1946 (1966).
38. A. C. Neumann and I. Macintyre, in Proceedings Fifth
International Coral Reef Congress, B. Delesalle, R.
Galzin, B. Salvat, Eds. (Ant. Museum-Ephe, Mo-
orea, French Polynesia, 1985), vol. 3, pp. 105-1 10.
39. National Oceanic and Atmospheric Administration,
Bathymetric charts of Hawaii to Midway (National
Ocean Survey, U.S. Department of Commerce,
Washington, DC, 1981).
40. H. W. Menard, Science 220, 913 (1983).
41. B. U. Haq, J. Hardenbon, P. R. Vail, ibid. 235,
1156 (1987).
42. J. E. Hoffmeister and H. G. Multer, Geol. Soc. Am.
Bull. 75, 353 (1964).
43. J. E. Bardach, Science 133, 98 (1961).
44. D. W. Kinsey, thesis, University of Hawaii, Hono-
lulu (1979).
45. L. Montaggione, Ann. S. Afr. Mus. 71, 69 (1976).
46. J. D. Milliman and K. 0. Emery, Science 162, 1121
(1968).
47. K. J. Hsu and E. L. Winterer, J. Geol. Soc. Lon.
137, 509 (1980).
48. S. 0. Schlanger, H. C. Jenkins, I. Premoli-Silva,
Earth Plan. Sci. Lett. 52, 435 (1981).
49. D. Q. Bowen, Nature 238, 621 (1980).
50. J. I. Tracey, Jr., and H. S. Ladd, in Proceedings of the
Second International Coral Reef Symposium, A. M.
Cameron et al., Eds. (Great Barrier ReefCommittee,
Brisbane, Australia, 1974), pp. 537-550.
51. T. Spenser, Coral Reefs 4, 59 (1985).
52. J. F. Campbell, Geo-Mar. Lett. 6, 139 (1986).
53. T. A. Davies, P. Wilde, D. A. Clague, Mar. Geol.
13, 311 (1972).
54. B. Parsons and J. G. Sdater, J. Geophys. Res. 82,
803 (1977).
55. W. C. Pitman III, Geol. Soc. Am. Bull. 89, 1389
(1978).
56. We thank S. V. Smith, I. Macintyre, T. Jones, and J.
Winterer for reading the manuscript and offering
helpful suggestions. R.W.G. thanks the National
Sea Grant Program (grant NA-81AA-D-00070) and
the State of Hawaii, Office of Ocean Resources,
Department of Business and Economic Develop-
ment for financial support of this research. D.E. was
supported by the Office of Naval Research and the
National Science Foundation. Hawaii Institute of
Marine Biology contribution No. 758, Departnent
Business Economic Development contribution No.
72.
31 August 1988; accepted 22 December 1988
The Scanning Ion-Conductance Microscope
P. K. HANSMA, B. DRAKE, 0. MARTI,* S. A. C. GOULD, C. B. PRATER
A scanning ion-conductance microscope (SICM) has been developed that can image
the topography of nonconducting surfaces that are covered with electrolytes. The
probe of the SICM is an electrolyte-filled micropipette. The flow of ions through the
opening of the pipette is blocked at short distances between the probe and the surface,
thus limiting the ion conductance. A feedback mechanism can be used to maintain a
given conductance and in turn determine the distance to the surface. The SICM can
also sample and image the local ion currents above the surfaces. To illustrate its
potential for imaging ion currents through channels in membranes, a topographic
image of a membrane filter with 0.80-micrometer pores and an image of the ion
currents flowing through such pores are presented.
T HE FAMILY OF SCANNING PROBE
microscopes (1-4) is broadening the
frontiers of surface imaging. These
microscopes scan various sharp probes over
samples to obtain surface contours-in
some cases at the atomic scale (2). We report
results from the SICM. It is designed specif-
ically for biology and electrophysiology in
that it can image soft nonconductors (such
as membranes) that are covered with an
electrolyte.
A schematic view of the SICM is shown
in Fig. 1. A micropipette is filled with
electrolyte and lowered through a reservoir
of electrolyte toward an insulating sample
surface while the conductance between an
electrode inside the micropipette and an
electrode in the reservoir is monitored. As
the tip of the micropipette approaches the
surface, the ion conductance decreases be-
cause the space through which ions can flow
is decreased. The micropipette is then
scanned laterally over the surface while a
feedback system raises and lowers the micro-
pipette to keep the conductance constant.
The path of the tip follows the topography
of the surface.
Department of Physics, University of California, Santa
Bar ara, CA 93106.
*Present address: Institut fur Quantenelektronik, ETH
Honggerberg, HPT, CH-8093 Zurich, Switzerland.
3 FEBRUARY I989
Preliminary experiments were performed
in our lab in 1986 by J. Saad and G.
Tarleton (5). They were able to measure the
topography ofmachined plastic pieces using
an eyedropper as a probe. These experi-
ments were not pursued fiuther because the
scan ranges then available with our micro-
scopes (6) were not much larger than the
openings in available micropipettes.
However, x,y,z piezoelectric translators
with larger scan ranges are now available,
and microscope design has evolved enough
that experimentation with various scanning
probes is relatively easy. The pipette of the
SICM is brought near the surface with two
fine screws that are adjusted by hand while
the separation is monitored with an optical
microscope (7). A third fine screw that is
driven with a stepper motor (7) brings the
pipette within range of a single-tube x,y,z
piezoelectric translator (8). This translator
has a 9.3-,um lateral range and a 3.0-~am
vertical range (9).
The micropipettes for early experiments
were made from 1.5-mm outer diameter
(OD), 0.75-mm inner diameter (ID) Ome-
ga Dot (10) capillary tubing. Later versions
were made with similar tubing (11) on a
Brown-Flaming (12) puller. Although these
pipettes are designed for measurement of
intracellular potentials and patch clamping,
similar micropipettes have been used in oth-
er scanning probe microscopes, namely the
near-field scanning optical microscope and
the micropipette molecule microscope (3).
We estimated our micropipette tip diame-
ters with a nondestructive bubble pressure
method developed by Mittman et al. (13).
S
C)
c
co
ac
Fig. 1. The SICM scans a micropipette over the
contours of a surface by keeping the electrical
conductance through the tip of the micropipette
constant by adjusting the vertical height of the
probe.
T
a 7%
A-
b 3%
b / ~ %T
c ~~~~~~T
0 4 8 12 16 20
Distance (probe I D)
Fig. 2. Resolution test for the SICM. A pipette
with an ID of 0.71 mm and an OD of 1.00 mm
was scanned at constant height over three
grooved plastic blocks with spacing of (a) four
times, (b) two times, and (c) the same as the ID of
the pipette. A 0.lM NaCl solution covered the
blocks and filled the pipette. Note that even the
grooves spaced by the ID of the pipette could be
resolved.
REPORTS 641
This method correlates the OD of the pi-
pette to the internal pressure required for
the pipette to produce a fine stream of
bubbles in a liquid bath. The OD/ID ratio
was essentially constant along the entire
length of the pipette (14). Thus IDs were
estimated from the OD/ID ratio of the
unpulled capillary tubing. For the images
we report the pipette tips had ODs of order
0.1 to 0.2 ,um and IDs of order 0.05 to
0.1 jm.
Samples were glued onto glass substrates
or directly onto electrodes (11) and then
covered with a few drops of 0.1M NaCl.
The micropipette tips were allowed to fill by
capillary action with a commercial kit (10).
The shafts were then backfilled with a sy-
ringe. In the pipettes we also used 0.1M
NaCl to avoid concentration cell potentials
and liquid junction potentials. Reversible
Ag/AgCl microelectrode holders and bath
electrodes (11) provided the necessary stabil-
ity for reliable current and topographic im-
aging.
We applied dc voltages of 0.03 to 0.4 V
to an electrode in the bath and measured dc
currents (typically 1 to 10 nA) flowing into
the pipette to determine the conductance,
which was generally 10-8 to 10-7 S. The
current was amplified by a preamplifier
(preamp) used in a commercial scanning
tunneling microscope (STM) (9). The
preamp converted the current to a voltage
with a 1 MQ resistor, and the voltage was
then amplified by a non-inverting amplifier
with an AD544 operational amplifier with a
voltage gain of 100. The overall current gain
was 0.1 V/nA. We found it important to
topographic images of the
same area on an acetate
film were taken a few min-
utes apart. The similarity
of the imnages demon-
strates that even soft sam-
ples are not damaged by
the SICM. The imaged
area is a square 4 ,um on a
side.
mount the preamp directly on the micro-
scope base to minimize stray capacitance and
noise pickup. We operated the microscope
with conductances 0.9 to 0.98 as great as the
conductance when the tip was far from the
surface. At smaller conductances the micro-
pipette tip at times actually pressed into the
sample surface.
We generated topographic images by
measuring the voltage that the feedback
system applied to the z-axis of the single-
tube x,y,z translator to keep the conduc-
tance constant. For ion current images, the
local current was monitored as the micropi-
pette was scanned over the surface at a
constant height z. A digital scanner supplied
the x and y scan voltages for both topo-
graphic and ion-current images. The z val-
ues or ion currents together with their x and
y coordinates were recorded on a video
cassette recorder (VCR) with a digital data-
acquisition system (15). The resulting image
was filtered and shading or color scales were
added to allow surface features to be more
easily seen (16). The method of statistical
differences (17), which enhances features on
their local background while suppressing
noise, was especially useful for processing
current images.
The resolution of the SICM as a function
of pipette diameter was measured with a
large-scale model (Fig. 2). A glass pipette
with an ID of 0.71 mm and an OD of 1.00
mm was scanned at a constant height over
plastic blocks with regularly spaced grooves
that were 0.71 mm deep. The height was set
by lowering the pipette until the ion con-
ductance decreased from 4.2 x 10-5 S, its
value when far from the surface, to
4 x 10-5 S. These conductances were mea-
sured at a frequency of 10 kHz. This resolu-
tion test showed that it should be possible,
at least in principle, to resolve features as
small as the ID of the micropipette if the
noise on the ion conductance signal could be
reduced to less than 1%.
In practice, we have resolved features a
few times the ID ofour micropipette, which
was 0.05 to 0.1 ,um. There is a compromise
between averaging the ion-conductance sig-
nal for a long time to reduce noise and
obtaining entire images in a reasonable time.
We chose to acquire our images in -5 min
and found that the smallest resolvable fea-
tures were of order 0.2 ,um (Fig. 3).
The practical resolution limits of the
SICM could be extended. As shown in Figs.
2 and 3, the SICM can resolve features on a
scale set by the ID of the micropipette
(although a large OD may prevent the tip
from probing into a narrow, deep groove).
B
C
Fig. 4. (A) A SICM topographic image of the
0.8-,urm diameter pores in a Nuclepore membrane
filter (24). (B) The same image presented in a top
view. (C) A SICM image of the ion currents
coming out through the pores. The false colors go
from black at the background level of current, 8
nA, up to white at the maximum level of -40 pA
above the background. The imaged area is 7.8 ,um
by 4.5 ,um for all three images.
SCIENCE, VOL. 24364.2
Micropipettes with IDs as small as 30 nm
have been produced for near-field scanning
optical microscopes (18), and aluminosili-
cate pipettes with IDs of less than 10 nm
have been made (19). The higher resistances
of these smaller IDs should not be a prob-
lem; STMs have been operated with resis-
tances thousands of times greater than our
present values of 10 to 100 MQ (20). The
most serious limitation we have faced is that
the smaller micropipette tips are extremely
fragile and often break during a scan.
Shorter taper pipettes may help with this
problem and allow resolutions of 10 nm to
be achieved.
The most promising application for the
SICM is not, however, just imaging the
topography of surfaces at submicrometer
resolution. The SICM can image not only
the topography, but also the local ion cur-
rents coming out through pores in a surface
(Fig. 4). Comparison of topographic and
ion current images can give a more detailed
picture of the type of surface features that
correlate with ion channels. In this model
system, the comparison is simple: ion cur-
rents come through the holes as we would
expect. Biological samples are more subtle,
of course, as not every hole is an ion chan-
nel.
For images of the local ion currents, the
micropipette was digitally scanned over the
surface at a preselected height without
movement in the z direction while the cur-
rent flowing into the pipette at each point
was measured (21). It was also possible to
hold the micropipette over various locations
on the imaged surface and measure local
electrical properties. Thermal drift was small
enough (-0.004 ,um/min) that we could
look, for example, at the time dependence of
the ion currents above a pore, which was
again simple for this model system (the
current was constant), but which would be
more subtle for biological samples.
The SICM offers both high-resolution
topographic and ion-current images of non-
conductors. Much of the necessary appara-
tus-micropipettes, microelectrodes, and
current amplifiers-are already used rou-
tinely by electrophysiologists (19). Most of
the positioning and feedback mechanism is
the same as for the STM and is available
commercially (22). Because the SICM oper-
ates in a saline solution or other ionic solu-
tions, the microscope is well suited for bio-
logical applications. It complements the vi-
brating probe system (23) that can measure
larger scale extracellular currents. An excit-
ing extension of this work would be to use
multiple-barrel micropipettes (10) with ion-
specific electrodes (19). The total current
into all of the barrels (or the current into
one barrel with a nonspecific electrode)
3 FEBRUARY I989
could be used for feedback, while the micro-
scope could simultaneously measure and
image the flow of different ions. We antici-
pate that this technique can be used in the
fature by electrophysiologists to combine
spatially resolved ion-flow measurements
and topological imaging of biological mem-
branes.
REFERENCES AND NOTES
1. For a review of scanning probe microscopes, see V.
Martin, C. C. Williams, H. K. Wickramasinghe,
Scanning Microsc. 2, 3 (1988). Other scanning probe
microscopes are described in (2-4).
2. Scanning tunneling microscope: G. Binnig, H.
Rohrer, Ch. Gerber, E. Weibel, Phys. Rev. Lett. 49,
57 (1982); atomic force microscope: G. Binnig, C.
Quate, Ch. Gerber, ibid. 56, 930 (1986).
3. Micropipette molecule microscope: J. A. Jarrell, J.
G. King, J. W. Mills, Science 211, 277 (1981); near-
field scanning optical microscope: A. A. Lewis, M.
Isaacson, A. Harootunian, A. Muray, Ultramicroscopy
13, 227 (1984).
4. Scanning tunneling potentiometry: P. Muralt and
D. W. Pohl, Appl. Phys. Lett. 48, 514 (1986);
scanning electrochemical microscope: A. J. Bard, F-
R. F. Fan, J. Kwak, 0. Lev, Anal. Chem., in press;
scanning thermal profiler: C. C. Williams and H. K.
Wickramasinghe, Appl. Phys. Lett. 49, 1587 (1986);
scanning capacitance microscope: J. R. Matey and J.
Blanc, J. Appl. Phys. 57, 1437 (1985).
5. J. Saad, G. Tarleton, P. K. Hansma, unpublished
results.
6. B. Drake et al., Rev. Sci. Instrum. 57, 441 (1986).
7. B. Drake, R. Sonnenfeld, J. Schneir, P. K. Hansma,
Surf. Sci. 181, 92 (1987); W. J. Kaiser and R. C.
Jaklevic, ibid., p. 55.
8. G. Binnig and D. P. E. Smith, Rev. Sci. Instrum. 57,
1688 (1986).
9. Digital Instruments, Santa Barbara, CA.
10. Frederick Haer & Co., Brunswick, ME.
11. World Precision Instruments, New Haven, CT.
12. Model P-77 from Sutter Instrument Company, San
Rafael, CA.
13. S. Mittman, D. G. Flaming, D. R. Copenhagen, J.
H. Belgum, J. Neurosci. Methods 22, 161 (1987).
14. K. T. Brown and D. G. Flaming, Neuroscience 2, 813
(1977).
15. This system had been previously used for an AFM
and is described in more detail in 0. Marti, S.
Gould, P. K. Hansma, Rev. Sci. Instrum. 59, 836
(1988).
16. Images were processed using a program developed
at UCSB by 0. Marti and S. A. C. Gould.
17. R. J. Wilson and S. Chiang, J. Vac. Sci. Technol. A.
6, 398 (1988); W. K. Pratt, Digital Image Processing
(Wiley, New York, 1978), pp. 323-324.
18. E. Betzig et al., Proc. Soc. Photo-Opt. Instrum. Eng.
897, 91 (1988).
19. K. T. Brown and D. G. Flaming, Advanced Micropi-
pette Techniquesfor Cell Physiology (Wiley, New York,
1986).
20. R. S. Becker et al., Nature 325, 419 (1987).
21. It is also possible to follow the topography with the
ac ion current from one electrode in the bath and
measure the dc ion currents from an electrode below
the surface (or, perhaps, inside a cell).
22. For example, Park Instruments, Palo Alto, CA;
Digital Instruments, Santa Barbara, CA; McAllister
Technical Services, Berkeley, CA; Microscience Inc.,
Braintree, MA; and VG Instruments, Danvers, MA.
23. L. F. Jaffe, Trends Neurosci. 8, 517 (1985).
24. Nuclepore Corporation, Pleasanton, CA.
25. We thank W. Stoeckenius and C. Bracker for sug-
gesting that we image a Nuclepore filter; F. Haer for
providing micropipettes and related equipment; E.
Widder for help in making the later versions of the
micropipettes; C. Bracker, J. Case, V. Elings, M.
Haugan, E. Martzen, J. Saad, J. Schneir, G. Tarle-
ton, and M. Wilson for their help; J. Belgum, C.
Bessemer, K. Prater, C. Quate, and T. Sleator for
useful discussions; K. Wickramasinghe and L. Ingle-
hart for opening our eyes to the potential diversity
of scanned probe microscopes; and G. Somorjai for
pointing out the importance ofstudying liquid-solid
interfaces. Supported in part by the Office of Naval
Research (P.K.H., B.D., and O.M.) and by the
Solid State Physics program in the Division of
Materials Research of the National Science Founda-
tion, under grant DMR8613486 (C.P., S.G., and
P.K.H.)
6 October 1988; accepted 23 November 1988
A Diet-Induced Developmental Polymorphism
in a Caterpillar
ERICK GREENE
Caterpillars of the spring brood of Nemoria arizonaria develop into mimics of the oak
catkins upon which they feed. Caterpillars from the summer brood emerge after the
catkins have fallen and they develop instead into mimics of oak twigs. This develop-
mental polymorphism may be triggered by the concentration of defensive secondary
compounds in the larval diet: all caterpillars raised on catkins, which are low in tannin,
developed into catkin morphs; those raised on leaves, which are high in tannin,
developed into twig morphs; most raised on artificial diets of catkins with elevated
tannin concentrations developed into twig morphs.
M ANY ORGANISMS OCCUR IN TWO
or more distinct forms. Develop-
mental polymorphisms (or poly-
phenisms) occur when phenotypic variation
is produced by differences in environmental
conditions rather than by differences in ge-
netic constitution (1, 2). Such developmen-
tal polymorphisms are conspicuous among
arthropods with life spans that are short
relative to the scale of environmental varia-
tion: examples are some color forms of
caterpillars, pupae, and butterflies (2),
winged and nonwinged morphs of water
striders (3) and planthoppers (4), sexual and
asexual forms of aphids (5), and caste sys-
Department of Biology, Princeton University, Princeton,
NJ 08544.
Present address: Department of Avian Sciences, Univer-
sity of California, Davis, CA 95616.
REPORTS 643