letters to nature
visible range 400±800 nm. A 100-W tungsten ®lament lamp
(observed distribution median1) emits 8 W in the visible (with its
®lament at 3,000 K), matching this nanosphere model at 1,350 K.
The l dependence of Si emission at 1,200 K matches a 1,700 K
blackbody pro®le in the visible, so such a ball would be reported as
`translucent white'. Lower temperatures may be reported as yellow
or red, thus covering most of the observed colours. Optical emission
may also come from molecules of salts vaporized from the soil.
Many soils are expected to yield SiO rather than Si. Amorphous Si
may also be formed, with a higher absorptivity than if crystalline25.
(We note that some lightning strikes (or man-made discharges) on
metal structures may also result in a long-lived oxidation of
networked metal nanospheres.)
The model of a nanoparticle network of slowly oxidizing Si that
we present here successfully explains all the salient observed features
of ball lightning in its most common environment. Future work
may usefully explore the lightning/soil interaction, the chemical
properties of the soil (or soil/wood mix) and lightning parameters
(strength, duration), to ®nd those which produce ball lightning in
the laboratory.
M
Received 1 September; accepted 10 November 1999.
1. Smirnov, B. M. The properties and the nature of ball lightning. Phys. Rep. 152, 177±226 (1987).
2. Barry, J. D. Ball Lightning and Bead Lightning (Plenum, New York, 1980).
3. Turner, D. J. The structure and stability of ball lightning. Phil. Trans. R. Soc. Lond. A 347, 83±111
(1994).
4. Singer, S. The Nature of Ball Lightning (Plenum, New York, 1971).
5. Hutchison, S. G., Richardson, L. S. & Wai, C. M. Carbothermic reduction of silicon dioxideÐa
thermodynamic investigation. Metall. Trans. B 19, 249±253 (1988).
6. Essene, E. J. & Fisher, D. C. Lightning strike fusion: extreme reduction and metal-silicate liquid
immiscibility. Science 234, 189±193 (1986).
7. Siegel, R. W. in Physics of New Materials 2nd edn (ed. Fujita, F. E.) 66±106 (Springer, Heidelberg,
1998).
8. Ludwig, M. H. in Handbook of Optical Properties Vol. II, Optics of Small Particles, Interfaces and
Surfaces (eds Hummel, R. E. & Wissmann, P.) 103±127 (CRC, Boca Raton, 1997).
9. Friedlander, S. I., Jang, H. D. & Ryu, K. H. Elastic behaviour of nanoparticle chain aggregates. Appl.
Phys. Lett. 72, 173±175 (1998).
10. Fortov, V. E. et al. Highly nonideal classical thermal plasmas: experimental study of ordered
macroparticle structures. JETP 84, 256±261 (1997).
11. Aleksandrov, V. Ya., Borodin, I. P., Kechenko, E. V. & Podmoshenskii, I. V. Rapid coagulation of
submicron aerosols into ®lamentary three-dimensional structures. Sov. Phys. Tech. Phys. 27, 527±529
(1982).
12. Cawood, W. & Patterson, H. S. A curious phenomenon shown by highly charged aerosols. Nature 128,
150 (1931).
13. Aleksandrov, V. Ya., Golubev, E. M. & Podmosheskii, I. V. Aerosol mode of ball lightning. Sov. Phys.
Tech. Phys. 27, 1221±1224 (1983).
14. Lowke, J. J., Uman, M. A. & Liebermann, R. W. Toward a theory of ball lightning. J. Geophys. Res. 74,
6887±6898 (1969).
15. Deal, B. E. & Grove, A. S. General relationship for the thermal oxidation of silicon. J. Appl. Phys. 36,
3770±3778 (1963).
16. Massoud, H. Z. & Plummer, J. D. Analytical relationship for the oxidation of silicon in dry oxygen in
the thin-®lm regime. J. Appl. Phys. 62, 3416±3423 (1987).
17. Jacobson, N. S. Corrosion of silicon-based ceramics in combustion environments. J. Am. Ceram. Soc.
76, 3±28 (1993).
18. Rayle, W. D. Ball Lightning Characteristics (Note TND-3188, NASA, Washington, 1966).
19. Andrianov, A. M. & Sinitsyn, V. I. Erosion-discharge model for ball lightning. Sov. Phys. Tech. Phys. 22,
1342±1347 (1977).
20. Pemberton, S. T. & Davidson, J. F. Turbulence in the freeboard of a gas-¯uidised bed. The signi®cance
of ghost bubbles. Chem. Eng. Sci. 39, 829±840 (1984).
21. Nagamori, M., Boivin, J. A. & Claveau, A. Gibbs free energies of formation of amorphous Si2O3, SiO
and SiO2. J. Non-Cryst. Solids 189, 270±276 (1995).
22. Littau, K. A. et al. A luminescent silicon nanocrystal colloid via a high-temperature aerosol reaction. J.
Phys. Chem. 97, 1224±1230 (1993).
23. Bohren, C. F. & Huffman, D. R. Absorption and Scattering of Light by Small Particles (Wiley, New York,
1983).
24. Palik, E. D. Handbook of Optical Constants of Solids (Academic, Orlando, 1985).
25. INSPEC Properties of Silicon 75, 938 (Institution of Electrical Engineers, London, 1988).
Acknowledgements
We thank the Electrical and Electronic Engineering Department at the University of
Canterbury, especially J. Woudberg, for the use of their high-voltage laboratory; C. Maslin,
B. Lane, P. Niamskul and T. Benson for experimental work, supporting that of J. Dinniss,
and D. Brown, N. Foot and R. Boyce for technical help. We also thank N. Andrews and
J. McKenzie of the Plant and Microbial Sciences Department for electron microscopy.
Correspondence and requests for materials should be addressed to J.A.
(e-mail: j.abrahamson@cape.canterbury.ac.nz).
NATURE | VOL 403 | 3 FEBRUARY 2000 | www.nature.com
.................................................................
Large-scale complementary
integrated circuits based on
organic transistors
B. Crone*, A. Dodabalapur*, Y.-Y. Lin*, R. W. Filas*, Z. Bao*, A. LaDuca²,
R. Sarpeshkar*³, H. E. Katz* & W. Li*
* Bell Laboratories, Lucent Technologies, 600 Mountain Avenue, Murray Hill,
New Jersey 07974, USA
² Lucent Technologies, 555 Union Boulevard, Allentown, Pennsylvania 18103,
USA
³ Present address: Department of Electrical Engineering, Massachusetts Institute
of Technology, Cambridge, Massachusetts 02139, USA
..............................................................................................................................................
Thin-®lm transistors based on molecular and polymeric organic
materials have been proposed for a number of applications, such
as displays1±3 and radio-frequency identi®cation tags4±6. The main
factors motivating investigations of organic transistors are their
lower cost and simpler packaging, relative to conventional inorganic electronics, and their compatibility with ¯exible substrates7,8. In most digital circuitry, minimal power dissipation
and stability of performance against transistor parameter variations are crucial. In silicon-based microelectronics, these are
achieved through the use of complementary logicÐwhich incorporates both p- and n-type transistorsÐand it is therefore
reasonable to suppose that adoption of such an approach with
organic semiconductors will similarly result in reduced power
dissipation, improved noise margins and greater operational
stability. Complementary inverters and ring oscillators have
already been reported9,10. Here we show that such an approach
can realize much larger scales of integration (in the present case,
up to 864 transistors per circuit) and operation speeds of ,1 kHz
in clocked sequential complementary circuits.
Our fabrication method for organic complementary circuits
spatially separates n-channel and p-channel transistors to facilitate
the separate deposition of electron-transporting and hole-transporting semiconductors. Other approaches are possible, and an
alternative method will also be outlined. In the experiments
described below, hexadeca¯uorocopper phthalocyanine (F-CuPc)
is used as the n-type semiconductor and a-sexithiophene (a-6T)
as the p-type semiconductor. The chemical structures of these
a
S
b
S
S
S
S
S
F
F
F
F
F
F
F
N
Cu N
N N
F
F
N
F
F
F F
F F
N
F
N
N
c
Organic semiconductor
Source
Drain
Dielectric
Gate
metal
Via
Metal
Substrate
d
S/D metal
Upper metal
Via
F- C u Pc
Gate insulator
Flexible PWB
Lower metal
Figure 1 Chemical structures of organic semiconductors and schematic layer structures
of circuits. a, Structure of a-sexithiophene (a-6T); b, structure of hexadeca¯uorocopper
phthalocyanine (F-CuPc). c, Schematic layer structure of the circuits used in this study.
The circuits are fabricated on rigid substrates, and metal lines and vias are de®ned by
photolithography and standard semiconductor processing techniques. The interconnection metal level is de®ned immediately above the substrate, followed by the gate metal
level and the source/drain (S/D) metal level. The interlayer dielectrics are not shown. The
organic semiconductor is deposited above the S/D metal. d, Schematic of the crosssection of an organic transistor circuit technology based on ¯exible printed wiring boards.
© 2000 Macmillan Magazines Ltd
521
letters to nature
materials are shown in Fig. 1a and b. Both F-CuPc (ref. 11) and a-6T
(refs 12, 13) are examples of relatively well ordered organic
semiconductors with mobilities in excess of 10-2 cm2 V-1 s-1 in
thin-®lm form. There are other possible choices of well ordered
organic and polymer semiconductors, particularly for forming the
p-channel transistors. Some alternatives include pentacene (ref. 3),
regioregular poly(thiophene) (refs 1, 2), and oligothiophenes other
than a-6T.
The vertical circuit layout that we employ is based on an `upsidedown' con®guration (Fig. 1c), in which the interconnect metal level
is de®ned ®rst, followed by the gate metal level, and ®nally the
source/drain metal level. The gate and source/drain metal levels are
also used for local interconnections. This layout con®guration
differs from that employed in Si-based circuits and in the polymer
circuits reported by Drury et al.5, wherein the transistors are formed
®rst followed by the interconnect metallization. Figure 1c shows the
schematic of the circuits formed on rigid substrates that we
employed for this work, whereas Fig. 1d shows the schematic of a
process that we have developed for circuits based on ¯exible plastic
substrates. Standard ¯exible printed wiring board processes allow
the fabrication of two levels of metal on either side of an insulating
substrate and through-board conductors (vias) which connect these
two levels. The application of a thin dielectric (which functions as
the gate dielectric) above the upper metal level, together with the
de®nition of a third metal level (source/drain, S/D, metal), results in
a layer structure with a cross-sectional topology similar to that
shown in Fig. 1c. This is an example of an approach to the
fabrication of ¯exible circuits that is essentially an extension of
existing technology and is both reliable and low-cost.
The operation of discrete devices as well as circuits depends
critically on the metallization employed for the sources and drains.
We have found that a combination of electroless Ni deposition
and immersion Au coating14 works very well for both n-channel and
p-channel thin-®lm transistors (TFTs). These metals are deposited
from appropriate solutions over pre-patterned materials that can be
sensitized for electroless deposition. Five-stage ring oscillators have
an oscillation frequency of 10 kHz for devices with channel lengths
of 7.5 mm, providing a useful index for the speeds of circuits that can
be made. Although this is the highest frequency reported so far for
ring oscillators based on organic semiconductors, it can be
increased further by shrinking the channel length and/or employing
higher-mobility semiconductors. Reducing device and stray capacitances will also be helpful in improving circuit speeds.
The design of logic circuits was based on a particular type of
complementary logic known as pass transistor logic15. The essential
circuit building blocks are an inverter and a transmission gate, as
shown in Fig. 2. These building blocks have been combined to form
latches and D ¯ip-¯ops (Fig. 2e). The use of pass transistor logic
a
b
P-TFT
Out
N-TFT
In
CLKN
e
CLK
c
results in space and power-dissipation reduction, as fewer transistors are required to implement most functions. The relatively simple
layouts of pass transistors logic are convenient for implementation
on ¯exible plastic substrates.
Complementary circuits that we have tested include row decoders
and shift registers. The largest circuit that was evaluated was a 48stage shift register with 24 output buffers. The total number of
transistors in this circuit is 864, and the channel length of each
transistor is 7.5 mm. Each stage of the shift register is a D ¯ip-¯op,
with the output of one stage connected to the input of the next. The
clock and its complement drive all the stages. Every second stage
has an output buffer, which consists of two inverters with large
transistors to facilitate probing without loading the circuit. Shift
registers are ubiquitous in digital systems and can perform many
functions. One function is to shift a `bit' in an orderly and
predictable manner from one stage to the next every clock cycle.
The operation of the shift register described above is illustrated in
Fig. 3, in which the clock, data, and output of the 24 output buffers
are plotted as a function of time. The data consist of a single bit,
which is sequentially shifted over all stages of the register. We have
operated two-stage shift registers at clock rates of up to 1 kHz.
The operating voltage is 80 V because of the thick gate dielectric
employed. By utilizing thinner or higher-k dielectrics16, the operating voltage as well as the power dissipation can be reduced.
The static current drawn by the 48-stage register, including the
output buffers, is 70 mA, and that drawn by a smaller two-stage
register (with 1 buffer per stage) is 7.6 mA. In comparison, two-stage
complementary shift registers based on NOR/NAND gates required
more than twice as many transistors as the two-stage shift register
based on pass transistor logic, and drew 20 mA. Simulations indicate
that the current drawn by shift registers based on p-channel
transistors alone is greater than that of complementary shift
registers of comparable transistor dimensions and speed. The
static current drawn in p-channel TFT registers depends strongly
on the nature of the load (whether enhancement or depletion),
the on-off current ratio, and the ratio of the dimensions of
the transistors that constitute the basic gates (NOR, NAND and
inverter). The static current drawn by complementary registers can
be further reduced by reducing the off-current of the transistors.
The lower power dissipation of complementary circuits will be an
important factor in an application such as radio-frequency identi®cation tags, which are powered by the ambient ®eld.
The schematic of a three-bit row decoder is shown in Fig. 4a. This
decoder is designed so that a single serial input activates one of eight
outputs. It consists of three D ¯ip-¯ops and a NOR array (Fig. 4b).
The output of the three ¯ip-¯ops and their complements drive the
NOR array which is con®gured so that for any given set of inputs
only one output is `high'. The characteristics of the decoder are
d
In
Out
CLK
CLKN
Q
D
CLKN
CLK
Figure 2 Representation of circuits and their constituents. a, A complementary inverter;
b, symbol of an inverter; c, layout of a complementary transmission gate; d, symbol of a
transmission gate; e, schematic of a D ¯ip-¯op showing the data input (D) and the
connections of the clock (CLK) and its complement (CLKN). Also shown is the output (Q).
522
Figure 3 Characteristics of the 48-stage shift register. The clock (500 Hz) and data are
shown along with the output voltages of the 24 output buffers as a function of time. The 24
outputs have been vertically offset for clarity.
© 2000 Macmillan Magazines Ltd
NATURE | VOL 403 | 3 FEBRUARY 2000 | www.nature.com
Clk (V)
letters to nature
a
D Q
Flip-Flop
D Q
Flip-Flop
D Q
Flip-Flop
QN
QN
QN
MSBN MSB
b
7
6
7
6
5
SBN SB
NOR Array
4
5
3
4
LSBN LSB
2
3
1
2
0
1
0
MSBN
MSB
SBN
SB
LSBN
LSB
Outputs (V)
D
Bits (V) D (V)
CLK
50
0
50
0
50
0
50
0
50
0
50
0
50
0
50
0
50
0
50
0
50
0
50
0
50
0
LSB
SB
MSB
'0'
'1'
'2'
'3'
'4'
'5'
'6'
'7'
0
10
20
30
40
Time (ms)
Figure 4 Row decoder design. a, Schematic of a three-bit row decoder with eight outputs
(0±7). Each square is a D ¯ip-¯op (Fig. 2e). The bits are designated as most signi®cant bit
(MSB), second bit (SB) and least signi®cant bit (LSB). Complements are labelled with a N.
b, Details of the NOR array showing the manner in which individual transistors are
connected.
shown in Fig. 5. The circuit used for the NOR array is signi®cant for
another reason: four transistors are serially connected between
supply and ground. Such `four-deep' connections are often
employed in silicon arithmetic and logical unit (ALU) circuits.
The fact that such a con®guration has been shown to work for
organic semiconductors suggests that complex logic gates such as
those used in ALUs and simple microprocessors can be constructed
out of organic TFTs.
There are alternative approaches to the formation of complementary circuits that differ from that described above. Instead of
spatially separating the n-channel and p-channel TFTs, it is possible
to create heterojunction organic transistors with two active layers,
one of which is electron-transporting and the other holetransporting17. Thus, individual devices can function either as an
n-channel or a p-channel device depending on bias conditions. A
hybrid organic/inorganic complementary circuit technology with
inorganic and organic semiconductors has been reported18,19. The
propagation delay per gate in a ®ve-stage ring oscillator with
amorphous Si/organic transistors has been measured as 5 ms (ref.
19), corresponding to an oscillation frequency of 20 kHz.
In addition to their lower static power dissipation, complementary circuits have other advantages which are important from the
perspective of system lifetime. Complementary circuits are more
robust against variation of transistor parameters than are p-FET
(®eld-effect transistor) or n-FET circuits. Further, complementary
circuits are expected to last longer as they draw less current. With
these many advantages, we anticipate that large-scale complementary logic circuits based on organic/polymer transistors will be
attractive for numerous applications.
M
Figure 5 Characteristics of the three-bit decoder. In descending order, the traces are as
follows. Clock, data, outputs of the three D ¯ip-¯ops (labelled LSB, SB, MSB on the righthand vertical axis). The eight remaining traces, labelled 0±7, show the output voltages of
the eight outputs: at any time, one output at most is high. Also shown (dashed lines) are
the simulated responses of the ¯ip-¯ops and the outputs of the decoder. The good
agreement between simulation and experiment may be noted. The simulations were
performed with a set of tools developed for organic/polymer transistor circuit design.
2. Dodabalapur, A. et al. Organic smart pixels. Appl. Phys. Lett. 73, 142±144 (1998).
3. Jackson, T. N. et al. Organic thin-®lm transistors for organic light-emitting ¯at-panel display
backplanes. IEEE J. Spec. Topics. Quant. Electron. 4, 100±104 (1998).
4. Brown, A. R. et al. Logic gates made from polymer transistors and their use in ring oscillators. Science
270, 972±974 (1995).
5. Drury, C. et al. Low cost all-polymer integrated circuits. Appl. Phys. Lett. 73, 108±111 (1999).
6. Ziemelis, K. Putting it on plastic. Nature 394, 619±620 (1998).
7. Garnier, F. et al. Thin ®lm transistors based on organic conjugated semiconductors. Chem. Phys. 227,
253±262 (1996).
8. Bao, Z. et al. High-performance plastic transistors fabricated by printing techniques. Chem. Mater. 9,
1299±1301 (1997).
9. Dodabalapur, A. et al. Complementary circuits with organic transistors. Appl. Phys. Lett. 69, 4227±
4229 (1996).
10. Lin, Y.-Y. et al. Organic complementary ring oscillators. Appl. Phys. Lett. 74, 2714±2716 (1999).
11. Bao, Z., Lovinger, A. J. & Brown, J. New air-stable n-channel organic thin-®lm transistors. J. Am.
Chem. Soc. 120, 207±208 (1998).
12. Dodabalapur, A., Torsi, L. & Katz, H. E. Organic transistors: two dimensional transport and improved
electrical characteristics. Science 268, 270±271 (1995).
13. Katz, H. E., Torsi, L. & Dodabalapur, A. Synthesis, material properties and transistor performance of
highly puri®ed thiophene oligomers. Chem. Mater. 7, 2235±2237 (1995).
14. Filas, R. W. in Advances in Electronic Packaging Vol. 1, 1265±1282 (Am. Soc. Mech. Eng., New York,
1997).
15. Rabay, J. M. in Digital Integrated Circuits: A Design Perspective Ch. 6 (Prentice Hall, Upper Saddle
River, New Jersey, 1996).
16. Dimitrakopoulos, C. et al. Low-voltage organic transistors on plastic comprising high dielectric
constant gate insulators. Science 283, 822±824 (1999).
17. Dodabalapur, A., Katz, H. E., Torsi, L. & Haddon, R. C. Organic heterostructure ®eld-effect
transistors. Science 269, 1560±1562 (1995).
18. Dodabalapur, A., Baumbach, J., Baldwin, K. & Katz, H. E. Hybrid organic/inorganic complementary
circuits. Appl. Phys. Lett. 68, 2246 (1996).
19. Bonse, M. et al. in IEDM Technical Digest 249±252 (Inst. Elect. Electron. Eng., Piscataway, New Jersey,
USA, 1998).
Acknowledgements
We thank B. Batlogg, E. A. Chandross, A. J. Lovinger, J. H. O'Neill, M. Pinto, V. R. Raju,
E. Reichmanis, J. Rogers, R. E. Slusher and P. Wiltzius for discussions.
Received 31 August; accepted 25 November 1999.
1. Sirringhaus, H., Tessler, N. & Friend, R. H. Integrated optoelectronic devices based on conjugated
polymers. Science 280, 1741±1744 (1998).
NATURE | VOL 403 | 3 FEBRUARY 2000 | www.nature.com
Correspondence and requests for materials should be addressed to A.D.
(e-mail: ananth@bell-labs.com).
© 2000 Macmillan Magazines Ltd
523