∆N (Electrons per probe pulse) Spin relaxation time T1 (ns)
Downloaded from www.sciencemag.org on November 3, 2010

REPORTS A

B
250

-10

200

5.5 T

150

-1

100

3.0 T

50

1.5 T

-0.1 0

200

400

600

Delay time ∆t (ns)

0

800

0

1

2

3

4

5

6

7

8

Magnetic field (T)

Fig. 4. Field and site dependence of the spin relaxation time. (A) Pump-probe measurements for different magnetic fields on an Fe-Cu dimer; solid lines are exponential fits. (B) T1 as a function of magnetic field for the two Fe-Cu dimers shown in the accompanying 5-nm by 5-nm STM topographs.

One of the Fe-Cu dimers in Fig. 4B always exhibits a larger T1 than the other. We speculate that this variation is due to differences in the nearby surface features as seen in the accompanying topographs. This observation emphasizes the capability of the all-electronic pump-probe technique presented here to resolve local variations in the spin relaxation time with atomic precision.
The pump-probe scheme we have described can be used to monitor the temporal evolution of any excitation provided (i) the excitation can be driven by tunneling electrons; (ii) the conductance of the tunnel junction exhibits a postexcitation time dependence; and (iii) the system evolves on an accessible time scale. Excitations fulfilling these requirements include long-lived vibrational excitations, conformational changes of molecules (26) such as in molecular motors (27), or fast localized heating (28). We emphasize that this pump-probe scheme can in principle be used to monitor the dynamical evolution of the excited state, not just its relaxation; with sufficient temporal resolution it should be possible to monitor the vibration of an atom or molecule and even the precession of a spin.

References and Notes 1. C. P. Slichter, Principles of Magnetic Resonance
(Springer, Berlin and New York, 1996). 2. D. Rugar, R. Budakian, H. J. Mamin, B. W. Chui,
Nature 430, 329 (2004). 3. J. R. Maze et al., Nature 455, 644 (2008). 4. G. Balasubramanian et al., Nature 455, 648 (2008). 5. M. Bode et al., Nature 447, 190 (2007). 6. A. J. Heinrich, J. A. Gupta, C. P. Lutz, D. M. Eigler, Science
306, 466 (2004). 7. N. Tsukahara et al., Phys. Rev. Lett. 102, 167203
(2009). 8. C. F. Hirjibehedin et al., Science 317, 1199 (2007). 9. X. Chen et al., Phys. Rev. Lett. 101, 197208 (2008). 10. C. F. Hirjibehedin, C. P. Lutz, A. J. Heinrich, Science 312,
1021 (2006). 11. M. J. Rost et al., Rev. Sci. Instrum. 76, 053710 (2005). 12. C. Durkan, M. E. Welland, Appl. Phys. Lett. 80, 458
(2002). 13. G. Nunes Jr., M. R. Freeman, Science 262, 1029
(1993). 14. O. Takeuchi et al., Appl. Phys. Lett. 85, 3268 (2004). 15. S. Loth et al., Nat. Phys. 6, 340 (2010). 16. R. Wiesendanger, Rev. Mod. Phys. 81, 1495 (2009). 17. Materials and methods are available as supporting
material on Science Online. 18. M. N. Leuenberger, D. Loss, Nature 410, 789 (2001). 19. D. E. Freedman et al., J. Am. Chem. Soc. 132, 1224
(2010). 20. A. Ardavan et al., Phys. Rev. Lett. 98, 057201
(2007).

21. T. Choi, C. D. Ruggiero, J. A. Gupta, Phys. Rev. B 78, 035430 (2008).
22. F. Komori, S.-Y. Ohno, K. Nakatsuji, Prog. Surf. Sci. 77, 1 (2004).
23. D. Gatteschi, R. Sessoli, J. Villain, Molecular Nanomagnets (Oxford Univ. Press, New York, 2006).
24. C. Sangregorio, T. Ohm, C. Paulsen, R. Sessoli, D. Gatteschi, Phys. Rev. Lett. 78, 4645 (1997).
25. W. Wernsdorfer, R. Sessoli, Science 284, 133 (1999). 26. K. Henzler-Wildman, D. Kern, Nature 450, 964 (2007). 27. W. R. Browne, B. L. Feringa, Nat. Nanotechnol. 1,
25 (2006). 28. B. Barwick, H. S. Park, O.-H. Kwon, J. S. Baskin,
A. H. Zewail, Science 322, 1227 (2008). 29. J. A. Gupta, C. P. Lutz, A. J. Heinrich, D. M. Eigler,
Phys. Rev. B 71, 115416 (2005). 30. We thank B. Melior for technical assistance. S.L., C.P.L.,
and A.J.H. acknowledge financial support from the Office of Naval Research, S.L. from the Alexander von Humboldt Foundation, and M.E. from European Science Foundation project FunSMARTs II.
Supporting Online Material www.sciencemag.org/cgi/content/full/329/5999/1628/DC1 Materials and Methods Figs. S1 to S3 References Movie S1
30 April 2010; accepted 4 August 2010 10.1126/science.1191688

Optical Clocks and Relativity
C. W. Chou,* D. B. Hume, T. Rosenband, D. J. Wineland
Observers in relative motion or at different gravitational potentials measure disparate clock rates. These predictions of relativity have previously been observed with atomic clocks at high velocities and with large changes in elevation. We observed time dilation from relative speeds of less than 10 meters per second by comparing two optical atomic clocks connected by a 75-meter length of optical fiber. We can now also detect time dilation due to a change in height near Earth’s surface of less than 1 meter. This technique may be extended to the field of geodesy, with applications in geophysics and hydrology as well as in space-based tests of fundamental physics.

Albert Einstein’s theory of relativity forced us to alter our concepts of reality. One of the more startling outcomes of the theory is that we have to give up our notions of simul-

taneity. This is manifest in the so-called twin paradox (1), in which a twin sibling who travels on a fast-moving rocket ship returns home younger than the other twin. This “time dilation” can be

quantified by comparing the tick rates of identical clocks that accompany the traveler and the stationary observer. Another consequence of Einstein’s theory is that clocks run more slowly near massive objects. In the range of speeds and length scales encountered in our daily life, relativistic effects are extremely small. For example, if two identical clocks are separated vertically by 1 km near the surface of Earth, the higher clock emits about three more second-ticks than the lower one in a million years. These effects of relativistic time dilation have been verified in several important experiments (2–6)
Time and Frequency Division, National Institute of Standards and Technology (NIST), Boulder, CO 80305, USA.
*To whom correspondence should be addressed. E-mail: chinwen@nist.gov

1630

24 SEPTEMBER 2010 VOL 329 SCIENCE www.sciencemag.org

1S0↔3P0 Transition probability Downloaded from www.sciencemag.org on November 3, 2010

REPORTS

and are accounted for routinely in satellite-based
navigation systems (7). The most accurate de-
terminations historically involve velocities near
the speed of light (8) and changes in elevation of 104 to 107 m (3, 4). Previously, small relativistic shifts (<10−16) could be observed only over short
distances in g-ray Mössbauer spectroscopy (5, 9)
and atom interferometry (6). However, they might
be detected over longer distances by clocks with
sufficiently high sensitivity, such as accurate
atomic clocks operating in the optical regime, or “optical clocks.” Here we report the detection of relativistic time dilation due to velocities of sev-
eral meters per second and, separately, due to a
change in height of 0.33 m by comparing two optical clocks based on 27Al+ ions.
With operating frequencies in the petahertz (1 PHz = 1015 Hz) range and natural linewidths
at the millihertz level, optical clocks have dem-
onstrated substantial improvements in stability
and accuracy over the current microwave time
and frequency standards (10). We compared two
optical atomic clocks based on individual trapped Al+ ions, with reported systematic frequency uncertainties of 8.6 × 10−18 (11) and 2.3 × 10−17
(12). For comparison, the lowest reported fre-
quency uncertainty of Cs fountain clocks is 3.4 × 10−16 (13). With the accuracy and the associated sensitivity of these Al+ optical clocks, frequency variations below 10−16 in the clocks due to
relativistic effects can be observed.
Due to the lack of an accessible optical transition in 27Al+ for efficient laser cooling and
state detection, precision spectroscopy of these
ions uses techniques developed in quantum information science. Here, an Al+ ion is sympa-
thetically cooled through its Coulomb interaction with an auxiliary “logic” ion that is simultaneously held in the same linear radio-frequency (RF) Paul
trap (14). The logic ion also helps prepare and detect the internal state of the Al+ ion via quantum logic protocols. In this work, the two Al+ clocks used a beryllium (9Be+) ion (14) and a magnesium (25Mg+) ion (11), respectively, as the logic ion. The Al+ 1S0↔3P0 clock transition with frequency f0 near 1.121 PHz has a narrow (∆f = 8 mHz) natural linewidth and a corresponding intrinsic quality (Q) factor of f0/∆f = 1.4 × 1017 that permits high sensitivity for detecting small
frequency-shifting effects. However, the observed
linewidth for the clock transition is limited by the
linewidth of the probe laser. We probed the clock
transition with a subhertz linewidth laser refer-
enced to a high-finesse optical cavity (15). In the
Al-Mg clock, with 300 ms probe duration we obtained a narrow, Fourier transform–limited linewidth, realizing a Q factor of 4.2 × 1014 with nearly
80% contrast (Fig. 1). This high-Q line provides
the basis for high-stability clock operation and sen-
sitivity to small frequency shifts. The two Al+ optical clocks were located in
separate laboratories and were compared by trans-
mitting the stable clock signal through a 75-m
length of phase-stabilized optical fiber. To observe time dilation due to motion, we set the Al+

ion in the Al-Mg clock in motion by applying a small static electric field that shifts the position of the ion slightly away from the null of the confining RF field (16). The ion is thereby subject to an RF electric field at fRF = 59 MHz (Fig. 2) and

undergoes harmonic motion. The ion velocity is adjusted by varying the applied static electric field. In the language of the twin paradox, the moving Al+ ion is the traveling twin, and its harmonic motion amounts to many round trips. Time dila-

Fig. 1. The 1S0↔3P0 tran-

1

sition in Al+ observed with

probe pulses of 300 ms duration. Each datum (blue

0.8

dots) consists of an aver-

age of 9 or 10 individual probes, and the sequence

0.6

of probe frequencies was

2.7 Hz (FWHM)

randomized to eliminate

0.4

skewing effects due to fre-

Q = 4.2 x 1014

quency drifts in the probe

laser. A Lorentzian fit to

0.2

this data (not shown in

the figure) produces a linewidth [full width at half maximum (FWHM)] of 2.30 T 0.51 Hz (uncertain-

0

−6 −4 −2

0

2

4

6

Frequency offset (Hz) near 1.121 PHz

ties correspond to 1 SD).

The Lorentzian linewidth is narrower than the Fourier transform–limited spectrum (smooth fitted red

curve) that is the result of the rectangular temporal waveform of the probe pulses. For the probe time used here, the Fourier limit corresponds to a linewidth (FWHM) of 2.7 Hz and a Q factor of 4.2 × 1014, which we

take as the achieved values. The yellow shaded band indicates the expected scatter of the data due to

quantum projection noise (34).

Fig. 2. Relativistic time dilation at familiar speeds (10 m/s = 36 km/hour ≈ 22.4 miles/hour). (Lower left inset) As the Al+ ion in one of the twin clocks is displaced from the null of the confining RF quadrupole
field (white field lines), it undergoes harmonic motion and experiences relativistic time dilation. In the
experiments, the motion is approximately perpendicular to the probe laser beam (indicated by the blue shading). The Al+ ion clock in motion advances at a rate that is slower than its rate at rest. In the figure,
the fractional frepquffiffieffiffinffifficffiy difference between the moving clock and the stationary clock is plotted versus the velocity (vrms = 〈v2〉) (rms, root mean square) of the moving clock. The solid curve represents the theoretical prediction. (Upper right inset) A close-up of the results for vrms < 10 m/s in the dashed box. The vertical error bars represent statistical uncertainties, and the horizontal ones cover the spread of measured
velocities at the applied electric fields.

www.sciencemag.org SCIENCE VOL 329 24 SEPTEMBER 2010

1631

REPORTS A
g

1
B
0.5
0

δ f/ f (10−16)
Downloaded from www.sciencemag.org on November 3, 2010

−0.5

−1

Fig. 3. Gravitational time dilation at the scale of daily life. (A) As one of the clocks is raised, its rate increases when compared to the clock rate at deeper gravitational potential. (B) The fractional difference in frequency between two Al+ optical clocks at different heights. The Al-Mg clock was initially 17 cm lower in height than the Al-Be clock, and subsequently, starting at data point 14, elevated by 33 cm. The net relative shift due to the increase in

−1.5 0

5

10

15

Measurement number

height is measured to be (4.1 T 1.6) × 10−17. The vertical error bars represent statistical uncertainties (reduced c2 = 0.87). Green lines and yellow shaded bands indicate, respectively, the averages and statistical uncertainties for the first 13 data points (blue symbols) and the remaining 5 data points (red symbols). Each data point represents about 8000 s of clockcomparison data.

tion from this motion leads to a fractional frequency shift for the moving clock of (17)

df f0

¼

〈gð1

1 − v∥=cÞ〉

−

1

ð1Þ

Here v|| is the velocity of the Al+ ion along thepwffiffiaffiffivffiffieffiffiffiffivffiffiffieffifficffiffitffioffi r of the probe laser beam g ¼ 1= 1 − v2=c2, c is the speed of light, v is the ion’s velocity with respect to the laboratory reference frame, and f0 is the ion’s proper resonant frequency. Angle brackets denote time averages. Because the induced Al+ ion motion is harmonic, its contribution to 〈v||〉 averages to zero; therefore, any observed change in the ion’s transition frequency is due to a change in g and corresponds to
relativistic time dilation (18). For v/c << 1, Eq. 1 can be approximated by df/f0 ≈ −〈v2〉/2c2 (17). We measured the frequency difference between the
two clocks (df/f0) while varying the velocity of the ion motion. The experimental results, which con-
firm the prediction of Eq. 1, are plotted in Fig. 2.
Differences in gravitational potential can be de-
tected by comparing the tick rate of two clocks. For
small height changes on the surface of Earth, a clock that is higher by a distance ∆h runs faster by

df f0

¼

gDh c2

ð2Þ

where g ≈ 9.80 m/s2 is the local acceleration due to gravity (4). The gravitational shift corresponds to a clock shift of about 1.1 × 10−16 per meter of
change in height. To observe this shift, we first compared the frequencies of the two Al+ clocks at the original height difference of ∆h = h(Mg-Al) − h(Be-Al) = −17 cm, which was measured with a laser level. Then we elevated the optical table on
which the Mg-Al clock was mounted, supporting
it on platforms that increased the height by 33 cm,
and compared the frequencies again. The two mea-

surements consist of approximately 100,000 s of
low-height data and 40,000 s of high-height data,
and the clocks exhibit (Fig. 3) a fractional frequency change of (4.1 T 1.6) × 10−17. When this shift is interpreted as a measurement of the change
in height of the Al-Mg clock, the result of 37 T 15 cm agrees well with the known value of 33 cm.
Although ideally 〈v||〉 = 0, small linear velocities of the Al+ ions can occur because of effects
such as slow electrical charging of insulating material in the trap. From Eq. 1, the clock’s frequency (that is, the frequency of the probe laser locked to the moving ion’s clock transition) exhibits a fractional frequency shift

df ≈ 〈v∥〉

ð3Þ

f0

c

if the Al+ ion is moving at an average velocity 〈v||〉 in the propagation direction of a probe laser beam. In the comparison measurements between the Al+ clocks, the Doppler effect was carefully constrained by alternate use of probe laser beams counterpropagating with respect to each other (11). Any motion of the ion is detected as a difference in the transition frequencies measured by the two laser beams. In the Al-Mg clock, we observed a fractional frequency difference of (1.2 T 0.7) × 10−17 between the two probe directions, which corresponds to the ion moving at a speed of (1.8 T 1.1) nm/s in the lab frame. However, the clock rate is not significantly affected by a velocity of this magnitude, because it is derived from an average of the two opposite laser-probe directions.
Small relativistic effects reported here have been observed with optical atomic clocks of unprecedented precision and accuracy. With improved accuracy, the sensitivity of optical clocks to small variations in gravitational potential might find applications in geodesy (19, 20), hydrology (21), and tests of fundamental physics in space

(22). The basic components for clock-based geodetic measurements were demonstrated here by comparing two accurate Al+ optical clocks through 75 m of noise-canceled fiber and measuring height-dependent clock shifts. In clock-based geodesy (23, 24), accurate optical clocks would be linked to form a network of “inland tide gauges” (25) that measure the distance from Earth’s surface to the geoid: the equipotential surface of Earth’s gravity field that matches the global mean sea level. Such a network could operate with high temporal (daily) and geospatial resolution at the clock locations. It would therefore complement geodetic leveling networks, whose update period is typically 10 years or longer, as well as biweekly satellite-generated global geoid maps.
For a network to be useful, clock accuracy must be improved to 10−18 or better (26–28) to allow for height measurements with 1-cm uncertainty. In Al+ clocks, improved control of the ion motion is needed to reduce the uncertainty of motional time dilation, and issues of reliability must be addressed, so that the clocks can operate unattended for long periods. High-quality links are also needed to connect the optical clocks. Realistic link demonstrations with telecommunications fiber akin to the links used in this work have shown that optical frequencies can be transmitted across fiber lengths of up to 250 km with inaccuracy below 10−18 (29–31), and continentscale demonstrations are in progress (30). However, intercontinental links may require the faithful transmission of optical carrier frequencies to satellites through the atmosphere, and this is an unsolved problem under active investigation (32, 33).
References and Notes 1. A. Einstein, Annal. Physik 17, 891 (1905). 2. B. Rossi, D. B. Hall, Phys. Rev. 59, 223 (1941). 3. J. C. Hafele, R. E. Keating, Science 177, 166 (1972). 4. R. F. C. Vessot et al., Phys. Rev. Lett. 45, 2081
(1980).

1632

24 SEPTEMBER 2010 VOL 329 SCIENCE www.sciencemag.org

REPORTS

5. R. V. Pound, G. A. Rebka, Phys. Rev. Lett. 4, 337 (1960).
6. H. Müller, A. Peters, S. Chu, Nature 463, 926 (2010). 7. N. Ashby, M. Weiss, NIST Technical Note 1385 (NIST,
Boulder, CO, 1999). 8. S. Reinhardt et al., Nat. Phys. 3, 861 (2007). 9. W. Potzel et al., Hyperfine Interact. 72, 195 (1992). 10. P. Gill, Metrologia 42, S125 (2005). 11. C. W. Chou, D. B. Hume, J. C. J. Koelemeij,
D. J. Wineland, T. Rosenband, Phys. Rev. Lett. 104, 070802 (2010). 12. T. Rosenband et al., Science 319, 1808 (2008). 13. T. E. Parker, Metrologia 47, 1 (2010). 14. P. O. Schmidt et al., Science 309, 749 (2005). 15. B. C. Young, F. C. Cruz, W. M. Itano, J. C. Bergquist, Phys. Rev. Lett. 82, 3799 (1999). 16. D. J. Berkeland, J. D. Miller, J. C. Bergquist, W. M. Itano, D. J. Wineland, J. Appl. Phys. 83, 5025 (1998). 17. See supporting material on Science Online.

18. D. Hasselkamp, E. Mondry, A. Scharmann, Z. Physik A 289, 151 (1979).
19. D. Kleppner, Phys. Today 59, 10 (2006). 20. B. D. Tapley, S. Bettadpur, J. C. Ries, P. F. Thompson,
M. M. Watkins, Science 305, 503 (2004). 21. R. Schmidt et al., Surv. Geophys. 29, 319 (2008). 22. G. Tino et al., Nucl. Phys. B Proc. Suppl. 166, 159 (2007). 23. A. Bjerhammar, Bull. Geod. 59, 207 (1985). 24. J. Müller, M. Soffel, S. A. Klioner, J. Geod. 82, 133
(2008). 25. O. Colombo, Rep. No. 296 (Department of Geodetic
Science, Ohio State University, Columbus, OH, 1980). 26. H. G. Dehmelt, IEEE Trans. Instrum. Meas. 31, 83 (1982). 27. D. J. Wineland, W. M. Itano, J. C. Bergquist, R. G. Hulet,
Phys. Rev. A 36, 2220 (1987). 28. M. Takamoto, H. Katori, S. I. Marmo, V. D. Ovsiannikov,
V. G. Pal’chikov, Phys. Rev. Lett. 102, 063002 (2009). 29. N. R. Newbury, P. A. Williams, W. C. Swann, Opt. Lett.
32, 3056 (2007).

30. G. Grosche et al., Opt. Lett. 34, 2270 (2009). 31. F. Kéfélian et al., Opt. Lett. 34, 1573 (2009). 32. B. Sprenger, J. Zhang, Z. H. Lu, L. J. Wang, Opt. Lett. 34,
965 (2009). 33. K. Djerroud et al., Opt. Lett. 35, 1479 (2010). 34. W. M. Itano et al., Phys. Rev. A 47, 3554 (1993). 35. This work was supported by the Office of Naval
Research and NIST. We thank N. Pavlis and M. Weiss for helpful discussions and J. C. Bergquist and N. Ashby for suggestions on the manuscript. Not subject to U.S. copyright.
Supporting Online Material www.sciencemag.org/cgi/content/full/329/5999/1630/DC1 Methods References
24 May 2010; accepted 5 August 2010 10.1126/science.1192720

Downloaded from www.sciencemag.org on November 3, 2010

Alkali-Stabilized Pt-OHx Species Catalyze Low-Temperature Water-Gas
Shift Reactions

Recently, Mavrikakis and co-workers showed that a carboxyl-mediated mechanism is responsible for the WGS reaction on Cu (5) and other metallic sites, including Pt (6). In both the redox and carboxyl mechanisms, the activation of H2O, leading to the formation of the species responsible for oxidizing CO, is a kinetically important

Yanping Zhai,1 Danny Pierre,1* Rui Si,1† Weiling Deng,1‡ Peter Ferrin,2 Anand U. Nilekar,2 Guowen Peng,2 Jeffrey A. Herron,2 David C. Bell,3 Howard Saltsburg,1
Manos Mavrikakis,2§ Maria Flytzani-Stephanopoulos1§

step (7). Furthermore, the supply of oxidizing species and the related WGS activity has been shown to be high on metals supported on reducible oxides (8), whereas no WGS activity has

We report that alkali ions (sodium or potassium) added in small amounts activate platinum adsorbed on alumina or silica for the low-temperature water-gas shift (WGS) reaction (H2O + CO → H2 + CO2) used for producing H2. The alkali ion–associated surface OH groups are activated by CO at low temperatures (~100°C) in the presence of atomically dispersed platinum. Both experimental evidence
and density functional theory calculations suggest that a partially oxidized Pt-alkali-Ox(OH)y species is the active site for the low-temperature Pt-catalyzed WGS reaction. These findings are
useful for the design of highly active and stable WGS catalysts that contain only trace amounts
of a precious metal without the need for a reducible oxide support such as ceria.

been reported for Pt supported on irreducible oxides, such as silica.
A recent key finding is that oxidized Pt (and Au) species in ceria are the catalytic sites for the WGS reaction (2, 9); Pt or Au nanoparticles are spectator species. This observation has also been made for Au on Fe3O4 catalysts (10, 11). On these oxide surfaces, nanoparticles, clusters, and atoms (ions) of Pt or Au coexist, but the chemistry takes place on the atomically dispersed Ptd+ or Aud+

The water-gas shift (WGS) reaction is a key step in all carbon-based fuel processing aimed at producing and upgrading hydrogen

The WGS reaction mechanism on Pt-CeO2 has been debated as either involving formate intermediates (3) or a simple redox chemistry (4).

species and involves the neighboring OH groups. Aberration-corrected high angle annular dark field (HAADF) scanning transmission electron micros-

(1). For use in low-temperature fuel cells, the WGS

catalysts must be nonpyrophoric, active over a wide temperature range, and stable during frequent shutdown-restart operation cycles. To meet these criteria, a reemergence of platinum-group metals has occurred in recent years, and various combinations are being evaluated as low-temperature WGS

Table 1. Physical properties and reducibility of parent and washed (deionized water–washed samples were heated to 400°C in air 4 hours) xPt-yNa-SiO2 samples (x and y are atomic percentages.). Bulk composition was measured by inductively coupled plasma optical emission spectrometry; surface concentration, by XPS; surface [O], from CO-TPR (5% CO–He, from room temperature to 400°C, at 10°C/min); and surface [OH] from CO-TPR using only the CO2 peak associated with the production of H2. The XPSmeasured value of surface Pt was used in calculating the [OH]/Pt ratio. Dash entries indicate nondetectable

catalysts (2). Nanoscale ceria is the support of amounts, and NA, not applicable.

choice because of its high concentration of oxygen

defects and stabilization of active Pt species (2).

1Department of Chemical and Biological Engineering, Tufts University, Medford, MA 02155, USA. 2Department of Chemical and Biological Engineering, University of Wisconsin, Madison, WI 53706, USA. 3School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA.
*Present address: Cabot Corporation, 157 Concord Road, Billerica,

Sample
3Na-SiO2 3Na-SiO2
washed

Bulk composition (atomic %)

Pt

Na

0

3.0

0

–

Surface concentration (atomic %)

Pt

Na

0

4.0

0

–

Surface [O]
(mmol/gcat)

Surface Na/Pt

Surface Na/O

Surface [OH]/Pt

500

NA

1.5:1

NA

–

NA

–

NA

MA 01821, USA. †Present address: Chemistry Department, Brookhaven National Laboratory, Upton, NY 11973, USA.

1Pt-1Na-SiO2 1Pt-1Na-SiO2

1 1

1.0 0.7

0.3 0.4

0.9 0.5

‡Present address: Chemical Sciences and Engineering Divi-

washed

210

3:1

0.7:1

2:1

185

1.2:1 1:1.5

2:1

sion, Argonne National Laboratory, Argonne, IL 60439, USA. 1Pt-3Na-SiO2

1

3.0

0.3

3.9

450

13:1 1.4:1

3:1

§To whom correspondence should be addressed. E-mail: maria.flytzani-stephanopoulos@tufts.edu (M.F.-S.); manos@ engr.wisc.edu (M.M.)

1Pt-3Na-SiO2 washed

1

1.6

0.4

1.2

245

3:1

1:1

3:1

www.sciencemag.org SCIENCE VOL 329 24 SEPTEMBER 2010

1633