Hysteresis in the field emission (FE) data of a chemical vapor synthesized carbon nanotube fiber cathode is analyzed in the regime where self-heating effects are negligible. In both the forward and reverse applied field sweeps, various FE modes of operation are identified: including Fowler-Nordheim (FN) tunneling and space-charge limited emission from the fiber tip and FN emission from the fiber sidewall. Hysteresis in the FE data is linked to the difference in the field enhancement factors in the different FE modes of operation in the forward and reverse sweeps and related to changes in the fiber morphology.

Advanced cathode emitters are needed to produce the stable, high current electron beams needed for compact, high power, high frequency vacuum electronic devices. Current state of the art thermionic cathodes require high operating temperatures, resulting in inefficient power consumption, poor reliability, and diminished lifetime. The addition of cooling components leads to additional device complexity and added weight, both of which are undesirable. New cathode materials are needed for high current density applications including microwave power amplifiers, miniature X-ray sources, micro-machines mass spectrometers, and ion propulsion, among others.1–3 

Carbon nanotubes (CNTs) have proven to be excellent field emitters due to their high aspect ratio as well as their high electrical and thermal conductivity.4,5 Recent research has focused on fabricating CNT sheets, yarns, and fibers in an attempt to transfer their exceptional physical properties from the nanoscale to the macroscale.6–8 CNT yarns that are spun from vertically grown CNT arrays have been studied for use as field emission (FE) sources,9–12 and there have been some recent reports on the exceptional FE properties of acid spun CNT fibers.8,13–16

There have been several observations of hysteresis in the FE data of CNT arrays, porous diamond like fibers, CNT fibers (CNFs), and graphene films.17–19 The onset of hysteresis has been linked to various phenomena, such as emission via intermediate electron energy states for nanocarbon films20 and the electrostatic alignment of various atomic and radical species on the emitting surface of porous diamond-like carbon21 and adsorption/desorption effects in CNTs.17,21,22 It has been shown that hysteresis in the FE data depends on gas exposures, growth conditions, and conditions of the cathode materials.21,23

In this report, we analyze the hysteresis properties of laser cut CNT fibers that were fabricated by direct online condensation from a floating catalyst chemical vapor deposition (CVD) reactor.24,25 In the forward and reverse sweeps of the cathode to anode bias, we identify FE modes of operation including Fowler-Nordheim (FN) tunneling and space-charge limited emission from the fiber tip and FN emission from the fiber sidewall, which starts from the tip and proceeds toward the base with increasing applied field. We show that the field enhancement factor in each regime of operation is related to changes in the fiber morphology.

The process for fabricating the CNT fibers has been previously covered15,24,25 and is briefly described here. Aligned CNT fibers were drawn directly from a floating catalyst chemical vapor deposition reactor as a one-step CNT generation process. Toluene was introduced into a hydrogen atmosphere accompanied by ferrocene. This gas mixture flowed through a reaction furnace (at 1200 °C) where the constituents broke down to form floating catalytic nanoparticles. From these nanoparticles, CNTs grew and formed collectively as an aerogel cloud which was mechanically drawn directly from the reactor and collected on a spinning wheel. This process allows for the synthesis of CNT fibers made from predominantly single-walled CNTs with a significant reduction of carbonaceous impurities compared to the material from the conventional process. The CNTs had single wall signatures with only a few chiralities present, as well as a high degree of graphitic crystallinity (Raman G:D ratio of approximately 15). The CNT fibers were cut with the use of a fiber laser (SPI Lasers, Model G3 SM) operating at 1062 nm ±3 nm with a repetition frequency up to 25 kHz, pulse duration of 200 ns, and average power of 20 W. This was coupled through a galvanometer scanner to a Photonics Devices 03–90FT-125–1064 lens (focal length 125 mm) and scanned at a speed of 150 mm/s.

FE characterization from a laser cut, 30 μm diameter CNT fiber was carried out in an ultrahigh vacuum chamber whose base pressure was 4.0 × 10−7 Pa. A fiber cathode was created by mounting a 5 mm long section of the fiber to a stainless steel sample holder with silver paint. A stainless steel anode probe tip (7 mm diameter) was aligned with the CNT fiber with the use of two orthogonally situated cameras looking through two different windows on the chamber. One camera was equipped with a long working distance objective to accurately determine the anode-cathode gap distance d. The gap d was varied with integrated stepper motors capable of 2.5 μm travel per step. With the gap distance set and the cathode grounded, the voltage on the anode was first increased, with a Keithley 2410 source meter, at a rate of 1 V per 10 s from 300 V to the maximum of 1000 V (forward sweep). After holding the potential fixed at 1000 V for 2 min, it was decreased to 300 V (reverse sweep) at the same rate as for the forward sweep. FE data were recorded at each voltage setting under LabView control; once a complete sweep (forward and reverse) was carried out, the gap distance d was reduced, and the data acquisition process was repeated. Thermal images of the fibers were captured during the FE experiments with an infrared camera that was attached to a long distance microscope mounted on the FE chamber. For a d = 1.75 mm at a maximum bias voltage of 1000 V, the temperature at the tip of the fiber was found to be less than 500 °C, indicating that self-heating effects were minimal for that distance d setting.

Figure 1(a) is a plot of the FE data for both forward and reverse sweeps at different values of the gap distance d. The size of the hysteresis loops in the FE data increases with decreasing gap. As will be discussed below, this is due to increasing changes in fiber morphology which affects the FE properties of the emitting CNTs. Scanning Electron Microscope (SEM) images of the CNTs on the tip and sidewall of the fiber are shown in the supplementary material accompanying this letter.26 For d = 1.5 mm, there is a sudden drop in the emission current around Eext = 0.63 V/μm. We attribute this drop to self-heating effects which lead to a sudden failure to emit for some of the CNTs forming the fiber. In fact, for d = 1.25 mm, shortly after applying the maximum voltage of 1000 V, the fiber abruptly stopped emitting entirely.

If FN FE is taking place, the current-external electric field characteristics should be well fitted by an expression of the form27 

I(A)=Aeff1.54×106φβ2Eext2e6.83×103φ3/2βEext,
(1)

where Aeff is the effective area (in μm2) of the emitting fiber, φ is the work function (in eV) of the emitting surface, β is the field enhancement factor, and Eext is the externally applied electric field (in V/μm), i.e., Eext=Vd, where V is the applied bias (in V) between anode and cathode (assumed to be same as between tip of the fiber and the anode since the potential drop along the fiber is negligible), and d is the distance (in μm) between the anode and the base of the CNT fiber cathode.

According to Eq. (1), all the FE data of the CNT fiber, when plotted as ln(I /Eext2) vs 1/Eext, should fall on the same straight line if FN emission prevails, and if there is no change in the fiber morphology with variations in d. Figure 1(b) shows the ln(I /Eext2) vs 1/Eext plots for the fiber for different values of d during the forward sweep. It can be seen that the plots for d = 2 and 1.75 mm (curves (a) and (b)) are fairly close to each other, except for large values of the applied voltage. We attribute the divergence of the two curves beyond 3.7 μm/V to back bombardment from the ionized desorbed species (H2 and CO15) damaging the CNTs on the fiber tip. The maximum emitted current (at an applied bias of 1000 V) was found to be 0.64, 1.15, 1.6, and 2.2 mA for the distance d equal to 2, 1.75, 1.5, and 1.25 mm, respectively. The FN plots for d = 1.5 and 1.25 mm (curves (c) and (d)) are markedly different from those for d = 1.75 and 2 mm and indicate a change of fiber morphology as the gap distance decreased. This indicates that some of the sharper features contributing to FE at the large gap distance were destroyed at the largest applied fields, leaving CNTs with lower field enhancement factors and thereby causing an increase in the threshold for FE in subsequent sweeps. This destruction can arise from a combination of self-heating effects and ion back bombardment. The importance of the latter is supported by residual gas analysis data described in the supplement.

Figure 2 is a closer look at the FE data for d = 1.75 mm, a distance representing the regime in which self-heating effects are negligible and in which the fiber should be operated to avoid permanent damage. For the forward sweep (Fig. 2(a)), the FE data are well fitted by the FN expression over the range of 1/Eext from 3.0 to 3.4 μm/V. This fit is labeled “FN1” in Fig. 2(a). This regime is followed by a rather flat region in the FE data over the range of 1/Eext from 2.5 to 2.9 μm/V. As shown in Fig. 2(a), the FE data in this regime are in good agreement with the Child-Langmuir (CL) equation28,29 indicative of space-charge limited transport. This is labeled “CL1” in Fig. 2.

We explain the FN1 and CL1 modes of operation as follows. Beyond threshold, FE occurs mostly from the fiber tip, where the density of electric field lines is the largest and the local electric field at the end of the CNTs on the fiber tip is larger than the applied electric field due to the enhancement factor. This leads to FN emission mostly from the tip. As the current through the fiber increases, the emission current eventually saturates due the limited electric conductivity of the fiber (8600 S/m), and CL space-charge limited emission from the tip prevails. In fact, we have observed that a portion of the tip exhibits a visible glow as the fiber enters the CL regime, and the intensity of the glow increases with applied bias, suggesting the formation of a plasma near the tip.

Also indicated in Fig. 2(a) are the values of the effective enhancement factor β extracted from Eq. (1) by forcing the FN expression to agree with the experimental emission current. As the external electric field increases in the FN1-CL1 regime of operation, β progressively decreases from 12 800 (at 1/Eext = 3.25 μm/V) to 10 800 (at 1/Eext = 2.5 μm/V). This reduction is due to the progressive shielding of the applied external field with the onset of space-charge effects on the tip of the fiber. Therefore, the CL regime of operation can be thought of as a FN mode of operation with an electric field dependent enhancement factor. Figure 2(b) shows a plot similar to Fig. 2(a) with a transition from CL to FN modes of operation during the reverse sweep.

Over the range of Eext displayed in Figures 2(a) and 2(b), the relation between β and Eext was found to be of the form 1/β = a + b Eext with (a) and (b) equal to (2 × 10−5, 2 × 10−4) and (2 × 10−5, 3 × 10−5) for the forward and reverse sweeps, respectively. This relation explains the weak dependence on Eext of the exponential factor in Eq. (1) once the CL mode of operation is reached. In the range of 1/Eext where the FN mode prevails, the β values are larger for the reverse sweep. If β is estimated using the formula h/r, where h and r are the average length and radius of the emitting CNTs, the larger β values for the reverse sweep can be explained as arising from the progressive alignment of the CNTs along the applied electric field during the forward sweep. As a result, past the regime CL1 of operation in the forward sweep, a larger fraction of the CNTs on the tip stand up and remain aligned during the reverse sweep, leading to the larger β values calculated in the FN mode of operation. Another potential effect increasing the field enhancement factor of some CNT tips is the effect of ion back bombardment which, if not too drastic, can help to open some of the CNT caps and sharpen their tips.

On the forward sweep, beyond the CL regime of operation, the density of the electric field lines near the tip tends to saturate and starts to increase along the fiber sidewall, progressing down the fiber shaft with increasing Eext. This assertion is supported by calculating the current emitted from the fiber sidewall (Isw) by subtracting from the measured emission current the value extrapolated from the CL fit beyond the electric field value 1/Eext = 2.0 μm/V, as illustrated by the inset of Figure 3(a). A FN plot of the FE emission data past this threshold in region 2 is shown in Fig. 3(a) for the forward sweep. A FN plot of the sidewall emitted current Isw vs Eext data is found to be in good agreement with the FN expression over the 1/Eext range from 1.85 to 2.18 μm/V (fit FN2 in Fig. 3(a)). This indicates FE from some CNTs on the sidewall of the fiber. At a maximum applied bias of 1000 V (0.57 V/μm), over 80% of the emission current was found to come from the sidewall as indicated in the inset of Figure 3(a).

We have also observed an increasing number of bright spots along the sidewall of the fiber past the CL1 regime of operation. On the forward sweep, the number of bright spots was found to increase and propagate down the fiber until the spots finally reached the fiber base at large values of the applied bias. Images of the fiber before the start of the FE measurements and at a voltage of 1000 V are shown in the supplement. As the number of spots increased and got brighter, the FE data were found to be in good agreement with the CL expression over the Eext field range from 1.75 to 1.85 μm/V (fit CL2 for forward sweep in Fig. 3(a)), indicative of space-charge limited transport from the surface emitting areas of the fiber sidewall. Figure 3(a) also contains the effective β values over the range corresponding to FE from the sidewall. The effective β values are about a factor of 2 smaller than in region 1 and increase with increasing Eext. This is attributed to an increase in the FN contribution from CNTs further away from the tip on the fiber sidewall as FE from the sidewall closer to its tip starts to be space-charge limited. The overall smaller β values in region 2 also indicate that different populations of CNTs emit from the tip and sidewall of the fiber. At large anode voltage, the space-charge limited mode of operation over the larger fraction of the fiber sidewall eventually leads to an increase in self-heating effects in the entire core of the fiber. Eventually, the fiber fails to emit if the gap distance is too small, as we have observed in Fig. 2(a). Figure 3(b) is a FN plot of the sidewall emission current on the reverse sweep in region 2 indicated in Fig. 1(a) showing no evidence of a CL mode of operation. In fact, the emission current was found to decrease by 0.1 mA over the 2 minute period during which a maximum voltage of 1000 V was applied. We attribute this decrease to a deterioration of the FE properties of the fiber tip. As a result, FE occurs over a larger fraction of the sidewall, and the FN mode of operation prevails.

In conclusion, chemical vapor synthesized CNT fibers were characterized as field emitters. FE data show that the fiber first emits in agreement with FN emission from its tip at low anode voltage, followed by three regimes of operation, including space-charge limited transport from the tip of the fiber, followed by FN emission from the sidewall of the fiber, and finally a regime of space-charge limited transport from the sidewall. The increased hysteresis in the FE data with reduced gap distance is indicative of a change in the morphology of the fiber due to a progressive destruction of the emitting CNTs as a result of increasing self-heating effects.30 Achieving higher maximum emitted current (several mA) should be attainable by reducing the electrical resistivity and increasing the thermal conductivity of the fibers.

This work was supported by Air Force Contract No. FA8650-11-D-5401 at the Materials & Manufacturing Directorate (AFRL/RXAP). The authors thank Gregory Kozlowski for helping to establish this collaboration and John Luginsland at AFOSR in addition to Scott Dudley and Victor Putz of EOARD for supporting this work

1.
W. I.
Milne
,
K. B. K.
Teo
,
E.
Minoux
,
O.
Groening
,
L.
Gangloff
,
L.
Hudanski
,
J.-P.
Schnell
,
D.
Dieumegard
,
F.
Peauger
,
I. Y. Y.
Bu
,
M. S.
Bell
,
P.
Legagneux
,
G.
Hasko
, and
G. A. J.
Amaratunga
,
J. Vac. Sci. Technol. B
24
,
345
(
2006
).
2.
H. M.
Manohara
,
R.
Toda
,
R. H.
Lin
,
A.
Liao
,
M. J.
Bronikowski
, and
P. H.
Siegel
,
J. Infrared Milli. THz. Waves
30
,
1338
(
2009
).
3.
S. H.
Heo
,
A.
Ihsan
,
S. H.
Yoo
,
G.
Ali
, and
S. O.
Cho
,
Nanoscale Res. Lett.
5
,
720
(
2010
).
4.
J. M.
Bonard
,
M.
Croci
,
C.
Klinke
,
R.
Kurt
,
O.
Noury
, and
N.
Weis
,
Carbon
40
,
1715
1728
(
2002
).
5.
N.
Perea-López
,
B.
Rebollo-Plata
,
J. A.
Briones-León
,
A.
Morelos-Gómez
,
D.
Hernández-Cruz
,
G. A.
Hirata
,
V.
Meunier
,
A. R.
Botello-Méndez
,
J. C.
Charlier
,
B.
Maruyama
,
E.
Muñoz-Sandoval
,
F.
López-Urías
,
M.
Terrones
, and
H.
Terrones
,
ACS Nano
5
,
5072
(
2011
).
6.
K. K. K.
Koziol
,
J.
Vilatela
,
A.
Moisala
,
M.
Motta
,
P.
Cunniff
,
M.
Sennett
, and
A.
Windle
,
Science
318
,
1892
(
2007
).
7.
R. M.
Sundaram
,
K. K. K.
Koziol
, and
A. H.
Windle
,
Adv. Mater.
23
,
5064
(
2011
).
8.
N.
Behabtu
,
C. C.
Young
,
D. E.
Tsentalovich
,
O.
Kleinerman
,
X.
Wang
,
A. W. K.
Ma
,
E. A.
Bengio
,
R. F. ter
Waarbeek
,
J. J. de
Jong
,
R. E.
Hoogerwerf
,
S. B.
Fairchild
,
J. B.
Ferguson
,
B.
Maruyama
,
J.
Kono
,
Y.
Talmon
,
Y.
Cohen
,
M. J.
Otto
, and
M.
Pasquali
,
Science
339
,
182
(
2013
).
9.
Al. A.
Zakhidov
,
R.
Nanjundaswamy
,
A. N.
Obraztsov
,
M.
Zhang
,
S.
Fang
,
V. I.
Klesch
,
R. H.
Baughman
, and
A. A.
Zakhidov
,
Appl. Phys. A
88
,
593
(
2007
).
10.
P.
Liu
,
Y.
Wei
,
K.
Liu
,
L.
Liu
,
K.
Jiang
, and
S.
Fan
,
Nano Lett.
12
,
2391
(
2012
).
11.
Y.
Wei
,
D.
Weng
,
Y.
Yang
,
X.
Zhang
,
K.
Jiang
,
L.
Liu
, and
S.
Fan
,
Appl. Phys. Lett.
89
,
063101
(
2006
).
12.
G.
Chen
,
D. H.
Shin
,
S.
Kim
,
S.
Roth
, and
C. J.
Lee
,
Nanotechnology
20
,
315201
(
2009
).
13.
D.
Shiffler
,
S.
Fairchild
,
W.
Tang
,
B.
Maruyama
,
K.
Golby
,
M.
LaCour
,
M.
Pasquali
, and
N.
Lockwood
,
IEEE Trans. Plasma Sci.
40
,
1871
(
2012
).
14.
V.
Guglielmotti
,
E.
Tamburri
,
S.
Orlanducci
,
M. L.
Terranova
,
M.
Rossi
,
M.
Notarianni
,
S. B.
Fairchild
,
B.
Maruyama
,
N.
Behabtu
,
C. C.
Young
, and
M.
Pasquali
,
Carbon
52
,
356
(
2013
).
15.
P. T.
Murray
,
T.
Back
,
M.
Cahay
,
S. B.
Fairchild
,
B.
Maruyama
,
N. P.
Lockwood
, and
M.
Pasquali
,
Appl. Phys. Lett.
103
,
053113
(
2013
).
16.
S.
Fairchild
,
J.
Bulmer
,
M.
Sparkes
,
J.
Boeckl
,
M.
Cahay
,
T.
Back
,
P. T.
Murray
,
G.
Gruen
,
M.
Lange
,
N. P.
Lockwood
,
F.
Orozco
,
W.
O'Neill
,
C.
Paukner
, and
K. K. K.
Koziol
, “
Field emission from laser cut CNT fibers and films
,”
J. Mater. Res.
29
,
392
402
(
2014
).
17.
C.
Li
,
G.
Fang
,
X.
Yang
,
N.
Liu
,
Y.
Liu
, and
X.
Zhao
,
J. Phys. D: Appl. Phys.
41
,
195401
(
2008
).
18.
V. P.
Mammana
,
T. E. A.
Santos
,
A. P.
Mammana
,
V.
Baranauskas, Helder
,
J.
Ceragioli
, and
A. C.
Peterlevitz
,
Appl. Phys. Lett.
81
,
3470
(
2002
).
19.
S. L.
Yue
,
C. Z.
Gu
,
C. Y.
Shia
, and
C. Y.
Zhib
,
Appl. Surf. Sci.
251
,
215
(
2005
).
20.
A. V.
Arkhipov
,
M. V.
Mishin
, and
I. V.
Parygin
,
Surf. Interface Anal.
39
,
149
(
2007
).
21.
S. C.
Lim
,
H. J.
Jeong
,
Y. S.
Park
,
D. S.
Bae
,
Y. C.
Choi
,
Y. M.
Shin
,
W. S.
Kim
,
K. H.
An
, and
Y. H.
Lee
,
J. Vac. Sci. Technol. A
19
,
1786
(
2001
).
22.
K. H.
Park
,
S.
Lee
, and
K. H.
Koh
,
Diamond Relat. Mater.
14
,
2094
(
2005
).
23.
V. V.
Zhirnov
,
C.
Lizzul-Rinne
,
G. J.
Wojak
,
R. C.
Sanwald
, and
J. J.
Hren
,
J. Vac. Sci. Technol. B
19
,
87
(
2001
).
24.
Y. L.
Li
,
I. A.
Kinloch
, and
A. H.
Windle
,
Science
304
,
276
(
2004
).
25.
C.
Paukner
and
K. K.
Koziol
,
Sci. Rep.
4
,
3903
(
2014
).
26.
See supplementary material at http://dx.doi.org/10.1063/1.4900787 for hysteresis during field emission from chemical vapor deposition synthesized carbon nanotube fibers.
27.
A.
Modinos
,
Solid-State Electron.
45
,
809
(
2001
).
28.
C. D.
Child
,
Phys. Rev. (Series I)
32
,
492
(
1911
).
29.
30.
P.
Vincent
,
S. T.
Purcell
,
C.
Journet
, and
V. T.
Binh
,
Phys. Rev. B
66
,
075406
(
2002
).

Supplementary Material