Microfabricated surface-electrode ion trap for scalable quantum information processing.
ABSTRACT Individual laser-cooled 24Mg+ ions are confined in a linear Paul trap with a novel geometry where gold electrodes are located in a single plane and the ions are trapped 40 microm above this plane. The relatively simple trap design and fabrication procedure are important for large-scale quantum information processing (QIP) using ions. Measured ion motional frequencies are compared to simulations. Measurements of ion recooling after cooling is temporarily suspended yield a heating rate of approximately 5 motional quanta per millisecond for a trap frequency of 2.83 MHz, sufficiently low to be useful for QIP.
arXiv:quant-ph/0601173v1 26 Jan 2006
A microfabricated surface-electrode ion trap for scalable quantum information
S. Seidelin,∗J. Chiaverini,†R. Reichle, J. J. Bollinger, D. Leibfried, J. Britton, J. H. Wesenberg,
R. B. Blakestad, R. J. Epstein, D. B. Hume, J. D. Jost, C. Langer, R. Ozeri, N. Shiga, and D. J. Wineland
NIST, Time and Frequency Division, Boulder, CO
(Dated: February 1, 2008)
We demonstrate confinement of individual atomic ions in a radio-frequency Paul trap with a novel
geometry where the electrodes are located in a single plane and the ions confined above this plane.
This device is realized with a relatively simple fabrication procedure and has important implications
for quantum state manipulation and quantum information processing using large numbers of ions.
We confine laser-cooled24Mg+ions approximately 40 µm above planar gold electrodes. We measure
the ions’ motional frequencies and compare them to simulations. From measurements of the escape
time of ions from the trap, we also determine a heating rate of approximately five motional quanta
per millisecond for a trap frequency of 5.3 MHz.
Recent interest in coherent quantum state control and
methods to realize practical quantum information pro-
cessing (QIP) has led to impressive developments in
quantum processing using several different physical sys-
tems . Single quantum bit (qubit) rotations, two-qubit
gates, and simple quantum algorithms have been imple-
mented. However, perhaps the most significant challenge
for any possible physical implementation of a quantum
processor is to devise methods that scale to very large
numbers of quantum information carriers.
The system of trapped ions is an interesting candi-
date for QIP because the basic requirements  have
been demonstrated in separate experiments , and sev-
eral schemes for scaling this system to large numbers of
qubits have been proposed [1, 3, 4, 5, 6, 7]. One approach
is based on a network of interconnected processing and
memory zones where ion qubits are selectively shuttled
between zones [3, 6]. Within this approach, miniature
linear trap arrays [8, 9, 10, 11] and a three layer T-
junction trap  have been demonstrated. Since the
speed of most multi-ion qubit gates is proportional to
the ions’ motional frequencies and these frequencies are
inversely proportional to the square of the dimensions,
we would like to decrease the size of these dimensions.
To do this robustly, microfabrication techniques are re-
quired. Three-dimensional traps have been demonstrated
with boron-doped silicon  and monolithically fabri-
cated gallium-arsenide electrodes . A significant sim-
plification in fabrication could be achieved if all trap elec-
trodes reside on a single surface and the ions are trapped
above this surface . In this case, the trapping elec-
tric fields would be the analog of magnetic fields used
in “chip” traps for neutral atoms (see Ref.  and ref-
erences therein). Surface-electrode ion traps have the
potential added benefit for scaling that micro-electronics
for electrode potential control can be fabricated below
the plane of the electrodes .
Recently, macroscopic charged particles have been con-
fined in a surface-electrode trap . Storage of atomic
ions, however, requires substantially different experimen-
electrode geometry where all electrodes reside in a single
plane, with the ions trapped above this plane.
(a) Standard linear RF Paul trap; (b) Surface-
tal parameters. In this letter we report the first demon-
stration of stable confinement of atomic ions in a surface-
electrode trap. The trap is constructed with standard
and scalable microfabrication processes. We load24Mg+
into this trap, measure the motional frequencies of the
ions, and find reasonable agreement with those deter-
mined from simulations. We also approximately deter-
mine a motional heating rate of the ion(s) that is low
enough to allow for high fidelity logic operations.
trap  consists of four parallel rods whose centers
are located on the vertices of a square (Fig. 1a). An
RF potential is applied to two opposing rods with the
other two (control electrode) rods held at RF ground.
This configuration creates a nearly harmonic pondero-
motive pseudopotential in the ˆ x-ˆ y plane. Longitudinal
confinement for a single trapping zone is obtained by
segmenting each control electrode along its length and
applying appropriate static potentials to the different
segments. Several variations on this design have been
demonstrated [8, 9, 10, 11, 12], but it is very desirable
to simplify their construction.
to modify the 3-D design of Fig. 1a is to place the four
rods in a common plane, with alternating RF and control
electrodes ; one version of this geometry is shown in
Fig. 1b. In this design, the rods are replaced with flat
A straightforward way
FIG. 2: Pictures of the surface-electrode trap. (a) The com-
plete trap structure, including out-lead wires (ribbons) and
filter capacitors. The directions of the laser beams (cooling
and photoionization) and atom flux are indicated. (b) Ex-
panded view of the trap region (center marked by ×). The
control electrodes are numbered for reference in the text. (c)
On-board meander line resistor.
electrodes, as shown in Fig. 2.
We can fabricate this electrode structure by means
of photolithography and metal deposition using evapo-
ration and electrodeposition. For the substrate we use
polished fused quartz, a material with low RF loss. A
0.030 µm titanium adhesion layer and a 0.100 µm cop-
per seed layer are first evaporatively deposited onto the
substrate. This deposition is uniform except for small
areas for resistors where the quartz is left exposed. Re-
sistors (∼1 kΩ) and leads are fabricated through a liftoff
process that entails patterning them with standard pho-
tolithography and evaporation of a 0.013 µm titanium
adhesion layer followed by 0.300 µm of gold. Resistors
are fabricated directly on the quartz substrate; leads are
fabricated on top of the copper seed layer. The gold elec-
trodes near the trapping region are electroplated onto the
copper seed layer after a second photolithographic pat-
terning step.Afterward, the exposed initial seed and
adhesion layers are etched away to isolate electrodes and
leads. The trap electrodes are plated to a thickness of ∼
6 µm so that the ratio of height to inter-electrode spacing
is relatively high (the inter-electrode spacing is ∼ 8 µm).
This should reduce alteration of the trapping potential
due to stray charges that may collect on the exposed in-
sulator between electrodes.
We create ions in the trap by photoionizing thermally
evaporated neutral magnesium atoms. The magnesium
source is realized by resistively heating a stainless steel
tube containing solid magnesium, which is sealed at both
ends and has a small slit from which evaporated magne-
sium atoms emerge. With a planar electrode geometry,
there is a risk of shorting electrodes to each other due
to magnesium deposited onto the trap structure. To re-
duce this risk, in a last processing step we perform a
controlled hydrofluoric acid (HF) etch of the central trap
region. The HF etches away a small part of the titanium
adhesion layer and the substrate, without affecting the
electrodes. The result is a ∼ 2 µm horizontal undercut
of the electrodes to help prevent shorting due to deposi-
tion from the magnesium source. As a further precaution,
we direct the magnesium flux nearly parallel to the sur-
face and avoid as much as possible having the channels
between electrodes be parallel to the flux (Fig. 2).
We use five independent control electrodes to provide
sufficient degrees of freedom to be able to overlap the
electric field null point of the static potential and the RF
pseudopotential minimum. We create a low impedance
path for the RF to ground on the control electrodes with
capacitors (820 pF) that are surface mounted directly
onto the chip (in the future, capacitors could be included
as part of the fabrication process). Gold ribbons for ap-
plying the electrode potentials are gap-welded to contact
The trap structure is mounted in a copper tube that
also serves as part of an RF transformer  and the en-
tire structure is surrounded by a quartz envelope. The
system is baked under vacuum prior to operation to reach
a base pressure below 10−8Pa with the use of an ion get-
ter pump combined with a titanium sublimation pump.
As we describe below, the trap well depth UT for a
surface-electrode trap is fairly shallow , not much
above the mean kinetic energy of the neutral atoms before
they are ionized. Nevertheless, we can load24Mg+ions
efficiently by resonant two-photon photoionization (PI)
at 285 nm . The PI laser, resonant with the 3s2 1S0
↔ 3s3p1P1electric dipole transition in neutral magne-
sium, co-propagates with a Doppler cooling beam tuned
approximately 400 MHz below the 3s2S1/2↔ 3p2P1/2
electric dipole transition in24Mg+at 280 nm. The laser
beams are parallel to the trap surface, and at an angle of
approximately 45 degrees with respect to the trap ˆ z axis
as shown in Fig. 2a.
Since the laser beam direction has significant overlap
with all principal trap axes, cooling will be efficient in all
directions . During loading, both the Doppler cool-
ing and PI beams have 2 mW power and waists of ∼ 40
µm. The atomic flux of magnesium intersects the laser
beams at the trap (Fig. 2a). The cooling beam is applied
continuously, while the PI beam needs to be applied for
only a few seconds to create ions in the trap. Ions are
detected by observing 3p2P1/2 → 3s2S1/2fluorescence
along a direction perpendicular to the trap surface with
a CCD camera as in the view of Fig. 2b. Despite the fact
that the center of the laser beam is only 40 µm above
the surface, which increases the risk of scatter from light
striking the trap electrodes, the signal-to-background ra-
tio for scattered light from the ions is greater than 100
when the Doppler cooling laser is tuned approximately 20
MHz (one-half linewidth) below resonance with intensity
slightly below saturation.
We measure the oscillation frequencies for a single ion
in the trap (equal to the center-of-mass mode frequencies
for multiple ions) by applying an oscillating field to a
control electrode and observing a change in fluorescence
rate when the frequency of the applied field is equal to one
of the motional frequencies, thereby heating the ion .
To excite the axial mode, we apply the oscillating field to
electrode 2, while both transverse modes can be excited
using electrode 1.
As an aid in initially determining the correct operat-
ing conditions, trapping potentials are determined us-
ing numerical solvers (boundary element method) sub-
ject to the constraint that the RF pseudopotential min-
imum overlaps the null points of the electric field from
the static potential, to minimize RF micromotion (see for
example ). For the experiments described here, the
static potentials on each control electrode, expressed as
a fraction of the potential V5on electrode 5 (Fig. 2b) are
V1= 0.320, V2= 0.718, V3= 0.738, and V4= -0.898.
The peak potential amplitude VRFapplied to the RF
electrode (at a frequency of 87 MHz) is difficult to mea-
sure directly. To determine it, we measure the three
mode frequencies for a fixed RF power and V5 = 5.00
V. We use the simulations with the measured electrode
dimensions to extract a best fit value of VRF = 103.5
V. The consistency between the experimental and sim-
ulated frequencies for this fit is shown in Table I. We
further check for consistency by comparing the measured
and predicted mode frequencies for two other values of
the control and RF potentials (also shown in Table I).
The value VRF= 46.2 V is determined by scaling the RF
power delivered to the trap. The agreement between pre-
dicted and measured frequencies is within a few percent.
Finally, we determine numerically the overall potential
depth UT in the pseudopotential approximation [21, 22]
for the different control and RF potentials (also shown
ments (Exp.) and simulated values (Sim.)) for three different
potential configurations explained in the text. The axial fre-
quency is denoted f?, while f⊥1 and f⊥2 are the frequencies
of the two transverse modes whose axes are indicated by the
cross in Fig. 3. The uncertainties in the experimental values
for the frequencies are approximately 0.10 MHz.
Oscillation frequencies (experimental measure-
(MHz) (MHz) (MHz)
−20 −10 10 20
ping potential (both RF pseudopotential and static poten-
tials included) for potentials corresponding to V5 = 5.00 V
and VRF = 103.5 V (UT = 177 meV). The cross indicates the
directions of the normal mode axes n⊥1 and n⊥2, and the ex-
pected position for an ion at the center of the trap in the ˆ x-ˆ y
plane. The separation between contour lines corresponds to
5 meV and the colors blue to red correspond to low to high
potential. The electrodes are depicted to scale in the lower
part of the figure and labeled as in Fig. 2b.
A transverse cross-section of the simulated trap-
in Table I). A transverse cross-section (the ˆ x-ˆ y plane
of Fig. 1b) of the trapping potential for V5 = 5 V and
VRF= 103.5 V near the central trap region is shown in
In Fig. 4, we show groups of ions for the VRF= 103.5
V and V5 = 2.00 V configuration. The separation of
the ions is related to the center-of-mass axial oscilla-
tion frequency , and the horizontal bars indicate the
expected ion separations according to the measurement
of this frequency (1.84 ± 0.10 MHz for this configura-
tion). When the number of ions becomes large enough,
the string breaks into a zig-zag configuration (see for ex-
Because24Mg+lacks hyperfine structure, we cannot
easily determine heating rates near the quantum limit
(without applying large magnetic fields) due to poor in-
ternal state discrimination . However, an approxi-
mate value of the transverse mode heating rate can be de-
termined by observing how long an ion remains trapped
in a shallow well in the absence of the Doppler cooling
light .If we assume the heating rate (the energy
change per unit of time) is constant until the ion can
overcome the saddle-point in Fig. 3, we can estimate the
initial heating rate as ?r? ≃ UT/(?τs?hf⊥1) quanta per
second, where ?τs? is the average survival time of the ion
in absence of cooling light, h is Planck’s constant and f⊥1
is the smaller of the two transverse mode frequencies.
To help ensure that the results are not influenced by
the ions thermalizing to the ambient temperature (∼ 300
ions loaded into the surface trap (red corresponds to high-
est fluorescence count rate). The length scale is determined
from a separate image of the electrodes whose dimensions are
known. The horizontal bars indicate the separation distance
between the ions as predicted from the measured axial oscilla-
tion frequency. The ratio between transverse and axial oscil-
lation frequencies makes it energetically favorable for the 12
ion string to break into a zig-zag shape (see for example ).
False-color images of one, two, three, six, and 12
K ≃ 25 meV) we perform the measurements of ?τs? with
a very shallow well depth, UT = 6 ± 1 meV (V5= 5 V,
VRF= 46.2 V, and f⊥1≃ 5.3 MHz). From these data,
we determine ?τs? = 53 ± 10 s, and find ?r? ≃ 5 quanta
Johnson noise in the resistance of the RC filters on the
control electrodes is a potential source of heating , but
we theoretically estimate it to contribute only 1 quantum
per second. Therefore the observed heating is apparently
dominated by anomalous heating as observed in previous
experiments . We note that in presence of Doppler
cooling light, the lifetime of the ion is several hours (in
the trap with UT = 177 meV), presumably limited by
chemical reactions .
Heating and decoherence rates for small ion traps
will be an important issue for future implementations
of quantum information processing, and will need to be
more thoroughly investigated. Therefore, an important
next step is to replace the24Mg+with an isotope that
allows for Raman sideband cooling, such as25Mg+or
9Be+, and to study heating rates near the ground state
of motion [24, 25]. A longer-term goal is to design and
fabricate a surface trap with more complex structures
involving T- or X-junctions with separate memory- and
We thank E. Langlois and J. Moreland for advice on
the micro-fabrication processes and Y. Le Coq for help-
ful comments on the manuscript. This work was sup-
ported by the Advanced Research and Development Ac-
tivity (ARDA) under contract MOD-7171.05 and NIST.
S. S. wishes to thank the Carlsberg Foundation for finan-
cial support. This manuscript is a publication of NIST
and not subject to U. S. copyright.
∗Electronic address: firstname.lastname@example.org
†Present Address: Los Alamos National Laboratory
 See for instance http://qist.lanl.gov/qcomp map.shtml.
 DiVinzenco, Fortschr. Phys. 48, 771 (2000).
 D.J. Wineland, C. Monroe, W.M. Itano, D. Leibfried,
B.E. King, and D.M. Meekhof, J. Res. Nat. Inst. Stand.
Tech. 103, 259 (1998).
 R.G. DeVoe, Phys. Rev. A 58, 910 (1998).
 J.I. Cirac and P. Zoller, Nature 404, 579 (2000).
 D. Kielpinski, C. Monroe and D.J. Wineland, Nature
417, 709 (2002).
 L.-M. Duan, B.B. Blinov, D.L. Moehring, and C. Mon-
roe, Quant. Inform. Comp. 4, 165 (2004).
 M.A. Rowe, A. Ben-Kish, B. DeMarco, D. Leibfried, V.
Meyer, J. Beall, J. Britton, J. Hughes, W.M. Itano, B. Je-
lenkovic, C. Langer, T. Rosenband, and D.J. Wineland,
Quant. Inform. Comp. 2, 257 (2002).
 M.D. Barrett, J. Chiaverini, T. Schaetz, J. Britton, W.M.
Itano, J.D. Jost, E. Knill, C. Langer, D. Leibfried, R.
Ozeri, D.J. Wineland, Nature 429, 737 (2004).
 W.K. Hensinger, S. Olmschenk, D. Stick, D. Hucul, M.
Yeo, M. Acton, L. Deslauriers, J. Rabchuk, C. Monroe,
 D. Stick, W.K. Hensinger, S. Olmschenk, M.J. Madsen,
K. Schwab and C. Monroe, Nature Physics 2, 36 (2006).
 D.J. Wineland, D. Leibfried, M.D. Barrett, A. Ben-Kish,
J.C. Bergquist, R.B. Blakestad, J.J. Bollinger, J. Britton,
J. Chiaverini, B. DeMarco, D. Hume, W.M. Itano, M.
Jensen, J.D. Jost, E. Knill, J. Koelemeij, C. Langer, W.
Oskay, R. Ozeri, R. Reichle, T. Rosenband, T. Schaetz,
P.O. Schmidt, S. Seidelin, in Proc. XVII Int. Conf. on
Laser Spectroscopy, Avemore, Scotland, 2005, ed. by E.
A. Hinds, A. Ferguson, and E. Riis (World Scientific,
Singapore, 2005), pp. 393-402 (quant-ph/0508025).
 J. Chiaverini, R.B. Blakestad, J. Britton, J.D. Jost,
C. Langer, D. Leibfried, R. Ozeri and D.J. Wineland,
Quant. Inform. Comp. 5, 419 (2005).
 R. Folman, P. Kr¨ uger, J. Schmiedmayer, J. Denschlag
and C. Henkel, Adv. Atom. Mol. Phys. 48, 263 (2002).
 J. Kim, S. Pau, Z. Ma, H.R. McLellan, J.V. Gages, A.
Kornblit, and R.E. Slusher, Quant. Inform. Comp. 5, 515
 C.E. Pearson, D.R. Leibrandt, W.S. Bakr, W. J. Mallard,
K. R. Brown, and I. L. Chuang, quant-ph/0511018.
 W. Paul, Rev. Mod. Phys. 62, 531 (1990).
 S.R. Jefferts, C. Monroe, E.W. Bell, and D.J. Wineland,
Phys. Rev. A 51, 3112 (1995).
 D.N. Madsen, S. Balslev, M. Drewsen, N. Kjærgaard, Z.
Videsen and J.W. Thomsen, J. Phys. B: At. Mol. Opt
Phys. 33, 4981 (2000).
 D.J. Berkeland, J.D. Miller, J.C. Bergquist, W.M. Itano,
and D.J. Wineland, J. Appl. Phys. 83, 5025 (1998).
 H.G. Dehmelt, Adv. Atom. Mol. Phys. 3, 53 (1967).
 P.K. Ghosh, Ion Traps (Clarendon Press, Oxford, 1995).
 Daniel H.E. Dubin, Phys. Rev. Lett. 71, 2753 (1993).
 C. Monroe, D.M. Meekhof, B.E. King, S.R. Jefferts,
W.M. Itano, D.J. Wineland and P. Gould, Phys. Rev.
Lett. 75, 4011 (1995).
 Q.A. Turchette, Kielpinski, B.E. King, D. Leibfried,
D.M. Meekhof, C.J. Myatt, M.A. Rowe, C.A. Sackett,
C.S. Wood, W.M. Itano, C. Monroe, and D.J. Wineland,
Phys. Rev. A. 61, 063418 (2000).
 K. Mølhave and M. Drewsen, Phys. Rev. A. 62, 011401