Mark W Keller | National Institute of Standards and Technology (original) (raw)

Papers by Mark W Keller

Journal of low temperature physics, May 19, 2024

Bulletin of the American Physical Society, Mar 12, 2008

(SET) devices, manifesting as a time-dependent instability. Through a compendium of drift measure... more (SET) devices, manifesting as a time-dependent instability. Through a compendium of drift measurements on SET transistors fabricated in five different laboratories, we can show that the drift is endemic in metal-based devices, but is absent in Si-based devices. Given that it is well-known that two-level fluctuators (TLF's) exist in Si devices, the question naturally arises: why is the long-term drift so much better in the Si-based devices? Our answer: the TLF's in Si devices are stable over time, thermal cycling, etc., whereas the TLF's in the metalbased devices are unstable, and exist in interacting glass-like state. Following these observations, we have developed a model based on the theory of heat evolution in glasses that quantitatively agrees with the rate of charge offset drift in metal-based devices. Finally, we suggest some particular directions for future fabrication that may eliminate this problem.

Applied Physics Letters, Aug 21, 2006

APS March Meeting Abstracts, Mar 1, 1997

ABSTRACT Fluctuations in the background charges of the tunnel junctions in the electron pump limi... more ABSTRACT Fluctuations in the background charges of the tunnel junctions in the electron pump limit the accuracy of the device over long times. We have measured the background charge on each junction as follows. We create a charge bias by applying voltages of opposite polarity to the gates on either side of the junction. As this bias is swept from 0 to e, the junction allows tunneling when the sum of the background charge and the applied charge is equal to 0.5e. Thus if tunneling occurs at an applied bias of 0.6e, we infer a background charge of -0.1e. Since the tunneling signal is thermally broadened, the measurement also yields the electron temperature of the junctions. We report several results based on this technique. 1) The rate and amplitude of the background charge fluctuations decrease over a period of about 2 weeks at 40 mK. 2) The electron temperature follows the refrigerator temperature down to 40 mK. This shows conclusively that electron heating cannot explain our earlier observation that pump accuracy is independent of temperature below 100 mK. 3) Using the known temperature, we infer the total capacitance of each junction and find nearly the same value for all junctions. These results significantly improve our ability to compare our observed error rates with theory.

Bulletin of the American Physical Society, Mar 18, 2013

In the context of a proposed new capacitance standard based on counting electrons, we discuss a m... more In the context of a proposed new capacitance standard based on counting electrons, we discuss a method to investigate its frequency dependence. We describe measurements of this frequency dependence using a technique that involves the same singleelectron tunneling (SET) devices used in the capacitance standard.

Physica E-low-dimensional Systems & Nanostructures, Mar 1, 2004

A single-electron transistor (SET) is used to detect tunneling of single electrons into individua... more A single-electron transistor (SET) is used to detect tunneling of single electrons into individual InGaAs self-assembled quantum dots (QDs). By using an SET with a small island area and growing QDs with a low density we are able to distinguish and measure three QDs. The bias voltage at which resonant tunneling into the dots occurs can be shifted using a surface gate electrode. From the applied voltages at which we observe electrons tunneling, we are able to measure the electron addition energies of three QDs.

Physical Review Letters, May 18, 1998

By measuring each junction of an electron pump in a single electron box configuration, we determi... more By measuring each junction of an electron pump in a single electron box configuration, we determine all quantities needed to test the standard theory of pumping error and leakage. Background charges are determined with an imprecision of 60.01e. Electron temperature is measured to 40 mK. Measured charging energies show all junctions have nearly the same capacitance. We find agreement with theory at 140 mK, but a disagreement of many orders of magnitude at 40 mK. We suggest that the excess error and leakage at 40 mK are due to photon-assisted cotunneling processes not included in the standard theory. [S0031-9007(98)06149-3]

Physical review, Apr 30, 2003

We have measured the counting errors of a seven-junction electron pump when charge is pumped agai... more We have measured the counting errors of a seven-junction electron pump when charge is pumped against a voltage difference. At finite bias voltages, we find that the errors increase exponentially with both pump voltage and temperature, in agreement with theoretical predictions. To compare experiment and simulation, all pump parameters were determined by independent electron-box experiments. Although we assume temperatures somewhat higher than those measured, simulations based on the ground-capacitance model yield excellent quantitative agreement with experiment and indicate that errors in the high-voltage regime are due to thermally activated tunneling. In addition, a surprising asymmetry between positive and negative voltages is explained by an asymmetry in the junction capacitances.

Physical review, Sep 15, 1999

Leakage and counting errors are explored experimentally in a well-characterized seven-junction el... more Leakage and counting errors are explored experimentally in a well-characterized seven-junction electron pump and compared with predictions of the orthodox theory, including cotunneling. Theory and experiment are in good agreement at intermediate temperatures, where errors are dominated by thermally activated, singlejunction processes. At low temperatures, however, the observed errors far exceed predictions, indicating that the orthodox theory omits an important error process.

Physical review, Dec 15, 2000

Computer simulations reveal that the lowest rates of leakage and counting errors observed in the ... more Computer simulations reveal that the lowest rates of leakage and counting errors observed in the electron pump can be explained by photon-assisted tunneling driven by 1/f noise. The noise power at microwave frequencies required to account for the observed errors is consistent with extrapolation of the low-frequency noise spectrum commonly recorded in single-electron transistors. Pump simulations, based on the groundcapacitance model, include cotunneling as well as single-junction photon-assisted tunneling. Quantitative agreement between theory and experiment is obtained for leakage and counting errors in pumps with four, five, six, and seven junctions in the limit of low temperatures and low counting rates. The effect of self-heating is explored.

Physical Review Letters, May 29, 2000

We measure photon-assisted tunneling in 4-and 6-junction electron pumps at photon frequencies up ... more We measure photon-assisted tunneling in 4-and 6-junction electron pumps at photon frequencies up to 60 GHz. We determine the microwave voltage at the pumps using noise thermometry. The standard theory of leakage in the electron pump, modified to include photon-assisted tunneling, describes our experiments well. From this test of theory we argue that, in the absence of external microwaves, photonassisted tunneling driven by 1͞f noise is an important error mechanism in electron pumps.

APS March Meeting Abstracts, Mar 1, 2003

We are studying the tunneling of individual electrons into self-assembled InAs quantum dots with ... more We are studying the tunneling of individual electrons into self-assembled InAs quantum dots with single-electron transistors. Aluminum single-electron transistors were fabricated over an epitaxial structure including self-assembled InAs quantum dots and an underlying n-doped layer. Sweeping the potential on the n-doped layer causes individual electrons to tunnel onto a quantum dot, and is detected capacitively with a single electron transistor.

Bulletin of the American Physical Society, Mar 7, 2007

arXiv (Cornell University), Dec 17, 2021

SPIE Proceedings, 2017

Tabletop-scale coherent EUV generated through high-harmonic generation (HHG) produces light in th... more Tabletop-scale coherent EUV generated through high-harmonic generation (HHG) produces light in the form of an attosecond pulse train that uniquely combines characteristics of good energy resolution (≈100-300meV) with sub-fs time resolution. This makes HHG an ideal source for studying the fastest dynamics in materials. Furthermore, using angle-resolved photoemission spectroscopy (ARPES), it is possible to extract detailed information about electron dynamics over the entire Brillouin zone. In recently published work, we combined HHG with ARPES to identify a sub-femtosecond excited-state lifetime for the first time. Photoemission occurs as a three-step process: 1) An electron is photoexcited from the valence band to far above the Fermi energy; 2) it transports to the surface, and 3) it overcomes the work function and exits. If the electron is promoted into a highlyexcited unoccupied band in the material (as opposed to a free-electron-like state), we observe the electron emission lifetime to increase in a measurable way-the Ni band 22 eV above the Fermi level has a lifetime of 212±30 attoseconds. Furthermore, by comparing photoemission from Cu and Ni, we reveal the influence of attosecond-timescale electron screening vs scattering by the electrons near the fermi surface. This work for the first time demonstrates the relevance of attosecond spectroscopy to the study of intrinsic properties and band structure in materials, as opposed to the strong-field induced dynamics studied extensively to-date.

Nature Physics, 2018

explains how the elementary charge will soon be reinstated in metrology-and why it got sidelined ... more explains how the elementary charge will soon be reinstated in metrology-and why it got sidelined in the first place.

Journal of low temperature physics, May 19, 2024

Bulletin of the American Physical Society, Mar 12, 2008

(SET) devices, manifesting as a time-dependent instability. Through a compendium of drift measure... more (SET) devices, manifesting as a time-dependent instability. Through a compendium of drift measurements on SET transistors fabricated in five different laboratories, we can show that the drift is endemic in metal-based devices, but is absent in Si-based devices. Given that it is well-known that two-level fluctuators (TLF's) exist in Si devices, the question naturally arises: why is the long-term drift so much better in the Si-based devices? Our answer: the TLF's in Si devices are stable over time, thermal cycling, etc., whereas the TLF's in the metalbased devices are unstable, and exist in interacting glass-like state. Following these observations, we have developed a model based on the theory of heat evolution in glasses that quantitatively agrees with the rate of charge offset drift in metal-based devices. Finally, we suggest some particular directions for future fabrication that may eliminate this problem.

Applied Physics Letters, Aug 21, 2006

APS March Meeting Abstracts, Mar 1, 1997

ABSTRACT Fluctuations in the background charges of the tunnel junctions in the electron pump limi... more ABSTRACT Fluctuations in the background charges of the tunnel junctions in the electron pump limit the accuracy of the device over long times. We have measured the background charge on each junction as follows. We create a charge bias by applying voltages of opposite polarity to the gates on either side of the junction. As this bias is swept from 0 to e, the junction allows tunneling when the sum of the background charge and the applied charge is equal to 0.5e. Thus if tunneling occurs at an applied bias of 0.6e, we infer a background charge of -0.1e. Since the tunneling signal is thermally broadened, the measurement also yields the electron temperature of the junctions. We report several results based on this technique. 1) The rate and amplitude of the background charge fluctuations decrease over a period of about 2 weeks at 40 mK. 2) The electron temperature follows the refrigerator temperature down to 40 mK. This shows conclusively that electron heating cannot explain our earlier observation that pump accuracy is independent of temperature below 100 mK. 3) Using the known temperature, we infer the total capacitance of each junction and find nearly the same value for all junctions. These results significantly improve our ability to compare our observed error rates with theory.

Bulletin of the American Physical Society, Mar 18, 2013

In the context of a proposed new capacitance standard based on counting electrons, we discuss a m... more In the context of a proposed new capacitance standard based on counting electrons, we discuss a method to investigate its frequency dependence. We describe measurements of this frequency dependence using a technique that involves the same singleelectron tunneling (SET) devices used in the capacitance standard.

Physica E-low-dimensional Systems & Nanostructures, Mar 1, 2004

A single-electron transistor (SET) is used to detect tunneling of single electrons into individua... more A single-electron transistor (SET) is used to detect tunneling of single electrons into individual InGaAs self-assembled quantum dots (QDs). By using an SET with a small island area and growing QDs with a low density we are able to distinguish and measure three QDs. The bias voltage at which resonant tunneling into the dots occurs can be shifted using a surface gate electrode. From the applied voltages at which we observe electrons tunneling, we are able to measure the electron addition energies of three QDs.

Physical Review Letters, May 18, 1998

By measuring each junction of an electron pump in a single electron box configuration, we determi... more By measuring each junction of an electron pump in a single electron box configuration, we determine all quantities needed to test the standard theory of pumping error and leakage. Background charges are determined with an imprecision of 60.01e. Electron temperature is measured to 40 mK. Measured charging energies show all junctions have nearly the same capacitance. We find agreement with theory at 140 mK, but a disagreement of many orders of magnitude at 40 mK. We suggest that the excess error and leakage at 40 mK are due to photon-assisted cotunneling processes not included in the standard theory. [S0031-9007(98)06149-3]

Physical review, Apr 30, 2003

We have measured the counting errors of a seven-junction electron pump when charge is pumped agai... more We have measured the counting errors of a seven-junction electron pump when charge is pumped against a voltage difference. At finite bias voltages, we find that the errors increase exponentially with both pump voltage and temperature, in agreement with theoretical predictions. To compare experiment and simulation, all pump parameters were determined by independent electron-box experiments. Although we assume temperatures somewhat higher than those measured, simulations based on the ground-capacitance model yield excellent quantitative agreement with experiment and indicate that errors in the high-voltage regime are due to thermally activated tunneling. In addition, a surprising asymmetry between positive and negative voltages is explained by an asymmetry in the junction capacitances.

Physical review, Sep 15, 1999

Leakage and counting errors are explored experimentally in a well-characterized seven-junction el... more Leakage and counting errors are explored experimentally in a well-characterized seven-junction electron pump and compared with predictions of the orthodox theory, including cotunneling. Theory and experiment are in good agreement at intermediate temperatures, where errors are dominated by thermally activated, singlejunction processes. At low temperatures, however, the observed errors far exceed predictions, indicating that the orthodox theory omits an important error process.

Physical review, Dec 15, 2000

Computer simulations reveal that the lowest rates of leakage and counting errors observed in the ... more Computer simulations reveal that the lowest rates of leakage and counting errors observed in the electron pump can be explained by photon-assisted tunneling driven by 1/f noise. The noise power at microwave frequencies required to account for the observed errors is consistent with extrapolation of the low-frequency noise spectrum commonly recorded in single-electron transistors. Pump simulations, based on the groundcapacitance model, include cotunneling as well as single-junction photon-assisted tunneling. Quantitative agreement between theory and experiment is obtained for leakage and counting errors in pumps with four, five, six, and seven junctions in the limit of low temperatures and low counting rates. The effect of self-heating is explored.

Physical Review Letters, May 29, 2000

We measure photon-assisted tunneling in 4-and 6-junction electron pumps at photon frequencies up ... more We measure photon-assisted tunneling in 4-and 6-junction electron pumps at photon frequencies up to 60 GHz. We determine the microwave voltage at the pumps using noise thermometry. The standard theory of leakage in the electron pump, modified to include photon-assisted tunneling, describes our experiments well. From this test of theory we argue that, in the absence of external microwaves, photonassisted tunneling driven by 1͞f noise is an important error mechanism in electron pumps.

APS March Meeting Abstracts, Mar 1, 2003

We are studying the tunneling of individual electrons into self-assembled InAs quantum dots with ... more We are studying the tunneling of individual electrons into self-assembled InAs quantum dots with single-electron transistors. Aluminum single-electron transistors were fabricated over an epitaxial structure including self-assembled InAs quantum dots and an underlying n-doped layer. Sweeping the potential on the n-doped layer causes individual electrons to tunnel onto a quantum dot, and is detected capacitively with a single electron transistor.

Bulletin of the American Physical Society, Mar 7, 2007

arXiv (Cornell University), Dec 17, 2021

SPIE Proceedings, 2017

Tabletop-scale coherent EUV generated through high-harmonic generation (HHG) produces light in th... more Tabletop-scale coherent EUV generated through high-harmonic generation (HHG) produces light in the form of an attosecond pulse train that uniquely combines characteristics of good energy resolution (≈100-300meV) with sub-fs time resolution. This makes HHG an ideal source for studying the fastest dynamics in materials. Furthermore, using angle-resolved photoemission spectroscopy (ARPES), it is possible to extract detailed information about electron dynamics over the entire Brillouin zone. In recently published work, we combined HHG with ARPES to identify a sub-femtosecond excited-state lifetime for the first time. Photoemission occurs as a three-step process: 1) An electron is photoexcited from the valence band to far above the Fermi energy; 2) it transports to the surface, and 3) it overcomes the work function and exits. If the electron is promoted into a highlyexcited unoccupied band in the material (as opposed to a free-electron-like state), we observe the electron emission lifetime to increase in a measurable way-the Ni band 22 eV above the Fermi level has a lifetime of 212±30 attoseconds. Furthermore, by comparing photoemission from Cu and Ni, we reveal the influence of attosecond-timescale electron screening vs scattering by the electrons near the fermi surface. This work for the first time demonstrates the relevance of attosecond spectroscopy to the study of intrinsic properties and band structure in materials, as opposed to the strong-field induced dynamics studied extensively to-date.

Nature Physics, 2018

explains how the elementary charge will soon be reinstated in metrology-and why it got sidelined ... more explains how the elementary charge will soon be reinstated in metrology-and why it got sidelined in the first place.