Full text of DOI:10.1557/JMR.2002.0375

Electron field emission from Ar+ ion-treated thick-film
carbon paste
Gillian A.M. Reynolds, Lap-Tak Cheng, Robert Bouchard, Paul Moffett, Howard Jones, and
Linda F. Robinson
DuPont Central Research and Development, Wilmington, Delaware 19880
S. Ismat Shah
Department of Materials Science and Engineering and Department of Physics and Astronomy,
University of Delaware, Newark, Delaware 19716
Daniel I. Amey
DuPont Electronic Technologies, Research Triangle Park, North Carolina 27709
(Received 14 January 2002; accepted 11 July 2002)
Ion bombardment was used to produce electron-emitting microscale features on
surfaces of thick films printed with carbon pastes. This technology can potentially
enable the development of large-area field emission displays. Systematic investigations
using microscopy and electron field emission experiments have demonstrated a close
link among paste formulation, ion processing parameters, and the development of
surface microstructures. These investigations were also useful in understanding the
fundamentals of microstructure formation under ion bombardment and the field
emission characteristics of the carbon-based emitters. Several device concepts aimed
toward achieving a low-voltage switchable triode were also tested with varying degrees
of success. In this paper, we discuss various technological issues related to the
materials, processes, and devices.
I. INTRODUCTION
on the order of several microns in length and exhibit
electron emissive properties under an external electric
field. The effects of such parameters as ion bombardment
time, substrate temperature, and oxygen exposure on the
morphology and resulting emission are also discussed.
There is intense research underway to develop vacuum
electronics and flat panel displays based on field emis-
sion. The original idea, as presented by Spindt,1 involved
Si microtips. The creation of Spindt tip field emitters
requires several lithographic steps, and not only are the
resulting tips susceptible to mechanical damage but they
can be easily contaminated by exposure to the atmos-
phere. These sensitivities require fabrication and particu-
larly operating pressures better than 1 × 10−6 torr. The
current field emissive material of choice is carbon in a
variety of forms including, for example, graphite, carbon
nanotubes, diamond, and various mixtures thereof.2–5
Several techniques have been used to create the nano-
structures associated with these carbon-based field emit-
ters. These include chemical vapor deposition,6,7 pulsed
laser deposition,8,9 epoxy mixing,10 etc. Structures are
also known to form on a variety of surfaces as a result of
ion bombardment. These structures take the form of
cones, hills, or whiskers. Feature formation has been ob-
served on ion bombarded metal surfaces that have been
seeded with another type of metal.11 Ion bombardment of
carbon surfaces has also been shown to produce conelike
features and carbon nanostructures such as nanotubes.12–15
In this paper, we discuss the formation of cones from
a thick-paste graphite film upon exposure to Ar+ ions.
The cones formed as a result of the ion bombardment are
II. EXPERIMENTAL
The conelike features and resulting electron emitters
were fabricated using a two-step process of screen-
printing and Ar+ ion bombardment of thick-film carbon
paste materials. The screen-printed samples were derived
from paste mixtures of inorganic powders (including
carbon-based powders like graphite and carbon black),
organic polymers, solvents, and surfactants. The inor-
ganic powders contained a relatively low melting glass to
ensure sintering of the composite and adherence to the
substrate. A standard paste composition of 50/50 by
weight inorganic to glass was used in this work. The
simplest emitter structure made was a diode. In this con-
figuration, a 1-mm-thick and 2.5-cm-square soda-lime
glass was used as a substrate. A silver layer was screen-
printed onto the glass substrate and prefired at 525 °C to
form a 6–7-␮m-thick electrode. The carbon paste was
then screen-printed on top of the silver electrode to a
thickness of 10–20 ␮m and fired at 475 °C in air. The
2590
J. Mater. Res., Vol. 17, No. 10, Oct 2002
© 2002 Materials Research Society

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
firing removed all the organic constituents of the paste
temperature at 100 °C. After ion bombardment, emission
currents from the 1-cm2 exposed areas were measured in
a diode configuration with the ion-treated graphite paste
acting as the cathode and an indium tin oxide (ITO)
coated phosphor screen acting as the anode. The diodes
were operated at 1 × 10−6 torr by applying pulsed nega-
tive voltages to the cathode at duty cycles in the range of
approximately 0.02% to approximately 1% (pulse widths
of 3 to 40 ␮s and frequencies of 60 to 100 Hz). For the
diode configuration, the cathode–anode spacing was on
the order of 1 to 1.25 mm.
The sample morphologies were studied using scanning
electron microscopy (SEM) and transmission electron
microscopy (TEM). For SEM [Hitachi S-800 field-
emission scanning electron microscopy (FE-SEM, Mito
City, Japan)], imaging was performed at 4-kV accelerat-
ing voltage at a 15-mm working distance and 2500 mag-
nification. The features produced on the samples’ surface
material and left behind only the inorganic layer. The
samples were then subjected to an Argon plasma using an
8-cm Kaufman source. During the ion bombardment, the
samples were covered with a graphite mask in order to
expose only a 12 × 12 mm (approximately 1 cm2) area
to the ions. Typically, samples were positioned at either
a 90° or 45° angle to the incoming ion beam and treated
under 1.5 × 10−4 torr of argon pressure for 30 min at
1200-eV beam energy, 120-mA beam current, approxi-
mately 2-mA/cm2 current density, and 40-V discharge
voltage under total beam neutralization conditions. A
heating stage was built into the system which allowed
control over the substrate temperature (TS) during ion
bombardment. Values for TS could be varied from 100 to
450 °C. Generally, without temperature control, TS
reached approximately 350 °C during ion bombardment.
Active cooling of the substrate was needed to keep the
FIG. 1. Morphology (a) before and (b) after etching for 30 min in 1200-eV argon ion. The morphology consists of cones of approximately 3 ␮m
in height. Features form in the direction of the incident ion beam.
FIG. 2. SEM and pulsed diode field emission (1.44-cm2 emissive area) of an etched graphite paste sample (3 ␮s, 60 Hz). The morphology consists
of whiskers supported by cones of approximately 2 ␮m in height. Features form in the direction of the incident ion beam.
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2591

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
(effective emissive area 1.44 cm2) running at 0.02% duty
cycle with a current density of 4.5 ␮A/cm2 at 3 V/␮m.
Figure 3 shows the reproducibility of the I–V curves for
five samples running at approximately 0.02% duty cycle
and etched under similar conditions for 45 minutes. The
turn-on field is in the range of 1.5–2 V/␮m, with a cur-
rent density of approximately 5 ␮A/cm2 at 3.5 V/␮m.
These turn-on voltages are comparable to carbon nano-
tube devices in the literature. The direct current (dc)-
equivalent current densities are higher than some carbon
nanotube diode devices cited in the literature.2,16 Using a
1-megapixel camera to expand the emission from a 1 ×
1 cm area, emission spots were not distinguishable, sug-
gesting emission site densities on the order of 104 to
105 sites/cm2. This estimate is reasonable when com-
pared to site densities obtained via scanning anode field
emission microscopy in the literature.17
The microstructure of the whiskers and cones was
studied by high-resolution TEM. Figure 4 shows the
HRTEM of a cone with a whisker extending from the tip
of the cone. The whisker is generally amorphous while
the supporting cone has an ordered outer layer (approxi-
mately 10-nm wide) covering an amorphous interior. The
ordered outer shell on the cone consists of graphene
planes which are oriented perpendicular to the cone
growth direction. Generally, in a carbon filter, graphene
planes grow parallel to the fiber growth direction. This
microstructure could point to regrowth as a possible
mechanism for the cone structure formation.
The cones form as a result of the surface chemical or
morphological inhomogeneities. Carbon has a very low
sputter yield.18 The glass frit around the carbon-rich
area sputters at a higher rate leaving behind carbon re-
gions that, under anisotropic etching, form conical
shaped regions. The orientation of the cones in the di-
rection of the ion beam also supports the preferential
etching process as a possible mechanism for cone for-
mation. In this case, the physical etching generated by
ion impact defines the direction of cone growth by ef-
fectively eroding cones that are growing off angle to the
ion trajectory. This results in the cones always growing,
as observed, in the direction of the ion beam. On the
other hand, the growth of whiskers, which are several
microns in length, could not be explained solely on the
basis of preferential sputtering. A simultaneous growth
process has to occur. Also, the presence of the oriented
crystalline outer layer (Fig. 4) cannot be explained as due
purely to a sputter-etching process. Most likely, cone and
whisker dimensions stem from a combination of etching
and regrowth processes. Thus, the physical etching, due
to preferential sputtering, defines the orientation and
shape of the cones and a concurrent surface atom migra-
tion and vapor-phase redeposition of sputtered carbon
assists in whisker formation. To further study the cone
formation process, ion bombardment was performed as a
FIG. 3. I–V curves (duty cycle approximately 0.02%) from sam-
ples treated with an Ar+ ion beam under similar conditions (1200 eV,
120 mA, 45 min) at a substrate temperature of 450 °C.
by ion bombardment were evaluated further by high-
resolution transmission electron microscopy (HRTEM)
(Philips CM-20 TEM, Eindhaven, The Netherlands) us-
ing an accelerating voltage of 200 kV. To achieve this,
surface layers of the ion beam etched carbon were gently
scraped using clean razor blades. The scraped material
was deposited directly onto perforated carbon films sup-
ported on 3-mm TEM Cu grids.
III. RESULTS AND DISCUSSION
A. Mechanisms of, and factors affecting,
morphology formation
Figures 1(a) and 1(b) are SEM images of the morphol-
ogy of the screen-printed carbon paste before and after
Ar+ ion bombardment, respectively. Before bombard-
ment, the morphology is inhomogeneous, with flakes of
glass frit and graphite visible on the surface as light and
dark regions, respectively. These untreated films exhibit
no electron emission. After ion bombardment, the mor-
phology consists of whiskers, when visible, supported by
larger triangular-shaped cones. The cones, which form in
the direction of the ion beam, are on the order of 2 to
4 ␮m in height, while whiskers can be tens of microns
long. Figure 2 shows the cone morphology (SEM) and
some of the earliest emission results from these struc-
tures. The emission is from a 2.5-cm-square sample
2592
J. Mater. Res., Vol. 17, No. 10, Oct 2002

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
function of time by treating samples for 15 to 45 min
cone density or height could be changed by external pa-
rameters, the system could be tailored to the desired
emission. To determine the extent to which the ion-
treated graphite paste morphology and resulting device
emission could be controlled, experiments were con-
ducted to study the effects of contaminants, substrate
temperature, and ion bombardment time on the properties.
We have found that cone formation is possible on a
variety of carbon surfaces such as graphite and carbon
black. However, the presence of contaminants (i.e., non-
graphitic components) can destroy cone formation.
Figure 6 shows the microstructure of variable width
stripes of graphite and glass frit (50/50 by weight) printed
on a pure glass background. Here the glass acts as the
contaminant. As the stripe width is decreased from 2 mm
to 254 ␮m, the degradation in cone formation is clearly
evident. Since the sputtering yield of low melting oxide
is much higher than that of carbon, for every carbon atom
under similar ion source conditions. Figure 5 shows the
resulting morphologies. For times as short as 15 min
[Fig. 5(a)], the inhomogeneous surface becomes pitted
and acts as nucleation sites for the emergent short cones
that are visible. As the ion bombardment time increases
to 30 and 45 min [Figs. 5(b) and 5(c), respectively], the
overall surface morphology remains inhomogeneous
while the cone densities and cone aspect ratios increase.
The overall morphology of the ion-bombarded graph-
ite paste material is fairly inhomogeneous. Cone heights,
thickness, separations, and radii of curvature vary within
and between samples. An ideal morphology would be
one in which all the cones were uniform in structure and
height with an even distribution over the emitting sur-
face. The ability to control the morphology to this degree
would result in an emitter where the majority of the sites
contributed to the emission process. If the variations in
FIG. 4. (a) High-resolution TEM of a whisker and cone. (b) Extent of the ordered outer layer of the cone (approximately 10-nm wide), as indicated
by the arrows.
FIG. 5. Showing the effects of Ar+ ion bombardment time on morphology. Samples were treated under similar conditions for (a) 15, (b) 30, and
(c) 45 min.
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2593

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
removed by ion etching, five or more oxygen atoms are
features and seed cones, thus decreasing cone density and
sharpness. As the ratio of exposed oxide/carbon surface
area increases, cone formation deteriorates rapidly. Also,
the excess glass, during ion bombardment, recondenses
on the cone surface, as shown in the high-resolution
generated. This source of reactive oxygen represents an
isotropic chemical etching process that competes effec-
tively with the anisotropic physical etching of ion bom-
bardment. Chemical etching tends to round off any sharp
FIG. 6. Degradation in cone formation with decreasing graphite paste stripe width. Graphite paste stripe width of (a) 2 mm, (b) 762 ␮m, and (c)
254 ␮m.
FIG. 7. High-resolution TEM of cones created from (a) an all-graphite paste surface and (b) a graphite surface near a pure oxide edge.
FIG. 8. Effects of contaminants on emission. The emission improves with decreasing amount of dielectric contaminant (bottom to top). Black
regions represent graphite, and white regions, dielectric.
2594
J. Mater. Res., Vol. 17, No. 10, Oct 2002

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
TEM of Fig. 7(b). This HRTEM image is compared to an
Spindt tips. An ion-treated sample can also tolerate re-
peated exposure to atmosphere without any observable
degradation in the emission properties.
all graphite paste cone surface, Fig. 7(a). In the all graph-
ite paste, the cone surface is free of a glass coating. We
have observed that surfaces such as those in Fig. 7(b) will
not emit as efficiently as those in Fig. 7(a). This con-
tamination effect is shown in Fig. 8, which consists of a
sample with 25–127 ␮m of exposed rings of dielectric
(white regions) surrounding graphite (black regions). For
the lines with large areas of exposed dielectric (contami-
nant), after ion bombardment, emission is absent from
the graphite holes. Emission occurs in the etched graphite
line in regions far enough away from the exposed dielec-
tric. The emitting line becomes more continuous as the
amount of exposed dielectric and its influence on mor-
phology contamination are reduced.
We have studied the effects of substrate temperature,
TS, during ion bombardment on the morphology and
emission. The results in Fig. 9 show the diode
pulsed emission from samples ion treated at TS ס 100,
200, 300, and 450 °C. There is an increase in emission
current and uniformity with increasing TS. There are no
distinct differences in the morphologies at the SEM level
in samples treated at different temperatures. It is possible
that the cone structures are becoming more graphitic in
the outer layer with an increase in temperature. The state
of the graphitic outer layer within the cones created in
these experiments is yet to be analyzed via HRTEM.
Tests done on many ion-treated samples have shown
phenomena such as the existence of a distribution of
emitter threshold voltages. In pulsed diode mode, as the
anode voltage is increased, emitters turn on according to
their relative work function characteristics. The emitters
generally remain turned on for about 2 V/␮m beyond
their threshold voltage. After this point the population of
emitters with lower threshold burn out. This burnout is
often accompanied by arc discharge at hot spots on the
sample which are associated with high current. Figure 10
shows the I-V data for a diode device for two voltage
ramps. On the first voltage ramp, the low turn-on emitters
burn out which results in a higher turn-on voltage for the
second voltage ramp. As long as the burn-out voltage is
B. Device characterization
The electron field emission properties of the ion
treated surfaces and the overall performance of the di-
odes made from these samples were characterized by
operating devices in a pulsed mode. In general, emission
from the ion-treated thick-film samples does not deterio-
rate after exposure to atmosphere. However, physical
contact with the ion-treated surface can cause damage to
the microstructure and the resulting emission. The emit-
ters can be operated at moderate vacuum at pressures on
the order of 1 × 10−5 torr, two orders of magnitude higher
in background pressure than that required by traditional
FIG. 9. Pulsed emission (duty cycle approximately 0.02%) from samples (1.44-cm2 emissive area) that were Ar+ ion treated with substrate
temperatures, TS, of (a) 100 °C, (b) 200 °C, (c) 300 °C, and (d) 450 °C.
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2595

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
not exceeded for the low turn-on emitters, the emission
characteristics of the samples can be retraced on subse-
quent voltage ramps.
4 kV. Triode switching in the form of bright lines against
a dark background was observed at cathode voltages (VC)
as low as −300 V (Fig. 12). At −400 V, an emission cur-
rent of about 14 ␮A was measured at 2% duty cycle.
Field emission lifetime studies have also been per-
formed on the diode emitter devices. When the emitters
are driven at high duty cycle (>5%) and with high elec-
tric fields, accelerated degradation of emission sites and
emission uniformity have been observed. However,
when the duty-cycle and drive field are maintained below
these limits, emission currents remained stable for sev-
eral hundred hours of operation. Figure 11 shows the
lifetime tests of a diode device with a copper anode op-
erating constantly at 4 kV at a duty cycle of 0.8% (40-␮s
pulse, 200 Hz). These parameters represent a high stress
test for the emitters, corresponding to about 30 times the
typical display end use duty cycle at peak voltage and
about 15,000 h of run time.
The feasibility of the emitters to perform as true field
emission devices was determined by fabricating a simple
wire-gate triode system. This design separates the triode
structure into two subassemblies: the cathode, containing
graphite emissive elements, and a wire gate electrode. A
1 × 1 cm all graphite surface was ion treated to develop
surface structures and was used as the cathode. The
sample was checked for diode emission. A 127-␮m-thick
polyimide tape was placed on two parallel edges of the
glass substrate to serve as stand-offs. Wire of 64-␮m
diameter and coated with 2 ␮m of varnish was wound
onto the glass substrate to serve as the gate electrode. A
phosphor-coated ITO glass was used as the anode with a
cathode to anode spacing of 2 mm. The cathode was
connected to a pulsed negative voltage power supply, and
the gate wire and anode were connected to earth ground
(via an electrometer) and to a positive-dc high voltage
power supply, respectively. The onset of background
emission was measured to be 4.5 kV. To keep the emit-
ters in the off state, the anode voltage, VA, was kept at
FIG. 11. Emission current as a function of time for a diode device run
at 4 kV and 0.8% duty cycle and at pressures of 1 × 10−6 torr.
FIG. 10. How the population density decrease on a second voltage
ramp leads to higher turn-on fields.
FIG. 12. Triode emission from a 7-wire sample.
2596
J. Mater. Res., Vol. 17, No. 10, Oct 2002

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
FIG. 13. (a) Diode emission and triode emission from a 7-segment display operating at 20 ␮s and 60 Hz with (b) 127-␮m gate to cathode spacing
with VC ס −200 V and VG ס 120 V, (c) 51-␮m gate to cathode spacing with VC ס −120 V and VG ס 30 V, and (d) 51-␮m gate to cathode
spacing with VC ס −90 V and VG ס 60 V.
To further study the wire gate triode concept, a 7-
segment alphanumeric cathode was also fabricated. A
wire-gate electrode was made with the gate to cathode
spacing defined by the thickness of the polymer film
spacer used, as described above. A 7-terminal relay was
built to facilitate the individual operation of each seg-
ment on the cathode. The cathode was connected to a
pulsed negative voltage power supply, while the anode
was connected to a positive-dc high voltage power sup-
ply. The anode to cathode spacing was 2.75 mm. To
reduce the switching voltage required at the cathode, the
gate was biased with a positive voltage, VG. Figure 13
shows the triode performance of this device at three cath-
ode to gate spacings. As this spacing was decreased from
127 to approximately 51 ␮m, VC decreased from −300 to
−90 V, thus demonstrating sub-100-V switching of a
field emission display (FED) triode device fabricated
with graphite thick-film paste technology. The seg-
mented device shows good contrast, stable operation, and
minimal image spreading despite the cathode to anode
spacing. Improvements need to be made with respect to
the uniformity of the segments and poor image edge
definition.
low anode voltages are achievable with this emitter
material technology. Aging studies of up to several hun-
dred hours at 30 times application duty cycle also
demonstrated considerable emitter durability. Using a
wire-gated concept, triode operation with sub-100-V
switching was observed.
ACKNOWLEDGMENTS
The authors particularly thank Gail Raty for her work
with the SEM images. We also acknowledge the help of
Dr. S. Subramoney (HRTEM), A. West, B. Farquhar, and
S. Harvey.
REFERENCES
1. C.A. Spindt, I. Brodie, L. Humphrey, and E.R. Westerberg,
J. Appl. Phys. 47, 5248 (1976).
2. F. Ito, K. Konuma, and A. Okamoto, J. Appl. Phys. 89, 8141
(2001).
3. W.A. de Heer, A. Chatelain, and D. Ugarte, Science 270, 1179
(1995).
4. V.I. Merkulov, D.H. Lowndes, and L.R. Baylor, Appl. Phys. Lett.
75, 1228 (1999).
5. D. Zhou, A.R. Krauss, T.D. Corrigan, T.G. McCauley,
R.P.H. Chang, and D.M. Gruen, J. Electrochem. Soc. 144, 224
(1997).
6. C. Wang, A. Garcia, D.C. Ingram, M. Lake, and M.E. Kordesh,
Electron. Lett. 27, 1459 (1991).
7. W. Zhu, C.P. Kochanaski, S. Jin, and L. Seibles, J. Appl. Phys.
78, 2707 (1995).
8. A.A. Voevodin and M.S. Donley, Surf. Coat. Technol. 82, 199
(1996).
9. G.T. Visscher, D.C. Nesting, J.V. Badding, and P.A. Bianconi,
Science 260, 1496 (1993).
IV. CONCLUSIONS
We have created an electron field emitter by using ion
bombardment to transform the morphology of screen-
printed graphite thick-film pastes. Adequate emitter site
density and uniformity, high brightness, and contrast at
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2597

G.A.M. Reynolds et al.: Electron field emission from Ar+ ion-treated thick-film carbon paste
10. P.G. Collins and Z. Zettl, Phys. Rev. B 55, 9391 (1997).
15. W.A. Solberg and I.L. Spain, J. Vac. Sci. Technol. A 8, 3907
11. G.K. Wehner, J. Vac. Sci. Technol., A 3(4), 1821 (1985).
12. J.A. Floro, S.M. Rossnagel, and R.S. Robinson, J. Vac. Sci. Tech-
nol. A 1, 1398 (1983).
(1990).
16. K.H. Park, K.M. Lee, S. Choi, S. Lee, and K.H. Koh, J. Vac. Sci.
Technol. B 19, 946 (2001).
13. Z. Wang, F. Zhu, W. Wang, G. Yu, and M. Ruan, Appl. Phys. A
17. L. Nilsson, O. Groening, P. Groening, O. Kuettel, and
71, 353 (2000).
L. Schlapbach, J. Appl. Phys. 90, 768 (2001).
14. J.A. Van Vechten, W. Solberg, P.E. Batson, J.J. Cuomo, and
S.M. Rossnagel, J. Cryst. Growth 82, 289 (1987).
18. S. Ismat Shah, in Handbook of Thin Film Process Technology,
edited by S. Ismat Shah and D. Glocker (IOP, Bristol, U.K., 1995).
2598
J. Mater. Res., Vol. 17, No. 10, Oct 2002