Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
+;bv pev. -pola t"
Demonstration of a Nanolithographic Technique
using a Self-Assembled Monolayer Resist
for Neutral Atomic Cesium
Ii. K. Berggren, R. Younkin, E. Cheutrg, M. Prentiss.
A. J. Black, G. N' I . Whitesides.
D. C. Ralph, C. T. Black, l l l . Tinkham*
[. ] Prof. M. Prent iss, I ( . I i . Berggren, R. \ounkin, E. Cheung
Prof. M. Tinkham, D. C. Ralph, C. T. Black
Department of Physics, Harvard University
17 Oxford Street, Cambridge, MA 02138 (USA)
Prof . G. M. Whi tes ides, A. J . B lack
Department of Chemistry, Harvard Universi t r .
l2 Oxford St reet , Cambr idge. NIA 02138 (USA )
[**] This rvork rvas supported in part by NSF grant PHY 93L2572 and made use of the
Harvard MRSEC shared facil i t ies. The authors rvould l ike to acknowledge Hans Robinson,
Dr. Hans Biebuyck,, Dr. James L. Wilbur, and Dr. Jabez N{cClelland. for assistance rvith
the init ial stages of this rvork. The authors rvould also l ike to thank Dr. Yuanchang Lu for
assistance in generating SEN'I images, Joseph Thyrvissen for a careful reading of the
manuscr ipt , and I ient Johnson, Dr. Andreas Bard, and Dr. John Gi l laspy for many helpful
discussions.
Abstract
This experiment demonstrates the formation of nanometer-scale features in a gold
substrate using a self-assembled monolayer (SAI{) of nonanethiolate as a resist for a
patterned beam of neutral cesium atoms. The mask used to pattern the atomic beam rvas
a si l icon nitr ide membrane perforated with nm- and plm-scale holes. A dose of ^,3
monolayers of cesium damaged the SAI,I suff iciently to al lorv penetrat ion of the damaged
regions by an aqueous etching solut ion. Etching transferred the pattern of damage in the
SAN,I laver in to the under ly ing gold substrate. Features of -70-nm s ize rvere etched in to
the gold substrate. Invest igat ions of the ref lect iv i ty of samples exposed to the atomic beam
rvi thout a mask and subsequent l r , 'e tched revealed that the res is t -etch svstem exhib i ts a
minimum threshold dose of cesium for damage; at doses lower than ^,3 monola-vers, the
damage was insuf f ic ient to a l lorv penetrat ion of the SAI I { by the etch ing solut ion.
Conventional l i thographic processes incorporate three basic elements: 1) u damaging
agent ( r .S.photons, ions, or e lec t rons) ,2) u pat tern ing mechanism ( r .g .a mask or focus ing
opt ics), and 3) u resist that is sensi t ive to the damaging agent ( typical ly a -1-pm-thick
organic layer). The relatively recent development of optics for neutral atomic beamsl has
yielded a patterning mechanism for l i thographic techniques that use neutral atoms.2-4
There is, therefore, interest in nelv resist systems that are sensit ive to exposure to neutral
atoms.s-8 The experiment described in this paper demonstrates a new method of making
nanostructures that uses a patterned beam of neutral cesium atoms to damage a
^, 1.2-nm-thick sel f -assembled monolayer (SA\1) resist of alkanethiolates on golds'r0 (see
Figure 1). The atomic beam rvas not patterned opt icai ly, but rv i th a contact mask
consist ing of a
;-rm-scale holes.
40 pm x 40pm si l icon ni t r ide (Si3i{4) membrane perforated with nm- and
Exposure to the cesium altered the SAN,{ layer sufficiently to allow
penetrat ion of a rvet -chemical e tch. resul t ing in ^ ,70-nm- lv ide features etched in to a
^,20-nm-thick gold la.ver on a si l icon substrate. With an appropr iate mask and a broad
atomic beam. patterning over larger areas (^,cm2) should be possible rv i thout increasing
exposure t imes or feature size. The pr imarv appl icat ion for th is technique is the fabr icat ion
of nm-scale structures, however, this method might also be used for high-resolution
detection of atomic density.
Lithography using neutral atoms has several desirable features: the theoreticaliy
achievable image resolution of patterns formed with neutral atoms is l imited by the size of
the atom (^,0.3 nm),11 and because neutral atoms are not subject to Coulombic forces, th is
technique avoids some of the instabi l i t ies inherent in ion- and electron-beam l i thography.
Cesium has several attractive features for atomic l i thographv: 1) Cesium has cycling
atomic resonances with large electronic susceptibil i t ies in regions of the spectrum (852 and
895 nm) that are easiiy accessible by relatively high powered diode (-100 m!V) and
solid-state (^,1 W) lasers; it can. therefore, be easily and inexpensively manipulated using
laser l ight . 2) A rvide variety of atomic optical elements have already been developed for
use in patterning atomic cesium, and in preparing high-quality beams.t 31 Thermal effusive
sources of cesium are in widespread use; thel/ are relatively inexpensive and simple to
fabricate.
Current techniques for neutra l a tom l i thography fa l l in to t rvo c lasses: i ) an opt ica l ly
pa t te rned a tomic beam ( r .g .sod ium.2 ch romium,3 and a luminum{ ) i s depos i ted d i rec t l - v
onto a substrate; or 2) the in ternal energ-v s tored in exc i ted metastable s tates of neutra l
atoms ( r .S.argonS and hel iumt ' t ) is used to damage a SAIU. l l le tastable argon can a lso be
used to create a res is t on the substrate us ing contaminant vapor in the vacuum system.6 In
pat tern ing a gold substrate by image t ransfer f rom a res is t o f SAIvI exposed to neutra l
ces ium atorns, lve demonstrate a r le \v technique for neutra i a tom l i thography that uses a
fundamenta l ly ' d i f ferent phy 's ica i mechanism to pat tern a res is t . L in l ike exper iments us ing
the in ternal energy of metastable atoms ( -10 eV) to damage the res is t , the damage
process for cesium is l ikely to be chemical because the energy stored in the thermally
occupied states in the ground state hyperf ine manifold 1^,4 peY ) and the kinetic energy
(-50 meV) together should be insuff icient to effect col l is ionai damage to the chemical
bonds in the res is t ( ^ ,2 eV) .12
This resist-based technique potentiaily incorporates several advantages of resist
systems for appl icat ion to neutra l -atom l i thographl , and h igh-resolut ion atomic detect ion:
1) Unlike direct deposit ion schemes that bui ld features up one atom at a t ime, the etching
process used to transfer the pattern from cesium into gold can ampli fy the depth of the
image, so a few monolayers of material can be used to make a tal ler structure (20 nm) in
Iess t ime. 2) Conventional resists are thick (^,1 pm), and therefore would not be
chemically sensitive to monolayer-level doses of neutral atoms. Because typical beams of
neutral cesium are relat ively low in f lux (^,3 min monolayer t ime), the thinness of SAMs
(-1.2 nm), and therefore thei r sensi t iv i ty to low doses of ces ium, rvas cr i t ica l to
accomplishing exposures in real ist ic t ime scales. 3) I f the dose profi le of a feature inciudes
some undesi red background, res is ts that exhib i t a min imum dose threshold for damage can
be used to narrorv a feature and irnprove i ts contrast after image transfer. We observed
evidence of such a threshold at a dose of ^,3 monolavers for mm-scale features formed in
samples exposed to a broad cesium beam rvithout a mask. 4) I f the amount of gold
remaining after etching depends l inearl l ' on dose,, and i f this l inearity is observabie at the
nm scale, carefu l exposure of a SAl l l res is t in th is dosing regime could detect atomic
densi ty rv i th nm-scale resolut ion: the re lat ive l ,v iner t pat terns created in the res is t or
substrate could be imaged outs ide of vacuum using convent ional techniques of
high-resolut ion microscopy. For mm-scale features,, there was a region, betrveen ̂ "3 and ^,7
monolayers of cesium, where such a l inear response lvas observed.
Figure 1 shorvs the processing steps used in this experiment. 20-nm-thick gold f i lms
were deposited by electron-beam evaporation onto a Si (100) substrate rvith a ^,2-nm-thick
native oxide la-ver using 1.5 nm of Ti as an adhesion promoter. The SAlvls were formed on
a solut ion of ̂ ,1 mN'I nonanethiol (CHr(CHr)rSH) inthe gold bv overnight immersion in
absolute ethanol ( t r toH ). After the
several t imes rvith EIOH, and rvere
samples \\ 'ere remove(l from solut ion, they \\ 'ere r insed
subsequently dried using nitrogen gas rvith part iculates
T2
Figure 1 : Schematic diagram of the experimental procedure (not to scale). A) A
Au/Ti/SiO2/Si substrate rvas first coated with a SAIvI of nonanethiolate. B) The substrate
was then exposed to a beam of neutral cesium atoms through a mask that lvas in close
contact (^,g pm) with the surface. C) In a f inal etching step, the pattern of damage
caused by the cesium rvas transferred into the gold layer.
/SAM(-1 nm)
+ Au(20 nm)/Ti(1 .5 nm)!-SiO2(-2 nm)/Si(500 pm)
I expose the SAM to a beam of neutralCs atoms through a physical mask
oooo
. t o t l t ooI O
. . . t o V: :- a o o o o
o '. . -+- physical mask
ooo
t j ./ affected region
etching removes gold in regionswhere the SAM was damaged
B. ^I or ' . t f
ot - t o
o
- o ta oo
: oo o o
-
o :.+'."IO _i , U :Yo :oo ooo
o o t a.:fljr
oa O
oO O
oo
o
o = Cs atoms
c
larger than 0.2 pm removed by f i l t rat ion. A patterned 50-nm-thick Sigl{.r membranel3 was
placed in contact (^,5 pm spacing) rvi th the substrate. The mask and substrate lvere held
in place with a steel cl ip. The samples were then put into the vacuum svstem and exposed
to the beam of cesium for t imes varying betrveen 5 min and 2 hours. Immediateiy upon
removal from the vacuum system (( 5 min delay) the exposed samples were etched for 7
min in a wet-chemical etch.14 When outside of vacuum, the samples tvere exposed to
ambient laboratory condit ions. The samples rvere anaVzed using either scanning electron
microscop-v (SEM) or optical ref lectometrt ' .
F igure 2 shorvs a schemat ic d iagram of the atomic beam apparatus used to expose
the sampies. The atomic beam \vas created b1' heating cesium in an oven to ^'290oC and
extracting atoms through a 1.6-mm-diameter circular aperture in the oven lvall. The beam
rvas col l imated using a 1.S-mm-diameter c ircularaperture placed 27.0 cm away from the
along the path of the atomic beam.oven, and the sample rvas p laced 17.0 cm fur ther
Sample entry was accompl ished through a load- lock chamber pumped rv i th a LN2-cooled
sorpt ion pump. Pressures in the main chamber dur ing exposure \ \ 'ere t1 'p ica l ly
5 x 10-8 torr . The f lux in the atomic beam n 'as determined b"v measur ing the opt ica l
th ickness of the beam for c i rcu lar ly polar ized laser l ight tuned to the 6St / , F-4 to 6P312
F-5 transit ion in atomic Cs. The typical f lux was 20 monolavers per hour, rvhere we define
a monolaver to have the surface ciensity of the SANI molecuies ( '1.6 x 10ra atoms/cm2).
Figure 3 shorvs a SENI of ^,70-nm- and ^,500-nm-wide features formed in a gold
surface by cesium exposure, as rvel l as the masks used to form these features. Typical
exposure t imes \ \ ,ere -15 minutes. corresponding to a dose of ^ ,5 monolavers. The
roughness of the edge of the damaged r lgions was -20 nm, rvhich rvas comparable to the
13
Figure 2 : Schematic diagram of the apparatus used for the exposure of samples of SAMs
on a gold surface to a measured dose of cesium from a neutral atomic beam. The sample
holder could be removed to al lorv for the determination of cesium f lux b-v the measurement
of absorption of resonant laser l ight by the atomic beam.
oo9.cI
3oo-tJ
E Oo=3='= 0 ,o =;.
J
(cI-o-FJ
ot-tsoCL6'CLoa
f taoq ,33rn ?t
o l 'o. -
go.o-
f
- a
r{ r{rJ rr+r
E O= i l or?r J= . +
g
rzo r l -A =1 a = .- =qJ rcroq)-l
J
rq gEgqil
'rI ==' q):: (oo.3
o!3.o
T4
Figure 3 : A) SEM image of 500-nm- and 5O-nm-scale holes in a Si3N4 mask that rvas
used to pattern the atomic beam. The fabrication of the mask is described elservhere.r3 B)
Image of features formed in Au after exposure to a dose of ^-,5 monolayers of monatomic
neutral cesium through the mask shown in A and subsequent etching. C) Cross-section of
SEM images shown in B. The secondary electron intensity across a typical feature is shorvn
in arbitrary units, averaged over I nm in the vertical axis for the 50O-nm-wide feature, and
11 nm in the vertical axis for the 7O-nm-rvide feature.
sis N4-l pmHoles
.t-,(-
a
Oq)
tU\-Cg
1OcoC)c)a
500 nm+
500 1 000 1500Position (nm)
N4 Holes
siq/si'..
70 nm{--*
100 200
frr. -'1.00
Position (nm)
granularity of the gold surface.
In order to investigate the damage to the surface as a function of dose, samples were
also exposed to the atomic beam without using a mask. Using the known flux profi le of the
atomic beam and the measured reflectivity profi le of damaged spots to normally incident
l ight at 632 nm, the damage to the surface was determined as a function of the dose of
cesium atoms. The reflectivity of the gold has previously been correlated, using atomic
force microscopy, to the amount of gold remaining on the surface.T Figure 4 shorvs the
transfer function from cesium dose into the reflectivity of the etched gold layer. Below ̂ ,3
monoiayers, no visible damage rvas observable. After ̂ ,3 monola,vers, increased dose
resul ted in further etching. Above -10 monolayers, , increased dose induced protect ion
against etching.
Control experiments demonstrated that damage lvas caused by atomic cesium and
not by some other agent , e .g .molecu lar ces ium, o ther chemica l contaminants , or thermal
radiation from the oven. Radiation pressure from laser l ight resonant oniy rvith atomic
cesium rvas used to distort the circular prof i le of the atomic beam.l None of the other
possible beam constituents rvould be deflected by the laser l ight. Figure 5 shorvs the
distorted damaged spot on an etched substrate. Damage observed outside of the or iginal
beam spot - where cesium had been pushed by the i ight - established that the cesium
atoms did damage. The lack of damage in regions that originally rvere in the beam
- rvhere cesium had been removed - indicated that cesium atoms rvere predominantly
responsible for the damage seen in these experiments. When longer exposures were used,
radiat ion pressure establ ished that the protect ion ef fect that rvas observed at h igh doses
(> 10 monolayers) was also due to atomic cesium.
l c
Figure 4 : Reflect ivi ty of an etched gold surface rvhose protective SAlvl resist had been
damaged by a measured dose of cesium atoms from an atomic beam rvithout a mask' The
reflect ivi ty rvas measured rvith incident laser i ight at 632 nm. Different symbois refer to
data obtained for dif ferent exposure t imes on the same sample. Reflect ivi t ies for a range of
doses were extracted for each exposure t ime bv examining the ref lect ivi ty along the
gradient of the dose profi le. The inset graph shorvs that for doses above -10 monolavers' a
slow increase in ref lect ivi ty rvas observed rvith increasing dose. \Ve define a monolaver to
have the su r face dens i t y o f the SA\ l ( ' 1 .6 x i0 ra a toms/cm?) '
A 20 min . 30 min o 40 min
80f ^r?
70
60
Q. aL
III
AIIID
=
()ooE
o()a)o_
70
65
60
55
a
a Oa
aa
a a o
A n
; €
t0
A 1 I
a . A o :
"r^h{:tII
..Er"r.l 3 Zc!'l
50
nodamageobserved
l inearregime
t 4 l
5
Dose (Monolayers
saturated damage reglme
10
of Atomic Cesium)
16
Figure b: Photograph of an etched sample made in a control experiment where the SAIvI
resist lvas exposed to an unmasked cesium beam that had been distorted by radiation
pressure. For exposures (not shorvn) without deflecting laser l ight., the resulting etched
spot rvas circular, as indicated by the dashed l ines in the figure. The observed distortion of
the damaged region for the sampie shorvn here proved that atomic cesium was responsible
for the damage effect seen in this paper.
Def lectingLight
Cesium isdeflected intothis region
Border of cesium exposurewithout deflecting light
The central result of this paper is that nm-scale patterns can be fabricated in a gold
surface by exposing a nonanethiolate SAM resist to a patterned beam of neutral cesium
atoms. Relatively low doses (-3 monolayers) of atomic cesium were sufficient to sensit ize
the SAM surface to penetration by u subsequent wet-chemical etch. When studied using
the reflectivity of the gold substrate as an assay, this technique showed a crit ical
thresholding effect. This effect might be used to improve the contrast or narrorv features
rvi th broad backgrounds.
The damage mechanism of the SANI rvas not explored in this paper; i t is l ikely that
the damage is caused by o chemicai in teract ion of the cesium rv i th some component of the
SAIVI surface or the SAIVI-Au interface.l5 Because SANIs incorporating a variety of surface
chemistr ies can be formed, this chemical damage mechanism suggests that the technique
time, or increasedescr ibed in this paper couid be exploi ted ei ther to opt imize the exposure
the range and ut i l i ty of possible structures created rv i th neutral atoms.
In control experiments that establ ished that only atomic cesium effected the surface
da,mage, the substrate was pat terned on the macroscopic length-scale by opt ica l def lect ion
of the cesium beam using radiat ion pressure forces. In past experiments, l ight forces from
an optical standing-rvave have been used to pattern similar materials on the nm
length-scale.2-4 The technique presented here could be used to transfer nm-scale features
formed by optical standing-wave focusing into gold features.
layer can then be
in th i s paper i s a
used as a mask to etch in to the under ly ing
f irst step torvard fabricating nanostructures
Because the patterned gold
si l icon.5 the resul t presented
in si l icon using opt ical ly
pat terned neutra l a torn ic beams.
Erperimental
The optical thickness of the cesium beam was determined b.v measuring absorption of
852-nm laser l ight16 on resonance. The incident laser was shaped by expanding the laser
beam to a ^,l-cm-diameter spot, and then aperturing to a diameter of 1.8 mm and l0pW
of power. Upon entering the a+-polarized laser beam, the atoms in the F-4 ground state
were quickly optically pumped into the mp - 4 Zeeman sublevel. A -5 Gauss magnetic
field lvas used to establish the quantization axis for the atoms. ltfeasurements of the
absorption were made before and after each exposure and averaged (dri f ts were tSrpical ly
less than I0% over I hour) . Anal -vs is of absorpt ion data a l iowed calcu lat ion of f lux to a
combined systematic and stat ist icai error " f i i ln. An oven temperature of ^,290"C
(measured using thermocoupies outside the oven) was used to calculate the longi tudinal
velocity distr ibution, rvhich lvas assumed to be thermal. The geometrical factor from the
non-uniform radial density prof i le of the atomic beam rvas included in the analysis. This
profi le lvas veri f ied experimental l-v by probing the atomic beam rvith resonant iaser l ight.
and re-imaging the f luorescence onto a CCD-arra.v.
The ref lect ivi ty of mm-scale regions of damage in the gold surface from the unmasked
atomic beam was measured using l ight from a hel ium-neon laser at 632 nm. The laser
beam was spatial ly f i l tered by focusing through a 100- pm pin-hole, and then focused using
a 5-cm lens to a I f e2 d iameter on the substrate of 20 pm. A por t ion of the specular ly
ref lected l ight rvas directed into a photodetector (assumed to be l inear over the porver
ranges used - 0.1 - 4 pIV) and compared to the incident laser polver. The 100%
reflect ir . ' i ty levei rvas cal ibrated using a high-quaii t .v mirror.
r0
References
[1] For a review of atomic optics and optical forces, see C. S. Adams, M. Sigel, J. I l ' l lynek,
Physics Reports 1994, 240,, 143.
[2] G. Timp, R. E. Behringer, D. M. Tennant, J. E. Cunningham, NL Prentiss, I(. K.
Berggren, Phys. Reu. Lett. 1992,69, 1636.
[3] J. J. NlcClelland, R. E. Scholten, E. C. Palm, R. J. Celotta. Science 1993, 262,887.
[4] R. W. N{cGowan, D. M. Giltner, and S. A. Lee, Opt. Lett. 1995, 20,2535.
[5] K. I(. Berggren. A. Bard, J. L. Wilbur, J. D. Gillaspy, A. G. Helg, J. J. IllcClelland, S. L.
Rolston. W. D. Phi l l ips, M. Prentiss, G. NI. Whitesides, Science 1995,269, 1255.
[6j K. S. Johnson, I(. K. Berggren, A. J. Black, C. T. Black, A. P. Chu, N. H. Dekker, D. C.
Ralph, J. H. Thywissen, R. Younkin, M. Prentiss, M. Tinkham, G. M. Whitesides, in press,
Appl. Phys. Lett.
[7] A. Bard, I(. K. Berggren. J. L. lVilbur, J. D. Gillaspy, S. L. Rolston, J. J. NfcClelland. !V.
D. Phillips, M. Prentiss, G. M. Whitesides, submitted to J. Vac. Sci. Tech. B,
[8] S. Norvak, T. Pfau, J. Mlynek. in press, J. Appl. Phys. B.
[9] R. C. Nuzzo, F. A. Fusco, D. L. Allara, J. Am. Chem. Soc. 1987, 114, 1990.
[10] C. D. Bain, J. Eval l , G. N{. Whitesides, ibid.,7155.
[11] Diffraction does not play a role in neutral atom lithograph-'- because the deBroglie
wavelength is - .01 nm. Electron and ion beam lithography exploit the same advantage to
make features as small as 5 nm.
f12] Sodium atoms have also been used to pattern a SAM resist of hexadecanethiolate on a gold
substrate. The pattern rvas subsequently transferred into the gold using a rvct-chemical
etch. H. Robinson, I(. I(. Berggren, H.'Biebuyck, M. Prentiss, G. M. Whitesides,
1 l
unpublished results.
[13] I(. S. Ralls, R. A. Buhrman, R' C. Tiberio. Appl. Phvs' Lett' 1989,55,2459'
[14] 0.001 M KaFe(CN)6, 0.01 M K3Fe(CN)6, 0.1 M I(zSzO:' and I i r l KOH; see Y. Xia, X.-NI.
zhao, E. Kim, G. M. lVhitesides, chem. IlIater. Lgg', 12,2332. For discussion of the
quality of the selectivity of sAMs against this etch, see also Y. Xia, X. -M. Zhao, and G.
M. lVhitesides, in press, Microelectronic Engineering'
[15] Prior to etching, exposed samples rvith mm-scale patterns of damage were analyzed using a
SSX-100 (Surface Science) X-ray photoelectron spectrometer rvith an Al Ka X-ray source,
and an analyzer pass-energy of 150 eV. Both cesium and oxygen rvere observed in the
exposed regions but not on the surrounding surface. Subsequent rvashing of the samples
using water and EtoH significantly reduced both the cs (3d5) and o (1s) photoelectron
peak intensit ies. Differences in the C (1s), S (2s.2p), and Au (4f) peak intensit ies between
damaged and undamaged rvashed regions rvere not observed. Ii. I(. Berggren, R. Younkin,
E. Cheung, M. Prentiss. A. J. Black. G. I I . Whitesides, unpublished results'
[16] For a description of the design of the diode lasers used in these experiments see C.
Wieman, L . Ho lberg, Reu. Sc i . Ins t rum. 1991, 62, I .