of 9
ARTICLE
Overcoming evanescent
fi
eld decay using
3D-tapered nanocavities for on-chip targeted
molecular analysis
Shailabh Kumar
1,5
, Haeri Park
1,2,5
, Hyunjun Cho
3
, Radwanul H. Siddique
1,2
, Vinayak Narasimhan
1
,
Daejong Yang
1
& Hyuck Choo
1,2,3,4
Enhancement of optical emission on plasmonic nanostructures is intrinsically limited by the
distance between the emitter and nanostructure surface, owing to a tightly-con
fi
ned and
exponentially-decaying electromagnetic
fi
eld. This fundamental limitation prevents ef
fi
cient
application of plasmonic
fl
uorescence enhancement for diversely-sized molecular assemblies.
We demonstrate a three-dimensionally-tapered gap plasmon nanocavity that overcomes this
fundamental limitation through near-homogeneous yet powerful volumetric con
fi
nement of
electromagnetic
fi
eld inside an open-access nanotip. The 3D-tapered device provides
fl
uor-
escence enhancement factors close to 2200 uniformly for various molecular assemblies
ranging from few angstroms to 20 nanometers in size. Furthermore, our nanostructure allows
detection of low concentration (10 pM) biomarkers as well as speci
fi
c capture of single
antibody molecules at the nanocavity tip for high resolution molecular binding analysis.
Overcoming molecule position-derived large variations in plasmonic enhancement can propel
widespread application of this technique for sensitive detection and analysis of complex
molecular assemblies at or near single molecule resolution.
https://doi.org/10.1038/s41467-020-16813-5
OPEN
1
Department of Medical Engineering, California Institute of Technology, 1200 E. California Blvd., MC 136-93, Pasadena, CA 91125, USA.
2
Image Sensor Lab,
Samsung Semiconductor, Inc., 2 N. Lake Ave. Ste. 240, Pasadena, CA 91101, USA.
3
Department of Electrical Engineering, California Institute of Technology,
1200 E. California Blvd., MC 136-93, Pasadena, CA 91125, USA.
4
Imaging Device Lab, Device & System Research Center, Samsung Advanced Institute of
Technology (SAIT), Suwon 16678, Republic of Korea.
5
These authors contributed equally: Shailabh Kumar, Haeri Park.
email:
hyuck.choo@samsung.com
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
1
1234567890():,;
W
hile enhancement of optical signals such as
fl
uores-
cence using plasmonic nanostructures promised
breakthroughs in areas such as single molecule
fl
uorescence-driven DNA sequencing
1
,
2
, rapid disease detection
3
5
, as well as observation of biological reactions
6
,
7
, their wide-
spread application towards bioassays has remained lacking. One
major reason for the lack of applicability of plasmonic
fl
uores-
cence enhancement remains the wide variability in enhancement
ef
fi
ciency for molecular assays. Fluorescence enhancement
obtained from plasmonic nanostructures is intrinsically depen-
dent on both the ef
fi
ciency of electromagnetic (EM)
fi
eld con-
fi
nement at the plasmonic hotspot as well as the distance between
the optical emitter (
fl
uorophore) and the plasmonic hotspot
8
12
.
These
fl
uorophores are typically attached to biorecognition ele-
ments such as antibodies or nucleic acid aptamers that recognize
and speci
fi
cally bind to other target molecules. Therefore the
position or distance of light-emitting
fl
uorophores with respect to
a plasmonic hotspot is dependent on size as well as number of the
molecules within the biomolecular complex, and can drastically
alter the plasmonic enhancement of the emitted signal due to
changes in the radiative and non-radiative
fi
eld components
11
14
.
Inconsistent or weak enhancement of signal due to these varia-
tions limits both the accuracy and ef
fi
ciency of molecular binding
analysis on chip. While several reports have discussed novel
nanoscale geometries that improve the con
fi
nement of EM
fi
elds
leading to strong
fl
uorescence enhancements
15
18
, engineering a
hotspot that resolves the distance challenge between the emitter
and the nanostructure surface has remained elusive.
Powerful
fl
uorescence enhancement within a nanostructure
independent of variation in molecule size and position can be
expected to rely on several important factors: (a) strong con-
fi
nement of electromagnetic
fi
eld (b) powerful coupling of the
emitter to the
fi
eld for enhancement of emission and (c) a hotspot
geometry, which generates uniform electromagnetic
fi
eld dis-
tribution. At the same time, the hotspot geometry needs to be
large enough for commonly used protein
protein binding assays
(i.e. larger than antibodies, ~15 nm). Metal
insulator
metal
(MIM) structures utilizing surface-plasmon-polariton (SPP)
propagation have been known to enable ef
fi
cient con
fi
nement of
EM energy
19
22
. Speci
fi
cally, waveguides with a 3D taper that rely
on adiabatic compression of the SPP mode inside the MIM gap
have been shown to provide extreme volumetric nanoscale con-
fi
nement of EM
fi
elds
23
27
. While extremely promising for energy
con
fi
nement and transfer, the potential of 3D-tapered designs for
coupling with
fl
uorescent emitters and bioassays had remained
unrealized primarily due to closed monolithic structures, which
prevented molecular integration with these devices. Furthermore,
as electromagnetic
fi
eld intensity on a plasmonic surface is
maximum at the metal
dielectric interface, close packing of
multiple metal
dielectric interfaces such as in very thin MIM
gaps, can result in integration of these multiple
fi
eld pro
fi
les
within the gap creating a more homogeneous
fi
eld distribution.
Therefore, a 3D-tapered structure provides unutilized potential
toward these goals, for enabling con
fi
nement of a large amount of
incident electromagnetic energy into a tiny MIM gap.
In order to take advantage of the previously known as well as
unexplored abilities of 3D-tapered MIM devices, we design and
fabricate a
fl
uidic channel-like 3D-tapered gap plasmon nano-
cavity, allowing ready access of the nanostructures to molecules
in solution (Fig.
1
a). We demonstrate that a 3D-tapered gap
plasmon nanocavity can overcome a long-standing limitation
for plasmon-enhanced
fl
uorescence demonstrating powerful
emission enhancement independent of the size or position of the
molecules within the nanocavity. The 3D-taper results in con-
fi
nement of the electromagnetic
fi
eld collected throughout the
body of the device into a tiny cavity (~3300× smaller in volume)
with a |
E
|
2
enhancement close to 500. Furthermore, our analysis
reveals that the 3D-taper geometry improves the coupling of
molecular emitters to the electromagnetic
fi
eld, delivering up to
28% improvement in the radiative decay rates and thus leading
to even stronger enhancement of
fl
uorescence. We speci
fi
cally
trap and observe single antibo
dymoleculesorarraysofmole-
cular assemblies within the device, as well as detect low con-
centration (10 pM) protein molecules diffusing in solution.
Signi
fi
cantly, optimizing the taper angle and tip geometry of the
device results in 40× improvement in uniformity of the elec-
tromagnetic
fi
eld volume compared with a bowtie nanoantenna.
Combination of the strong electromagnetic con
fi
nement, pow-
erful coupling of the emitter to the con
fi
ned
fi
eld and a
homogenous electromagnetic
fi
eld volume result in experi-
mental enhancements of ~2200 compared with glass chips for
molecular heights ranging from few angstroms to 20 nm, which
can be further extended to 50 nm using the current design
approach. Overcoming the molecule placement limitation for
plasmonic enhancement of
fl
uorescence such as presented in
this manuscript can allow this technique to be reliably and
widely applicable for a broad range of biological assays including
complex molecular assemblies.
Results
Device fabrication and design optimization
. The 3D-tapered
gap plasmon nanocavities were fabricated in gold- and silica-
coated silicon substrates as shown in Fig.
1
a, b. Detailed
description of the fabrication process can be found in the
methods section. The sidewalls that are ion-milled in the gold
layer and the exposed top surface of the SiO
2
base, together form
a
fl
uidic MIM nanocavity that tapers vertically and laterally into a
nanoscale tip (Fig.
1
c). The hydrophilic SiO
2
base attracts
fl
uid
into the 3D-tapered nanocavity channel, promotes ef
fi
cient
molecular delivery, and provides surface site-speci
fi
c molecular
binding. Repeating the fabrication process, we also produced 3D-
tapered gap plasmon nanocavity arrays with 20-nm wide and
500-nm long tips integrated onto a larger
fl
uidic channel as
shown in the scanning electron microscope images of Fig.
1
e
g.
The design of the device was optimized for (a) ef
fi
cient
coupling of excitation light into the device body, (b) optimal
transversal con
fi
nement of EM
fi
eld- high and uniform |
E
|
2
of the
guided mode- through the taper, and (c) ef
fi
cient longitudinal
con
fi
nement of EM
fi
eld at the tip. We chose 750-nm as the target
wavelength for
fl
uorophore excitation. Finite-difference time-
domain (FDTD) simulations were utilized to accomplish ef
fi
cient
con
fi
nement of the fundamental anti-symmetric (AS) SPP mode
at the tip of the 3D-tapered gap plasmon nanocavity. In the new
design that allows hotspot access to
fl
uids, a pair of Au walls
separated horizontally on the SiO
2
substrate support the AS mode
whose electric
fi
eld is aligned parallel to the substrate (Supple-
mentary Fig. 1a, b). Parameters were set to
w
body
=
150 nm,
h
body
=
150 nm, and
l
body
=
3
μ
m in order to accomplish ef
fi
cient
coupling of excitation light and low-loss guidance of the AS mode
inside the body with high ef
fi
ciency (Supplementary Note 1).
The taper angle (
α
) was set to 20° in order to achieve ef
fi
cient
transversal con
fi
nement of EM
fi
eld, which results in large and a
uniform |
E
|
2
pro
fi
le at the tip (Supplementary Note 2, Supple-
mentary Fig. 2). Figure
2
a shows the pro
fi
les of EM energy
density
u
at the cross-sections of the body (top) and tip (bottom)
at
α
=
20°. The total transversal EM energy stored inside the body
(
U
A_body
)isef
fi
ciently con
fi
ned inside the tip with minimal loss
(
U
A_body
~
U
A_tip
), which signi
fi
cantly increases the average
transversal EM energy density
u
A
at the tip (Fig.
2
b, Supplemen-
tary Note 2). The 3D-tapered gap plasmon nanocavity showed
greater con
fi
nement of EM energy (greater
u
A
at the tip) than the
ARTICLE
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
2
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
2D-tapered nanocavity and the tip-only structure (MIM structure
without a taper) of the same tip size due to a larger cross-sectional
area of the body being capable of storing greater
U
A_body
(see
Methods
). This extreme con
fi
nement of EM energy through the
3D taper suppresses evanescent-
fi
eld type decay inside the 20 ×
50 nm
2
tip (MIM gap), which is generally observed in high-|
E
|
2
plasmonic structures such as a bowtie nanoantenna
15
,
28
or tip-
only structures (MIM waveguides without a taper)
29
(Fig.
2
c, d
and Methods). The optimized 3D taper achieved four times
greater
u
A
and 40× improvement in uniformity
σ
|
E
|2
than a
bowtie nanoantenna of the same gap size (Fig.
2
e). All of the gap
plasmon structures studied in the FDTD simulations -3D-, 2D-
tapered nanocavities, and the tip-only structure showed a trend
that a greater
u
A
enabled a more uniform hotspot (a smaller
σ
|
E
|2
)
and greater average |
E
|
2
(
E
jj
2
). As a result, the optimized 3D-
tapered nanocavity provided a large (
E
jj
2
=
230), uniform
(
σ
|
E
|2
=
0.05) |
E
|
2
pro
fi
le at the tip along with an 11% net
coupling ef
fi
ciency. Experimental analysis of the taper angle and
body width also matched the trend predicted by simulations, as
maximal emission output was obtained from devices with taper
angle
α
~20° and
w
body
=
150 nm (Supplementary Fig. 3).
The volume and |
E
|
2
pro
fi
le of the hotspot can be further
optimized for capture of targeted number of molecules by varying
l
tip
. A shorter
l
tip
allows the coupled EM energy to be more
densely packed longitudinally inside the tip, which results in
greater |
E
|
2
within a smaller hotspot volume. A 20-nm long tip
provided a
E
jj
2
magnitude of 550 within a 20 × 50 × 5 nm
3
hotspot (Fig.
2
f, g, Supplementary Note 3, Supplementary Fig. 4).
Based on the simulations, experimental results and
considerations, devices were fabricated with the aforementioned
dimensions (as shown in Fig.
1
) and
fl
uorescence enhancement
was studied.
Volumetric optical con
fi
nement and biosensing
. We experi-
mentally tested coupling of incident light into the nanocavity and
gap plasmon-mediated volumetric con
fi
nement at the tips using
surface-linked molecular layers and
fl
uorescent labels. A mole-
cular monolayer of biotin was assembled along the exposed silica
surface of the 3D-tapered nanocavity, using silane-polyethylene
glycol-biotin (SPB) as the reagent to form silane
silica covalent
linkages (Supplementary Fig. 5a). Streptavidin linked with Alexa
Fluor 750 (S-AF 750) was then used as a
fl
uorescent label for
detection of biotin in the tips, taking advantage of the well-known
strong and highly-speci
fi
c molecular interaction between biotin
and streptavidin
30
,
31
. This two-step binding reaction allows the
formation of a monolayer of
fl
uorescently-tagged streptavidin on
the silica surface along the length of the 3D-tapered nanocavity.
We observed the capture of diffusing streptavidin molecules at
high concentrations (1 μM) and resultant enhancement in
fl
uorescence once a molecule binds to the tip region (Supple-
mentary Fig. 5b, c). Detection of molecules diffusing in solution at
micromolar concentrations has been an important target for
biological tracking and has been demonstrated using nanos-
tructures with zeptoliter detection volumes
16
. However, these
detection methods lacked molecular speci
fi
city commonly
required for bioassays. The silica base of our device serves as a
targeted functionalization region inside the gap plasmon nano-
cavity, allowing speci
fi
c capture of molecules diffusing in solution
e
Fluidically accessible waveguide array
SiO
2
Au
b
FIB milling
Metal
deposition
FIB milling
Si
SiO
2
Au
c
d
500 nm
150 nm
20 nm
150 nm
Waveguide array
a
Tip
f
SiO
2
Au
g
Au
SiO
2
H
2
O
Tip
Au
SiO
2
Tip
x
y
z
Fig. 1 Fabrication of 3D-tapered gap plasmon nanocavities. a
Overview image of the device with plasmonic wave propagation and con
fi
nement at the end
of three-dimensionally tapered tip. Open cavity allows
fl
uidic delivery and surface functionalization enabling molecular capture.
b
Device fabrication using a
silicon wafer with 1
μ
m thick thermally grown silicon dioxide, gold deposition, and FIB milling.
c
Zoomed-in view of the 3D-tapered nanocavity tip.
d
Schematic showing an array of 3D-tapered nanocavities with a hydrophilic silica base.
e
SEM image showing fabricated arrays of 3D-tapered nanocavities
with hydrophilic silica base. Scale bar is 25
μ
m.
f
A single 3D-tapered nanocavity. Scale bar is 2
μ
m.
g
Sharp 3D-tapered naocavity tip. Scale bar is 200 nm.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
ARTICLE
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
3
and differentiating them from the background through enhanced
fl
uorescence signal.
After molecular binding, chips were washed to remove
unbound molecules and interrogated using tail-end and full
illumination modes (Fig.
3
a, b). The tail-end illumination mode
involves illumination of the back-end of the 3D-tapered
nanocavity, whereas full illumination mode allows the complete
device to be placed under illumination (Supplementary Note 4).
Excitation light from a near-infrared light-emitting diode light
source (750 nm) was incident on the 3D-tapered nanocavities and
050
Tip-only
0
100
200
300
400
500
600
Bowtie
3D nanocavity
–10
10
x
(nm)
y
(nm)
0
200
400
600
800
|E |
2
|
E
|
2
|
E
|
2
|
E
|
2
|
E
|
2
|E |
2
Max
Min
l
tip
=
l
tip
= 500 nm
l
tip
= 20 nm
l
tip
(
nm)
20
500
200
250
300
350
400
450
500
550
600
0
0.5
1
1.5
2
2.5
3
3.5
Hotspot volume (nm
3
)
×
10
5
600
0
x = 0
y = 0
Bowtie
3D-tapered
Nanocavity
Tip-only
600
0
100
0
25
0
650
700
750
800
Wavelength (nm)
10
–24
10
–26
10
–28
U
A
(J/m)
10
–11
10
–10
10
–9
10
–9
10
–10
10
–11
10
–8
10
–7
10
–6
1 × 10
–10
0
u
1 × 10
–9
0
u
3D-tapered nanocavity body
3D nanocavity tip
Tip-only
Nanocavity body
3D-tapered nanocavity tip
10
–2
10
–1
10
0
10
3
10
2
10
1
|E |
2
|E |
2
Bowtie
Tip-only
2D-tapered
nanocavity tip
3D-tapered
nanocavity tip
2D nanocavity tip
3D
2D
y
x
z
x
ab
cde
fg
u
A
(J/m
3
)
u
A
(J/m
3
)
E
Fig. 2 Design optimization of the 3D-tapered nanocavity using FDTD simulation. a
Comparison between the cross-sectional views of the EM energy
density pro
fi
les in the body (top) just before the 3D-taper (
α
=
20°) and at the tip (bottom).
b
The total transversal EM energy
U
A
and average transversal
EM energy density
u
A
of the body and tip of the 3D- and 2D-tapered nanocavities and a tip-only stru
c
ture.
c
Cross-sectional view of the |
E
|
2
pro
fi
les of the
3D-tapered nanocavity (top), a bow tie (middle), and a tip-only structure (bottom) of the same gap size (20 nm × 50 nm).
d
|
E
|
2
inside the gaps of the
structures at
y
=
0 (top) and
x
=
0 (bottom).
e
Hotspot uniformity
σ
|
E
|2
and average |
E
|
2
at cross-sectional areas of the 3D- and 2D-tapered nanocavities, a
tip-only structure (circle), and a bowtie (triangle).
f
Top view of the |
E
|² pro
fi
les in the devices with
l
tip
=
20 nm, 500 nm, and
.
g
|
E
|² and hotspot volume
at the tips of the devices with varied
l
tip
.
ARTICLE
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
4
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
a visible hotspot in the form of enhanced
fl
uorescence was
observed from the sub 20-nm tip region as a result of volumetric
fi
eld con
fi
nement at the tips (Fig.
3
b). Due to improved coupling
and light collection through the body of the device, about an
order of magnitude improved intensity (~9×) at the tip was
obtained using full illumination mode as compared with tail-end
illumination mode, which agrees with the calculation based on
the FDTD simulations (~10×) (Supplementary Fig. 6). The role of
the 3D-tapered nanocavity body towards
fl
uorescence enhance-
ment was further veri
fi
ed by fabricating dimensionally-varying
structures (Supplementary Note 5). Fluorescence intensity
comparison between the samples indicated that stand-alone taper
and tip structures have a signi
fi
cantly weaker performance for
optical molecular analysis as compared with the complete device
(Supplementary Fig. 7). This observation can again be attributed
to improved coupling ef
fi
ciency for the full device (Supplemen-
tary Fig. 8), in addition to light collection through the device
body. For subsequent molecular detection and analysis of
plasmonic enhancement, full illumination mode was
implemented.
Limit of detection for biomolecules on the device was
examined using two types of sensing experiments. First,
detection of low concentration protein molecules in solution
(10 pM
1 nM) was performed, testing device suitability for on-
chip diagnostics of rare disease-speci
fi
cbiomarkers(Supple-
mentaryFig.9).Thedetectionlimitinthiscaseisgovernedby
diffusive transport of molecules to the plasmonic hotspot and
can be accelerated by improving molecule transport using
b
Hotspot
Light
spot
Full illumination
Edge of chip
Light
spot
Hotspot
Tail-end illumination
Tail-end
d
Anti-biotin antibody-AF 750
20 nm
Edge of taper
Edge of taper
500 nm
Anti-biotin antibody-AF 750
Light
spot
Hotspot
Tail-end illumination
Tail-end
Light spot
Hotspot
Full illumination
a
c
Max
Min
Max
Min
Max
Min
Max
Min
Fig. 3 Molecular
fl
uorescence enhancement and single molecule capture. a
Illustration and
b
fl
uorescence images of streptavidin capture inside the
nanocavities using tail-end (left) as well as full (right) illumination modes. Hotspot is visible at the end of taper within the 20 nm-wide tip. Scale b
ars are 1
μ
m.
c
SEM image (left) of a 3D-tapered nanocavity with tip length 500 nm and
fl
uorescence image (right) obtained from the tip after antibody
immobilization. 1D-array of antibodies is expected within the tip. Scale bar is 100 nm.
d
SEM image (left) of a 3D-tapered nanocavity tip with length 20 nm
and
fl
uorescence image (right) obtained from the short tip after antibody immobilization. Volumetric-limitation enables speci
fi
c capture of single antibody
at the end of the tip. Scale bar is 100 nm.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
ARTICLE
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
5
convective
fl
ow, magnetic or dielectrophoretic trapping. Second,
we examined the capture and detection of individual or small
arrayofmoleculesatthehotspot,whilehighconcentrationof
molecules were present in solution. User-controlled analysis of
single or small array of molecules remains an important target
for high-resolution analysis of protein function and behavior
32
34
. This is especially important in cases where ligands and
biomolecules are physiologically present at higher (
μ
M
mM)
concentrations
35
. We control the number of molecules captured
within the tips by altering the tip length. After formation of the
biotin monolayer within the 3D-tapered nanocavities, we
utilized an anti-biotin IgG primary antibody (tagged with
DyLight 755 ~spectral response similar to AF-750) for
performing
fl
uorescence assays
36
. In order to compare effect
of tip length on capture of molecules, we performed these
antibody binding experiments using devices with
l
tip
=
500 nm
as well as
l
tip
=
20 nm (Fig.
3
c, d). The dimensions of a 3D-
tapered nanocavity with a short tip (
l
tip
=
20 nm,
w
tip
=
20 nm)
compared with those of an IgG antibody (length ~15
20 nm)
36
,
37
indicate that a single antibody should be speci
fi
cally
trapped within the tip region. We performed atomic force
microscopy (AFM) which demonstrated uniform monolayer of
PEG-biotin and antibodies on
fl
at silica surfaces and indicated
that only one antibody should speci
fi
cally occupy the 20 nm ×
20 nm tip (Supplementary Fig. 10). We obtained 1.5% variation
between expected and measured
fl
uorescence intensity for a
single molecule at the short tip, further indicating presence of a
single antibody at the short tip-e
nd. The calculation was based
on integrated
fl
uorescence intensities from the tips and
simulation-predicted |
E
|
2
pro
fi
les (Fig.
2
a) along with an
assumption of packed molecular a
rrangement within the tips.
EM
fi
eld decay-resistant plasmon-enhanced
fl
uorescence
.We
analyzed the enhancement of
fl
uorescence experienced by an
emitter within the tip, as compared with a non-enhancing sub-
strate (silica or glass surface). As discussed previously, the tip
region of the device exhibits a hotspot with uniformly distributed,
highly con
fi
ned electromagnetic
fi
eld (Fig.
4
a). The emission
enhancement was calculated using FDTD simulations (Fig.
4
b, c),
where the net
fl
uorescence enhancement experienced by a
fl
uorophore is de
fi
ned as a product of EM
fi
eld intensity and
quantum yield gain (Supplementary Note 6, Supplementary
Figs. 11, 12). As discussed in detail in Fig.
2
and Supplementary
Fig. 2e, the 3D-tapered gap plasmon nanocavity provides superior
|
E
|
2
uniformity within the gap compared with conventional MIM
structures. Our analysis also shows that the 3D-tapered gap
plasmon nanocavity allows greater radiative decay rate that
results in up to 28.2% enhanced quantum yield gain along the
width of the channel as compared with the tip-only structure,
thereby enabling ~500× greater net
fl
uorescence enhancement
(Supplementary Fig. 13, Fig.
4
b, c). For a 500 nm long tip, the
expected
fl
uorescence enhancement is >1000 for about 70 % of
the channel width. The full-width half-maximum (FWHM)
covers 95.5% of the
x
-axis with indication of
fl
uorescence
quenching very close to the metallic sidewalls (Fig.
4
b). Fluores-
cence enhancement is very uniform (~1000) along the
y
-axis and
does not decline below half of the maximum enhancement
throughout the channel height (Fig.
4
c). This is especially
important as variation along the y-axis represents the increase in
fl
uorophore height from the bottom silica surfaces. This can be a
result of the size of molecule the
fl
uorophore is attached to or
based on number of molecules in the assembly. The simulations
indicate that this variation should have minimal impact on the
enhancement experienced by the emitter, which has been a long-
standing challenge for plasmonics-enhanced
fl
uorescence.
In order to experimentally verify the optimized
fl
uorescence
enhancement response predicted in Fig.
4
b, c, we performed tests
using diversely-sized molecules including dye molecules, apta-
mers, smaller proteins and antibodies. These molecules were
speci
fi
cally bound to the 3D-tapered nanocavity, and
fl
uorescence
enhancement was analyzed compared with non-structured SiO
2
control samples. While the dye molecules were covalently bound
to the bottom silica surface, all other molecules were speci
fi
cally
bound to their respective recognition agents or antigens
preassembled on the surface (Methods and Supplementary
Note 7). The expected height of the
fl
uorophore from the silica
base of the nanocavity is indicated in Fig.
4
d and varies from <1
nm to ~20 nm. The experimentally calculated enhancement
performance of the device showed a uniform response for various
molecular shapes and heights, which agrees with the simulation
results shown in Fig.
4
c. We expect this device to maintain the
same enhancement for molecular assemblies with height up to 50
nm, which is de
fi
ned by the height of the tip for this device. The
level of enhancement was dependent on the tip length as
expected. For tips with length 500 nm, the enhancement was close
to 950 whereas shorter tips (
l
tip
=
20 nm) provided higher
enhancement (EF~2200) due to stronger |
E
|
2
enhancement at
the tip region for diverse molecular sizes (Fig.
4
e, Supplementary
Fig. 14). These results matched the trend and values predicted
earlier by simulation-based analysis (Supplementary Fig. 3c,
Supplementary Note 8).
As
fl
uorescence enhancement is dependent on the quantum
yield of the
fl
uorophore used, we also used another metric,
enhancement
fi
gure of merit
which normalizes the enhance-
ment with respect to the dye quantum yield and allows
comparison of device performance to other nanostructures
(Supplementary Table 1)
18
. The device showed an enhancement
fi
gure of merit close to 260, which is one of the highest values
obtained for
fl
uorescence enhancement obtained using plasmonic
nanostructures, and uniquely provides this enhancement inde-
pendent of molecule size within the nanocavity tip.
Discussion
We have demonstrated 3D-tapered gap plasmon nanocavities,
which provide one of the highest enhancement of
fl
uorescence
obtained by plasmonic nanostructures (EF: ~2200 with
fi
gure of
merit ~260), independent of the size of the molecular assemblies
used in the assay. Overcoming molecule size and placement-
dependent extreme variation in plasmonic
fl
uorescence
enhancement has been a major challenge restricting widespread
application of this method in bioassays. The nanostructure geo-
metry presented in this work demonstrates a way to overcome
this limitation thereby improving the consistency and range of
plasmon-enhanced emission for diversely sized assembly of
molecules. Simultaneously, we also demonstrate capture and
visualization of single antibodies at the tip as well as sensing of
proteins at low concentrations (10 pM). These advantages can be
readily transferred towards applications such as highly sensitive
biosensing using molecular labels of varying sizes and analysis of
single molecules or tightly controlled arrays of molecules for
protein orientation, protein function, and biological polymer
formation studies
32
34
. While the device geometry promises
several bene
fi
ts, weaknesses of the current fabrication process
include low throughput using focused ion beam lithography and
high footprint of the device compared with smaller nanos-
tructures. Future improvements can target the replacement of
current fabrication process with wafer-scale methods including
nanoimprinting, e-beam lithography, and template stripping for
reproducible manufacturing of nanocavities in combination with
anisotropic etching methods for the tapered portions. We may
ARTICLE
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
6
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
also see further improvement in device performance after repla-
cing ion beam milling with alternative methods mentioned above,
which are known to yield smoother device surfaces, and have
shown such improvements in the past
18
. The presented device
design can also provide advantages in other areas of nanopho-
tonics for optical con
fi
nement, data-transfer, quantum optical
communication, and molecular sensing in mid-infrared and ter-
ahertz domains.
Methods
3D-tapered nanocavity fabrication
. Single-side polished silicon wafers with
thermally grown SiO
2
(thickness: 1 μm) were purchased from University Wafers,
Boston, USA. E-beam evaporation was used to deposit 50 nm gold (Au) on the
wafers. 3D-tapered nanocavity patterns were milled through the gold and silica
using a FEI Nova 600 dual beam system as shown in Fig.
1
. Au (50 nm) was
deposited again using e-beam deposition. Second round of milling was performed
using Nova 600 to remove gold from the bottom of the substrates, exposing the
silica and to mill the tip.
Simulations
. We used a commercial software developed by Lumerical Inc. for the
FDTD analyses. In all analyses, the mesh size was 1 nm and uniform throughout
the device area. For the |
E
|
2
and uniformity analyses, a 750-nm dipole source was
placed at the tail-end of the 3D-tapered nanocavity body and the intensity of the
guided AS mode was monitored across the cross-section of the tip.
E
jj
2
at each
α
was calculated by averaging the |
E
|
2
pro
fi
le over the cross-sectional area (20 × 50
nm
2
) of the tip.
σ
|
E
|2
was calculated by normalizing the |
E
|
2
enhancement pro
fi
les
(setting 1 as the highest) and calculating 2D standard deviation of the pro
fi
les over
the cross-sectional area. 2D-tapered nanocavity was designed to provide only lat-
eral con
fi
nement (body cross-sectional area: 150 × 50 nm
2
) along with the same
taper length of 500 nm and tip cross-sectional area (20 × 50 nm
2
) as the 3D-tapered
nanocavity. The tip-only structure was a simple MIM waveguide without a taper
that provided the same cross-sectional area (20 × 50 nm
2
). The bowtie antenna was
a set of two equilateral Au triangles (side length 140 nm, thickness 50 nm) sepa-
rated by 20 nm on a 3-nm thick Cr layer on top of a 25-nm thick ITO
15
,
28
, which
provided the same cross-sectional area at the gap (20 × 50 nm
2
). For the coupling
ef
fi
ciency into body analyses (Supplementary Figs. 6, 8), a 1.4-NA Gaussian source
centered at 750-nm was focused onto either the tail-end or varied longitudinal
locations (
z
) throughout the body. In both cases, transmitted power through the
cross-section was calculated at each location of incidence. For the
fl
uorescence
enhancement analysis, a 750-nm dipole source was placed inside the tip and
transmitted power through a closed box enclosing the dipole and tip was mon-
itored in order to obtain radiative and non-radiative decay rates (see Supple-
mentary Note 6 and Supplementary Figs. 11,12).
Dye (AF-750) binding
. 3-Aminopropyl-triethoxysilane (APTES) was obtained
from Sigma-Aldrich, USA. The nanocavity chips were cleaned with acetone,
methanol, and isopropanol prior to binding experiments. A 1% APTES solution
was prepared in toluene (anhydrous) and allowed to bind on the chips for 30 min.
This step functionalizes the silica surface with amino groups available for further
binding. The substrates were washed with toluene to remove weakly bound APTES
molecules. The chips were baked at 110 °C for 30 min and then cleaned again with
DI water for 15 min. The samples were then dried with nitrogen. Alexa Fluor 750
NHS ester was purchased from Thermo
fi
sher Scienti
fi
c USA and dissolved in
DMSO (1 mg/ml). The solution was added to the substrates with stirring and dye
molecules were allowed to bind to the functionalized substrates for one hour. The
chips were then cleaned well with DMSO and water. A drop of water was placed on
the chips, covered with a coverslip, and then imaged.
ab
Max
Min
|
E
|
2
x
y
SiO
2
Au
Au
x
= 0
y
= 0
x
(nm)
10
0
10
1
10
2
10
3
Fluorescenceenhancement
10
0
10
1
10
2
10
3
Fluorescenceenhancement
3D nanocavity
Tip-only
0 2 4 6 8 10
0 1020304050
y
(nm)
d
c
Aptamer
Streptavidin
Antibody
Dye
Molecular height
3D nanocavity
Tip-only
20 nm
50 nm
FWHM
FWHM
FWHM
e
0
300
600
900
1200
0
5
10
15
20
25
Fluorescence enhancement
Molecular height (nm)
x
0
1000
2000
3000
4000
Fluorescence enhancement
500 nm tip
20 nm tip
Fig. 4 Overcoming electromagnetic
fi
eld decay for molecular analysis. a
Cross-section of the tip at the end of taper showing EM
fi
eld intensity and
placement of a
fl
uorophore with distance along the
x
and
y
axis for a 500 nm long tip. Variations in net
fl
uorescence enhancement, calculated as a product
between the EM enhancement and quantum yield gain, of the 3D nanocavity (solid line) and tip-only structure (dashed line) as a function of molecular
position along the
b
x
- and
c
y
-axes. Metal-induced quenching effects are visible near the metallic sidewalls of both structures. The 3D nanocavity shows
uniform
fl
uorescence enhancement for about 95.5% (FWHM) of the
x
-axis and 100% of the
y
-axis.
d
Experimentally obtained enhancement factors
calculated for various molecular assemblies using a tip with length 500 nm, demonstrating highly-enhanced
fl
uorescence from all samples as a result of
their placement within the plasmonic hotspot. The estimated height of the molecules from the silica surface has been shown. The near uniform mean
enhancement indicates suitability towards using molecular assemblies of different heights, overcoming evanescent
fi
eld decay typically associated with
plasmonic nanostructures. Plot shows datapoints, mean and s.d. for three devices at each condition.
e
Increase in mean enhancement factor with decrease
in tip length to 20 nm. Plot shows datapoints, mean and s.d. for six devices at each condition.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
ARTICLE
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
7
Aptamer sensing
. Insulin from bovine pancreas was purchased from Sigma
Aldrich (USA). Insulin was mixed in PBS at a concentration of 10 μM. Solution was
added to the chips and left undisturbed for an hour to allow physisorption of the
peptide hormone to the surfaces. BSA (0.1%) was then added to the chips and
allowed to interact with the surface to account for any nonspeci
fi
c binding. An
insulin-binding aptamer (sequence: 5
GGT GGT GGG GGG GGT TGG TAG
GGT GTC TTC 3
) with AF-750 conjugated to the 5
end was obtained from IDT
technologies (St. Louis, USA). Aptamer was dissolved at a concentration of 1 μM in
a folding buffer (Tris: 10 mM; MgCl
2
·6H
2
O: 1 mM; NaCl: 100 mM; pH: 7.4) and
added to the substrate for an hour. The samples were washed with folding buffer
after an hour and then imaged.
Surface biotinylation
. SPB (MW: 1000 and 5000) were purchased from Laysan
Bio Inc. (USA). For surface functionalization, SPB powder was dissolved in 50%
ethanol: DI water. This solution was added to substrates with exposed silica surface
and left undisturbed for an hour. The substrates were then washed with DI water.
Streptavidin binding
. Streptavidin Alexa
fl
uor 750 conjugate (SAF-750) was pur-
chased from Thermo
fi
sher Scienti
fi
c (USA). Streptavidin in PBS (pH 7.4, con-
centration: 0.1 mg/ml) was added to chips with biotin monolayer and left
undisturbed for an hour. The chips were then washed with DI water and imaged. A
layer of water was maintained at all times on the chip through molecular func-
tionalization and imaging. For Streptavidin bioassays, the concentration of strep-
tavidin in PBS was varied while maintaining the duration of incubation (1 h).
Antibody binding
. Anti-biotin antibody (Mouse monoclonal (Hyb-8), IgG1, MW:
244 kDa) conjugated with DyLight 755
fl
uorophore (response similar to Alexa
fl
uor
750) were obtained from Novus Biologicals (USA). The antibody was diluted in
PBS at a concentration of 0.05 mg/ml and added to the chips. The solution was left
undisturbed for 2 h. The samples were then washed with DI water.
Atomic force microscopy (AFM)
. Topology and phase images of dried antibody
monolayer on silica substrates were obtained on an AFM system (Bruker
Dimension Icon, Santa Barbara, CA, USA) using a 100 μm long monolithic silicon
cantilever (All-In-One-Al, NanoAndMore USA Corp (BudgetSensors), Watson-
ville, CA, USA). All the experiments were conducted under ambient laboratory
conditions using tapping mode with a resonance frequency of about 350 kHz.
Images were analyzed afterward by commercial software Nanoscope Analysis.
Imaging
. Fluorescence imaging was performed using a Leica DMI 6000 wide
fi
eld
fl
uorescence microscope. Scanning electron micrographs were taken using FEI
Nova 600 and 200 dual beam systems.
Image analysis
. Images were analyzed using Fiji (ImageJ) software
38
. Graphs were
created in Matlab and Graphpad Prism 5.01.
Data availability
The data that support the
fi
ndings of this study are available from the corresponding
author upon reasonable request.
Received: 18 June 2019; Accepted: 27 May 2020;
References
1. Eid, J. et al. Real-time DNA sequencing from single polymerase molecules.
Science
323
, 133
138 (2009).
2. Flusberg, B. A. et al. Direct detection of DNA methylation during single-
molecule, real-time sequencing.
Nat. Methods
7
, 461
465 (2010).
3. Ambrosi, A., Airo, F. & Merkoci, A. Enhanced gold nanoparticle based ELISA
for a breast cancer biomarker.
Anal. Chem.
82
, 1151
1156 (2009).
4. Chen, S., Svedendahl, M., Van Duyne, R. P. & Ka
̈
ll, M. Plasmon-enhanced
colorimetric ELISA with single molecule sensitivity.
Nano Lett.
11
, 1826
1830
(2011).
5. De La Rica, R. & Stevens, M. M. Plasmonic ELISA for the ultrasensitive
detection of disease biomarkers with the naked eye.
Nat. Nanotechnol.
7
,
821
824 (2012).
6. Acuna, G. et al. Fluorescence enhancement at docking sites of DNA-directed
self-assembled nanoantennas.
Science
338
, 506
510 (2012).
7. Yokoe, H. & Meyer, T. Spatial dynamics of GFP-tagged proteins investigated
by local
fl
uorescence enhancement.
Nat. Biotechnol.
14
, 1252
1256 (1996).
8. Lakowicz, J. R. Radiative decay engineering 5: metal-enhanced
fl
uorescence
and plasmon emission.
Anal. Biochem.
337
, 171
194 (2005).
9. Dragan, A. I. et al. Distance dependence of metal-enhanced
fl
uorescence.
Plasmonics
7
, 739
744 (2012).
10. Feng, A. L. et al. Distance-dependent plasmon-enhanced
fl
uorescence of
upconversion nanoparticles using polyelectrolyte multilayers as tunable
spacers.
Sci. Rep.
5
, 7779 (2015).
11. Anger, P., Bharadwaj, P. & Novotny, L. Enhancement and quenching of
single-molecule
fl
uorescence.
Phys. Rev. Lett.
96
, 113002 (2006).
12. Bharadwaj, P., Anger, P. & Novotny, L. Nanoplasmonic enhancement of
single-molecule
fl
uorescence.
Nanotechnology
18
, 044017 (2006).
13. Kühn, S., Mori, G., Agio, M. & Sandoghdar, V. Modi
fi
cation of single
molecule
fl
uorescence close to a nanostructure: radiation pattern, spontaneous
emission and quenching.
Mol. Phys.
106
, 893
908 (2008).
14. Bardhan, R., Grady, N. K. & Halas, N. J. Nanoscale control of near
infrared
fl
uorescence enhancement using Au nanoshells.
Small
4
, 1716
1722 (2008).
15. Kinkhabwala, A. et al. Large single-molecule
fl
uorescence enhancements
produced by a bowtie nanoantenna.
Nat. Photonics
3
, 654
657 (2009).
16. Punj, D. et al. A plasmonic/antenna-in-box/
platform for enhanced single-
molecule analysis at micromolar concentrations.
Nat. Nanotechnol.
8
, 512
516
(2013).
17. Chikkaraddy, R. et al. Single-molecule strong coupling at room temperature in
plasmonic nanocavities.
Nature
535
, 127
130 (2016).
18. Flauraud, V. et al. In-plane plasmonic antenna arrays with surface nanogaps
for giant
fl
uorescence enhancement.
Nano Lett.
17
, 1703
1710 (2017).
19. Volkov, V. S. et al. Nanofocusing with channel plasmon polaritons.
Nano Lett.
9
, 1278
1282 (2009).
20. Schnell, M. et al. Nanofocusing of mid-infrared energy with tapered
transmission lines.
Nat. Photonics
5
, 283
287 (2011).
21. Gramotnev, D. K. & Bozhevolnyi, S. I. Nanofocusing of electromagnetic
radiation.
Nat. Photonics
8
,13
22 (2014).
22. Bermúdez-Ureña, E. et al. Coupling of individual quantum emitters to channel
plasmons.
Nat. Commun.
6
, 7883 (2015).
23. Pile, D. & Gramotnev, D. K. Adiabatic and nonadiabatic nanofocusing of
plasmons by tapered gap plasmon waveguides.
Appl. Phys. Lett.
89
, 041111
(2006).
24. Gramotnev, D. K., Pile, D. F., Vogel, M. W. & Zhang, X. Local electric
fi
eld
enhancement during nanofocusing of plasmons by a tapered gap.
Phys. Rev. B
75
, 035431 (2007).
25. Vedantam, S. et al. A plasmonic dimple lens for nanoscale focusing of light.
Nano Lett.
9
, 3447
3452 (2009).
26. Choo, H. et al. Nanofocusing in a metal-insulator-metal gap plasmon
waveguide with a three-dimensional linear taper.
Nat. Photonics
6
, 838
844
(2012).
27. Bao, W. et al. Mapping local charge recombination heterogeneity by
multidimensional nanospectroscopic imaging.
Science
338
, 1317
1321 (2012).
28. Ko, K. D. et al. Nonlinear optical response from arrays of Au bowtie
nanoantennas.
Nano Lett.
11
,61
65 (2010).
29. Cai, W., Shin, W., Fan, S. & Brongersma, M. L. Elements for plasmonic
nanocircuits with three
dimensional slot waveguides.
Adv. Mater.
22
,
5120
5124 (2010).
30. Chaiet, L. & Wolf, F. J. The properties of streptavidin, a biotin-binding protein
produced by Streptomycetes.
Arch. Biochem. Biophys.
106
,1
5 (1964).
31. Grubmüller, H., Heymann, B. & Tavan, P. Ligand binding: molecular
mechanics calculation of the streptavidin-biotin rupture force.
Science
271
,
997
999 (1996).
32. Park, S. H. et al. Programmable DNA self-assemblies for nanoscale
organization of ligands and proteins.
Nano Lett.
5
, 729
733 (2005).
33. Brodin, J. D. et al. Metal-directed, chemically tunable assembly of one-, two-
and three-dimensional crystalline protein arrays.
Nat. Chem.
4
, 375
382
(2012).
34. Doyle, A. D., Wang, F. W., Matsumoto, K. & Yamada, K. M. One-dimensional
topography underlies three-dimensional
fi
brillar cell migration.
J. Cell Biol.
184
, 481
490 (2009).
35. Levene, M. J. et al. Zero-mode waveguides for single-molecule analysis at high
concentrations.
Science
299
, 682
686 (2003).
36. Tan, Y. H. et al. A nanoengineering approach for investigation and regulation
of protein immobilization.
ACS Nano
2
, 2374
2384 (2008).
37. Werner, T., Bunting, J. R. & Cathou, R. E. The shape of immunoglobulin G
molecules in solution.
Proc. Natl Acad. Sci. USA.
69
, 795
799 (1972).
38. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis.
Nat. Methods
9
, 676
682 (2012).
Acknowledgements
Device fabrication was performed at the Kavli Nanofabrication Center at California
Institute of Technology. Fluorescence imaging was performed at the Beckman Imaging
center at California Institute of Technology. Funding for this research was provided by
HMRI Investigator award and Samsung GlobalResearch Outreach (GRO) program.
ARTICLE
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
8
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
Author contributions
S.K., H.P., and H.C. conceived the study. S.K
. performed the device fabrication and the
molecular
fl
uorescence experiments. H.P. perfor
med the theoretical device design
optimization and numerical simulations. HJ
.C. and D.Y. helped establish device fab-
rication protocol. R.H.S. helped with numer
ical simulations and device optimization.
V.N. assisted with data visualization and validation. S.K., H.P., and H.C. wrote
the paper.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information
is available for this paper at
https://doi.org/10.1038/s41467-
020-16813-5
.
Correspondence
and requests for materials should be addressed to H.C.
Peer review information
Nature Communications
thanks the anonymous reviewer(s) for
their contribution to the peer review of this work. Peer reviewer reports are available.
Reprints and permission information
is available at
http://www.nature.com/reprints
Publisher
s note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional af
fi
liations.
Open Access
This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article
s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
article
s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit
http://creativecommons.org/
licenses/by/4.0/
.
© The Author(s) 2020
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16813-5
ARTICLE
NATURE COMMUNICATIONS
| (2020) 11:2930 | https://doi.org/10.1038/s41467-020-16813-5 | www.nature.com/naturecommunications
9