MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES...

8
MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES LALITA UDPA Michigan State University East Lansing, Michigan INTRODUCTION Four years after its invention in 1982 (1), the scanning tunneling microscope (STM) was awarded the 1986 Nobel Prize for physics, one of only four such prestigious awards given for a truly significant contribution to scientific instrumentation. Since then, the family of scanning probe microscopy (SPM) techniques, which includes scanning tunneling microscopy, atomic force microscopy (2–4), mag- netic force microscopy (5), near-field optical microscopy (6), scanning thermal microscopy (7), and others, has revolu- tionized studies of semiconductors, polymers, and biologi- cal systems. The key capability of SPM is that, through a controlled combination of feedback loops and detectors with the raster motion of piezoelectric actuator, it enables direct investigations of atomic-to-nanometer scale phe- nomena. Scanning probe microscopy is based on a piezoelectric- actuated relative motion of a tip versus sample surface, while both are held in a near-field relationship with each other. In standard SPM imaging, some type of tip-sample interaction (e.g., tunneling current, Coulombic forces, mag- netic field strength) is held constant in z through the use of feedback loops, while the tip relative to the sample under- goes an xy raster motion, thereby creating a surface map of the interaction. The scan rate of the xy raster motion per line is on the order of seconds while the tip-sample interaction is on the order of nanoseconds or less. The SPM is inherently cable of producing surface maps with atomic scale resolution, although convolution of tip and sample artifacts must be considered. Scanning tunneling microscopy is based on a tunneling current from filled to empty electronic states. The selectiv- ity induced by conservation of energy and momentum requirements results in a self-selective interaction that gives STM the highest resolution of all scanning probe techniques. Even with artifacts, STM routinely produces atomic scale (angstrom) resolution. With such resolution possible, it would be highly desir- able to apply STM to investigations of molecular biology and medicine. Key issues in biology and medicine revolve around regulatory signaling cascades that are triggered through the interaction of specific macromolecules with specific surface sites. These are well within the inherent resolution range of STM. The difficulty when considering the application of STM to molecular biology is that biological samples are non- conductive. It may be more accurate to describe biological samples as having both local and varying conductivities. These two issues will addressed in this article, and examples of conditions for the successful use of STM for biomedical imaging will be discussed. We begin with an overview of successful applications of STM in biology and medicine. 516 MICROSCOPY, SCANNING TUNNELING Encyclopedia of Medical Devices and Instrumentation, Second Edition, edited by John G. Webster Copyright # 2006 John Wiley & Sons, Inc.

Transcript of MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES...

Page 1: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

516 MICROSCOPY, SCANNING TUNNELING

Encyclopedia of Medical Devices and Instrumentation, Second EditionCopyright # 2006 John Wiley & Sons, Inc.

MICROSCOPY, SCANNING TUNNELING

, edited by John G. Webster

VIRGINIA M. AYRES

LALITA UDPA

Michigan State UniversityEast Lansing, Michigan

INTRODUCTION

Four years after its invention in 1982 (1), the scanningtunneling microscope (STM) was awarded the 1986 NobelPrize for physics, one of only four such prestigious awardsgiven for a truly significant contribution to scientificinstrumentation. Since then, the family of scanning probemicroscopy (SPM) techniques, which includes scanningtunneling microscopy, atomic force microscopy (2–4), mag-netic force microscopy (5), near-field optical microscopy (6),scanning thermal microscopy (7), and others, has revolu-tionized studies of semiconductors, polymers, and biologi-cal systems. The key capability of SPM is that, through acontrolled combination of feedback loops and detectorswith the raster motion of piezoelectric actuator, it enablesdirect investigations of atomic-to-nanometer scale phe-nomena.

Scanning probe microscopy is based on a piezoelectric-actuated relative motion of a tip versus sample surface,while both are held in a near-field relationship with eachother. In standard SPM imaging, some type of tip-sampleinteraction (e.g., tunneling current, Coulombic forces, mag-netic field strength) is held constant in z through the use offeedback loops, while the tip relative to the sample under-goes an x–y raster motion, thereby creating a surface mapof the interaction. The scan rate of the x–y raster motionper line is on the order of seconds while the tip-sampleinteraction is on the order of nanoseconds or less. The SPMis inherently cable of producing surface maps with atomicscale resolution, although convolution of tip and sampleartifacts must be considered.

Scanning tunneling microscopy is based on a tunnelingcurrent from filled to empty electronic states. The selectiv-ity induced by conservation of energy and momentumrequirements results in a self-selective interaction thatgives STM the highest resolution of all scanning probetechniques. Even with artifacts, STM routinely producesatomic scale (angstrom) resolution.

With such resolution possible, it would be highly desir-able to apply STM to investigations of molecular biologyand medicine. Key issues in biology and medicine revolvearound regulatory signaling cascades that are triggeredthrough the interaction of specific macromolecules withspecific surface sites. These are well within the inherentresolution range of STM.

The difficulty when considering the application of STMto molecular biology is that biological samples are non-conductive. It may be more accurate to describe biologicalsamples as having both local and varying conductivities.These two issues will addressed in this article, andexamples of conditions for the successful use of STM forbiomedical imaging will be discussed. We begin with anoverview of successful applications of STM in biology andmedicine.

Page 2: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

MICROSCOPY, SCANNING TUNNELING 517

Figure 2. STM image of portion of Hetero A-T double helix ofshowing base pairs. (Reproduced from Ref. 9, used withpermission.)

SCANNING TUNNELING MICROSCOPY IN BIOLOGY ANDMEDICINE: DNA AND RNA

The STM imaging for direct analysis of base pair arrange-ments in DNA was historically the first biological applica-tion of the new technique. An amusing piece of scientifichistory is that the first (and widely publicized) images (8–12) of (deoxyribonucleic acid) DNA were subsequentlyshown to correspond to electronic sites on the underlyinggraphite substrate! However, more careful investigationshave resulted in an authentic body of work in which thebase pairings and conformations of DNA and RNA aredirectly investigated by STM. One goal of these investiga-tions is to replace bulk sequencing techniques and crystaldiffraction techniques, which both require large amounts ofmaterial, with the direct sequencing of single molecules ofDNA and RNA. Two examples of DNA and RNA investiga-tion by STM are presented here. One is an investigation ofDNA and RNA structures, and the other is an investigationof DNA biomedical function.

Recently reported research from the group at The Insti-tute for Scientific and Industrial Research at Osaka Uni-versity in Japan (13) has shown detailed STM images ofwell-defined guanine-cytosine (G-C) and adenine-thymine(A-T) base pairings in double- and single-stranded DNA.Four simple samples involving only G-C and only A-T basepairs in mixed (hetero) and single sided (homo) combina-tions were chosen for analysis (Fig. 1). These were depos-ited on a single-crystal copper (111)-orientation [Cu(111)]substrate using a technique developed specially by thisgroup to produce flat, extended strands for imaging. AnSTM image showing the individual A-T base pairs in thehetero A-T sample is shown in Fig. 2. Images of the overallstructures indicated repeat distances consistent with inter-pretation as the double helix. Images from mixed samplesof hetero G-C and hetero A-T are shown in Fig. 3. Thelarger structure is interpreted as hetero G-C and thesmaller as hetero A-T, which is consistent with X-raydiffraction data that indicates the A-T combination is morecompact.

Only the double helix structure was observed for thehetero G-C samples. However, the homo G-C structures,

A

T

A

T

A

T

A

T

G

C

G

C

G

C

G

C

A

T A

TA

T A

T

G

C G

C

G

C

G

C

(a) (b)

(c) (d)

Figure 1. (a) Homo A-T, (b) Hetero A-T, (c) Homo G-C, and (d)Hetero G-C. (Figure adapted from Ref. 9, used with permission.)

hetero A-T structures, and homo A-T structures wereobserved in two types, and the spot spacings and sizes ofthe second type would be consistent with interpretation assingle-stranded DNA. The observed presence or lack ofsingle-stranded configurations among the samples is con-sistent with the fact that hetero G-C has a higher melting(unraveling) temperature than the homo G-C and thus ismore difficult to unwind. Both hetero and homo A-T pairshave lower melting temperatures than either of the G-Cpairs. Images of both hetero A-T and Homo A-T samplesoften showed sizing and spacings consistent with inter-pretation as single-stranded DNA, in addition to observeddouble helix specimens. Thus, the presence/lack of single-stranded versus double helix images is consistent withknown melting temperature data for the C-G and A-T basepairings.

The same group has also reported successful STMinvestigations of transfer-ribonuclic acid (t-RNA) (14). In

Figure 3. Hetero G-C and Hetero A-T mixed sample. The largerspecimens are identified asHetero G-C, and the smaller specimensare identified as Hetero A-T. Both are in a double helixconfiguration. (Reproduced from Ref. 9, used with permission.)

Page 3: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

518 MICROSCOPY, SCANNING TUNNELING

TΨC arm

(a) (b) (c)

TΨC arm

D armD arm

anticodon armextra loop

anticodon arm

accepter stem accepter stem

6 nm

UU

U

A

A5′

3′

CG

AG

G

GC

G

U

A

C

G

G

C

U

UU

U

UUA

A AC

C G

GG

GC

CG

C

G

G

C

19.5 nm

C

ACC

GCG

CGGCAU

A

A

A

UU

AUAU

U

UU UU

AAA

UG

C

C

G

Figure 4. (a) Model of t-RNA L-shapedconformation. (b) Model base pair arrangementin L-shaped conformation. (c) STM image of L-shaped conformation at physiological pH.(Reproduced fromRef. 10, usedwith permission.)

Figure 5. (a) STM image of retinoic acid on a graphite substratecompared with its molecular model showing the aliphatic ringhead and polymeric tail. (b) STM image of retinoic acid binding tot-RNA with molecular model overlay. (Reproduced from Ref. 13,used with permission.)

RNA, the base pairing is adenine-uracil (A-U) instead ofadenine-thymine (A-T). Also the backbone sugars areribose rather than deoxyribose, but are still linked byphosphate groups. The RNA is very difficult to synthesizeas a single crystal and consequently there is a very limitedamount of X-ray diffraction data available for RNA. Littleis known about its variations, and therefore direct inves-tigations of single molecule RNA would add much to ourknowledge.

Transfer RNA is a small RNA chain of � 74–93 nucleo-tides that transfers a specific amino acid to a growingpolypeptide chain at the ribosomal site of protein synthesisduring translation (15). It has sites for amino acid attach-ment, and an anticodon region for codon recognition thatbinds to a specific sequence on the messenger RNA (m-RNA) chain. It has a partial double-helix structure eventhough it has only one chain, because the single RNA chainfolds back, and loops back, on itself, as shown in Fig. 4a.

X-ray diffraction studies (16) have indicated that thet-RNA structure may often assume an L-shaped conforma-tionwith a long and a short arm. Amodel of the EscherichiaColi lysine t-RNA macromolecule used by the group for itsSTM studies is shown in Fig. 4a and b. It shows both theL conformation and the underlying loop and base pairchemistry.

Using STM, the group was able to directly image the Lconformation as shown in Fig. 4c. In addition to the firstdirect statistical data on the lengths of the long and shortarms, obtained from analysis of several STM images, ananalysis of the influence of pH on conformation was alsocarried out. Current investigations are focusing on biofunc-tion research issues in addition to structural researchissues, using STM to directly image the coupling of theimportant amino acid molecules at specific t-RNA sites.

The STM investigations of nanobiomedical rather thanstructural issues are an important emerging research area.One example is the recently reported research from theUniversity of Sydney group in which the local binding ofretinoic acid, a potent gene regulatory molecule, to plasmidp-GEM-T easy (596 base pair Promega) DNA fragments ona single-crystal graphite substrate, was directly imagedand reported (17). Retinoic acid has been documented asresponsible for a number of profound effects in cell differ-entiation and proliferation, and is known to accomplish itsfunctions through selective site binding during the tran-scription process. The STM images of retinoic acid by itself

on a single-crystal graphite substrate were investigatedfirst. These showed sizes consistent with the retinoic acidmolecular structure, and a bright head area with a darkertail area. A molecular model of retinoic acid, also shown inFig. 5a, shows its aliphatic carbon ring head and polymerictail. For reasons further discussed below, the aliphatic ringhead may be expected to have a higher tunneling currentassociated with it than the polymeric tail, and therefore theobserved bright and dark areas are consistent with theexpected structure.

At low concentrations, retinoic acidwas observed to bindselectively at minor groove sites along the DNA, with someclustering of retinoic acid molecules observed, as shown inFig. 5b. High resolution STM imaging provided directevidence for alignment of the retinoic acid moleculeshead-to-tail structure edge-on with the minor groove andalso in steric alignment with each other. From STM heightstudies, it could also be inferred that the aliphatic ringhead was attached to a ring partner along theminor groovesurface, but that the tail was not attached. This maysuggest a loosely bound on–off functional mechanism. Athigh concentrations, retinoic acid was observed to bindalong the whole length of the DNA double helix, but againselecting the minor grooves. These first direct studies ofselective site binding of retinoic acid with the minor grooveof DNA should serve as a template for further directinvestigations of other known minor groove binders,thereby opening up the direct investigation of an entire

Page 4: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

MICROSCOPY, SCANNING TUNNELING 519

Electronenergy is lessthan barrierenergy

Tunneling Current

I ~ V exp–c d

0 d

Figure 7. A particle penetrating into and through a wall.

class of regulatory molecule–DNA interactions. The inter-actions of related structures that are candidate therapeuticdrug molecules could be receive similar direct investiga-tion.

Note that both of the above groups have also madeimportant contributions to sample preparation techniquesfor successful STM analysis of DNA and RNA. Thesesample preparation techniques will be discussed belowin the context of the basic physics of the STM interaction,and the basic chemistry and conductivity of DNA and RNAsamples.

BASIC PHYSICS OF THE STM INTERACTION

The STM is based on tip–sample interaction via a tunnelingcurrent between filled electronic states of the sample (or tip)into the empty electronic states of the tip (or sample), inresponse to an applied bias, as shown in Fig. 6. The biasmaybe positive or negative, and different and valuable informa-tion may often be obtained by investigation of the how thesample behaves in accepting, as well as in giving up, elec-trons. In STM imaging, it is important to recognize that thefeaturemap or apparent topography of the acquired image isreally a map of the local density of electronic states. Brightdoes not correspond to a raised topography; it corresponds toa region with a high density of electronic states. Therefore,in STM imaging of biological samples, an important con-sideration is that a differential conductivity will be observedfrom regions, such as rings (usually high) versus regions,such as alkane backbones (usually low). As in all SPMtechniques, a z-direction feedback loop maintains someaspect of the tip samples interaction constant (Fig. 6).The readily available choices on commercial machines areto hold either the tunneling distance d constant (constantheight mode) or the magnitude of the tunneling currentcontent (constant current mode).

The current in question is a tunneling current, which isa quantummechanical phenomenon. It is well documentedthat all electrons within the atomic planes of any materialare in fact in such tight quarters that they display thecharacteristics of a wave in a waveguide, in addition to

Feedback loop

Pie

zoel

ectr

ic s

cann

er

V

A d

Tip

Z

X, Y

Controllerelectronics

I ~ V exp–c d

Figure 6. Important features of an STM system.

their particle-likeness. An electron at the surface of amaterial faces a wall (barrier) created by the dissimilarmaterial (e.g., air, vacuum, or a liquid). While a particlewould run into a wall and bounce back, a wave can pene-trate into and indeed through a wall (as light goes throughglass). This is illustrated in Fig. 7. Additionally, all mate-rials have precise energy levels within them, and therefore,electrons will move by going from one energy level atlocation 0 to another at location d, meeting conservationof energy requirements.

In STM, a tip with empty electronic states is broughtphysically close to a sample surface. The electrons aregiven a direction through the application of the bias (posi-tive in this example). Because they are wavelike, whenthey reach the sample surface, they can tunnel through thebarrier created by the 0-to-d gap and reach the emptystates of the tip, where they are recorded as a currentproceeding from sample to tip. A tunneling current has theknown mathematical form: I � V exp�cd, where I is thetunneling current,V is the bias voltage between the sampleand the tip, c is a constant and d is the tip-sample separa-tion distance. The tunneling current depends sensitivelyon the size of the 0-to-d gap distance. To observe a tunnel-ing current, the gap must be on the order of tens ofnanometers. This is the case in any commercial STMsystem. It is remarkable, that with the addition of a simplefeedback loop, a tip can be easily maintained within nan-ometers of a sample surface without touching it. TypicalSTM tunneling currents are on the order of 10�9–10�12A.With special preamplifiers, currents on the order of 10�14 Acan be detected.

Because STM is a current-based technique, some situa-tions that can interfere with its current will be brieflydiscussed. Very common in STM imaging of biologicalsamples is for the tip to acquire a layer of biologicalmaterial, possibly by going too close to the sample surfacewhile passing over an insulating regionwhere the feedbackloop has little to work on. This usually just introducesimage artifacts, discussed below, but it can sometimesinsulate the tip from the sample, thus terminating thetip–sample interaction. The problem can be minimizedthrough careful consideration of the expected chemistryand local conductivity of the biological specimen to beinvestigated.

CHEMISTRY, CONFORMATION, AND CONDUCTIVITYOF BIOLOGICAL SAMPLES

Consideration of the basic chemistry involved in a biologi-cal sample can help to determine its appropriateness forSTM imaging. The building blocks for DNA and RNA are

Page 5: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

520 MICROSCOPY, SCANNING TUNNELING

NH

NHCH3

O

O

THYMINE

NH

N

NH2

O

CYTOSINE

NH

NH

O

O

URACIL

N

N

NH2

ADENINE

N

NH

N

NH

O

GUANINE

N

NH

NH2

OCH2

H

H

O

H

H

Base

H

O

OCH2

H

H

O

H

H

Base

H

PO

O

O

OCH2

H

H

O

H

H

Base

H

PO

O

O

PO

O

O

(a) (b)

Figure 8. (a) The deoxyribose (ribose) sugar/phosphate backbone for DNA (RNA) isnegatively charged due to phosphate groups.(b) DNA and RNA bases are nitrogenous ringsystems.

shown Fig. 8. The sugar-phosphate backbone containsnegatively charged phosphate groups for both DNA andRNA. The bases adenine, thymine, uracil, guanine, andcytosine are all nitrogenous ring systems. Thymine, cyto-sine, and uracil are six-member ring pyrimidine systems,and adenine and guanine are purines, the fusion of a six-member pyrimidine ring to a five-member imidazole ring.Successful STM imaging of monolayers of the individualbases has been reported (18,19). Examples of the highresolution STM imaging that is possible for monolayersof the individual bases are shown in Figs. 9 and 10.

The nitrogenous ring systems, like the classic benzenering system, which has also been imaged (20), contain p-orbital electrons above and below the ring structure plane,which create a conductive electron cloud. Hence, the suc-cessful STM imaging of the DNA and RNA systems by theOsaka University and University of Sydney groups mightbe expected from the charged phosphate groups in thebackbones and the ring systems in the base pairs.

However, there are also very difficult issues to resolve inmaking the local conductivity of, especially, the signatureDNA and RNA base pairs available to the STM tip. Theseare enclosed within the sugar-phosphate backbones, andonly partially exposed by the twisting of the helix, as shownin Fig. 11a and b (21,22). Also, the choice of substrate willpowerfully influence the molecular structure deposited onit, especially if it is small. An example of this is shown inFig. 12, taken from Ref. 16. The behaviors of pyridine (a

Figure 9. STM images of (a) guanine, (b) cytosine, and (c) adeninemonolayers on a single crystal (111)-orientation gold substrate.(Reproduced from Ref. 14, used with permission.)

single-nitrogen close relation to pyrimidine) and benzeneon a single crystal (001) orientation copper, Cu(001), sub-strate were investigated. The pyrimidine monolayers (thy-mine, cytosine, and uracil) in Figs. 9 and 10 had ringsoriented parallel to the substrate surface, but individualpyridine molecules on Cu(001) had rings perpendicular tothe surface, due to the strong nitrogen-copper atom inter-action, as shown in Fig. 12a. Also, if a single hydrogen atomwas dissociated from the pyridine molecule, as can happenduring routine scanning, the molecule would shift its posi-tion on the copper substrate (Fig. 12b). The STM imaging ofan individual benzene molecule indicated a ring systemparallel to the copper substrate (Fig. 12c), but hydrogendissociation would cause the benzene molecule to becomeperpendicular to the substrate surface (Fig. 12d). There-fore both the substrate choice and interactions with theimaging tip can influence the conformation of the biomo-lecule and whether its locally conductive portions arepositioned to produce a tunneling current.

Now consider the situation of a molecule with a differ-ence in local conductivity, like retinoic acid. The aliphaticring head would similarly be expected to have a high localconductivity, and separate investigations of just retinoicacid by the University of Sydney group confirmed that thisis the case (Fig. 5a). The polymeric tail is basically analkane systemwithout any p-type orbitals. Its conductivityis therefore expected to be less than the ring system andthis is experimentally observed. However, results such asthose shown in Fig. 13 from a group at California Instituteof Technology, demonstrate that high resolution STMimaging even of low conductivity alkane systems is possible(23–26). Therefore, one aspect of STM biomolecular ima-ging is that there may be large differences in the conduc-tivities of two closely adjacent regions. It then becomes anissue of whether the STM feedback loop will be able tosufficiently respond to the differences to maintain the tip-sample tunneling current interaction throughout theinvestigation. Prior consideration of the imaging para-meters necessary for successful STM imaging of the leastconductive part of the bio molecule can help.

Page 6: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

MICROSCOPY, SCANNING TUNNELING 521

Figure 10. STM images of (a) guanine, (b)adenine, (c) uracil, and (d) thyminemonolayerson (e) a single crystal (0001)-orientationmolybdenum dissulfide substrate. (Adaptedfrom Ref. 15, used with permission.)

Figure 11. The three-dimensional conformation of DNA. (a) Thebase pairs are positioned between the sugar-phosphate backbones.(b) The overall structure is a double helix. (Reproduced from Refs.17,18, used with permission.)

Pyridine on Cu(001)

–H

(a)

Benzene on Cu(001)

–2H

(c)

(b)

(d)

Figure 12. Influence of the sample-substrate interaction onsample orientation. (a) An individual pyridine molecule on acopper (001)-orientation, (Cu(001)) substrate is perpendicular tothe surface due to the strong nitrogen–copper atom interaction. (b)An individual pyridine molecule from which a hydrogen atom hasdissociated is also perpendicular to a Cu(001) surface but has ashifted location. (c) An individual benzene molecule on a Cu(001)substrate is parallel to the surface but (d) may becomeperpendicular if hydrogen dissociation occurs. (Adapted fromRef. 16, used with permission.)

Biomolecules, with only nanometer dimensions, alwaysshould be deposited on atomically flat single-crystal sub-strates. Substrates can also be selected to supply electronsto the biomolecule, for positive bias scanning, or to manip-ulate the biomolecule into a desired position. Anotherimportant sample preparation issue is that biomoleculesoften have multiple available conformations, includingglobular conformations that self-protect the moleculeunder nonphysiological conditions. While STM imagingmay be performed in vacuum, air, and even in a liquid-filled compartment (liquid cell), the best resolution may beachieved in vacuum, which is a nonphysiological condition.The less physiological the imaging conditions, the more itwill be necessary to use special molecular stretching tech-niques to investigate an open conformation. A specialpressure jet injection technique was developed by theOsaka University group to deposit stretched DNA andRNA on single-crystal copper for vacuum STM imaging,without giving them the chance to close into globularconformations (13,14).

Figure 13. High resolution STM images of an alkane(pentatracontane) monolayer on graphite. (Reproduced fromRef. 19, used with permission.)

Page 7: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

522 MICROSCOPY, SCANNING TUNNELING

Figure 14. Clockwise from upper left: (a) synthetic true image(b) Gaussian PRF, (c) degraded measurement, and (d) restoredimage.

IMAGING ARTIFACTS AND DATA RESTORATIONUSING DECONVOLUTION

Examination of Fig. 5a shows the ring head of retinoic acidas a large blurred bright spot. Greater resolution of detailwould clearly be desirable. As in all SPM imaging systems,tip artifacts versus the surface features will limit theresolution of the experiments performed. This is often citedas an ultimate barrier in STM studies of macromolecularstructures and in scanning probe microscopy in general(27). It is therefore necessary to develop techniques fordeconvolution of STM tip artifacts for enhancing the reso-lution of measured STM image data.

A commonly used approach for data restoration oreliminating the smearing effect of tip sample interactionis to assume that the observed signal is a convolution of thetrue image and the probe response function (PRF). Thefollowing equation gives a general degradation model dueto the convolution of tip artifacts with true data resulting inthe measurement g(x,y). Neglecting the presence of theadditive noise, the data can be modeled as

gðx; yÞ ¼ f ðx; yÞ � hðx; yÞ ¼X

n;m

f ðn;mÞhðx� n; y�mÞ

where g(x, y), f(x, y), and h(x,y) are the observed or rawsignal, true image, and PRF, respectively. One can thenuse deconvolution methods to extract the true image fromthe knowledge of measured data and probe PRF.

Theoretically, the probe response function is derivedfrom the underlying physics of the tip sample interactionprocess. Hence, there is a need for a theoretical model forthe tip sample interaction. Recent advances in formulationandmodeling of tip sample interactions allow developmentof accurate compensation algorithms for deconvolving theeffect of tip-induced artifacts.

Figure 14 shows an example of applying a deconvolutionalgorithm on synthetic degraded images. The degradedimage in Fig. 14c is generated from a synthetic image inFig. 14a blurred by a Gaussian PRF in Fig. 14b. Figure 14dshows the enhanced result obtained using deconvolution.Although the theoretical treatment of STM and relatedSPM techniques provide major challenges because theatomic structures of the tip and sample have to be modeledappropriately, its potential is clear and this is a stronglydeveloping research area at the present time.

CONCLUSIONS

The STM imaging has the highest resolution of all SPMimaging techniques. As such, it would be highly desirableto apply STM to investigations of molecular biology andmedicine. An often described difficulty when consideringthe application of STM to molecular biology is that biolo-gical samples are nonconductive. It would bemore accurateto describe biological samples as having both local andvarying conductivities. Design of STM experiments inwhich ring systems are exploited, and/or imaging para-meters are set for the least conductive portion of thebiomolecules may help produce successful imaging results.New research in applications of powerful deconvolution

techniques to STM imaging will also open up the field ofdirect STM investigations of the structure and function ofimportant biomolecules.

BIBLIOGRAPHY

1. Binning G, Rohrer H. Helv Phys Acta 1982;55:726–735.2. Hansma HG, Oroudjev E, Baudrey S, Jaeger L. TectoRNA and

kissing loops: Atomic force microscopy of RNA structures. JMicrosc 2003;212:273–279; Sitko JC, Mateescu EM, HansmaHG. Sequence-dependent DNA condensation and the electro-static zipper. Biophys J 2003;84:419–431; Hansma HG,Vesenka J, Siegerist C, Kelderman G, Morrett H, SinsherimerRL, Bustanmante C, Elings V, Hansma PK. Reproducibleimaging and dissection of plasmid DNA under liquid withthe atomic force microscope. Science 1992;256:1180.

3. Hartmann U. Magnetic force microscopy. Annu Rev MaterSci 1999;29:53–87.

4. Paesler MA, Moyer PJ. Near-Field Optics: Theory, Instru-mentation, and Applications. New York: Wiley-Interscience; 1996.

5. Majumdar A. Scanning thermalmicroscopy. Annu RevMaterSci 1999;29:505–585.

6. Beebe TP, Jr., et al. Science 1989;243:370–371.7. Lee G, Arscott PG, Bloomfiled VA, Evans DF. Science

1989;244:475–478.8. Driscoll RJ, Youngquist MG, Baldescwieler JD. Atomic scale

imaging of DNA using scanning tunnelling microscopy.Nature (London) 1990;346:294–296.

9. Tanaka H, Kawai T. Visualization of detailed structureswithin DNA. Surface Sci Lett 2003;539:L531–L536.

10. Nishimura M, Tanaka H, Kawai T. High resolution scanningtunneling microscopy imaging of Escherichia coli lysine trans-fer ribonucleic acid. J Vac Sci Technol B 2003;21:1265–1267.

11. Wikipedia, The free encyclopedia. Available at http://en.wi-kipedia.org/wiki/RNA.

Page 8: MICROSCOPY, SCANNING TUNNELING - Virginia Ayres · MICROSCOPY, SCANNING TUNNELING VIRGINIA M. AYRES ... the family of scanning probe microscopy (SPM) techniques, …

MICROSCOPY, SCANNING TUNNELING 523

12. GiegeR, Puglisi JD, Floentz C. In: CohnWE,MoldaveK, eds.,Progressin Nucleic Acid Research and Molecular Biology,Vol. 45. Amsterdam: Elsevier; 1993.

13. Hadi Zareie M, Lukins PB. Atomic-resolution STM structureof DNA and localization of the retinoic acid binding site.Biochem Biophy Res Commun 2003;303:153–159.

14. Otero R, et al. Proceedings of the 5th Trends In Nanotech-nology (TNT04) CMP Cientifica; 2004. Available at http://www.phantomsnet.net/files/abstracts/TNT2004/AbstractKeynoteBesenbacherF.pdf.

15. Available at http://biochem.otago.ac.nz/staff/sowerby/period-icmonolayers.htm. Multiple references listed.

16. Lauhon LJ, HoW. Single molecule chemistry and vibrationalspectroscopy: Pyridine and benzene on Cu(001). J Phys ChemA 2000;104:2463–2467.

17. Image credit in Fig. 11(a): U.S. Department of EnergyHuman Genome Program, Available at http://www.ornl.-gov/hgmis. This image originally appeared in the 1992U.S. DOE Primer on Molecular Genetics.

18. Image credit in Fig. 11(b):Mathematical Association of Amer-ica, Available at http://www.maa.org/devlin/devlin0403.html.

19. Claypool CL, et al. Source of image contrast in STM images offunctionalized alkanes on graphite: A systematic functionalgroup approach. J Phys Chem B 1997;101:5978–5995.

20. Faglioni F, Claypool CL, Lewis NS, Goddard WA III. Theo-retical description of the STM images of alkanes and sub-stituted alkanes adsorbed on graphite. J Phys Chem B1997;101:5996–6020.

21. Claypool CL, et al. Effects of molecular geometry on the STMimage contrast of methyl- and bromo-substituted alkanesand alkanols on graphite. J Phys Chem B 1999;103:9690–9699.

22. Claypool CL, Faglioni F, Goddard WA III, Lewis NS. Tunnel-ing mechanism implications from an STM study ofH3C(CH2)15HC¼C¼CH(CH2)15CH3 on graphite andC14H29OH on MoS2. J Phys Chem B 1999;103:7077–7080.

23. Villarubia JS. Algorithms for scanned probe microscopeimage simulation, surface reconstruction and tipestimation. J Res Nat Inst Standards Technol 1997;102:425–454.

See also BIOSURFACE ENGINEERING; MICROSCOPY, ELECTRON.