of the surface atoms.

TJIf' frct' gas and the adsorhed gas are IU dynamic equilibri1\llI. The fractiollal
coverage of th(' surface. fJ, increases vvith decreasing temperature and illcreasing pressure. The variation of () with pressure at a chosen temperature is called the ad"'()l"plio/l
isotllf/'ln (or desorption isotherm). This is a measure of the molar quantity of ga.s 11

takC'n up, or released. at a constant temperat.ure T by all initially ('\('all solid surfan'
as a f11nction of gas pressure P.
11

= f(P)

( I .S )

If the temperature is below the critical temperat.ure of the gas. it IS convent.iollally
ex pressed as

( \.6)

:31
wlwre po is tllf' sat.uration vapor prf'ssure of the adsorptiv('2. j}/ j}O is call<'d r<'lati\"('
preSSllrf'. Prr!. ranging from 0 to 1.
Adsorptioll isotherms gelH'rally follow olle of six forllls as showll ill Figure I.!l.
]t is hased 011 an earlif'r classification hy Bnlll(BDDT). while the sixth is a recent addition.
Type I is charactf'ristic of adsorption Oil ll1icroporous adsorlwllts (solid witll pores
less thall :20 A).
TYlw II clescribes ac\sorpt.ioll on nOll porous solids or adsorlwll1.s witll Illacropo)"{'s
(po)"('s with width larger thall ,)00 A).
Type III alld V isot.herms arc characteristic of weak gas-solid interact iOlls [:·):3J.
Tll<' l\pe III is givell by a lion porous or macro]>orous solid and til(' Type V isot h('rl1l
hy a mesoporous solid.
Typf' IV represf'nts the adsorpt.ion isotherm Oil adsorhents with ll}('sopores (pores
with width l)f'twf'f'1I 20 A and ,:)00 A).
Typc VI. which was uot iucluded ill til(' Brllllaue)" classificatioll. illustrates the
)"f'lativf'ly rare adsorpt.ion isotherms with Illult.iple steps.

1.4.1

Languluir IsotherUl

This simplest isotherm model is hased 011 thn'e assumptions [:}·t]:
1. Gases form onl~' oue molf'cular layer 011 a solid.

2. The surface of a solid is composed of a t.wo-dilllellsional alTay of f'Jwrgct.icaliy
homogelleous sites.

:t Tlwre is uo iuterac1.iou among t.he adsorheu1. 1ll0leCliles. either through repulsive
or attractive forces.
2 General

terminology

Adsorbent t.he solid mat.erial
Adsorbate the material actually adsorbed by the solid
Adsorptive t.he gelH'ral t.erm for ti]P material in t.he gas phase which is capable of being adsorbed.

III

'0
Q)

..Q
Ul

tJI

.w
r::::

;:::l

Relative pressure

Figure 1.9: The IUPA \'C classification of adsorptioll isotllPrllls.

:n
Tlw Langmllir isotherm call \w df'rived h.\· considering t llf' kind ics of C011(k11S;-ltiull
and evaporatioll of gas molecules at a Illlit solid surface. If () represents tite fractio11 of
t lw adsorhent surface covered by a mOllolayer of adsorhate. tite ral<' of ("011 <\e11 s;-11 iOIl

of gas molecules ont.o the surface is proport.iollal to t.he fractioll of Yanl.llt sites (I - 0).
and tllf' absolute pressure of the gas. P. wit.h rate const.ant 1.-,,:

dO
(t

-I = h:"P(l - 0)

The rate of (,Yaporatioll from t.he surface is proport.iollal 1.0 O. wit II rat.e cOllst.allts 1",/:
dO = -I.- ()
dt

Tlw cnthaJp.v of tllf' adsorbed lllolecule is lower hy ~II thall til<' dc'sorhed 1IIokcule.
so the desorpt.ioll rat.e is gi ven as
~H

1.'(1 = I.: a eXI)(--)
I.-T
At equilihrium. the net rate of adsorption is zero. and solving for 0 gives the Lallgmuir
isotlwrm:
( 1,7)

where l\.' = "'" / kd'
Equation 1.7 can he rearranged t.o the linear form

With 0 = liLx,. where \. is the volullwofgas adsorbed at pressll!'<' I). aud \~x is 11lf'
volume corresponding to complet.e coverage, we have

-: = -.- + -.-.

\ 'x

/\ \ 'x'

( I.K)

11
Thus. if the equatioll applies and valut's of PI\' arc plott('d agaillst P. a st relight lille
slJOlild rrsuJ t . with the slop 1I \ :x. and in t.ercppt I lId'".

1.4.2

BET Isothernl

BrUllauer. Emlllett. and Tell('J" [56] ext (,1H1('d the LangnlUir llIodel t.o illclllde llIultila.Vf'r adsorpt iOll phf'IlOmella. The essential assumpt.ions of t II<' BET 1lI0del arc that

the forCt's adi\'{' in the cOlld('nsatioll of gases also arc respollsihle for tlw bindillg
('llerg~' ill Illulti-molecular adsorption. and any given layer need Ilot \)(' cOIlIpl('t(' hefore

su bseq uellt layers can form. By equating tlw ra te of cOlldellsa t iOIl of gas Illolccu I('s
onto all alH'ad~T adsorbed layer t.o t.he rate of evaporation from t.hat la~Ter and Slll1l1l1illg
for all infinit.e number of layers. they ohtained the expressioll
(f 111011

cP"
(I - Pr )[ 1 - (1 - c) P,,]

(un

\',here Pr = PI PO. po is the saturation pressure of the gas, Om"" is th(' nlll11\wr of
l1loles of gas adsorhed COlT(,SPOlldiug t.o ll1onolaj'er coverage. a is tllf' amoullt of gas
adsorhed al pressure P, anel c is a const.ant. which is large when t.he t'nt.halpy of
desorption from a 111onolayer is large compared to the f'lltha.lpy of vaporisatioll of thE'
liquid adsorhate:

""here ~j{,f". is tile enthalpy of desorpt.ion from tllf' first la~T('r. alld ~JI""I' IS 1Ilf'
('Bthalpy of liquefaction of the aelsorptin'.
BET isotherm corresponds t.o practice of surface area lllf'aSU)"('IlH'nt.. as will 1)('
described later in section 2.2.:3.

:~.')

Bibliography
[1] R. D. McCart.y ..J. HoreL and H. 1\1. Hoeler. Sel('ct.ed Properties of Hydrog('n.
N at.ional Bureau of Stalldards l\lonograph l(i~. \Vash ingt.on. J)(':N BS (19~ I ).
[2] C. Borusbay and T. Nejat Veziroglu Hydrogell as a Fuel for Spark Ignitioll
Engines, Alternativf' Energy Sources VIII. VOlllnlf' 2. Research ami Df'wloplllf'nt.
Hemisphere Publishing Corp .. l\ew York (l9~9).
[:~]

W. Novis Smith anel .J. G. Santangelo. Hydrog('n: Production and Markdillg.
ACS Symposium. Washingt.on (1980).

[4] B. D. James. C.N. Baum, F. D. Lomax ..Jr., C. E. Thomas. and I. F. Kuhn .
.Jr.. '·Compa.risoll of Ollboard H~'clrogell Storag(' for Fuel ('('II Vehicles." Task
4.2 Final Heport under Subcont.ract 4i-2-H:H 14~. prf'pared for Ford ~Iolor (~o.
under DOE Prilllf' Contract DE-A( :02-94CE50:~8!). Directed Techllologi('s. Inc..
19!:l6.
[5] R. I~. Kunz and R. P. Golde, "High-pressure conformable h~'drogel1 st.orage for
fuel cell vehicles," PI'O(Hdi1!g.<; of thf 1999 (T.,S. DOE Hydrogf11 Program HCl'if 1/'.
[6] K. L. Yan. B. G. Sellars . .J. Lo. S . .Johar, and M. I~. l\iurthy, Int. J. lJydrogm

b'nfrgy. 10, 51i (1985).
til D. Linden Handbook of Batteries and Fuel Cf'lls. McGraw-HilL Inc.. N('w 'y'ork
(1984).
[8] Fuel Choice for Fuel (:ell Vehicles: All Overvi('w by the DOE H.vAdvisory Panel. May 1999.
[9] S. Yoshimura and R. P. H. Chang. SUI)f'rcarbon-S~'llthesis. Propert.ies and
Applications, Springer-Verlag Berlin Heidelberg, New York (1998).

[10] H. O. Pierson. Handbook of CClrhon. (;raphitc,. Dialllo1\d a1\d )-"IIII('1"('1\('s. ;-.io.\"('s
Publications. Park Hiclge. l\'ew Jersey (UHl).

[II] I\J. S. Drf'ssf'lhaus. G. Dresselhaus. and P. C. Eklund. Scit'nce of FIII1cn'll('s alld
('arboll Nanotll]ws. Academic Press. Sail Diego (19%).
[I~]

Brandt,

I\1.Chudinov.

Semi metals 1. Graphite and its Compounds.

alld

POllOIllCl n'v.

Sciellce

PII hI isl)('rs.

Ya.

Elsevi('J"

~ew York (1988).

[n] G. 1\1. .Jf'nkins and I":. Kawamura. PoI~'meric Carhons. Cambridge U ni versi ty
Press. Camhridge. TTK (19TG).

[14] H. A. W('ssling. Polyvinyliclelle Chloride. Gordon l\~ Breach. New York (I !)TT).
[15] A.

l\loort,.

Walker,

.Jr..

and

Thrower

(f'ds).

Ch(,lllistr~' alld Physics of Carboll. I\larc('1 Dekker. Inc.. New York (1981).

[IG] L.

Peebles.

Carbon Fibers: Format-ioll. Sf rllctllre. and Propertit>s.

('IH'

Press. Boca Rat.oll (199.1).
[17] M. Inagaki. I":. Muramatsu. y'. Maeda. and I":. l\laekawa. S!}lIth. J1JrI .. 8. :3:35

( 198:3).

[18] .J. S. Mattsoll and H. B. Mark . .Jr.. Activated Carbon - Surface Chelllisfr.v and
Adsorption from Solution. Marcf'1 Dekker. IItL. New York (1971).

[19] .J. \V. Hassler, Purification wit.h Activated Carboll. ('bemical Publishing Co ..
Inc.. New York (19T4).
[~O] H. W. 1":1'01,0 • .J. H. Heath. S. C. O'Brein, R. F. Curl, and H. E. Smallt>y. NaIll/'(

(Lolldoll) 318. IG~-IG:3 (198.1)).

[21] W. I":riitschmer, K. Fostiropoulos, and D. H. Huffman, in Dll8fy Objrct.'; i1l til(
Uniuf/\<;(:, Procef'dings of the 4th Intern. Workshop of the Ast.ronomical Obser-

vat.ory of Capodimonte, Capri, Italy (1989).

:3'1

[22] W. l":rat~chmer, L. D. Lamb. K. Fo~tiropoulos. and D. n. Illlfrlllall. Yolli/"(
(London), 347, :354 (1990).

[2:3] H. Taylor. .J. O. Hare, A. K. Ahdul-Sada, alld H. W. I~roto. J. ('1/(/1/. Soc. ('/If III.
('01111111111..20,142:3 (1990).
(24] H. W. Kroto, lVnfun (Lolldoll), 329. !)2!) (19~7).

[2.5] T. G. Schmalz. 'vV. A. Seitz. D ..J. Kif-ill. ami G. E. Hit.f', J. /1111. ('Ii'1I1. Soc.,
110. 111:3 (1988).

[26] \V. I. F. David, R.IVI. Ibberson,.J. C. 1\lattltewllIall. I~. Prassid('s. T .. J. S. Dellnis.

J. P. Hare. H. 'vV'. Droto, H. Taylor. and R. 1\J. Walton. /Vallln(Lolldoll), 353.
147 (1991).

[27] R. D. Johnson, G. Meijer, and D. S. Betlnllle. J. Am. (,hnn. Soc .. 112. 898:3
( HmO).
[28] R. D . .Johnso11, D. S. Bpt.hunf', and C. S. Yannolli, ACCOIlIlJ.'i of ('h( III. H( .'i .. 25.

169 (1992).
[29] S. Sait.o and A. Oshiyama, Phys. Rev. B, 44. 11532 (1991).
[:30] H. D. Bf'ckhaus. C. Ruchart, 1\1. has, F. Dif'derich. and C. S. Toot.f'. AlI!}f!/'.
ChUI1. illl. Edn. (1191. 31. 61 (1~)!)2).

[11] A. Tokmakoff, D. R. Haynes, and S. 1\1. Gt'orge.

('//(/1/.

Ph!}s. L(II., 186, 4.'>0

(1991 ).

P2] A. R. I":ortan, N. I~opylov, S. H. Clarum. R. M. Flemillg, F. A. Thiel. and n. ('.
Haddon, iValun (London), 355, .529 (1992).

[:n] P. 'viV. Stephens, L. Mihaly, P. L. Lee. R. L. Whet.1.ell, S. M. Huang, R. I~aller.
F. Diederich, and I~. Hokzer, Nail/I'f (Londo1l), 351. 6:32 (U)91).

[34] M. B. .lost, N. Troullier, D. M. Poirier. .J. L. Martins, .J. H. Weaver. L P. F.
Chibante, and R. E. Sma.lley. Ph!}.';. Rfv. B, 44,1966 (1991).

:3~

[:~.1]

J. E. Fischer, i\latrrials ,1..;6(11('( (/lId £1I.'}ill((r;lI.'/, B19, 90 (l!)<):n.

[:lG] .J. E. Fisc\H'), awl P. A. Heillcy . .J. Ph!),';. Chull. Soli(k 54.17:):2. (l()<):q.
[37] C. Pan, 1\1. P. Sampson, Y. Chai. H. H. Hauge. and J. L. 1\1 0 rgro\'c, .I. /)"y.~.

Ch( 111 •• 95, 2944 (U)91).
[38] S. Iijima, Natlln 354, .16 (1991).
[:39] L. D. Lomb and D. R. Huffman, J. PhY8. Chu1I. Solids, 54, H):3.1 (199:3).

[40] R. Sait.o, G. Dresselhaus, anel M. S. Dresselhal1s. PhysicaJ Propert.ies of (:01'1>011

Nallot.uhes, Imperial College Press, London (1 ~l98).
[41] l\'. Hamada, S. Sawada, and A. Oshiyamo, Phy.s. Ru'. Lflt .. 681:,)79-1!)81 (1992).

[42] A. Thess, R. Lee, P. Nikolaey, H. Dai, P. Petit,.J. Rohert, C. Xu, Y. H. Le(', S.
G. I~illl, A. G. Rinzler, D. T. Colbert. G. E. SCllseria, D. Tomanek, J. E. Fischer
anel R. E. Smaliey. 'Crystaliille ropes of 1lH't.aliic carhon llanot.lIl)('s', /)('i(lIrf,
273,48:3 (1996).

[4:3] T.

I~azu.voshi,

Tokio,

Science and Technology of Carbon Nallotulws.

and

Elsevier SciellcP

I\('nichi,
Lt.d..

New

'York (199!)).
[44] 1\1. Yudasaka. T. homatsu. T. Ichihashi, Y. Achiba, and S. IijiJllo. J. /)hy.<;.

(,hfm. B, 102, 4892 (1998).

[4!)] T. Guo, P. Nikolaev, A. G. Rinzler, D. Tomanek, D. T. Colbert. anel R. E.
Smalley. J. PhY8. Choll., 99, 10694 (199.1).
[46] 1\1. S. I~im, N. 1\1. Rodriguez, and R. T. h. Baker. J. Catal., 134, 2!):l (1991).
[47] H. I~a'yser, Wied. Ann .. 14. 4.51 (1881).

[48] .J. W. McBain, Z. Phy.5. Chnn .. 38, 471 (1909).

:39
[49J S . .1. Gregg and 1\:. S. W. Sing. Adsorption. Surface Area alld Porosit~·. 21ld Ed ..
Npw York (19~2).
[.50J P. A. Wpbh and C. Orr. Allal~,tical l\Ipthods ill Fine Part icl(' T('chnolog~"

l\li-

cronlf'ritics Instrument ('orp .. Norcross. GA (1 ~J9'I).
[.11J P. W. Atkins. Physical Chpmistry. 4th Ed .. Npw York. FrpPlllall (1990).

[.')2J S. Brunau('l'. The Adsorpt.ion of Gasps alld Vapors. Vol. I. PllysicClI Adsorptioll.
Princeton lTniversity Prpss, Princf'ton, N.J. (194:3).

[:):3] A. V. hisdev, J. Colloid i1l.tu:f([('( Sci. 28, 4:30 (19()~).
[.14] 1. Langmuir. J. Anlfr. ChrlTl. Soc .. 40, 16:31 (191~).
[55J 1. Langmuir. J .. Amfr. Chnn. Soc .. 38, 2221 (HH6).
[.1G] S. BruualIPr. P. H. Emmett., amI E. Teller. J. AlTIfr. CllfTII. Soc .. 60, :309 (Un~).
[.17] S. Ono and S. Kondo, Molpndar Tllf'ory of Surface Tellsion ill Liquids. SpringeI'.
Berlin (1960).

Chapter 2

Experimental Techniques,

Equipment, and Data Analysis
2.1

Sample Preparation and Modification

2.1.1

Nanofiber Synthesis

Carbon nanofiher samplf's for this invpstigat.ioll Wf'J"(' preparpd by t.1l(' catalytic
df'col1lposition of hydrocarbons. as introduced in 1.:3.0. In order t.o g<'llf'rat.e a range
of fi\)f'r sizps and morphologies. Wf' uSf'd Ff' - eu catalysts of difTf'rf'nt compositions.
producf'd by either mechanical alloying. gas condellsatioll or wet cliemicall1)('thods.
Mechallically alloyed catalysts W<'l"f' jHf'pal"f,d hy high ellNgy hall III ill ing ina
SPEX I 8000 shaker mill, using Ff' and eu powdC'rs in appropriatC' proportiolls [I].
The grain size of the catalyst pm\'(if'r is essentially rplau,d to the duratioll of ball
milling. It can be monitored by the broadening of X-ray diffract.ioll peaks.
TIl{' chemical method consisted of reduction of Fe and (·u nitrate precursors using the generic conditions of Rodriguez and Baker for the syntlwsis of catalysts that
produce high yields of graphite nanofibers [2, :3. 4]. Fe and eu was introduced onto
the support by impregnation from an aqueous solution of the mixed Illetal nitrates.
(:akination in oxygen at about :300°(' resulted in the conversion of tllf' met.al salts to
mixed oxides. an essential step if olle is to prevent volatilizatioll of the catalyst precursor. Following this relatively mild oxidation trf'at.menL the lIH'tal oxide particles
wer<:> reduced to the alloy state by reduction in hydrogell at. ahout 400°(:.
A variation of thf' gas condensation Il\f'thod was also used to produCt, catalyst,
when' nanosize particles were producf'd by f'vaporation into an illcrt. gas [I)]. Mixed
f'p and eu met.al samples were evaporated by electrical resistance heating into a gas
1 SPEX is a trademark of SPEX Indust.ries,

Edison, N.].

41
of N2 or He at a pressure of about :3.5 Torr.
('atalysts Wf'rp placpel in a tuhf' furnacf'. anel t1lf'ir surfac<' oxidc was reduccd
using a 7 vol (1<', H2 in Ar mixture at. either 250 or 5.')0°(' for 1 hr. TIl<' flhns wcre t hCII
grown by passing a mixtun' of pthylene (C 2 H 4 ) and H2 gasps OVf'r the cat.alyst.s at a
temlwrat.ure of 600°('.

2.1.2

Kditscluuer-Huffulan Method - Synthesis of Fullerenes

Several different cOlllmercially available fullerene sam pips were obtained frolll Alfa
Aesar and l\lER Corp .. including pure (\\0, pure C iO , and mixtures of tlwsp two
refprred as flllhritc
TIlf' fullerpne sample from MER Corp. was originally prppared about fiV(' years
ago. It was made by a carbon arc discharge ill helium (I\:riitschmer- II uffmal\ lllf't,hod)
[6]. The carbon soot was treatpd in toltwl1e. The fullprite was solu hlf' ill tohlf'w'
at the limit of 2 gm/liter and the solution was then spread ont.o a Teflon tray and
quickly evaporat.ed at temperat.ures from 480 to !i200( '.
TIlf' fullerite sample from Alfa Aesar was also producf'd wit.h fllf' classic 1\ ratschmerHuffman method. The raw material obtained from the carbon arc discharge was also
purified wit.h t.oluene. However. the precipitate from the tOIU<'llf' solution was ohtaillf'd
after slow evaporation of the toluene. TIlf' precipitate was subsequently \vas!ted \vit.h
pdroleull1 ether.

TIl(' material then underwent. drying t.o relllove the rest of tlw

solvent. [7].
Bot.h fullerite samples ment.ioned above cont.ained about. 72-7!i(X, ('em. 22% ('iU and
1-:3(!(', higher fu llerPllP molecules.

The pure C fiO (or ( 70 ) fullerenes were obtained by chromatograhic methods. and
t.he purity was 99.9% for C 6U and ",98% for Cil).

2.1.3

Single-Walled Nanotubes Production

Samples of sillgle-walled na.not.ubes (SWNT) ma.terials were prepared and purified
by the Smalley group at Rice University [8].

4:2
2" flow tube
inner 1" tube

NT--+
heated zone

Figure 2.1: The configuration in the :2" system vvith about 1 g/day product.ion rate
(from [8]).

A dual pulsed laspr vaporization (PLV) t.echniquf' was emplo,ved for the gf'nf'ratioll
of SWNT.
As shown ill Figure :2, L a 2"-diamder horizontal flow tu be with ill a tu IH' furllace
was held at elevated temperature and arranged t.o maint.ain all argo1l atlllosphf'r<'
flow at a controlled pressure. Laser pulsps from t.\VO Spectra Physics G( ~H-2!)0 lasers,
each running at :30 Hz. entered tIl(' Hem: tube through a Brewst.er angle willdo\\' 011
tIl(' froll t. flangf' and propagated coaxially down tllf' tu 1)(' in the samc d i rpclioll as
til(' Ar flow. The target sitting coaxially in the flmv tul)f' was a l"-dianwtf'r, l"-long
balance carbon cylinder with Co and Ni of 1 at. (/(' each. TIl(' ablation occlllTf'd frolll
t.he target's circular end face, and laser pulses were rastered across t.he f'nt.ire fac(',
S\VNTs condensing from t.he laser vaporization plume were entrained ill tlw AI' flow
to 1)(' swept. downstream and deposit on the quartz tube walls outsidp the Iwatpd
zone. In order to increase til(' yield of S\VNTs. a 1"-diamtcr quartz t.ube was acldt>d
coaxially with the r

tube extending from tlw frollt Rangf' to wit.hi1l ,j 111111 of the

target face. This 2" apparatus was capable of generating about Ig/day of mat.('rial
containing 60-90 voJ.f,?(, S\VNTs. TI1f' production conditions v,wrf': 1:200°(', 100 SCCIll
flowing argon, 500 Torr, first laser pulse 5:32 nm, 490 m.l/P. 6-Il1111-diameter spot.,
sccolld laser pulse 1064 nm. 550 m.l/P, over-lapping f)-IllIll spot. with 42 liS delay
between pulses.
The raw material was first purified by a reflux in 2-3 M nitric acid for 45 hours.

Typically 1 liter of acid was used for every 10 g or sample. \VeiglJl loss is ahollt 10(/..
after 2el hours ,vith little further weight loss after this timC'. The resultillg l,lac\.; sollltioll was cf'ntrifugf'(L leaving a black sediment at thc bottom of the centrirug(' bot t Ie
and a clear. brO\vnish-ydlow supcruat.allt acid. which is decanted off. TIl<' H'Jllaillillg
acidic s('diment is \\'aslwd with df'ionized wat.er to rc-susp('lId it. alld cClltriruged repeatl)' until the solutioll is nearly neutral. The post-acid solids are thell filtn \\'asl1('d
with mildly basic solution. usually a pH 11 NaOH solution with a stalldard J1lethod of
hollow-fiber. cross-flow filtration (CFF). Filt.er washing with deionized water followed.
Tllf' SVVNT solution was vacuum filtered off tllt' liquid through a PTFE IlwllIbrallc.
\VIH'n a sufficiently thick SWNT lay('\" was forllled. it might readily 1)(' ]>('eled off
t.he Jllembrane to produce a freestanding mat called "bucky paper'. The yidd was
typically 10-20%. depending upon the quality of the initial raw material.
Despite the dramatic improvement in the SWNT purity. high-resolution TEl\l
images showed that the material still contained a significant quantity or impurities.
In order to relllove these, further oxidizillg acid treatments were ('lllployed. A. (:l: 1)
mixture of sulfuric (!cI8/(') a1l(1 nitric (IO/(') acids was stirred alld mailltained at 10°(' ill
an oil bath for 20-:30 min. This is followed by allot her ( TF cycle as descri bed aho\'e.
The final purification is done with a (4: 1) mixture of sulfuric acid and h.vdrogen
peroxide following the same procedure as ''''ith tlH~ sulfuric /llitric mixture. A vacuulll
bake at 1200°(: was used for final purificat.ion of t.he bt1ck~' palWr.

2.1.4

Sonication and Filtration

To modify the morphology of the SWNTs and cut the SWNT's. the samples \\"('re
sonicated in dimethyl formamide [9].

The as-received SWNT sample from Hice's

group was in the shape of a piece of black paper. Thell it. was cut into lllllCh smaller
pieces of ahout a fraction of cm 2 • A N.N-Dinlf'thyl-Formamide (lH'ON(,J-hh) with
purity of 99,8(1r, was used as the media in a ultra-sonic bath. The ratio or the sample
to the solvent was about (0.1 mg) : (1 ml). It took 1110re than ten hours before the
sample was completely detached and suspended in the solvent. A vacuum filtration

syst.em was set up t.o separate the black, cloudy suspellSlOll frUlIl Ilw sol \"('111 • as
showlJ schematically ill Figul'f' 2.2. A Biichner fu II lie] with 10 microll filter was IIsed.
A wat.er trap is necessary to prevent allY so]v(,lll froll1 jwiug accid(,lIt,all~' sucked illt.o
the vacuum line. After it. was compkt,f'ly dried. t.he modified S\\,NT salllple was
carefully peeled off from til(' filt.er of tilt' Biicllllcr fUlllWI. The ~'i('ld of tltis proc('ss
was about 80%.
top view of Buchner funnel
showing Ibe filter paper

glass tubing altached to a pie
of rubber tubing cI~ by

a pinch or ~rew clamp
(A s(opcock may be used
in place of Ule clamp,)

Buchner funnel

filtration flask

a typical trap

Figure 2.2: Vaccum trap for filtrat.ioll (from [10]).

2.1.5

Catalyst hllpregnation

Background
Support.ed catalysts have been widely used in chemical enginf'ering. There an~ a
number of advantages in depositing catalytically act.iw llJ('tals Oll a porous support..
The metal can be highly dispersed in the form of smaIl cr~'stallit.es, thus a large actiw
metal surface is produced relative to the weight of metal used.
Catalysts can be prepared by different techniques including impregnation, adsorption from solution, deposition and precipitation/co-precipitation. The preparation of

imprf'gnated catalyst.s commonly involves sew'ral steps:
inlpregnation step: cont.acting a dr.v or wet support. ,,,it.b all illlpr('gnal in~ solutioll
wihich consists of a compound of tlw desired ('at.al~,t.i(' const.it.uenl dissolved ill
a liquid
drying step: drying t.he resulting mixture from impregnat.ion st.ep
activation step: calcining and/or reducing if lIecessary
The cat.alytic activity may he very sensitive to t1lf' details of the procc,dures. SlIch
as solutioll cOllcentration. cont.act time. washing. lIlf'thod and cOllditioll of calcillation
and reduction. which may influence the degree of llIetal dispersioll awl unifol'lllit.~, of
distribution over the support. Changing a preparat.ive variable may yar.\' the calalyst structure and its characteristics (metal area. crystallite size. distribution and
dispersioIl). which are direct.ly related to t.he perfol'lnancf' of t.he ca1.al),s1.
Some commonly used cat.alysts and t.heir cOIll]>otlllds used to prepare the catalysts
arc listed in Tahle 2.l.
Metal

Metal compounds

Pt
Pd
Ni
Co

H 2 PtCl". [Pt(NH 1 )4]CI 2 • [Pt(NH 1 )4](OHh
PdCI 2 • [Pd(NH 3 )4]CIz
Ni(N0 1 b·6H 2 0
Co(N0 3 b·6H 2 0
Cu(N0 3 k6H 2 0
RhCh, (NH 4 hRhCI 6
RuCh
AgN0 1 .Ag(NH:3 bN0 1
HAllCLl

eu
Rh
Ru
Ag
Au

Table 2.1: Selectioll of nwt.als ami llwtal compounds.

Factors Influencing Impregnation Profiles
Each step of preparation process has its impact on the impregnation profile. During impregnation of t.he solut.ion, t.he concentration of the solutioll. the Ilwtal COIll-

46
pounds' readiness to 1)(' absorbed onto the substratt' and capillarity will all contrihute
to tlw il1Jprrgllation profilt'. Thf' take-up of tlw solutioll is gov(,rTwd h.v the porous

structure of the support.

During the subsequent drying step. the segregatioll alld

evaporation speed should lw t.aken into consideratioll. For the activat.ioll step. COIIditions of calcination and reduction. such as t.emperature and gas composition. IlPt-'d
t.o be controlled properly.

Preparative Techniques
Catalysts are cOIllmonly produced by liquid-phasr impregnation ill which a dry
or wet pellet of the porous support is impregnated with a solution of a compoulld of
th(' desi red cat.a.iytic consti tuent.
Nickf'lous Bitrate. 6-Hydrate (Ni(N0 3 b·6H 20) solutions vv-ith differ<'nt concent.rat.ions ,,,-ere prt'pared. Designated amounts of solut.ion Were added into tllf' support.
sample. The mixture \vas stirred cont.inuously until a smooth past.e was obtailled.
Super sonic bath was employed if necessary. to mix the solut.ion alld the support.
part.icles thoroughly. It is believed that sonicat.ion treat.nwnt is l)('neficial t.o the rf'activi(v of solid catalysis by the creatioll of surface defects. til{' reductioll of particlt'
siz('. awl tlw dispersion of the cata.lyst particles on thr su pport. [15].
Let the wet pastE' sit inside t.1l{' hood for at least. several hours until t.he carboll
powder was complet.ely dry.

This was to ellsurE' that t.lw illlpregJlat.ed COIIII)()Il<'t.

remained in the pore system instead of migrating t.o the exterior surface of the support.
A quartz chamber was set up for the reduction process. The sizt' of t.hr chamber
was specially designed t.o fit to a Lindberg furnace. and it. was good for a slllall alllotlllt
of powder sample. up t.o:3 grams. The powder was put into the middle of chambN with
quartz wool plugged OlltO both ellds to ])l'f'vent the sample from Illigrat.illg aroulld.
Before raising the t.emperature. the chamber was purged with N2 gas for about 1('11
minut.es t.o drive away the oxidative atomosphere. H2 wa.s used as the reducing gas,
flowing through t.he carbon powder t.o reduce nickelous nitrate int.o nickel at elevat.ed
temperatures (,...., 400 - 500 0 ). The flow rate of hydrogen was monitor('d using a water
t.rap of the outlet ga.s by about ol1e bubblf' per second. The reduction process t.ook

47
OIlf' to two hours. Tlwll tlw flm'\'iIlg gas was switched hack to l\' 2 allel t h(' fUrIJoc(' was
tUfllf'd

ofT. Tllf' sample was cooled c10WII to ambient tf'll1»eraturf' slowly' illside thc

fll l"llace.
The nickel crystallites were characterized hy X-ray diffractollwtry aud trallsilllissioll electron microscopy.

2.2

Sample Characterization

2.2.1

X-ray Diffraction

X-ray diffractomet.ry (XRD) was used to obtain informat.ion about sample COlllpOsitioIls, crysta.J structure and lottice parameters, phase t.ransforlllat.ioIl, alld catolyrst
particle size.

Diffraction patterns were collect('c\ by an hlf'1 CPS-120 diffractolll(,-

tel' sysj,('m using Co I~o radiation (A =

1.788~6!)A)

ami a curvcd posit.ion sC1lsitiv('

detector (PSD) spanning 127 0 in 28.

Basic Physics
\Vhen a monochromatic, unpoiarized electric field is incident onto a cryst.allinf'
lattice, the scattered wave \II(~k) can be cakulated as [11]

\II(~k) = S(~k) F(~k)

(2.1 )

where S(~k) is cailed shape factor, which IS a Stllll OVf'r all the latt.ice sit.es of t.he
crystal (all unit cells):
1,,/1 iCE

S(~k)=

L exp(-27ri~k·(rg-s))
rg

(2.2)

and F(~k) is knowll as tlw st.rllct.urp fact.or, which IS a Slllll ovel' the atollls ill thr
basis (all atoms in the unit cell):
bnsis

F(~k) =

L f(Jtom(rdexp(-2rri~k·rd

(2.:3)

diffract ion vector (hkl)

deviation vector

unit cell positions in the defect-free crystal

unit cell basis vector

fntom:

at.omic scattering factor of atom at rk

\Vhen the diffraction condit.ions are satisfied, the relation between the illcidrIlt
wavelength and crystalline lattice spacing meets t.he Bragg law as

>. = 2d h1d sin ()

(2.4 )

wlwre dHI is the crystalliIlf' latticf' spacing. For cubic crystals with lat.t.in' paral1lf'ter
au, the interplanar spacing dhkl can 1)(' calculatf'd as

(2 ..))

.- -1

Sometimes, Q(A

) IS used as abscissa unit in XnD pattern. It may easily 1)('

con vert cd from 2().

_ 2rr _ 4rr . ()
Q - -d - -sm
>.

(2.G)

X-ray Line Broadening
Tlw SdH'l'l'er equation relates the l1lean dimension D of tht' powder crystallit.es
and the X-ray diffraction broadening /J as
( .)-.1-)

where ,\ IS the radiation v\;avelf'ngth, () is tlH' Bragg allgle, alld /\0 is III<' SCII<'ITf'r
COllstant approxi1l1atel~0 equaJ to unity and related both to crysUdlik size

way Ii and D are defined.

In order t.o evaluate ii. il is lI(,c(,sS

observed broadening B for instrument.al effects.

Ii IS the full-H:idth-at-1I1

(F\VHl\l) ill radius. Stokes showed that tlH' instrllllH'lltal COlTt'ction was hpst dOI\(' Iw
a deconvolut.ion analysis using Fourier lllf'thods aIld the inst.nlIlH'IIt.from a sam pie in which intrinsic sources of broac\eni ug (size aud sl rai n) A practical method to derive the true diffraction breadth uses valid for Gaussian functions:
( .)_.

when' b is the illstrumentaJ broadening and B is th(' ohs('r\'('d hroadC'uiug of tile X-ra

peak. \Vagner and Aqua's correctiOil [1:3J

(2.9)

proviC\f's an improvellH'nt for typical X-ray linesha)H's.
For a given crystal dimeIlsion D. the IH'ak Im'adth 1IlCl"<'ases as ("os f) de(Te(lsf's.
Hence the particle size broadening becomes more pronounced at large values of ().

2.2.2

Translllission Electron Microscopy

The morphologies of the samples w<'re studi<,d by transmissioll electrou Illicroscopy'
(TEM) with either a Philips E1\1 420 or a Philips EM ·no microscope. The Philips
EM 420 wa.s operatt'd at 120 k V, \,.; hi Ie EM 4:30 was operated at 1:30 k V.
The technique of preparing the TEM samplps was quite simple.

Most of t.lw

samples examined wt'rt' in the form of a black powdf'r. ThE' procedure of pr<'parat.ion
illciud(>s the following steps: Mix a ff'w drops of iso-propanol iuto Ow powder alld
stir for half a minute. Dip the copper TEM grid into tlw suspensioll and takp it out
quickly. Let it dry in the air. Before loading the samplf' into the vacuum chamber
of microscope, blow off the large partides, which are loosely attached to the grid, t.o

50
avoid cont.amination of t.he microscope.

2.2.3

BET

TI)(' surface area is detenninf'd by Ilwasul"lllg tilE' amount of gas adsorbed ill a
l1lollolayf'r. Thf' t.otal surface arf'a .';" is obtailwd from tlw product of two qualltitif's:
the number of molecules needed to cover thf' samplf' 'vvit.h a lIIollolayer of adsorbate
(tm

]\'A

and the amount of surface occupif'd b~' a llloleClIle of a particular adsorlwd

(:L 10)

where am IS the 11t1lll1wr of moles of gas adsorhed 111 Uw IllOllolayer c\lld j"A IS 111('
Avogadro number. The values of am call Iw found from jJ)(' IlH'ClSll)"('III('flt of Uw
amount of adsorbed gas and an adsorption isotherm. as discussed in Chaptf'r 1.4. For
monolayer adsorption. tllf' sil1lplf' Lamgmuir isotherm is givell as
(/

((711

1 + 1\" P"

(2.11 )

where P" = PI po and a is the aIllouut of gas adsorbf'd at Prf'ssure P.
In practical work, tllf' semi-empirical f'quat.ion of Brunauel". Ellllllet and Tellf'l"
(1.9) is used:
((

cP"

(1 - P,·Hl - (1 - c)p,.]

(2.12)

Thf' BET equation contains two unknown constant.s am and c. Hencf' IIIf'aSUrement.s of a at two pressures are enough to find am and. consequently, S. SOllwt.illH'S.
for adsorbed gases c » 1. the BET equation simplifies to

(/ = - - -

1- P,.

provided Pr » lie. Thus it requires measurement. only at olle point.. Usually c::::; L
since the heat of adsorption and evaporation are usua.lly a.lmost equaL In this case.
the more general BET equation must be used [14J.

:)J
2.0~--------------------------~---------------------------.

1.5

BET

ctl 1.0

Langmuir

0.5

Figure 2.:3:
isotherm.

Physical adsorption isot.lwrms:

Langmuir (I\=:W) alld BET (c=:W)

For many systellls, the BET equation holds in the rangf-' P,. = PI J{)

'"V

O.OS - O.:~.

Any condensable inert vapor can be used in tllf' BET method. Howevf']'. for reliablf'
measurements. tllf' molecules should he small and approximately spllf'rical. P,. values
of 0.OS-0.:30 are conveniently attainable. The choice is lIsua.lly llitroW'Il. ill which
casf' measurement.s are executed at cr~'ogenic temperatures. usillg liquid nitrogell as
coolant.

The boiling point of nitrogen at atmospheric pressure is -I %.KO( '. The

effective' cross-sect.ional area of an adsorbed molecule of nitrogen is usuall~' t.akel) t.o

1)(' 0.162 Ilm 2 .
A Microl1leritics ASAP 2010 BET surface analysis apparatus was elllplo~'f'd. usillg

N2 gas. The ASAP 2010 syst.em consists of all analyzer, a COllt rol Illoclule for enterillg
analysis and report. options, and an int.erface cont.roller. which cont.rols analys('s. As
shown in Figure 2.4, the analyzer contains two sample preparation ports and Ollt'
analysis port.

In-line cold traps are located between the vacuum pump and the

manifold in bot.h the anaJysis and the degas syst.em.

::illdlng
!jn I £'1 j

__ _

:lample
PreparatIon ~
Fort5
~-

S.rur 11111: 11
Prt!~~Ult' Tubt!

Vacuum
Pump
Enclosure

AnBI-,'SI5

Balh Dewer
Df'\\'iH E ISliater

Figurf' 2.4: Micromeritics ASAP 2010 BET surface analysis apparatus.

III general, a sample with a total surface area of 40 to 120 squarf' 111e1f'rs providf's
best. results for nitrogen analysis.

Smaller areas may cause variabilit.y of results:

considerably more than this unnecessarily extends the time required for anal,vsis.
The sample was degased at about 150 to 200°(, in the sample tuhe with a llf'aJillg
mant.le for more than 10 hours hefore the mf'asuremf:'llt was jwrforl1led.
The BET met.hod is unlikely to yield a value of thf' t rtlf' snfan' area if t.lw isotherlll
is of either Typf' I or Type III (1.4.2): 011 the other halld, hoUI Type II a1l<\ T~'l)(" 1\'
isotherms are, in general, amenable to the BET analysis.

:):3

2.2.4

Fourier Transfornl Infrared Spectroscopy

Infrarf'd slwctroscop~' is tllP study of tlw illt.f'ractioll of iufrar('d li~ht with Illattn.
\Vhen infrared radiatioll iuteract.s with matter it call 1)(' absorbed. cansillg tit(' clWlllical bonds in the material to vibrate. Chemical structural fragllH'llts within llIokculf's.
knowlJ as flll1cfio1l.u/ groups, tend to ahsorb infrarf'd radiatioll in tIl(' Salll(' waV('llUI\1her rallgf' rf'gardless of t.he struct.ure of t.he ff'st of t.he molf'Clilf'. The cOITela/ion
bet.wf'en t.llf' waveuumbf'rs and molecule st.rueturf' makes it possible t.o idclltify tlte
structurf' of unknown molecules. For instance. the peaks aroulld :3000 cm- I ar(' d 11('
to CH bond stretching. Infrared spectroscopy is a useful chemical analysis tool.
Pure vi brational sJwctra arf' observed in the range between 10 2 cm -I ( 10 2 JlIll)
and

j 04

cm- I (1 pm).

Alt.hough vibrational spectra are observed experiment.ally

as infrared or Raman spf'ct.ra. the physical origins of t.hf'Sf' t.wo t.ypes of spect.ra are
different. Infrared spectra originate in transi t.ions jwtWf'f'I1 two vi brat-ional If'vf'ls of tlte
molecule in the electronic ground state and are usually observed as ahsorptioll spectra
in t.he infrared region. On the othf'1" hand, Raman sJ)('ctra originate in tlw f'1f'ct.rollic
polarization caused by ult.raviolf't or visible light.

If a molecule is irradiated b~'

monochromatic light of frequf'ncy //. then. because of elf'ctronic polarization induced
in thf' molecule by this incident light, light of frf'quellcy 1/ (Haykigh scatkring) as
Wf'll as of 1I ± IJi is emitted (I/i rep1"(:'sf'lIts a vibrational frf'qllf'1I Cf' ) [1 ~J.
The Wal'fllllmbff' of a light wave is defined as the rf'ciprocal of tIl(' wavelength

{~. =

1/)..

where).. is the wavelength. W is usually ff'portf'd as cm- l .
To obtain the transmission spectrum, samples were prepared int.o potassiulll bromide pellet.s, which is a suit.able sampling technique for powder samples. I>\Br is an
illert., infrared transparent. material. It acts as a support and a diluent for thf' sample.
Approximately 1 mg sample and 200 mg I>\Br were ground anel mixed ill the mort.ar.
The sample/KBf mixture was t.hen placed in a press and pressed into a t.ransparent.
pellet. The pellet was then placed in a pellet holder to obt.ain the spf'ctrum. Also,

fund iOllal groups
or I1loleniles

absorbance
wavenllmher (cm-l)

H2 O
CO 2

:3900-:3400
:2:350
:lOOO- :2900
4:395
2200
1428
1585

HH
NiH

ello

CTO

1~')0-1 :{.)o
GG7

1927
1183
14tn

refel'f'11Cf'S

[17]
[17, l~]

[17]
[18]
[18]

Table 2.:2: Absorbauce wavellumbers of some fllllctiollal groups alld lIlol('cules.
background spectrum was obtaiued on the empty pf'llf't holdel'.
KBr is a hygroscopic materiaL which meaus it will ahsorb water dirf'ctI.\· from the
atmosphere. Thus it is difficult to stucl~· tllP OH b01ld strecbing ahsorballce pf'ak,
aroulld :l400 cm- I .

2.3

Sample Measurement and Data Analysis

2.3.1

Sievert's Apparatus

All automated Sivert's apparatus "vas lIsed to perform a.11 lhe isotil('rJlJ lll<'aSUrements with hydrogen.

Our Sieverts' apparatus (i.e" a volumetric system for quantitative I1Wasurellwnt
of gas ahsorption and desorption by solids) used metal sf'als, an oi 1- fref' vaCU1I1lI
pumping systelll and research purit.v hydrogen gas [19J. Pllf'umatic valvf' ope·rat.ion
and pressure and tf'II1)wl'ature data monit.oring werf' computer controlled, permitting
aut.omatic isotl)(,l'm data collection. The system was t.horoughly leak cIJf'cked at :WO
bar and calibrated to ensure reliahle determination of the hydrogen storage propertif's.

A schematic drawing of Sievert\; apparatus is shown in Figure 2.5.
All tubing and fittings used were 1/4" :HG stainless steel with Swagelok VCR
fittings. l\lanual valves were Nupro bellows va.lv('s, VS51 ('. Pneumatic valw's WPJ'('

:,).)

V150 = 151.4400

55 ReMlnr
""28.<100

55 ReMlnrw/
long e>tension
"'31.2 cc

shelf
V500

Figure 2.5: Schematic drawing of Sievert's apparat.lls.

06
actuated h.v small soknoid valvf's. Pressurf> lIlpasurell](,lIts were made wit h SETHI\
pressure gauges. Isotherms could be measured at high pressUl"f'S (up to :W()() psia) at

0.1 psi accuracy or at low pressurcs (100 psig) at 0.01 psi accuracy. These gauges W('I"('
calibrat.ed by mounting a 600 psig mallual J-h'ise gauge OJ] the systPlIl. Z('ro pressure
was read by a Granville-Philips cOllvect.rOIJ gauge. givillg a zero H'adillg of < I x 10--'
torr.

Reactor and room temperatures were Illcasured wit h 1\-I.Vlw t.lH'rJllocollples

from Omega.
The manifold a.Bd reator VOIUIlWS were calibrated with high-purity heliull1 gas. A

1.50 ml volume. \\so. wa.s initially calibrated at Aerojet Aerospace Corp. awl used as
t.he rderence' volul1w in gas-expansion volul1w measurCllwnts. It was possihle 1.0 var~'
the manifold volume by opening manual valves isolation calibrat.ed volulllf's. The
hydrogell used for I11f'aSl1rClllents was l\laJheson {T LSI (i.0 hydrog('11 (9!UJ9!J9(';{). It
was cOIlllected t.o the Sieverts' apparat.us with electropolished SS tubing to a brass
regulator. A tribodyn oil-free shaft driven mechanical pump Illallufactured h~' Danielson 'vas llsed to evacuate gas from the system. A hot air gUll \vas lIsed to heat t.hp
sample reactor to a maximum of :~OOO(' for samplc'> bah'-ollt. \\'11('11 ext.ellded hours of
degassing was necessary. a heat.ing tape was employed to obt.ain the' elevat.ed t.elllfwraturc by wrapping it. around the reactor. It is possibk to maintain t/w t.f'nljwrat,ure
at designated range, ac('uratf' to ± 1.0°('. Isot.herms at low t.emperat.ures wert' lIwasllred \\'it.h the reactor immersed int.o liquid nitrogell in a df'wf'r with VOiUIlIe' of ahout.
:) lit.rf's. To minimize t.he fluctuation of prc'ssurf' nitrogeu. a special connection part df'sigllewit.h a thin and long t.ubing. which part was around the iut.erfan' of liquid Ilitrogf'l\
and atmosphere during the isotherni nlf'asurements. TIlliS. evell though thf' lev('l of
liquid nit.rogn challged due t.o evaporation. the volume of gas \Willg aff('ct('d was relatively slllall since the inner diameter of the cOllllC'ct.ing tubing is very small. With
a soft plastic lid. one clewer of liquid nitrogen could last for at i<'ast. 8 hours and t.Ile
fluctuation of pressure was less than 1 psi at t.he pressure of 1500 psia.
The a.nalog readings from the pressure a.nd tempera.tur€' gauges were converted

!ji

h.Y a Strawberry Tre(' PCI board interfaced to a l\lacilliosh SE compllte!". 1'1)(' SE

read and rpcorc\('d tilt' prf'ssur(' data alld Ol)f'ratec\ t.Ilf' rf'lay-solelloid-plwlIIllat ic-yah'('
chain used to cont.rol t.he pressure of t.lw hydrogell gas ill t.he apparatus. TIl(' program
used t.o operat.e t.he equipment. and acquire data was writt.ell ill the (' programllling
Iangltagf'.
Tllf' sample reactor was a double ,valled t.ul)f' m;HI(' of 1/ j G" COj>j>f'l" pipe' elf'ctrollbeam weldpd to a :3/r :316 stainless stepl male Swagelok \"('1{ flangf'. Sf'veral \TH
porous l1lPt.al filter gash'ls with porf' sizf' 0.5 jim were used 10 IHPvpnt. tlw sa III pie
frolll migrating into t.he syst.f'1ll and cont.aminatillg t.he valves.

2.3.2

Hydrogen COlllpressiblity and Ideal Gas Correction

The ideal gas equation of st.at.e gives:
p\ .
11 = - -

HI'

V,' here n is tilt' llul1lber of moles in a

(:U1)

known volunw at a kllOWll (ellllwraturf'.

At very low densit.y, hydrogen approaches ideal gas behavior. At. higher densitif's,
the hf'havior may deviate substaJ1tiall~· from the above equation. Thus, the ohsf'rved
pressure of Hz is too high t.o give correct results whell used ill ideal gas equation t.o

find 11.
To obtain correct n, t.he observf'd ]H'eSSUl'e Po lllllSt he divided h~· 1Iw comprf'ssioll
factor Z t.o yield a correctf'el pressurf' Pc,

p _ Po
c Z

(2.15)

Z is givell by
(2.1G)

The coefficients Ai depend on the ga.s and tf'lllperature and decrease rapidly wit.h
n. I use only the first tlue(> A's and th(>y ar(> given as functions of temperaturf' (P
must he in atmosphere),

.40 =

1.000547 - 6.07 x 10-' x T

A\ =

0.000912 - 1.06:):3 x lO- fl X T

(7.:ri3407 - 0.01901T) X 10-'.

.42

.·h =

for 1'<100°('

for 1'> 1000('
Figure 2.6 shows a det.ailf'd cOlllpressibility chart for hydrogen.

Not.e t.ha\ at

temperatures of 300 1\ and above. the comprf'ssibility factor is near ullity lip to
pressure of 10 MPa..

::l
....

::l

f---

011

O.~

0..

::;-

0'
'"1

8 06

o'"1

U.40

16,"

14,1"
U.,)t

11."

U ...

ll.n

11.90
n.14

n.n

I..... .,
U.l'

.0''''7
)ItO

115.'
101.0
111.0

0.4

_lit.

'.f".

.0'''2

...•••
1)0.7

,OJiO.
.01200
.0101'

0.6

Il hf-

71' .•

tu .•

toO

.OJ461

440
it'D
480

.01111

.04.,..
.011"

.0'955
.000llS
.000JU

.037U

].0
lIO

,0)")

.OJlJt
.0)416

lOG

.aun

.QlGn

.0111'

1100
UO
110
200
UO
140
UO
JIG

.01251
.01(0)'
,DUll
.01'"
.01911
.OZU7
.Ol.lU

10
100

,CIOn')

.Clon

.00n91
,00.,.,

11.5

"n.'

44 ....

. f !j'±-1 i1

0.2

.0011,.
.00J'"

.0001'"

I ~.t.

.GOOn"

.0101'"

ptlcc

D eNllt,

...L

f·

,.26.'

".n

50.t1

.... 7.

36.11

17.11

n.lI

",Ot

':.JII

u.t1

••• .16
10.10
14.7&
10.09

.,.U

112.1
101.9

U4.'

160.2
140.:11

'''.9

210.'
U4.'

In.'

1121
Stt.1

124)

J2',21

)4.11

n.n
n.D'
n.ll

42."

".n

55."
50.51

",51
61.'1

".47

18'.4
Uf,l
LlLl
92.71

17'.1

te/_'-

t14n.
Ull

-o-nH', • . ./~ • 4_U"

(il 0.7

i-)

$U.)

'-<

Illl

eete-

....C1.UC, "'01_

iii

)101
IUt

u ••

h_Ul(

:J 08

I-

t--

t--+ !..c-r-:~'H.!~

t--+-

.ct:

pj~

8":
......

S!:a

(J)

I-

'"1

'-'

,11

,'-',

C'l

t-,,:,

I'!I

'"1

::::1
oq

0.8

.--

f--+L

t-+ ,-F'

--I 1-'

LlQUIO

r--

!---

f·-

f--

",?"

;=:::r

t±.

110

IDO

?;'l

1- .

t--

t)(

~--

200_

plia

PRESS~E,

PRESSURE, atmospher.s

'~
1""

_ 400

Ji"

-t ..

f~·-

L.~

'i:"

eIl()~ _IQQC>

200

400

ot 511111.otd.
£",1" • .,1"0 l..ailarotOl',
IIouIW.Cofou

zaunlN
• PV! RT

102

103

eoo eoo

It±i±ttF$MW:
NalianOI
C,ya,."ic

1.6

a..

iii

::J
in

I-

.., HO~

SOLID LINES ~ 'Ioth.rml
(T ; C)
BROKEN liNES - Ilomet rlCI (V" C

ff.',

,[t

.f-.llA'

l1OOO

l'~jj I

4000

,1_

COMPRESSIBILITY FACTOR
FOR NORMAL HYDROGEN

100

IjNOTE.

2000

Yr. '.tl ~.,.'tth# J'!ll ·7
1:It:::JL ,i~rT~ H

Bibliography
[1] B. Fultz, C. C. A1111, S. Spooner, L. B. Hong, .1. Eck('rt., alld W. L .. Johnsoll.
Met.all. Trans. 2 7 A, 20:34 (H)96).
[2] N. 1\1. Rodriguez and H. T. h. Bakf'r, '·Stora.g(' of Hydrogell ill Lay('rf'd Nallost.ruct.ures:' 1J. S. Patent #5.6.13.9.1)1. Issued Aug . .1. 1997.

[:3] N. 1\1. Rodriguez, J. Malcl'. Rrs., 8, :~2:3:3 (I ~H)).
[4] N. 1\1. Rodriguez, A. Chamlwrs, and R. T. h. Baker, .J. PhYfi. (,hflll. n. 10 2.
425:3 (1998).
[5] B. Fult.z. C. C. Aim, E. E. Alp. W. St.mhahn, and T. S. Toellnf'r, Pllys. Hc\,.
Lf'tt. 79,937 (1997).
[6] \V. I\ra.tschmer, L. D. Lamb, 1\. Fost.iropoulos, alld D. R. Hllffl11nll,

/Yn!u/,(

(Londoll), 347, :354 (1990).

[7] Information was obtained frol11 Alfa Aesar Co.
[8] A. G. Rinzler. .J. Lill, H. Dai, P. Nikolaev. C. B. Huffman. F ..J. Hodrigu(';1,Macias, P. J. Boul, A. H. Lu, D. Hf'ymcu11l, D. T. Colbert, R. S. Lee, .J. E.
Fischer, A. M. Rao, P. C. Eklund, and R. E. Smalley, Appl. Phy,o.;. A 67. 29
( 1998).

[9] .J. Liu, unpublished result.s.
[10] R.

.J.

Ff'ssenden

and

Ff'sseudeu,

Organic Lnborat.ory Tpcitniqllf's,

Brooks/Cole Publishing Company, Pacific Grove (1993).
[11] B. Fultz and .J. Howe, Translllission Electron Microscopy and Diffractolllet.ry of
Materials, (in preparation).

61
[12J A. R. Stoke::;, Proc. Phys. Soc. Londo'll, 61, :382 (1948).
[1:3] C. N . .1. Wagner and E. N. Aqua, Arlu(lI!. X-ray AlInL 7. 46 (UJ6·1 ).

[14] R. .J. Wijngaarden, A. I-':rollberg, and 1-':. H. Wf'stf'rtf'rp, Indllst.rial C'at,a l.Ysis,
WILEY- VCH Verlag GmbH, WeillIlf'im, 19!J8.
[1.'5] S. Ley and C. Low, Ultrasonics in Synthesis, Springer-Verlag, Bf'rlill Ikiclrll)('rg
(1989).
[16] A. ('hamlwrs, C. Park, R. T. r-.:. Bakf'r, and N. M. l1odriguC'z, J. Ph!!s. Ch(lll.
B, 102 ,425:3 (L998). Langmuir, 11 , :3862 (J9!l5).

[II] B. C. Smith, Fundamentals of Fourif'r Transform Infrared Spect.roscopy, el1e
Prf'ss. Inc., Nf'w York (1996).
[18] L

Nakamoto, Infrared ancl l1atnan Spectra of In organic and Coordillation

Compounds, 4t.h edit.ion, .J ohn Wile.v 8.:. SOilS Inc. ('anada (UJ8G).
[19] R. C. Bowman, Jr., C. H. Luo. C. C. Ahn, C. 1-':. Witham, and B. Fultz,.J. Allo.ys
('omp., 217.1 8.'5 (199.5) .
[20] R. A. Wf'ssling, Polyvinylic!f'nf' ('hlorick, Gordon 8.: Brf'ach, Nf'w York. I !J77.
[21] Y. Zholl and L. ZhOll, Sri. in China B 39. !)!J8 (]tJ%).
[22] M. Nielsen, .J. P. McTague. and W. Ellenson ..J. Phys. 38 C4/ 10-('·1/18 (1977).
[2:3] R. (,hahinf' and T . K. 80sf', In t. .1. Hydrogf'H Etlf'rgy 19, WI (1994).
[2.1] S. D. i\1. Brown, G. Drf'ssf'lhalls, and !'vi. S. Dn'ssf'lIlaus, !'vlat.. Bf's. Soc. Proc.
497,1.57 (1998).

Chapter 3

Hydrogen Desorption and

Adsorption Measurements on Graphite
Nanofibers
3. 1

His tory and C o mmercial Inte rest s

Diminishing worldwide rf'Sf'rves of fossil fuels, and tile rf'cogllitioll o[ t.he global
environmelltal impact. of combustion b.vproduct.s, 11l0tivatf's the sf'arch for practical
altf'rativf' fuels.

\tVhilf' hydrogen has a low energy df'lIsit.y 011 a per volume hasis .

and while tlwre is as of yf't no infrast.ructure for its dist.ribution, it. rf'1l1aillS t.he
most attractive source for use ill f'ither direct. internal combustion. or for fllf'l Cf'11
applications, as watf'r is tllf' primary byproduct.. TIl(' lHOSt. serious illlp('dinH'llt. 1.0 t.lle
lISf' of hydrogf'll as a tra.nsportat.ion fu e L however. is the lack of a suit.able means of Ollboard st.orage. Compressed gas storage is bulkv and requires the use of high strf'lIgt.1t
containers.

Liquid storage o[ hydrogen requirf's tf'mperaturf's of 201\ and efficielll.

illsulation. Solid state storage orrers the advantage of saf(-'r aud more efficiellt. ha.lldlillg
of hydrogen, but promises at most ,(J() h.ydrogf'l1 by wf'ight aud morf' t.ypicall.v 2%.
Tlwre has 111('rf'[orf' lwell cOllsiderablf' intf'rf'st ill rf'cent reports by Nf'lly Rodrigllf'z's
group at Nort.IH'astern Universit.y [1] t.hat cf'rtain carbon graphite nallofllwrs [2] can
absorb and rf'tai 11 6'/ wI. % of h.yclrogf'1l gas at. am bif'l1 t Lf'Ill\wrat.u re all d 1110df'rat.e
prf'ssurf'S (i.f'., up to 2:3 standard litf'rs or :2 grams of hydrogen Iwr gram of carboll at

50 to L20 bar).
The nanofilwrs produc('d werf' df'scrilwd as "t,Ilf' graphit.f' plat.f'If't.s arf' arrangf'd
parallei. perpendicular. or at. a.n allgle wit.h respf'ct. t.o t1lf' fiber axis." Result.s 011 sevf'ral 11l0rphologif's of graphit,f' nallofil)f'rs. illcludillg t.ubldar. Iwrringbo l1f', a.lld platekt..
Wf'rf' rf'portf'd [I]. None of t1wrn showf'd I('ss Ulall LI Wt.(7, storagf' ca.pacity. The Itiglt-

63
etit levels reported would mean that a practically-sized fuel container could give a
vehicle an 8000 km range.
When purified graphite nanofiber samp les were allowed to interact wit.h hydrogf'1l
at 2!)OC aud an initial pressure of 112 atm, a drop ill the pressurf' was reportf'd to
bf' observed over a period of 24 hours. Upon equlibriuIn, hydrogf'll was df'sorl)pd bv
opelling a rf'gulating val Vf'. It. was rf'ported that. t.IlPl"f' was a difference in the a ill ount
of hydrogen adsorbed and desorbed at room temperature, and heating was required
for complete rf'leasing of hydrogen.
Such claims are especially noteworthy, given that, until recently, til(' typical I)pst,
value of hydrogf'1l adsorption in carbon materials has bef'n on the ordf'r of 4 wt!A), or

O.!) HIe' atom (although there is also a reCf'llt claim that up t.o 10 wt,l7c, wati achieved
for H storage in single wall nanotubes[:1]). Owing to the potential import.ance of new
materials with high hydrogen storagf' capacity for t.he world-wide enf'rgy ('COnOIIlY,
transportation systems and interplanetary propubion sYtitems, we havf' synt llf'sizf'd
graphitic struct.ures of appropriatf' morphology in order to make Ollr own lllf'aSllrelllf'nts of hydroge n absorption and desorption.

3.2

Catalysts and Sample Preparation

Sf'vera.l graphitf' nanostrllctun-'c\matf'ria.ls vVf're preparf'd using Fc'-('ll cat.alyst.s of
differf'nt compositions, in order to gf'llPrat.e a range of fil)pr siz<'s alld morphologies.
As descrilwC\ in Chapter 2.1.1. mechanical alloying, gas condellsat.ioll or cllf'lllica.l
I1lPthods were elllploYf'd to produce the catalysts. 'vVe prepared a t.otal of tell differellt.
mate-rial:; and they are outlined ill Tab le :3.1.

3.3

Sample Morphology

Tllf'rf' Wf'rf' a wide variety or 1l10rphologif'S showl! in the graphitf' nanofii>f'r samples.

Therf' was somf' residual of Na iIllpurit.ies ill cat.alyst for CNF sal1lp lf' #1

and fibrous morphology ror carbon. TEM imagf's start.ed to show herringbone mi-

64
Salll pie

Catalyst.
composition

Catalyst.
Prep. Method

GNF#1
GNF#2
GNF#3
GNF#4
GNF#!)
GNf#6
GNF#7
GNF#8
GNF#9
GNF#IO

FeT5 C11 2S
FeTO C11 30
Fe iU C' u 30
Ff'SO C11 2U
Ff'goCu 10
Ff'lFegoC'l1JO
FegoCuJO
Ff'soNi20
FesuNi20

chemicall1w thod
chemical met.hod
chemical method
dwmical l1lf'tilod
l1lf'cilanica.J alloying
mecilallical alloyi Ilg
gas cOlldensation
gas condensation
mechanical alloying
gas condpnsatiol1

Nrt.nofi bel'
C; rowt II Ti Ill€'
(Ius)

lh·d uction
Temp.
(OC)
!)!)O
550
!)!)O
.1.10
.150

Composit.ion of
Reactant. Gas
(H 2 :C 2 H 4 )
(80:20)
(80:20)
(20:80)
(20:80)
(20:80)

.').')0

(20:80)

Tj!)O

(20:80)
(20:80)
(20:80)
(20:80)

:~

2!)0
250
2!)0

:3
:3
:3

Table :3.1: Bange of catalyst composit.ions and n 'ad.ant. gases IIsed t.o producp g raphit.e
nan 0 filw rs.
crostructurf' in some carboll fibers from samp le #2 (Figure :3.1) and #:3. Salllpip #.1
was more uniform and had smaller diameter of carboll fibers. It showed a range of mi crostructures incilldillg corkscrews, tu lws and a significant fraction of fibers with the
"herringbone" morphology as shown in Figure :3.2 and Figure :3.3.

Figure :3.4 shows

a high resolution micrograph of a single fiber , l'f>vealing a cavit.y at t.he (,op of the fiher
where the Ff'C'u catalyst originall y was, and from which the gra.phitic platelets grew.

In some samples, there were carbon nanotubes as well as fibers f'vident in TEM, as
shown in Figure :3.!).

3.4

Isotherm Measurement and Data Analysis

Before measurements of hydrogen adsorption and desorption, t llf' sa.mp les wen'
vacuum annealed at 900 ° (' in order to '"activate" t1w nallofibers.
The Sif'verts' apparatus was thorough ly leak checked at 200 bar and calihrated to
ensun' l'f'liable deterlllina.tion of the hydrogf'1I storage propert.ies.
Desorption mea.suremellts were performed at. 77 and :WO 1\ by first. placing about
0.:3 gram to 0.6 gram of sample in thf' reactor. H2 gas was aclll'lit.t.ed illLo (.llP evacuated
reactor to achieve a typical pressure of 4.5 or 80 bar for the 77 1\ rUlls or 180 bar
for til{' :300 I\ l'uns. This pressure was maint.ained for l!) hI' to allow the salllplf' to

65

Figure :3.1: High magnificaLion TEl\1 shot of graphit.e nanofiber sample #2 showing
herringbone structure.
reach equilibriulll and to check [or leaks in t.he systelll. The reactor was valv('d of[
from the rest of the system and the system was evacuated again. Tlw desorlwd H2
was t.hen lllf'asured by a pressure t.ransducer. The rwer(' accquired by a Macintosh SE computer.
For comparison, we also performed ll1eaSUrelllent.s Oil a "sarall" carbOll. a pure.
dense. porous material with high surface area. [orllled by t.he pyrolysis of polyvill.vlidene chloride as described in Chapter 1.3.1.

The microstructure of t.his mat.erial

consists of graphite microcrystals in an amorphous carbon matrix [9].
A lllacro program for the software package Igor by 'vVavemetrics was us('d 1.0
calculate the quantitive mole number of hydrogen before and after each desorptioll
step based on the pressure and the volume of t.he system. The cUlllulat.ive differellce
(~x)

was assumed to be the hydrogen relpaspd by the sample inside tlw reactor.

[den tical l1leasurements and calculations were also performed 011 an elll pty reactor
so that tllf' data could be properly corrected for inst.rument effects. A Sf't. of t.ypical
Illf'asu l'f'lllents for bot. h t he sample alld t. he ('Ill pty reactor are shown ill Figu re :LT.
Figure :3.8 shows the actual hydrogell df'sorpt.ion capacity of saran carbon after t.he
correct ion.
In addition to 112 desorption I1lPasllrpment.s. sample surface areas were mf'asurf'd

66

Figure :3.2: Low magnification TEM shot of graphitf' llanofilwr sa.lllplf' # ~).

Figure ~3.:3: SEM micrograph of graphitf' nanofibers in sample #S.

67

Figure :1.4: High resolution image from end of graphitf' nanofibf'r showiJlg Ilf'rrillghollf'
Illorphology. lnsf't at lower left shows latt.ice planf's from boxf'd rf'gioll.
using a Micromf'ritics ASAP :2000 BET surface analysis apparatus wit.h

3.5

2 gClS.

Results

Table :3.2 summarizf's tIl<" results of BET surfacf' arf'a measurf'lllf'nts and hydrogell
desorption data. For most samples, we did Illultiplf' runs on each daLa point t.o check
the reproducibility of the results. Some isotllf'rms a.re highly reproducible, as showlI
in Figlll"f' :1.9, sincf' the hydrogen acborption capacity is relatively high due t.o tllf'
high specific surface arf'a of saran carbon and the low t.emperature. At the maxilllum
prf'ssme of:3 bar at. 77 r--:. the hydrogf'1l adsorption capa.city of the saran carbon is 2.4

wI/It The desorption of H2 in the graphit.e nanofilwr samples is small but Illf'asurable.

68

.'.

100 nm
Figure :3.5: TEM micrograph of carbon nanotulws ill GNP samplp #5.
Figure :3 .10 shows a set of runs of desorption isot.herrll for G N F sam pip #4 at. room
tf'l1lperaturp. For the size of our sa.lllplps, tit(' s('llsitivit.y was b('ttPr t.han 1(X, accuracy
on a Iwr atom basis. Results frol11 fiv(' spts of rllns from t.hp saran carbon and sampl('

#5 and #8 ,up shown in Figure :~.12.
Data of a higb Iwrformancp activat.pd carbon d(,llOl,('d as AX-21 w('t"(' obtained
from the literature [10] and arP included in Tablp :3.2 for comparison.

AX-21 was

commercially availablf' and has a specific surface arf'a of :3000 ml/g and porf' volllllH'
of 1.f) lllL/g.

3.6

Discussion

As measured on a per atom basis, our graphite nanonbers may seem t.o show
adsorpt.ion Iwyond what one might. ('xpect. from normal surface adsorption . 'vVI1f'n
comparison is Illadf' to tllf' saran carbon, the rat.io of hydrogen covpragp t.o surface
area seems high for the graphit.e nallofibprs. \Vp would expp("\. n, challgf' in slope of
such an isotllf'rm, but t.his was not obs('["v('c1. \V(' Iwlipvp t.his res lllt.s froll] 1,11f' presenc('

69

70

30

68
66

60

28
64

50
62

Iii'
.."'.

26

1st desorption step
60

40

CD

.....

OJ

24

>-<

;oj

Empty Reactor

Ul

co
OJ
>-<
p.,

>-3

CD

.EJ

30

0.6467 g

'"c

r1"

.....

CD

GNF#4

[2
22

20

20

10

100

200

300

400

time [min]

Figure 3.6: Raw data of desorption steps at room U'lll]Wrat HrE'.

Eo

ca

--+- DelX for Saran carbon

-60

--&- DelX for empty reactor
---$f-- Difference in DelX

20

40

60

80

100

120

140

160

Pressure [bar]

Figurf' :~.7: Ca\culatf'd ~x (room U'lllperaturf') for saran carbon sample and the
empty reactor.

70

0.25

Saran carbon (LN 2 )
___ Saran carbon (300 K)

0.20

--Hl- Saran carbon (77 K. 15t run)

___ Saran carbon (77 K. 2nd run)

Eo
ro

E 0.15
.§.

<..)

0.10

Saran carbon (room temperature)
0.05

20

40

60

80

100

120

140

160

Pressure [bar]

Figure :~.8: Desorption isotherms for saran carbon at. room t.emperaturf' a.\ld liquid
nit.rogen t.(,lll])f'rat.ure (77 1\).

Carbon
Sample

AX-21 data from [10]
Saran Carbon
GNF #4
GNF #r)
GNF #7
GNF #8

Specific
Area
(m 2 /g)
:3000
1600

2!)

Hie
(~

(desorlwcl )
77 1\', 4)') bar
0.24
0.29 (:~ bar)

0.02

Hie
(clesorbcd)
let.' :300 1\', 160 har

0.06 (70 bar)
O.O!)
O.O2:3±O.OO:3
O.0:3±0.00:3
0.029±0.OO:3
O.02!)±0.00!)

Hydrogcn
( 'overage
(m 2 jg)
ir)O
i:W
r)f)

80
7r)
70

Table :L2: Comparison of surface area as measured by I3ET, desorhed at.omic rat.io of
H to carbon. and total Hz covPrage assuming diameter of solid 11l01f'Clilar Hz of 0.:3r) i
1l1ll[11].

71
0.30
EO:

.w
cO 0.25
.........
EO:

.w
cO

0.20

.........

::r:
s::
0.15
Q)
ry
1-1

'0

6;' 0.10

'0

Q)

.Q
1-1 0.05
til

Q)

0.00
0.5

1.0

2.0

1.5

2.5

3.0

Pressure [bar)

Figure :3.9: A set of runs of df'sorptioll isotherm for saran carbon sample at liquid
nitrogen t.emperature.

20

15

:::r::

10

~ GNF sample #4,

1 st run

-s- GNF sample #4, 2nd run
~ GNF sample #4, 3rd run

20

40

60

80

100

120

Pressure [bar]

Figllrf' :3.10: A Sf't of runs of desorption isotherm for GNF samplf' #1 at room telllperat.lll'e.

of surface irregularities in t1w graphite nanofilwrs that are not df'tecLed by the larger
N2 molecules in our BET measurement.s. Support for this viewpoint is found ill the
shapes of the isotherms in Figure :3.12. At a given tempera(urp the isotlWl"lllS for tllP
graphite nanofibers and the saran carbon have a similar shape, but a vertical o[fse\.
This difference by a scaling factor implies the same isosteric heat of adsorption for
both types of carhons. but there are more availahle sites for t.he sarall carboll.
None of the present hydrogen adsorption or desorption llwasurellwnts performed
on any of the carbon nanofiher materials has indicated a hydrogen storage capacity
that eXCf-'ed the values previously reported for various activated carbons [10, 12]. III
light of our results, the results of Chambers, et al. [1] are especially surprising. Their
claim of:2 gms of th per gill of C storage imply that 16 H2 molf'o11es are st.ored in
t.his material per (' atom. If we assume that these H molecules are all stored within
tlw graphite structure, then 48 monolayers of iucolllmensurately packed II:2 must 1)('
accommodated hetwf'en each pair of graphene planes (using a hard spllf'rf:' Illodd and
a value of 0.289 nm kinetic diameter for H 2 ) in order to account for tlw reported
adsorptioll. This amounts to each graphitic plane, with an a-b spacing of norlllall~'
0.:34 nm. being separated to over 14 nm. Their data for H:2 adsorption in graphite
of 4 ..1 wt(;{, implies that even this material accomlllodates H:2 beyond the 2.7 wtvalue that one would achieve with the comnwnsurate J1 struct.ure.

Fur1.lwrmore,

their reported hydrogen capacity for graphite at room temperat.ure is over Clll order of
lllClgnitude greater than the values determined by ot.hers [10, 12] for activated carbons
at 298 J\. The bE'st of these carbons yielded H:2 adsorption in the range of rv.1 vd%
ouly when cooled t.o below 100 l\.

3.7

Conclusions

As a result of our analysis, it seems unlikely that carbon in nanofilwr form shows

H:2 adsorpt.ion/desorption properties that would have the spectacular illlpact as a
solid state storage llwdium claimed by Chamlwrs, et al [1]. :\ more realistic analysis
of the limit.s of physisorpt.ion of hydrogen ill graphite is considereci by Browll, et. al

,:3
[l:~J, who forward arguments on the basis of geometry to show the likely upper limits

of adsorption in various forms of carbon. They note that under conditions where III
molecules are able to form two dosed packed layers within each graphite plane, the
atomic ratio of H: C approaches 1: 1 (rv 8 wt %).

74

35x10-3

30

25

....(1j

.......

20

....0(1j

"-'

.......

15

10

-s- GNF sample #5
-4f- GNF sample #7
~ GNF sample #8 (1st run)

-&- GNF sample #8 (2nd run)

20

40

60

80

100

120

140

160

Pressure [bar]
Figure :~.ll: Desorption isotherms for GNF sample #:>, #7, and #8 at room t.emperature.

Saran (R.T.)

...........

......

-ca
..........

()

()

ro

E 10-2

Pressure [bar]
Figure :~.l 2: Log-log plot of i i and :300 I\: isotllf'rlll dat.a showing amount of adsorbed
hydrogen/carbon for GNF sample #'f) ami #~, as a fUllctioll of pressure. Wh('11
multiple rUllS wefe taken. error bars arC' shown. with ollly the top half of the error
hars drawn for clarity. Traces from saran carbon are a.lso shown for COlli parison.

76

Bibliography
[1] A. Chamlwrs, C. Park, R. T. K. Baker, and N. 1\1. Rodrigu('z, J. Ph!Js. Chnll.

B, 102. 42!):~ (1998).

[2] N. 1\1. Rodriguez and R. T. h:. Baker. "Storage of Hydrogen in Layered Nanostrllctures," U. S. Patent #5,65:3,951. Issued Aug. 5,1997.

[:3] A. C. Dillon, I\". M. Jones, T. A. Bekkedahl, C. H. I\"iang. D. S. Hethunp. alld
1V1. .1. Heben, Nature (London), 386, :377 (1997).

['1] N. 1\1. Rodriguez, J. Mater. Res., 8, :3:2:3:3 (199:3).

[!)J N. M. Rodriguez, A. Chambers, and R. T. K. Baker. Langm,uir, 11. :3862 (199.1).
[6J B. Fult.z. C. C. AIm, S. Spooner, L. 13. Hong . .1. Ecknt. and \V. L. Johllsoll.
Ahtall. TrailS., 27 A, 2~M4 (1996).

[7J B. Fultz, C. C. AIm, E. E. Alp, W. Sturhahn, and T. S. Toellner, Ph.//s. Ufl'.
Lfft., 79. ~):37 (1997).

[8] R. C. Bowman, Jr.. ('. II. Luo, ('. ('. AIm, C. K. Witham, and B. Fultz. J. Alloys
Comp., 217,18.') (199.')).

[9] R. A. Wesslillg, Polyvinylidelle Chloride, Gordon ,~ Breach, New York, 1977.
[10J Y. Zhou and L. Zhou, Sri. ill China B39, .198 (1996).

[11 J 1\1. Nielsen, .1. P. McTague, and W. Ellenson. J. Phys., 38 (,4/10-( '·1/ 18 (1977).
[1:2] R. ChahinE' and T. I\". BosE', Int. J. Hyr/mg(1) Enf1:r;y. 19. WI (I~H)4).
[1:3J S. D. M. Brown, G. Dressplhaus, and M. S. Dress('lhaus. ,Hat. Rrs. Soc. PI'oc.,

497. 157 (1998).

II

Chapter 4

Hydrogen Adsorption and

Cohesive Energy of Single-Walled Carbon
Nanotubes
4.1

Background

A few years after the report of the Coo molecule by Krot.o, f't a!.

[1], Iijillla

discovf'red tllf' tubular form of carbon [2]. Singlf'-wallednanotu]ws (SWNT's) aTf' the
simplest of these struetures, being but a single graphit.e plane rolled into a thin tube

[:3,4]. Methods for the synthesis of SW'NT's do not produce a rnonodisperse product,
and the large scale purification of SWNT's has been achievf'd only rec(,IIt.ly [0]. The'
cohesion of these molecular crystals occurs through van der \Vaa.ls intf'ractions and
perhaps otllf'r effects of elf'ctron correlation [6, I, 8, 9], and it is widf'ly ohsf'rvf'd
that the individual SWNT's coales('f:' into rope-lih' strands [10]. Many propert.ies of
condensf'cl SWNT's aTf' now topics of intensive study; howevf'r, tllf' strength of the
cohesive energy of crysta.lline S\VNT's remains poorly undf'rst.ood.
There is a recent report that crystalline S\VNT's have a capacity for hydrogen
sorpt.ion of 5-10 wt.% at pressures less than 1 bar near room temperature [11]. Such
a hydrogen storage capacity would be a significant advance for the mw of hydrogen
as a fuel wllf'n a high gravimetric density of hydrogf'll is a figurf' of Tllf'rit. llntil
rf'cently, the best value of hydrogen adsorption in carbon materials has b('cl1 :).:~
wt.%, or 0.64 Hie, at a temperature of II IO\" [12, 1:3]. (A rf'cent claim that graphite
nanofibf'rs have a capacity of 24 HIe' at :300 K [1.1] has not bef'n corroborated [I!)].)
\Ve Wf're motivated to perform measurements S\VNT material of high purit.y Iwcause
the prf'violls llIeaSllrelllf'nts were made Oil dilute S\VNT's, so the analysis required a
large corrf'ction for more than 99% of lIlaterial that was assullH'd ilH'rt [11J.

78

Figurp 4.1: 1\ low resolution transmission cipctron microgra.ph of t.llP S\VNT llla1c>rials.
showing the rope structure.

4.2

Experiments

Threp batches of nanotube material werp preparpd [lG] and purifipd [!), 10]. To cut.
tlw SWNT's, about 0.2 g of mat.prial was sonicated for LO h in dimdhyl formamidf'
at a concentration of 0.1 mg/ml until the sample was COlllp lptply suspP Il(lf'd ill 1.lte
so lvpnt [17]. The modified SWNT material was then pxtracted by vacuum filtration
using a cpramic filter.
Desorption and some adso rp tion isotherms werp measured on samp lps of app roximatf'ly 200 mg with a computer controllf'cl Sieverts' apparatus. Aft.pr vaCUUlll degass in g at 220°(: for 10 h, the measurellwnt tf'mpf'ratur p was at.tained and hydrogf'l!
gas of 99.9999% purity was admitted into the reactor to a clesiredmaximulll prpssurp

(lGO bar at :300 I~, and no. 70. 4.5 or o.!) bar at 80 r~). This pressure was maint.ailled
for L!) It t.o a llow the adsorpt.ion 1.0 pquilibratp and to check for leaks in 1.hf' sys1.elll.
T llP amo un t of hycirogpn desorlwd from t.he samp le was determined in stpps by IJlf'a-

79
sUflng the pressure 011 the samplf' beforf' and afkr the sample reactor was opened
to an evacuated reff'rence volume. To COl'l'f'ct for instrumental f'ffect:,;, Wf' perfOrll1f'd
identical measurements on an empty reactor aft.er each sample Jlwasuren]('nt.. Surface'
area was measured with a Min:omerit ics ASA P :2000 BET surface analy:,;is instrument using nitrogen gas. Thf' surfacf' areas of the as-prf'pared S'vVNT material, tlw
mat.erial after isotherm measurement, and the matf'rial after sOllication, we're foulld
to be 285±5 m 2 jg. Phase contrast transmis:,;ion electron microscopy was perforllled
with a Philips EM4:30 transmission electron microscope opf'ratf'd at 200 kV. X-ray
powder diffractometry was performed using Co I\O' radiation wit.h an 111f'1 CPS-120
position sen:,;itive detector.

4.3

Isotherm Results and Cohesive Energy Calculation

Transmission ("If'ctron microscopy of th(" as-pr("parf'd material showed df'llse bUlldlf's. or "roPf'S" of crystallized SWNT's. The ropf' diamet("rs varif'd from 6-12 II Ill.
Tlw high resolution imagf' (Figurf' 4.2) shows circular rings of approximatf'ly L.~3 11m
in diametf'r, consistent with the dominant (10,10) SWNT st.ructure. Wit.h hexagonal tube coordination. a rope of 10 nm diameter would contain about. .10 tulws,
and would be S t.imes the diametf'r of a singlf' tube. The spf'cific surfacf' area of a
rope would be about. 6 times less t.han t.he Olltf'l' surface arf'a of a :,;inglf' t.ubf' (1:300
m 2 jg). The surfacf' arf'a measured by BET (285 m'2jg) is eviderJt.Jy a l1l<"asurf' of the

out.f'r slll'facf' arf'a of the ropes, not the total surfacf' arf'a of tIl<" illdividual l.u]ws.
The ropes Iwrpend icular to tlw elf'ctron ]wall1 showf'cl a II u mbf'r of Sf't.s of in temal
fringe spacings of 0.:34 nm. Fringe terminations Wf're observed within the rope:,;, indicat.ive of misalignments or terminatiolls of individual tubes. The rope dianwtf'rs of
the son icated ll1atf'rial wt'l'(, comparablt', bu t. wi th a broader size dist.ri bu t.ioll. The
sonicated mat.erial showed a more irrt'gular patchwork of fringf's. prf'sull1ably ]wcause
of more' t.errnillations of individual S'vVNT's in t.he ropes. On the ot.her hallci. it. call

80

Figure 4.:2: High resolution transmission electron micrographos of til(' SWNT 1l1aLerials. (a) as-prepared. showing cross sections of tubes towards lower cenLer. and (b)
after sonication in dimethyl formamicle.
be seen from Figure 4.3. X-ray cliffractometry showed no significant dif[erf'llces after
sonica.t ion.
For SWNT's (:28.1 m 2 jg), and high surface area saran carbon (1600 1l1 2 jg) [IK].
thf' hyclrogf'll adsorptions (ratio of H atomsjC atoms) obt.ained at :3.:2 bar at. 80 K
were 0.040. and 0.:28. respectively. These results show the ex]wcted proportionality
between surface area as measured by BET and the hydrogell adsorption, as did the
hydrogen adsorptions at 160 bar at :300 I\. The low pressure compositioll-pressl!rf'
isotherms at 80 f\ alld the high pressurf' isotherms at :300 I\ had similar shapf's for
all carbons. They were descrilwd adequately with the Langmuir adsorption isotherm.

81

3000

SWNT

>.

.w

-r<

2000
la)

Q)
-W

1000

SWNT (sonicated)

20

40

60

80

Ib)

100

120

Figure 4.:3: X-ray diffraction pattern of single-walled materials (a) as-prf'pared awl
(b) after sonication in dimethyl formamide.
for which the fractiollal coverage, f, is:

f = -,-----:-:-:-:=-t(E-li)/kT + 1

(-Ll )

The chemical potential of a hydrogen molecuh" in the gas is Il, allel its energ.v of
adsorption is E (E < 0). lTsing a tabulated function for cllPlllical potential versus
pressure [19] (Figure 4.4), we fit the saran carbon isotherm with f. At low coV<'rage
we found E = :l8 meV, in excellent agreement with the results of Pace and Siebt'rt

[21]. The isotherm of the saran carbon material in Figure -t,!"i is similar in siJalW to
isotherms from other carbons of high surfacE' area at 80 K [12, 1:3, 22].
Equation 4.1 can be written conveniently in terms of the gas pressul'f' [20]

-.f =
I~, + p

(L2)

-20

:>QJ

- 40

-60

20

40

60

80

100

P (atm.)

Figure 4.4: Chemica.l potential of hydrogen gas at 77 K.
For hydrogen on graphite at :300 I\:. Po rv 4 bar. Equation :1.2 shows that a smaller

P" is expected at lower temperatures. At SO K. e) rv 1 bar. amI adsorptioll occurs
at lower pressures of hydrogen. This is evident for the isotherm of the sarall carbon
material in Figure 4.5. and this is described well by Equation 4.1.
Also shown in Figure 4 ..1 is an adsorption curve calculated from the saran carboll
data by reducing it in proportion to the lower surface area of the S\VNT material as
measured by BET. This curYf'. scaled by the factor 285/1600. accounts approximately
for the adsorption of the SWNT material (lalwled "SWNT" in Figure 4 ..1) at low
pressures. but fails at pressures greater than about 20 bar.
At high hydrogen pressures at 80 I\:. the curves labeled "SWNT" in Figure 4 ..1
show a ratio of hydrogen to carbon atoms of about 1.0 (8.25 wt.(,1c»). anel suggest that
higher concentrations may occur at pressures lwyond ol1r experiment.al capabilities.
To our knowledge. this is the highest hydrogen storage capacity yet. measured on an
activated carbon material. (A coke material processed \vith KOH. denot<,d ":\:'\-21:'

1.0

SWNT
initial

\,/

0.8

SWNT

sonica\

SWNT

..........

.8
CO

---

0.6

+-"
CO

'--'

()

Saran

0.4

l.

0.2

fI

Saran x 3/16

-/ -0.0

20

40

60

80

100

120

Pressure [bar]
Figure ,1.5: Isotherms of composition versus })l'pssure at 80 K for sampips of asprepan'd SWNT materiaL til(" SWNT matprial after sOllication in dillwt.hyi forma.rnide. and a high surface arpa saran carboll. Adjacent pairs of CllrVf'S (lal)f'kd
"SWNT") were sequputial runs on the sallle sample. Also shO\vn is tlw curvP of the
saran carbon scaled to lower Hie ratio by the surface area ratio of 285/1600.

84
reaches a peak 0.64 Hie at ii K at an optimal pressure of :30 bar [12, 1:3].) Our results
are inconsistent with the report of Dillon, et a\., that such capacities are attained at
:300 K and pressures well below 1 bar [11]. The kink at 40 bar and the steep slope
of the hydrogen adsorption isotherm of the SWNT material at IH"C'SSlUes from 40 t.o
100 bar is unique for hydrogen adsorption on a carbon matf'ria.\. This shape call
not be obtained from a Sllm of concave-downwards isotherms such as f, even with a
distribution of adsorption energies, E. The shape of the SWNT isotherm is similar
to isotherms of metal-hydrogen systems which form a hydric\r- phase by a first-order
phase transition [23]' although the mechanism of hydrogen absorption is certainly
diffNent.
The low pressure adsorption of hydrogen on SWNT material saturates like that.
of the saran carbon, scaled by the surface area of the ropes. At pressures above ClO
bar, however, the isotherms indicate that the SvVNT material undergoes a transition
to a new state of hydrogen coverage. From til(' large HIe' ratio at high pressures,
we deduce that the surface area increases by about all order of Illagllitude. This
is consistent. with the hydrogen permeating into the ropes, separating thelll into individual SWNT's with full exposure of their outer and possibly innt'r surfaces. and
physisorbing onto the carbon surfaces. This high density phase must involve tub('
decohesion, since the hydrogen coverage is high and the attractive van der vVaals alld
exchange forces are attenuated when the tubes are separated by short distances. The
thermodynamic driving force for this tllbe decohesion is the high cllf'lllical pot.ential
of the hydrogen gas at high pressure. The hydrogen molecules that adsorb on the
surface of the SWNT's undergo a decrease in chemical potential. The equality of
chemical potential in t.wo-phase equilihrium requires the reduction of chemical potential of hydrogen gas in the high capacity phase to be equal to tilt' loss of van der
Waals cohesive energy [2/1].
The data of Figure .1.!) indicate an average pressure of iO bar for the phast' transition, which corresponds to a chemical potential of - L1 meV per hydrogen molecule
[19]. This should be compared to the characteristic chemical pot.ential for hydrogell
physisorptioll of -:~8 meV. corresponding to a decrease in chemical potelltial upon

a.dsorption, !1p, of 27 meV per hydrogen molecule. This ~p can be used to determine the cohesive energy of the SWNT's in a rope. When the adsorbed hydrogen
molecules are commensurate with the carhon atoms on a graphene plane, t.he COlllposition would be C:3H on one surface of t.ht' plane [2.'), 26]. For the l1If'asUl"f'd surface
area of the saran carbon of 1600 m 2 /g, this st.ructure ,vol1ld provide H/(:=OAO, ill
good agreement with the saran carbon isot.herm at ahout 70 bar. {ising this sallie
hydrogen coverage for the SWNT material at 70 bar, the ~p of 27 nwV/H 2 or (I:L.')
meV /H) provides a cohesive energy for the SWNT's in a rope of 4 ..') l1lf'V/C atolll.
An alternative estimate can he made by noting that the characteristic midpoint of the
phase transition corresponds to a composition of H/C=0.4:3 at 70 bar, which provides
a cohesive energy of 6 me V / (' atom.
The slope of thf' SWNT isotherm in Figure 4.G is not infinite, probably Iwcause
thert' is a distribution of cohesive energies in the material. To obtain an upper limit
on cohesive elwrgy, Wf' note that S0111f' of the tul)f's sf'parate at pressures of 100 har at
a macroscopic H/C=0.8. This corresponds to a colwsiw f'lwrgy of 1"1 I1wV/(' atom
for a minority of the material. Some of tlte clta ngt' in II/ (' at higher pressures COli Id
be caused by additional coverage on tubes that had separated at lower pressures,
however. Allowing for such a change in coverage in the majority of tlw material will
reduce this upper limit, perhaps cOllsidf'rahly.

4.4

Discussion

Our experimental value for the cohesive energy, .1 llWV for much of the materia\,
is smaller than the 17 meV Ie atom calculated by Tersoff [7], tlw :3.') meV /C atom
calculated by Benedict, f't al., for large tubes with tlat al"f'a of contact [9], or tiH' 22
nw V / (' atom calculated by Cagin and Goddard [27].
We are not surprised to see this reduction ill cohesive energy. The van der \Vaals
intf'raction and other electron-electron correlation pffects arp especially s('l1sitivt' to
distancp. \Ve can acconnt for small red \let iOlls ill colwsi ve f'lwrgy wi th featurf's of the
ropt' morphology ohservpc\ by TE1'vi.

86
Surface energy is one contribution. 'liVe assume the number of lwighbors will be
reduced from 6 to :~ for the tubes at the surface of the ropes. \Vithin a rope with a
radius of 6 tubes, the average number of first.-nearest neighbors of tube will be reduced
from 6 to .5, so the cohesive energy will be 5/6 of its value for a large crystal. This
means that the reduced coordination at t.he rope surfaces will reduce the cohesive
energy by only about 15(/(,.
The elastic energy of tube curvat.ure will be another contribution.

Figurf' 4.2

shows significant curvat.ure of the ropes, which provide for typical maximulll st.rains

in the tube walls of 0.5(7<'1 or so. However, with a calculated elastic modulus [28],
we obtained a typical maximum elastic energy of less than 0 ..5 meV per atol1l in tile
tubes. Thus the elastic energy of the obserw'd tube curvature is slllall and does not.
dominate the cohesive energy.
Since then' is a rapid reduction with distance of tIl(' van der Waals interaction and
other electron-f'lectron correlation effects [f'spollsible for S'vVNT cohesion, deff'cts ill
the close-packed triangular lattice of SWNT's should cause a large reduction ill t.ile
cohesive energy. The cohesive energy of S\VNT crystals is expected to be reduced
when the inter-tube distances are disrupted by tube terminations, misalignments, and
dislocations within the cryst.alline rope.
The following experimental evidence indicates that these defects canse a large
change in cohesive energy. The curve labeled "S\VNT initial" ill Fig11l'e ,\..5 was a
first run on a fresh sample, and first. runs on two other samples exhibited such high
reversible adsorption at. lower pressures. All second and subsequellt desorptiolls occurred at higlwr pressures, as shown in Figure 4.5 for the group of four C11l'V(,S labeled
"SWNT." We believe that the first adsorption/desorption cycle caused the SWNT's
to reorganize ill structure, perhaps into a IllOrf' perfect triallgular lattice, causing
subsequent desorptions at higher pl'f'SSUres.

More convincing are the data on the

sonicated material. The pressures of desorptioll for this material show a considnahle
breadth, indicating a distribution of cohesive energies of the t.ubes. Fllrt.hermort', for
all hydrogen concent.rations, the pressures for the sOllicated Illatf'rial are suppressed
with respect to the as-prepared samples. A reduction in pressure is consist.ellt with

87
additional defects in the crystalline ropes of sonicated material causing a reduct.ion
in the cohesive energy of the rope structure.
Finally, we note that the first order phase transition causes tllf' SWNT's to adsorh
and desorb over a narrower range of pressures, overcoming an engineering challenge
for hydrogen storage systems [22].

4.5

Conclusion

A phase transition between crystal SWNT and a new hydride phase was found
at high pressures at SOK. The phase transition was of first order, and involved the
separation of the individual tubes within a rope, exposing a high surface area for
hydrogen adsorption. From the change in chemical potential of the hydrogen gas upon
adsorption, we were able to calculate the cohesive van der \Vaals energy between til{'
tubes as 5 meV Ie atom. This is much smaller than expected from previous theoretical
work, and shows that defects in the crystal structure cause large suppressions of 1he
cohesive energy. We were able to altpr this cohesiv(' energy by changing the state of
t he material.

88

Bibliography
[1] H. W. Kroto . .1. R. Heath. S. C. O'Brien, R. F. Curl, and R. E. Smalky, Natur•.
318, 162 (198.5).

[2] S. lijima, Nature (London), 354. 60:3 (1991).
[3] S. Iijima and T. khihashi, Nature, 363, 60:3 (199:3).

[41 D. S. Bethune, C. H. I\:iang. lVI. S. Devries, G. Gorman. R. Savoy. .1. Vazqlwz.
and R. Beyers, Natun (London), 363, 605 (1993).

[.5] A. G. Rinzler, .1. Liu, H. Dai, P. Nikolaev, C. B. Huffman. F ..J. RodriguezMacias, P . .1. BouL A. H. Lu, D. Heymann, D. T. Colbert, R. S. Lee, .J. E.
Fischer, A. M. Rao, P. C. Eklund, and R. E. Smalley, Appl. Phy.-;., A67, 29
( 1998).

[6] R. S. Ruoff. .1. Tersoff. D. C. Lorellts. S. Sllbramoney. and B. Chall. Nature
(London), 364. 514 (1~)9:3).

[7] .J. Tersoff and R. S. Ruoff, PhY8. Rei'. Lfft., 73, 676 (1994).
[8] S. N. Song, X. K. Wang, R. P. H. Chang, and .J. B. I\:etierson. et a!.. Ph!) ..... fi(l'.

Lett., 72. 697 (1994).
[~)]

L. X. Benedict, N. G. Chopra, M. L. Cohell, A. Zettl, S. G. Louie and V. H.
Crespi, ('hem. Ph.lJ8. Lett., 286. 490 (1998).

[iO] .J. Liu, A. G. Rinzler. H. Dai, .J. H. Hafner. R. K. Bradlf'Y, P . .J. Boul. A. Lu.
T. IVf'rson, I\:. Shelimov, C. B. Huffman. F. Rodrig\lez-~lacias. 'y"-S. Shon. T. H.
Lee. D. T. Colbert.. and R. E. Smalley. ,I,'Ci(nCf, 280. 12.5:3 (1998).
[11] A. C. Dillon, I\:. lVI ..Jones. T. A. Bekkedahl. C. H. Kiang, D. S. Bethullf', and

M . .1. Heben, Nature (London), 386, :377 (E)97).

89
[12] R. Chahine and T. I\:. Bose, Int. J. Hydrogw Enfl:qy, 19, 161 (1994).
[13] Z. Yaping and A. Li, 8cimce in China, B 39, .598 (1996).
[14] A. Chambers, C. Park, R. T. Io\:. Baker, and N. M. Rodriguez . ./. Phys. ChrTII .. B
102, 42.5:3 (1998).

[15J C. C. Ahn. Y. Ye. B. V. Ratnakumar. C. Wit.ham. R. C. BowllIan ..Jr.. and B.
Fultz, Appl. Phys. Lett., 73. :3:378 (1998).
[16] A. Thess. R. Lee. P. Nikolaev. H. Hai. P. Petit ..J. Hobert, C. Xu. Y. H. Lf'e. S.
G. I\:im. A. G. Rirder. D. T. Colbert, G. E. Scuseria. D. Tomanek. J. E. Fischer.
and R. E. Smalley. Srlfll.cf, 273. 48:3 (1996).
[17J .1. Liu, unpublished results.

[I 8] Saran carbon is a pure. dellse. porous material with high surface al'('a. formed
by the pyrolysis of polyvinylidene chloride.
[19] R. D. McCarty..J. Horcl, and H. M. Roder, "Selected Propert.ies of Hydrogen"
NBS :Monograph 168 (US Government Printing Office. Washington. 0.( ' .. 1981)
pp. 6-1:30 - 6-269. W. E. Forsythe. editor, Smithsonian Physical Tahlf's 9t.h cd.
(Smithsonian Institute. vVashillgton, 1964).
[20] C. l\:itteL Int.roduction to Solid State Physics. 7t.h edit.ion. John \Viky & Sons.
Inc. New York (1996).
[:21] E. L. Pace and A. R. Siebert, J. PhY8. Chfm.. , 63, H98 (19!)!J).
[22] S. Hynek. W. Fuller. and.1. Bentley. IHt. J. Hydrogu) R1/.rrgy, 22. 601 (1997).
[2:3] L. Schlapbach. in Hydrogen in IntermC'tallic Compounds L L. Schlapbach. ed ..
Springer-Verlag. Heidelherg, 1988. p 1.
[24J \Ve assumf' that the ent.halpy of adsorption for a hydrogcll molecule on a curved
S\tVNT is the same as on flat graphite. The shape of the SWNT isotherm at low

90
pressures is nearly the same as that of high surface area ca,rbons, supporting
this assumption. \Ve also ignore the entropy gained during the separation of tllf'
tubes because this will be small on a per atom basis.
[25J M. Nielsen, J. P. McTague, and W. Ellenson, J. d( Physique, 38, eel-tO (E)77).
[26J H. Freimuth, H. Wiechert, H. P. Schindberg, and .1 ..J. Lauter, PhY8. Jl(P., B 42,
587 (1990).

[27] T. Cagin and W. A. Goddard, IlL unpublished material.
[28] G. Gao, T. Cagin, and W. A. Goddard, III, Nanotechnology, 9, 18:~ (1998).

91

Chapter 5

Hydrogen Adsorption and

Phase Transitions in Fullerites
Hydrogen desorption and adsorption properties of fullerene materials e nO , and C 70
and fullerite (a mixture of

e60 and ( 70 ) were measured volumetrically using a Siev-

ert's apparatus. Over severa'! cycles of isotherm llleasurements at 77 I\:, the hydroge\l
storage capacities of one of the fullerite samples increased from an initial value of 0.,\
wt(/CI for the first cycle to a capacity of 4.2 wt% for the fourth cyclf'. Correspo\ldingly, the surface area increased from 0.0 m 2/gm to 11 m 2 /gm, and showf'd a phase
transformation, characterized by X-ray powder diffraction. In comparison, two other
fullerite samples, prepared hy a different procedure. showed no such behavior. Pl11'e
C tiO and pure C iO were cycled and exhibited small and constant capacities of 0.7 wt(Y<,
and 0.:3:3 wt%, respectively, as a function of number of cycles. The enhanced storage
capacity of fullerite material is tentatively attributed to the presence of (\;0 oxide.

5.1

Background

Crystalline fcc

eno has been observed to absorb H2 in octahedral interst.ices, pro-

viding a storage capacity of only 0.28 wt70 at !)O°C [1].
In this chapter, I present the results of studies on physisorption of H2 into fullerelles.
In the fullerelle family, the closed cage nearly spherical C\;o and the ellipsoidal rugbyball-shaped C 70 are the most stable molecules [2, :3]. and thus show high relative'
abundance.

We performed measurements OIl pure ('tiO, pure ('iO and two sets of

a mixture of these two commonly referred to as "fulleri I.e." We report a surprising
increase in the amount of H2 that fullerite adsorbs and desorbs after a few cycles.

92

5.2

Experiments

Several different commercially availahle fullerene samples were obtained froIll Alfa
Aesar and MER Corp. All these fullerenes were made by a carbon arc discharge ill
helium (Kratschnwr-Huffman method). The carbon soot was tlwn treated in tolll('1}('.
The fullerite is soluble in toluene at the limit of 2 gm/liter and for til(" first fullerit(,
sample (denoted fullerite #1) from MER Corp., and obtained in 1992. The solution
was then spread onto a Tefton tray and evaporated at room temperatures. For the
second fullerite sa.mple from Alfa Aesar (denoted fullerite #2) and obtailwd ill 1999,
the precipitate from the toluene solution was obtained after slow evaporation of the
toluene. The precipitate was then washed with petroleum ether and underwent drying
to remove the rest of the solvent. A third sample of fullerite was obta.ined from MER
in 1999 (denoted as fullerite #:3) and was processed similarly to fullerite #2. All
fullerite samples typically contained about 72-75% e no , 22(;r, C70 alld

1-:~(1r,

higlH'r

fullerene molecules. The pure em (or ('70) fullf'rE'nes were obtailled by chromatograh ic
methods, and the purity was 99.9% for (\;0 and rv9S(Yr, for ('70.
Desorption and some adsorption isotherms were l1lf'asured on samples of 600 mg
with a computer-controlled Sieverts' apparatus at :300 K and 77 I--:. After VaClllllll
degassing at 200°(, for 10 h, the measurement temperature was attained and hydrogell
gas of 99.9999%, purity was admitted into the reactor to a maxil1lulll pressure of
about 120 bar. This pressure was maintain('d for 1!) h to allow the adsorptioll to
equilibrate and to check for leaks in the system. To correct for instrullwntal effects, we
performed identical volumetric measurements on an empty reactor after each salllple
lllf'aSUn'lIH'tlt. Til is procedure was used for every adsorption / desorption cycle, w hell
multiple cycles were taken on the identical sample.
Surface area was measured vvith a Microllleritics ASAP 2000 BET surface ar('a
analysis instrument using nitrogen gas.

Phase contrast transmission electron mi-

croscopy was performed with a Philips EM1:30 transmission electron microscope operated at 200 k V. X-ray' powder diffractollletr,v was perfornwd 011 samples, both asreceived and after-cycling, IIsing an hwl CPS-120 powder diffractollwt,('r using ('0

9:~

0.6
(.a)

0.5

Saran Carbm
(1000 m /g)

0.4

ro

--§ 0.3

Q<::::.

ro

--

()

rd run
th run

--

()

0.2

0.1
0.0

(b)

0.5

0.4

7th 6th

--rot5

Q<::::.

ro

--

()

--

nd
1st

o.

00

80

100

120

Pressure [bar]
Figure -1.1: Desorption isotherms of composition versus pressure at ii 1\ for two
different batches of fullerite sample #1 of C nO -C 70 fuUeritf' materials. The upper set
includes a trace for Saran carbon. The lower set shows identical isotherm hehavior
as a function of adsorption/desorption cycle number.

94
1\0 radiation. High performance liquid chromatography (HPLC) measurements \ver<'

performed at MER Corp.

Sample

Composition

(\)()

99.9+%
98+%
75%('00 22%(',u
1.5 7r,( :00 oxides
75%C 60 22(!(!C,o
0.2%('60 oxides
77%(\;0 21 %(',0
0.6(;("(\;0 oxides

C iU
Fullerite # 1
Fullerite #2
Fullerite #:3

H2 Capacity
wt% :300 K
abs
des
0.07
0.08
0.12
0.12

H2 Capa.city
wt(;('1 77 I":
des
abs
0.70
0.83
0.:33
0.!j8-4.:~8
4.00
0.:38-0.60

Surface area
(mz /g)
as~rec.
cyckd

O.!)

11

0.80~0.9!)

0.2~0.:~

Table 5.1: Il.! storage capacities and BET surface areas of fullen'nes.
The hydrogen adsorption/desorption and BET measurement ITsttlts are sumllla~
rized in Table 5.1. The hydrogen storage capacities of pure Coo and C,o are r('pro~
d uci ble and are consistent with results of others [1]. On tl1f' other halld, fu lleri te
sample #1 from l\IER and originally obtained ill 1992 exhibit.ed unusual isotherm
behavior. As shown in Figure 5.1 (a). the capacity during the first desorption run 011
the as-received fullerite #1 (labeled 1st run) is small, comparable to the capacity of
pure em or pure C,o. The capacity then increased dramatically with f'ach subsequent
isotherm cycle. In the fourth cycle (labeled 4th run), it reached a maximum of 4.4
>vt,(X, at

120 bar at 77 I":. This value is consistent with complete Hz adsorption onto

the surfaces of the fullerene molecules. assuming that Ib moiPcules of diameter :~.!j A
form a close packed shell around each fullerene molecule. At the fifth cycle. the ca~
pacity dropped back to about 2.5 Wt.(J(I. Samples of the samf' material were analysed
for their surface area by the BET method. The BET surface area changed from 0.9
rn 2 /g for the as~received fullerite #1 to 11 m.! /g after !j isotherm cycles.
In order to verify these results, Wf' tested a 2nd batch of fulleritc # 1. and oh~
tained the same isotherm behavior as a function of cycle lltll1lber. These results are
presented in Figure 5.1 (b). A total of seven cycle~ was performed 011 tlIP 2Ild hatch
of fullerite #1 and cycles!) to 7 showed identical isot.herm traces. 011 the othn hand.

95

a similar Sievert's apparatus was used to measure one point of the isot.llf'rm at 771\
at about 50 bar, but this measurement showed only 0.05 wt. (7r, hydrogf'11 sorpt.ion.
Sample handling procedures were similar although not identicaL the reason for this
discrepancy in these measurements is still unclear. Fullerite #2 behaved similarly to
the samples of pure C 60 or pure C m . and showed only small differcnces as a function
of cycle number as shown in Table 5.1. Fullerite #:~ which was obt.aillf'd recently frolll
MER and used as a cOl1trol sample. showed even smaller desorpt.ion capacities than
fullerite #'2 at 77K. For comparison. we also measured t.he hydrogen desorption of
high surface area saran carbon [4].

HPLC measurements of tllf'se materials showed only slight variations in tilt' ('60(\0 rat io.

Therefore. we believe that. the iIll portant differf'nce }wt.w('en fulleri tes # I.

#2 and #:3 is the amount of oxidized C hO found in as-received fullf-rite # I. The
oxidized ('60 accounts for rv 2% of the as-received fullerite. The oxidized component
was absent in the material after it had been cycl{'d with H2 gas. The oxidation of
fuUerite #1 occurred presumably over the 7' year time span from its synthesis to tllP
tinlP the desorption experiments "verf' perfornlf'd.
A pair of transmission elecron microscope (TEM) micrographs shown in Figure !).2
illustrates tllP instability of fullf'rite #1 under eif'ctron beam irradiation. The large
particle in the left side of the upper bright field image is the rf'siclue of a larger fuUcrite
particle. The remnants of this particlf' reCOndf'IlSed on the holey carbon support grid.
The higher magnification dark field (OF) image at. the bottom of Figure ·1.2 shows
a nanocrystalline microstructure of this recondensed fullerite. The t.ypical grain siz(,
for the rf'conclellsed nanocrystalline fullerite is rv I 0 Tlm. Some instabilit.y Hnder the
f'lectron beam was also observed for fullerite sample #'2 and for ('TO' The pure em
sample wa.s stable under the electron beam.

96

Figurf' 5.2: Transmission electron m icrograph of fu llf'rite # L ('fiUjC 7U . (a) low magnification bright field, and (b) higl1f'r magnification dark fif'ld of iUSf't arf'a showillg
recondensed fullerite nanocrystals.

5.3

X-Ray Diffraction Pattern and Phase Transformation of Fullerite

As introduced in Chapter l.3 .:3, the C GO 11101f'nilf's crystallize inLo a facf'-cf'lltf' rf'd
cubic structure with a latice constant of 14.17 A [5, 6]. At 1'00111 Lemperature, t.he
C HO molecules have Iwen shown by l1uclea,r magnetic resonancf' (N MR) [7, 8] to 1)("
rotating rapidly with three degref's of rotational freedom.
The crystal structure of ('70 is more complex than that of C HO . The structural
phase transitions and orientational ordering in G iO at d ifferent temperaLurf' and prf'Ssure haw been studied extensively for the past decade [10, 11, 12, 13, 14] . Although
thf're is still no agreement about the details of the structures, therf' is gelleral agrf'e-

97

ORIENTATlONAl
ORDERING
TRANSFORMATION
Monoclinic
Distorted

CONTINUOUS
TRANSFORMATION

SHEAR
TRANSFORMATION

HCP-1

..

..

Hep

HCP-2 _ _ _
..

Rhombohedral

..

.. FCC

~337 K

~27& K

cta=1.82

..

Cla:l.63-1.66

n=88-89°

t 0&~

[0101m

..

008

[0101m

t~

10001]hl

[0001] "2

i 0o~O i W
CJJ)
000

11 I 1] In

..

a = goo

[111] FCC

l@ l~

000

Figure 5.:3: Phase diagram showing the various phases of ('70. the possihle phase
transitions. and the stacking of the molecules [14].
ment that there are three regimes, the high temperature regime (above 1~l[ =:J40I\ J.
the intermediate temperature regime (275 1\ < T < :~40 1\). and the low temperature
regime (below T02 =275 1\). as shown in Figure G.:3. It is generally agreed t.hal the
structure is isotropic about TOI and becomes more anisotropic as T is decreased [9].
Five phases have been observed. including fcc [12]. rhombohedral [10]. ideal hcp (c / a

= 1.63) [12, 10], deformed hcp (c/a = 1.82) [16. 15] and a monoclinic phase [IOJ. Some
phase transitions occur over a wide temperature range and exhibit large hysteresis
which depends on the thermal history of the sample [10, 17J.
The X-ray diffraction patterns of all the samples are presented III Figure G.G.
Tlw diffradion pattern of the pure C 60 is indexed successfully as an fcc strllcture
with a = 14.21\, as expected [5, 6]. and it \vas nicely overlapped with t.1l(' simulated
pattern by Dr. Danut Dragoi with a software package Cry8tallographica by Oxford
('ryosystems. as shown in Figure 5.1.
For fullerite samples. the broad peaks between thE'

em (220) and (:ll 1) peaks are

98
contributed by C;o phases. As discussed above, thf' crysta.l structure of C;o if.; very
complex and some phase transitions may occur over a wide temperature range durillg
isotherm process.
Figure .5 ..5 shows that the diffraction pattern of C iO in fullerite #1 changed after
several isotherm cycles. Due to the broadening and the low intensity of the C;o pf'akf.;,
it is not possible for us to identify reliably the phases ill the samples. Howf'ver, t.1lf'
change in the diffraction pattern from q= 1.:3 to 1A A-1 indicates that sOllle parts of
the material have undergone a phase change after hydrogen cycling. Furthermore,
the diffraction peaks from the C 60 in fullerite # 1 \wcanlf' shaq)f'r aftcr cycling. 011
the other hand, the C 60 diffraction peaks from fullerite #'2 sharpellcd after cyding.
but therf' was little change in the region of q from 1.:3 to 1..1 A-I.
We believe tilf' high hydrogen adsorption of fullerite sample #1 IS a consequence
of a hydrogen-induced structural transition in the fullerite.

Carbon f.;ingle-walkd

nanotuhes (SWNT/s) undergo a hydrogen-induced structural phase transition [I xl.
Three energies are invol vecl. One energy is the energy of adsorption of the Il2 molecule
on the surface of the carbon. For S\VNT's this energy for hydrogen ph~/sisorption was
approximately the -:~8 meV that is characteristic of adsorption on graphite, and we
expect this adsorption energ.'r' to be similar for hydrogen adsorption on ('llO and (',0,
The second energy is the van del' Waals energy of cohesion of the (\lO and (',0 crystals.
Evidently these van del' \Vaals interactions in pure (\;u and (',0 arc sufficiently strollg
so that the crystals remain intact and the hydrogen sorption is limited t.o absorptioll
in interstitial sitf'S and adsorption OIl 1.1)(' relat.ively few surface sites.

The third

"energy" is the chemical potential of the hydrogen l1lolecules, which increases wit.h
the ]HeSf.;ure of hydrogen gas. It is possible to reduce this contribution to the total free
energy by surface adsorption of some of the hydrogen. The amollnt of adsorptioll will
increase with pressure as modeled by the Langmuir isotherm, for example. The phase
t.ransition in the S\VNT material was driven by this reduction in hydrogen cilPlllical
potentia.! during physisorption, which was sufficient to overcome the van del' vVaals
attraction between the tubes in a rope, separating thf'1l1 into individual tubes with a
large surface area for h.ydrogell adsorption.

99
Evidently this phase transition does not occur in the samples of pure (\m or <\·0,
and these materials remain intact because their van der Waals attractions are strong.
The van der Waals interaction and other electron-electron correlation effects responsible for cohesion decrease rapidly with distance, however. It is likely that crystallirlf'
defects in the C 60 /C TO fullerite #1, perhaps induced by tlw phase transitions ill the e TO
regions, or by oxidation, could reduce the cohesive energy of the fullerite so that t11f'
process of hydrogen adsorption became more competitive energetically. TIl(' shal)f'S
of the isotherms of sample #1 were not reproducible, indicating that its cohesive
energy was altered after a hydrogen sorption/desorption cycle. The x-ray diffraction
patterns of sample #1 also showed a change in structure, and this sample also showf'd
an unusual increase in surface area. (Although the increase in surface area is itself too
small to account for significant physisorption, we note that the surface area llleasul'nl
by BET was not for material in contact with high pressure hydrogen gas.
The isotherms of sample #1 are difficult to interpret because their shape changes,
but some of their features are noteworthy. The second isotherm run in Figure 0.1 a
may show some of tlw characteristics of the S\VNT isotherms, which ulldf'rWellt a
steep increase ill hydrogen adBorptiol1 once a critical pressure was attained. This
second run suggests a critical pressure in excess of 100 bar. The third, fourth, alld
fifth runs have some similarities to cOllventional adsorption isotherms, although t.heir
shapes are irregular and their characteristic adsorption pressure is not well defined.
This irregularity could be a consequence of the heterogeneity of the hydrogen-induced
structural phase transformation in impure fulleritp, and an accompanying change in
surface area resulting in a combination of lllacroporollS to microporous adsorbant
behavior [19]. It is of interest to note that the maximum va.!ue of adsorpt.ion attained
in the fourth isotherm corresponds to complete surface adsorption by all of the C no and
('TO

molecules. This implies that further improvements in the hydrogen adsorption

capacities of mixed fullerite materials lllay be difficult to achieve.

100

5.4

Discussion

\Ve suggest two possible origins for lower cohesive enf'rgy of fullerite # 1 than fullerite #2. The microstructural distribution of Coo. C 70 • and higher fulierf'ne molf'Cltles
may differ owing to differences in material preparation. ("ausing difff'H'ncf's in tiw
strudural transformations under tf'mperaturf' and hydrogen 1H"f'SSUIT.

Hf'call that

the chemica.! potential of hydrogf'Il gas is proportiona.l to the logarithm of its IHf'SSlHf'. so modest changf's in the cohesive elwrgy could cause large changes ill the
pressure for the hydrogen-induced phase transition. Differences in the breadth of til('
fcc egO peaks in the diffraction patterns are evidence for a microstructura.! difff'l'ence
betwf'en the fllllerites #1 and #2. and these peaks underwent an observablf' changf'
after fu lleri te # 1 was cycles.
The cohesive energy could also differ between fullerites # 1 and #2 becausf' of the
more extensive oxidation of the C 60 in fullerite #1. The cyclic exposure to hydrogen
gas reduced the oxidized ('60 in fulleritf' #1, perhaps causing an increase of its ("ohesivf'
f'nergy and the observed reduction of hydrogen storage capacity aftf'r 4 cycles. The
('60 oxide was absent aftf'r five cycles, but the hydrogen adsorpt.ion capacity of fllllnitf'

#1 rf'lllained large, so we do not attribllte all of tilt' difference between fllllerite
samples to oxidation.

\Ve do note, however, that oxidation has 1wen observed to

infhwncf' structural phase transitions in pure em [20] and sef'n to affect mechani("al
properties in enD [21] and electronic properties [22] in carbon nanotu\ws.
Comparf'd to S\VNT materials. mixed fulleritf's have much lower cost.. If t.he fl1lIf'riff' instabilities could be controlled to provide a largf' change in hydrogen adsorption
over a narrow range of pressure, a.s is possible for S\VNT materials, rulkrites may
be candidate materials for some hydrogen storage applications. The relationship betwef'll thf' structure and properties of the mixed fullf'rite samples remains uncf'riain,
howf'ver. Furthf'r inVf'stigation will be requirf'cl to identify the origin of tIl(-' structural
instability of fulleritf's likf' fulleritf' # 1.

101

160x10 3

140

311

120
422
222

100

333
331
420

-~

(b)

80

...!::

60

40

20
(a)

1_0

2_5

Figure 5.4: X-ray diffraction pattern of (b) pure C HO UJ9.9+<7c!) and (a) simulated with
Crystallographica.

'"rj

J.

-::;

::

.-

J.

'"'
'11

."tl

>-;

-c

.......,

'--'

.......,

gO

r")

:ll

::::11
>-;

0-

'<

:ll

>-;

't
U-.
U-.

c>-;

()q

CD
.......

en

.......

1.2
1 .4

1 .6

Q (A- 1 )

1 .8

C 70 98°/0

2.0

2.2

fullerite #1
(as-received)

fullerite #1
(cycled)

2.4

fullerite #2
(as-received)

f-"

Bibliography
[1] X. Lu, R. O. Loutfy, E. Veksler, and J. C. Wang, Pmc. of S!}mp. 011 RfcOl.t Ad!'.
in the Cfum. and Phys. of Fllilerenes and Related Matuial8, 823 (1998).

[2] H. W. Kroto, .J. R. Heath, S. C. O'Brien, R. F. Curl, and R. E. Smalley. Na/lln
(London), 318, 162 (U)85).
[:~] S. Saito and A. Oshiyama, Ph!}.'>. RfP. B, 44,115:32 (1991).

[4] C. C. AIm, Y. Ye, B. V. Ratnakumar, C. C. Witham, R. C. Bowman, and B.
Fultz. Appl. Phys. Lett. 73, :3:378 (1998).

[5] P. W. Stephens, L. Mihaly, P. Lee, R.. Whetten, S. M. Huang, H. I\:aller, F.
Deiderich, and I\:. Hokzer, Nahll'p (London), 351, 6:~2 (1991).

[6] A. R. I\:ortan, N. Kopylov, S. H. Glarum, E. M. Gyorgy, A. P. Ramirez, R. 1V1.
Fleming, F. A. Thiel, and R. C. Haddon. Natlll'e (London), 355, 529 (1991).
[7] Q. M. Zhang, .J. Y. Vi, and J. Bernhok, Phys. Rfl'. Lfft., 66, 26:n (1991).

[8] R. D. Johnson, C. S. Yannoni, H. C. Dorn, J. R. Salclll, and D. S. Uf't.hulle,
Science, 255, 12:35 (1992).

[9] M.

S.

Dresselhaus,

c.

Dresselhaus,

Science of Fullerenes and Carbon Nanotllbes,

and
Academic

P.
Press,

c.

Eklund,
San

Diego

(1996 ).

[10] A. R. lVIcGhie, J. E. FisdlPr, P. W. Stephens, R. L. Cappplletti, D. A. N<'lImaH\l,
W. H. l\Iueller, H. Mohn, and H. U. tel' i'vIepl'. Phys. Ret'. B, 49, 1261:1 (19!H).
[11] G. van Tpncleloo, S. Arnelinckx, M. A. Verheijell, P. H. M. van Loosdrecht., and
G. Meijer, Phy.':>. Rfl'. LEtt., 69, lOG!) (1992).

104
[12] G. B. M. Vaughan, P. A. Heiney, J. E. Fischer, D. E. Luzzi, D. A. Richetts-Foot.

A. R. Mcghie, U. W. Hui, A. L. Smith, D. E. Cox, W. J. Romanow, B. H. Allen.
N. Coustel, J. P. Mccauley, .Jr., and A. B. Smith III. Sciellce, 254. 1350 (1991).

[13] G. B. M. Vaughan, P. A. Heiney, D. E. Cox, .J. E. Fischer, A. 11. McGhie, A. L.
Smith, R. M. Strongian. M. A. Cichy, and A. B. Smith III. Chem. Phys. , 178.
.')99 (199:3).

[14] M. A. Verheijen. H. Meekes, G. Meijer, P. Bennema, .J. L. de Boer, S. van
Smaa.len, G. van Tendeloo. S. Amelinckx, S. lVluto and .J. van Landuyt. CII( In.
Phys., 166, 287 (1992).

[15] S. van Smaalen. V. Petricek ..J. L. de Boer. M. Dusek, M. A. Veriwijen, and G.
Meijer. ('hem. Phys. Lfft., 223. :tn (1994).

[16] E. Blanc. H. B. Burgi, R. Restori. D. Schwarzenbach. and Ph. Ochsenlwill. Frt.mphys. Lett., 33, 205 (1996).

[17] .J. E. Fischer and P. A. Heiney, J. Phy .... ClIrm. S'olids. 54. 1725 (1~)9:n.

[18] Y. Ye, C. C. AIm. C. Witham, B. Fultz,.J. Liu, A. G. RillZler. D. Colhert. T\". A.
Smith. and R. E. Smalley. Appl. Phys. Lfit .. 74, 2:307-2:309 (1999)

[19] l\I. D. Donohue and G. A1. Aranovich, Alit!. Colloid IntllJaCf Sci, 76-77 J:~7.
(1998 ).

[20] F. Yan and Y. Wang. J. Phys. Chr17l. Sol., 61, 100:3 (2000).
[21] Ll\lanika. J.lVIanika, and J. I\alnacs. Carboll. 36.641. (1998).
[22] P. G. Collins, 1\. Bradley, 1\1. Ishigami, and Z. Zettl, ScimCf, 287. 1801. (2000).

105

Chapter 6

Carbon Supported Catalysts

It has been reported that transition metals dispersed on the surface of activated

carhon may increase the hydrogen storage capacity. This occurs at a pressure of 011('
bar or greater at temperature near II K [1].
In recent years, a number of studies have been devoted to the phenomenon of
8pillm)f1', which is defined as the migration of adsorbed species from one solid phase

where it is easily adsorbed, onto another solid phase in contact with the first, where
it is not directly adsorbed. Hydrogen adsorbed on the metal crystallites can migrate
to the metal support in an activated form [6]. This effect could be extremely sensitive
to the impregnation procedure and to modifications due to the impregnation process.
Recently, a group of researchers at the Oak Ridge N at.ional Laboratory and t.he
Materials & Electrochemical Research Corporation claimed that t.llf'Y achieved 6 wt. %
hydrogen storage capacity around :200°(' using fullerenes and SOllle liquid organic
hydrides by adding liquid catalysts or doping alkali ions into fullerell{'s [8, 9]. This
capacity corresponds to C50H4s.
The work described in this chapter involved the preparation of nickel catalyst.s supported on fullerenes and activated carbon samples. Their hydrogen adsorption/desorption
isotherms were measured at various tempreatures.

6.1

Ni Catalyst Particle Size Control and Calculation

A fullerite sample obtained from MER in 1999 was used as the starting fullerite
for the support of impregnation. (This fullerite sample was denoted as fullerite #:3
in Chpater .5.) An activated carbon from Darco (I\:B-B-100), denoted as DAC. was
also used as the medium of a catalyst surport. The surface area of DA(' is 1550 m 2 /g

106
as measured by BET. The support was impregnated with a solution of nickelous
nitrate to obtain a designated metal loading. After impregnation, the catalyst. was
dried overnight in a fume hood at room telnperature and reduced in a furnace uuckr
hydrogen flow.
During impregnation of the solution, the concent.ration of the solution, the lllf'ta.l
compounds' readiness to be absorbed onto the substrate and capillarity will all contribute to the impregnation profile. During tllf' subsequent drying step, tlw s<'gregation and evaporation speed are important paranwters. For the activation st.ep.
conditions of calcination and reduction, such as temperature and gas composition,
need to be controlled properly. The detailed preparation techniques arf' described in
section 2.1.5. A total of ten NijDAC samples and several Nijfullerite samples Wf'l"e
prepared. The preparation conditions are listed in Table 6.l.

sample

NijDAC #1
NijDAC #:3
NijDAC #4
NijDAC #.1
NijDA(' #7
NijDAC #8
NijDAC #9
NijDAC #10
Ni/Fullf'rite

solution
concentration
(gjml)

Ni
composition
(wt. (1c1 )

reduction
tem perat ure
(Oe)

particle
diameter
(nIll )

0.1
0.05
0.025
0.02.1
0.0.5
0.1
0.05
0.1
0.05

rv 10
1.0

550°C
400°(,
500°('
·1GO°('
500°('
500°C
500°('
.500°('
500°C

rvGO
rv.1
rv5
rv9
rv 15
rv 10
rv I
rv8
rv 20

.)

Table 6.1: Preparation and characterization of carbon supported Ni catalyst particles.
To characterize the nickel particles, clark field transmission electron microscopy
was performed with a Philips EM4:30 transmission electron microscope operated at
200 kV. X-ray powder diffractol11etry was perfornlf'd on samples with an [nel CPS-l20
powder diffractometer using ('0 1\" radiation. The particle sizes were determined by
X-ray line broadening and direct meaSHrf'lllellt in t.he TEM illlages.

107

6.1.1

Nil Activated Carbon

A dark field TEM image of Ni/DAC #5 is shown in Figure 6.l. Nanosized nickel
particles were distributed on the surface of activated carbon . Although t he pore structure of the carbon is unclear, we expect there are numerous microposres, consistant
with the high specific surface area (1550 m 2 /g).

Figure 6.1: Dark field TEM imagf's of Ni catalyst particles supported on tllf' Darco
activated carbon.
X-ray diffraction patterns of Ni/ DAC samples arf' shown ill Figurf' 6.2 and t hf'
calculated part.icle sizes from t.he line shapes are summarized in Table 6.1. The large
particle size of Ni/DAC #1 was due to the high Ni composition. Difff'rillg rf'ductioll
temperature also resulted in the change of particle size and distribution. This suggests that Ni composition and rf'ductioll t.emperatu re arf' two clominant factors ill thf'
cat.alyst preparat.ion process.

108

Ni I Carbon #5

Ni I Carbon #4

Ni I Carbon #3
(111)
Ni I Carbon #1

(220)

(311)

20

40

100

80

120

25x10"

20

>.

"-'
......

15

#9

tI)

s:::

"-'

s:::

t-l

10

#8

1>7

20

40

80

60

100

120

28 (")

Figure 6.2: X-ray diffraction patterns of Ni/activated carbon samples.

6.1.2

Ni/Fullerite

The solubility of fullerenes in water is very low, thus the fullerit.e sample was 110t
readily to adsorb the nickelous component from its aquC01\S solut.ion. An ultrasonic
bath was employed to agitate and mix the sample thoroughly. Sonication treat.ment
is beneficial for catalyst preparation, catalyst activation, and cata.lytic reaction illvolving solid catalyst/reagents. It is normally considered that. t.he improvnnellt of
reactivity of solid catalysis was caused by tilE' creation of surface defects, t.he reduction of particle size, and the dispersion of the cata.lyst particles on the support [7].
The X-ray diffraction patterns of a set of Ni/fullerite samples are shown in Figure

109
6.:3. The particle size of the Ni catalyst was smaller after sonication t.rf'atment, whf'n
all other preparation parameters were the same.

(c)
(b)

(a)

'"

3000

(c)

2000

(b)

(a)
20

.00

2Theta

Figure 6.:3: X-ray diffraction patterns of Ni/fullerite sampks. (a) ful1erite frOl11 "esea.
(b) fullerite from MER, and (c) fullerite from MER with sonication used for catalyst
preparation process.

6.2

Results and Analysis

6.2.1

Isothenll Measureillent

Before mf'asuremf'nts of hydrogen adsorption and df'sorption. the samplf's were
degased at :200°(' oVf'rnight. The Sieverts' apparatus was thoroughly leak-checked
at :200 bar and calibrated to ensure l"f'liable deknnination of tlw hydrogen storage
properties.
Desorption measurements were performed at II h, :)00 K and 450 I~. H2 gas was
admitted into the evacuated reactor to achieve a typical pressure of 80 har. This
pressure was maintained for 15 hr to allow the sample to reach equilibrium ami to
check for If'aks in the system. The rf'actor was valvf'd off from the rest of the system
and thf' system was f'vacuatf'd again. The df'sor1wd Hl was tllf'1l df'terll1inf'd from t.he
system volume a.nd data measured by a pl"f'ssure transducer. Desorption isotherms of
fullf'rite and Ni/fullerite samples are shown in Figure G.!) and G.!).

110

--&- C 60 ,C70 (1 st run)
-fB- C SO ,C70 (2nd run)
-S- C SO ,C70 (3rd run)
~ Ni/C 60 ,C70 (1st run)
..... Ni/C so ,C 70 (2nd run)

0.4

~0.3

<.)
___ 0.2

0.1

0.0

20

40

6Q

Pressure [baq

80

100

Figure 6.4: Desorption isotherms of composition versus pressure at. 771\ for fullerite
and Nijfullerite samples.
Adsorption isotherm measurements were also performed manually. The sample
was first degassed at :200°(' and til(' manifold was evacuated. The reactor was valved
off before H2 was introduced into the rest of tlw syst.elIl and the presnre was recorded.
A valve was opened to allow the sample to make coutact with th, and the pressure
was recorded again after tequilibrium was achieved. This was one complete step of
the adsorption measnrenwnt. The reactor was valved off and the pr(,SSl1r(' of the
manifold was raised by introducing more hydrogen int.o the syst.elll. Tlwse steps w('re
repeat.ed and t.he pressure before and after the valve was opened was recorded unt.il
the pressure of the whole system reached about lOO bar. Some adsorption isot.herllls
are shown in Figure 6.6.

6.2.2

Isotherlll Results

The results of desorption and adsorpt.ion isot.herms at 77 I\", :300 I\", alld!!)O I\
are summarized in Table 6.:2. From Table 6.:2, we see several t.reuds. The desorption
capacit.v of the sample was lower than the adsorption cavacit.y of the same sample
at the same isotherm condit.ion. By adding Ni part.icles 011 the sample, the hydrogen

111

0.10

0.08

~0.06

0.04

__ Ni/fulierite (450K)
___ Ni/fulierite (300K, 1st run)
......... Ni/fulierite (300K, 2nd run)
""""*""""" fullerite (300K)

0.02

20

40

60

80

Pressure [bar]

Figure 6 ..5: Desorption isotherms of composition versus pressure at :~OOI~ and 4!)01~
respectively, for Ni/fullerite samples.
storage capacity of fullerit.e sample was increased. This was also observed for the
DAC samples. The Ni particles may adsorb some hydrogen, but this can llOt. account.
for t.he tot.al increased capacity even by assuming complete coverage of hydrogcll
molecules on the Ni particle surface. Assuming 5 wt. % of Ni particles are aggregat.ed
as spheres of 20 nm diameter; the complete coverage of hydrogell on the Ni particle
surface is equivalent to 0.01 wt.(J(, hydrogen storage capacity. We suggest that t.he
hydrogen adsorbed and dissociated on tlw surface of Ni part.icle may migrate onto
the surface of supporting carbon, which is called hydrogen :;pillol'rr.
The capacities of both Ni/fullerite and Ni/DAC samples at elevated temperature

Sample

OAt'
Ni/OAC
fullerite
Ni/fullerite

Hz Capacity
1(]77K (wt.%)
abs.
des.

Hz Capacit.y
1·(tJ:300K (wt%))
des.
abs.
0.2

o'r
.~I

0.2-0.:3
0.:3-0.4

0.1

0.06
0.08

Hz Capacity
1(]4!)OK (wt(X))
abs.
des.
0.04
0.02
0.4
0.12

0.2

0.1 L

Table 6.2: Hydrogen storage capacities of carbon samples with or without catalysts.

112
0.5
___ Ni/fullerite (300 K)
..... Ni/fullerite (455 K)
-+- Darco acrivated carbon (300 K)
Ni/OAC (300 K)
......... Ni/OAC (455 K)

0.4

"""*""

#0.3

.....

3:
..........
S:20.2

0.1

20

40

Pressure [bar]

60

80

Figure 6.6: Adsorption isotherms of composition versus pressure at :JOOI--: and 4})OK
respectively, for Darco activat.ed carbon, Ni/DAC, fullerite and Ni/fullerite samples.
(4})0 K) increased by a factor of two compared to those measured at amhient temperature. From the thermodynamics of adsorption, tht' fractional coveragt' of the sllrfacc
increases with decreasing temperature and increasing pressure. However, t.he fullerit<,
sample with Ni catalyst adsorbed more hydrogen at c!.1O K. This call be explained as
the better performance and higher acti vi ty of Ni catalysts at high tem peral. ure.

6.2.3

Fourier Transfornl Infrared Spectronletry

Since the amollnt of hydrogen desorbed was less than adsorbed, some hydrogen
molecules must have remained on the sample after the desorption.

To determine

tllf' mechanism of adsorption and check if there were any hydrogen carbon bonds
formed, infrared spectrum were measured with a 860 l'vIaglla series FTIR spectrollleter. Infrared spectroscopy is a st.andard clwmical analysis tool to idf'nti fy molecular
structures through molecula.r vibrations. The peaks around 2900 cm -1 are clue to (,H
bond stretching.
To obtain the t.ransmission spectrum, samples were prepared into potassium bromide pellets. Approximat.ely l mg sample and :WO mg KGr were ground and mixpd

in the mortar and pressed into a transparent lwllet in a press. The pellf't. was thell
placed in a pellet holder to obtain the spectrum. A background spectrum was also
obtained from an empty pellet holder.
A strong absorption at 2900 cm- 1 was observpd for the as-recf'ived salllpks of
both fullerite and Darco activated carbon. This indicatps that therf' lIlay he' sonw
contamination and/or a portion of CH bonds 011 the surface of carholl samples. It ,vas
report.ed that therf' are several types of hydrogen bOllds in coal, formed hy hydroxyls
with various hydrogen-bonding acceptors [10, 11].
After isotherm measurement at ambif'ut temperature. the fullf'rite sample sliovv<'d
a weaker absorption peak at 2900 ern- 1 . The absorption intensity of as-prf'pared
Ni/fullerite sample was also weaker than that of as-recf'ived sample. There is no significant absorption for the Ni/fullf'rite sample after isotherm at elevated temperaJurf'.
Before each hydrogf'n isotherm measnrenwnt. the sample was degassed at about

:200°('. This process may have driven off the contamination.

The reason for the

disa.ppeara.nce of CH bonds after high temperature hyclrogf'n isotherm is unclear.

6.3

Conclusion

The hydrogen storage capacity of fullerite and activatpd carbon sam pips was illcreased by adding Ni particles onto the sample. The adsorption of hydrogf'1l on Ni
particles cannot account for the total incrf'asf'd capacity. even by assuming complete
covprage of hydrogen 11101po11es on the Ni particle surfacp. The increasing of adsorptiOll capacity was much more significant at ('Ievaied temlwrature, suggesting improv('d
catalytic activity rather than improved thf'rmodYllamic adsorption.

114

,;-

Q)

·m>

LO

(.)

.-..

't

Q)

....

"0

@J
....E

Q)

....
CIS
a.
Q)
....
a.

Q)

L:

.!Q

rJ)

....

CIS

!~

':!(

"0

()

0......

~~

()

::::

....
Q)
.::::

Z~
':!(

rJ)

....

Q)

.0

CIS

(.)

Q)

III

CIS

..;J
(.)

:::l

c:
CIS
Q)

CIS

....(.)
CIS

<'l

III
<'l

III

III

c:i

a::lueqJOsqv

Figure 6.7: Fourier transform infrared spetrulTl of Darco activat.ed carbon alld
Ni/DAC samples.

IV,)

C')

co
,...

f/

~-:'<'1

-~-~"""~'7

It)

<;'

<\

tl

oC')
@J

LO

@J
"'"

1/

i/

C\J

'E

.s.....

(/l

....
Q)

!!

. ;:

Q)

Q)

.;:

.~

Q)

.D

Q)

o ::J
o c:

::J

Q)

Q)

Q)

;t:

It)

C\J

::J

.2

--Z

Q)

3:
-0

c:

.8~

It)

C')

C\J

C>

ci

<:I"

ci

C\J

ci

0<:1"

ci

a::>ueqJOsq\f
Figure 6.8: Fourier transform infrared spf'trum of fullerite sample and Ni/fullerite
samples.

116

Bibliography
[l} .J. A. Schwarz, "Metal Assisted Carbon Cold Storage of Hy#4,716,736, Issued .Jan. 5, 1988.

[2] 'l'u. A. Borisov, Yu. A. Zolotart'v, E. V. Laskatelev, and N. F. Myasoedov.
"Quantum-chemical calculation of a spillover model on a graphite support." Russian Chon. BlIllftin., 46.428 (1997).

[:3] A. Septilveda-Escribano. F. Coloma, and F. Rodrfguez-Reinoso, Appl. eatal. A:
(;fHfm/173 247 (1998).

[4] Z. X. Cheng, S. B. Yuan, .1. W. Fan,
III

Q. M. Zhu. and M. S. Zhen,

Spillover and Migration of Surface Species on Catalysts.

Elsevif'1'

Science

( 1997).

[.5] I.

Nakamura.

A.

Zhang.

Y.

Fan,

and

Spillover and Migration of Surface Species on Catalysts.

K.

Fllj illloto,
Elst'vier

III

Scienct'

( 1997).
[6] B. Mahipal Reddy. S. T. Srinivas. and P. K. Rao. hl.dioll J. of ChfTlI .. 31A.
957( 1992).

[7] S. Ley and C. Low, Ultrasonics in Synthesis. Springt'r-Verlag. Berlin Heiddberg
( 1989).
[8] .J. C. Wang, R. W. lVIurphy. F. C. Chen. R. O. Loutfy. E. Veksler. and W. Li.
"Hydrogen Storage in Fullerelws and in Organic H.ydrides." in Prorffdillg .... of II/(
1998 U. S. DOE Hydrogm Progmm Rfvifw, NREL/CP-570-269:38.

[9] .J. C. \Vang. R. W. Murphy, F. C. (,lwn, R. O. LOlltfy, E. Vekslt'r.. a.nd A. Singh.
"Hydrogen Storage in Ful\erenE's and Liquid Organic Hydrides." in Procudillgs

of the 1.9.9.9 U. S. DOE Hydl'Ogfll. Pl'Ogmm RfI'ifw, NHEL/CP-570-25:n5.

117

[1 OJ C. Chen ..J. Gao, and Y. Yan, Energy [i FuEls, 12,446 (1998).
[I1J R. W. Snyder, P. C. Painter, .1. R. Havens, and .J. L. Koenig, Appl. SPfctro .... c.,
37, 497 (198:3).

118

Chapter 7

Conclusions and Outlook for

Further Work
7.1

Conclusions

The hydrogen storage properties of some traditional carbon materials (activated
carbon) and novel ones (fullerenes, graphi te nanofibers and ca.rbon nanoi ubes) vvere
explored. The results are summarized in Table 7.1.

Carbon
samples
AX-21
Saran car bOll
GNF
SvVNT
C 60

Specific
Surface Area
(mz/g)
:3000
1600
20-:30
285

liz Capacity
wt% 77K
ads.
des.
2.2
0.15
4-7
0.8
0.7
O.:~

C/O
Fulleri te # 1
Fullerite #3
Nijfullerite
DalTO activated C
NijDAC

0.9-11
3-4
1550

0.6-4
0.2-0.:3
0.:3-0.4

lh Capacity
wt% :300K
ads.
des.
0.5
00'1
0.1-0.2
0.8
0.07
0.08
0.12
0.12

O. t

0.06
0.08

0.2

o'r
.~I

H2 Capacity
weYr) 450I~
ads.
des.

0.2
0.04
0.4

0.11
0.02
0.12

Table 7.1: Summary of hydrogen storage capacities and surface areas of carbon samples.
As a result of our analysis, it seems unlikely that carbon in nanofiber form shows
Hz adsorption/ desorption properties that would have the spectacular impact as a
solid state storage medium claimed by Rodriguez's group at Northeastern University.
Thus there is no point in following up with graphite nanofibers.
A phase transition bebveen crystal SWNT and a new hydride phase was found
at high pressures at 80K. The phase transition was of first order. and involved t.he

119

separation of the individual tubes within a rope, expOSIng a high surface area for
hydrogen adsorption. From the change in chemical potential of the hydrogell gas IIpOll
adsorption, we were able to calculate the cohesive van der Waals energy between the
tubes as .5 meV Ie atom. This is much smaller than expected from previous theoretical
work, and shows that defects in the crystal structure cause large suppressions of the
cohesive energy. We were able to alter this cohesive energy by ,hanging the state of
the material.
Hydrogen desorption and adsorption properties of fullerene materials C 60 , and
C TO and fullerite were measured. Over several cycles of isotherm measurements at

77 1\, the hydrogen storage capacities of one of the fullerite samples in(Tea~·wd from
an initial value of 0.4 wt% for the first cycle to a capacity of 4.2 wt% for the fourth
cycle. Correspondingly, the surface area increased from 0.9 m 2 /gm to 11 \11 2/glll.
and showed a phase transformation, characterized by X-ray powder diffraction. In
comparison. two other fullerite samples, prepared by a different procedure, showed no
such behavior. Pure C no and pure ('TO were ,ycled and exhibited small and constant
capacities of 0.7 wt% and 0.:3:3 wt%, respectively. as a function of number of cyelps.
The enhanced storage capacity of fullerite material is tentatively attributed t.o the
presence of C60 oxide.
Compared to SWNT materials. mixed fullerites have much lower cost. If the fullerite instabilities could be controlled to provide a large change in hydrogen adsorption
over a narrow range of pressure, as is possible for SWNT materials, fulkrites may
be candidate materials for some hydrogen storage applications. The relationship between t.he structure and properties of the mixed fullerite samples remains uncert.ain,
however. Further investigation will be requirt'd to identify the origin of the structural
insta.bili ty of fulleri tes like fulleri te # 1.
By adding Ni particles onto the sample, the hydrogen storage capacity of fullf'rite
and activatf'd carbon sample was increased. The adsorption of hydrogen on Ni part.icle
can not account for the total increased capacity even by assuming complf'te coverage
of hydrogen molecules on the Ni particle surface.

120

7.2

Further Work

Theoretically, the likely upper limits of hydrogen adsorption in various forms of
carbon can be considered as t,hat under conditions where H2 molecules are ahle to
form two close-packed layers within each graphite plane, that the atomic ratio of 11:('
approaches 1:1 (rv 8 wt%). This value is consistent with the capacity of SWNT at II
K and it is the best capacity among all the materials listed in Table 7.l.
However, capillary effect is possible for the condensation of hydrogen into nanotube. By altering the diameter of SWNT, one may find the capillarity varies. thus
optimizing the hydrogen storage capacity. Adding catalyst particles onto SWNT may
even improve its hydrogen storage property.
Carbon materials are not promising for the H2 storage at :~OO I\:. it seems unlikely
that carbon nanofiber shows any spectacular H2 adsorption/desorption properties,
thus there is no point in following up with graphite nanofiber works. SWNT's are
interesting but expensive for hydrogen storage at 77 I\. Fullerites arc cheaper and
showed some performance. However, sample-to-sample variation. which was caused
by oxidation, will require further work to understand. Carbon supported catalyst system may have promise to service at elevated temperatures. The amount of hydrogell
storage is small. but there seems to be an effect worth pursuing in future work.