The effect of a local external electric field on the barrier potential of a tunneling gap is studied utilizing an emerging technique, synchrotron x-ray scanning tunneling microscopy. Here, we demonstrate that the shape of the potential barrier in the tunneling gap can be altered by a localized external electric field, generated by voltages placed on the metallic outer shield of a nanofabricated coaxial metal-insulator-metal tip, resulting in a controlled linear modulation of the tunneling current. Experiments at hard and soft x-ray synchrotron beamlines reveal that both the chemical contrast and magnetic contrast signals measured by the tip can be drastically enhanced, resulting in improved local detection of chemistry and magnetization at the surface.

As the boundaries of advanced materials development are pushed toward the nanoscale, and still even further into molecular-scale multi-component systems, scanning probe microscopes provide highly valuable and essential localized information about the surface electronic, atomic, and magnetic structure of emerging electronic and spintronic technologies.1–4 In recent decades, efforts at synchrotron facilities around the world have been ongoing to develop promising next generation probe microscopy techniques, like the synchrotron x-ray scanning tunneling microscopy (SX-STM) and similar methods, that will provide scientists with not only atomic and electronic information, but also with the ability to chemically fingerprint material systems with near-atomic spatial resolutions.5–11 The SX-STM technique can chemically fingerprint a material's surface by combining the chemical, electronic, and magnetic sensitivities of synchrotron x-rays with the high spatial resolutions of scanning tunneling microscopy (STM).12,13

One serious obstacle in SX-STM experiments has been the introduction of additional x-ray induced photo-currents measured at the STM tip, during synchrotron x-ray illumination. The x-ray induced photo-currents degrade STM resolution. To minimize the degradation effects of photoelectron currents on STM spatial resolution, while the sample is under synchrotron x-ray illumination, insulator-coated (IC) smart tips were developed.14–16 The nano-fabricated metal-insulator-metal (MIM) coaxial smart tips have a metallic, PtIr core fabricated and etched from 250 μm thick PtIr wire, and the core is surrounded by an outer insulating oxide layer (∼650 nm). In addition, an outer conducting Au/Ti (250 nm/200 Å) shell is deposited onto the surface of the insulating oxide (outer insulating/metal layer thickness totals, ∼1 μm). Near the apex of the tip, a focused-ion beam (FIB) is used to expose the metallic core (PtIr) of the MIM tip's apex. The outer conducting layer (Au/Ti) serves a dual purpose: (i) as an electrical contact for grounding and preventing undesired charging effects, and (ii) acts as a “shield,” reducing the magnitude of x-ray excited electron current detected at the tip, when biased with a voltage VS, creating a repulsive (or attractive) E-field around the tip apex. In addition for this study, we demonstrate the effects of a locally generated, external electric field (E-field) in synchrotron x-ray scanning tunneling microscopy (SX-STM) experiments. The external E-fields discussed modulate the current measured at the STM tip (tunneling and x-ray induced) in SX-STM experiments.

Hard and soft x-ray beamlines, 26-ID and 4-ID-C, respectively, at Argonne National Laboratory's Advanced Photon Source were utilized throughout the course of the study.17,18 All spectra collected were taken with coaxial MIM smart tips that have an exposed tip apex of ∼30–40 nm. Simultaneous measurements of the current signals measured at the tip apex ISXSTMtip, the sample ISXSTMsample as the total electron yield (TEY), and the shield, ISXSTMshield are accomplished through the lock-in detection.19 During SX-STM measurements, when the tip is brought into close proximity (tunneling) with the sample surface and the tunneling gap is illuminated with x-rays, additional contributions to the conventional tunneling current, in the form of photocurrent, are present. Under x-ray illumination, the resulting current measured at the tip ISXSTMtip, now, contains two-contributions, (i) a conventional tunneling current contribution, and (ii) an x-ray excited tunneling and photocurrent contribution

ISXSTMtip=ItunnelSTM+Iexcitedtunnel+Iexcitedejected,
(1)

where ItunnelSTM and Iexcitedtunnel are the conventional and x-ray excited tunneling current contributions, respectively, and Iexcitedejected is the x-ray excited electron photocurrent exchanged across the local tunneling gap.8 The x-ray excited tunneling current Iexcitedtunnel arises from core-level electrons that do not have enough energy to leave the sample surface, but are excited by x-ray illumination into states just above Fermi. Upon these states being filled, additional electrons are available for tunneling. The Iexcitedejected contribution in ISXSTMtip arises from photo-excited electrons (resulting from primary, secondary, and Auger processes) that have enough energy to leave the sample surface and be detected at the MIM tip.

In SX-STM experiments, the x-ray excited portions of ISXSTMtip (Iexcitedtunnel and Iexcitedejected) constitute a chemical signal. This chemical signal in ISXSTMtip is separated from the conventional tunneling current ItunnelSTM, utilizing a modified experimental set-up. In the modified set-up, the sample-tip gap is illuminated by an incident x-ray beam chopped on and off at a frequency, ω, and chopped electronic (current) signals measured at the tip, shield, and sample, separately, and are fed into three separate lock-in amplifiers (LIAs). The outputs from each LIA contain chemical information, a result of electron excitation from x-ray illumination. The conventional tunneling current signal ItunnelSTM (beam off) is separated from the chemical signal, utilizing a topological filter (3rd-order low-pass filter & differential gain amplifier). As a result, four separate electronic channels are measured and separated in the SX-STM experiment: one is a pure conventional tunneling current channel ItunnelSTM measured at the tip and fed back into the STM electronics for tip stabilization during tunneling; and three separate chemical channels measured at the sample ISXSTMsample, tip ISXSTMtip, and shield ISXSTMshield. Details on the full SX-STM experimental set-up are published elsewhere.8,19

For hard x-ray experiments at beamline 26-ID, a clean Cu(111) substrate was used to collect spectroscopic chemical information from the x-ray photo-excited electrons. Soft x-ray experiments conducted at beamline 4-ID-C utilized Co thin films grown on Cu(111) to collect spectroscopic chemical information. The Cu(111) substrate was cleaned in ultra-high vacuum using repeated Ar+ sputtering and annealing cycles. STM scans with a bare PtIr tip verified flat Cu(111) substrate terraces >200 nm in width (not shown). Spectra of the clean Cu(111) surface were collected with the incident beam energy tuned to the near-region of the Cu K absorption edge (8.979 keV). Magnetic information was collected on 20 monolayer thick Co film samples deposited on Cu(111), via e-beam evaporation. A topographic image of the as-grown, Co film is shown in Fig. 1. The STM image shows the epitaxial Co film growth on the Cu(111) substrate and substantiates the stability of the MIM fabricated tip for SX-STM experiments. The incident beam energy on the Co film was tuned to near the region of the Co L3 (778.1 eV) and L2 (793.2 eV) absorption edges.

FIG. 1.

STM topographic image of the epitaxially grown Co film surface grown on the Cu(111) substrate surface (237 nm × 194 nm), collected with a nano-fabricated metal-insulator-metal coaxial smart tip. Scanning parameters I = 1 nA, V = −1 V.

FIG. 1.

STM topographic image of the epitaxially grown Co film surface grown on the Cu(111) substrate surface (237 nm × 194 nm), collected with a nano-fabricated metal-insulator-metal coaxial smart tip. Scanning parameters I = 1 nA, V = −1 V.

Close modal

Left- and right-circularly polarized x-rays (LCP/RCP) were utilized to collect magnetic information about the Co/Cu(111) surface, by examining the differences in measured spectroscopic intensity between RCP and LCP spectra for the Co L3 (778.1 eV) and L2 (793.2 eV) absorption edges.20–22 For all measurements, the tip was held at ground and the sample bias set to VB = −1 V, and the shield bias, VS, was varied from −5 V to 5 V. All spectroscopic information was collected with the apex of the coaxial smart tip positioned ∼300 nm from the sample surface (far-field) or with the tip positioned in tunneling (∼1 nm, near-field). When the tip is positioned under the near-field condition, ISXSTMtip=ItunnelSTM+Iexcitedtunnel+Iexcitedejected, as given by Eq. (1). In near-field (tunneling) condition, the tip current data were collected with the tip stabilized at 1 nA. With the tip under the far-field condition, contributions from tunneling (ItunnelSTM and Iexcitedtunnel) are negligible and ISXSTMtip reduces to ∼Iexcitedejected. The position of the tip (near-/far-field) above the sample surface will be specified throughout the text.

In addition, z-spectroscopy measurements were taken to demonstrate chemical sensitivity in the ISXSTMtip(chemical) channel, when incident x-rays were turned off versus when the incident x-rays, near the Cu K absorption edge (8.979 keV), were turned on. During z-spectroscopy measurements, the tip is first stabilized under tunneling conditions (near-field) with ItunnelSTM ∼1 nA, above the sample surface. Once the tip current (conventional, ItunnelSTM) has been stabilized, the feedback loop is turned off and the tip is ramped a few nanometers away from the substrate surface into the far-field, while spectroscopic chemical information from the current ISXSTMtip is recorded.

The z-spectroscopy measurements [Fig. 2(a)] display a clear difference in the spectroscopic chemical information contained in ISXSTMtip with z-displacement when the incident x-rays are “on” compared to when the x-rays are “off”. When the incident beam illuminates the sample (closed circles), ISXSTMtip exhibits an exponential dependence with z-displacement and demonstrates that under x-ray illumination, ISXSTMtip contains two contributions: an x-ray excited tunneling current contribution, exponentially dependent on z, and an x-ray excited photo-current contribution.23 As the tip is pulled away from the Cu surface, the tunneling current exponentially approaches zero, leaving only photo-excited electron current at large z. When the beam is off (open circles), no x-ray induced current (chemical signal) is measured in the ISXSTMtip channel. So the current measured when the beam is off is relatively small compared with beam on, as expected.

FIG. 2.

(a) Excited x-ray current [chemical signal, near Cu K-edge (8.979 keV)] taken from the output of LIA measured at the tip as a function of z-distance from the sample surface, with hard x-rays illuminating the sample surface (black closed circles) and with the hard x-rays turned off (green open circles). The sample bias was set to VB = −1 V and ItunnelSTM stabilized at ∼1 nA. Close to the sample surface, the tip current is dominated by tunneling electrons (near-field). As the tip z-distance from the sample surface is increased beyond ∼1 nm, the tunneling current contribution reduces significantly and current measured at the tip is dominated by photoelectron current (far-field). (b)–(d) Show the far-field photoemitted electron current from the coaxial MIM tip (shield, tip) and the sample. The spectra are measured across the Cu K-edge (8.994 keV). (e) and (f) The magnitude of the Cu K-edge peak intensity for the shield (Pshield) and tip (Ptip) as a function of VS, respectively. [Inset] The SEM image shows the nanofabricated MIM tip utilized in the experiments; the diagram indicates components of the coaxial MIM tip structure: the tip (PtIr) which is grounded, an insulating layer (SiO2), and an outer conducting shield layer (Au), which is independently biased from the sample.

FIG. 2.

(a) Excited x-ray current [chemical signal, near Cu K-edge (8.979 keV)] taken from the output of LIA measured at the tip as a function of z-distance from the sample surface, with hard x-rays illuminating the sample surface (black closed circles) and with the hard x-rays turned off (green open circles). The sample bias was set to VB = −1 V and ItunnelSTM stabilized at ∼1 nA. Close to the sample surface, the tip current is dominated by tunneling electrons (near-field). As the tip z-distance from the sample surface is increased beyond ∼1 nm, the tunneling current contribution reduces significantly and current measured at the tip is dominated by photoelectron current (far-field). (b)–(d) Show the far-field photoemitted electron current from the coaxial MIM tip (shield, tip) and the sample. The spectra are measured across the Cu K-edge (8.994 keV). (e) and (f) The magnitude of the Cu K-edge peak intensity for the shield (Pshield) and tip (Ptip) as a function of VS, respectively. [Inset] The SEM image shows the nanofabricated MIM tip utilized in the experiments; the diagram indicates components of the coaxial MIM tip structure: the tip (PtIr) which is grounded, an insulating layer (SiO2), and an outer conducting shield layer (Au), which is independently biased from the sample.

Close modal

Far-field spectra for the shield, tip, and sample channels are shown in Figs. 2(b)–2(d) with the shield bias set to ground, VS = 0 V. The spectra show the extended x-ray fine structure of the clean Cu surface. The Cu K absorption edge in the ISXSTMshield and ISXSTMtip exhibits a current increase for electrons arriving at the tip and shield. However, the peak intensity in the ISXSTMsample channel is inverted (negative) compared with ISXSTMshield and ISXSTMtip. The inversion observed in ISXSTMsample is due to the net loss of electrons from the sample in our experiment, generating an inverted current compared with ISXSTMshield and ISXSTMtip.

The Cu K-edge peak intensity is defined as the measured difference in the x-ray excited current above the Cu K-edge at 8.994 keV and the average intensity, below, on the pre-edge near 8.960 keV. The magnitudes calculated for the shield, Pshield, and the tip, Ptip, are plotted as a function of VS in Figs. 2(e) and 2(f), respectively, ramped from −5 V Vs+5 V in 0.2 V steps. The magnitude of the measured peak intensity is calculated as the difference in photo-excited electron current measured [see Fig. 2(b)]. For Pshield, as VS starts from negative biases up to a positive bias of +2.8 V, the ISXSTMshield remains flat. However above VS = 2.8 V, a sudden increase in Pshield is observed.

The behavior of Ptip, shown in Fig. 2(f), is similar to Pshield at relatively large negative biases, as Ptip shows only little increase for Vs −2.2 V. However for VS > −2.2 V, the onset of a linear VS-dependence is observed in Ptip. As VS is increased above 0 V, the slope of this increase changes, until Ptip saturates at VS = 3.2 V. The observed deviation of Ptip from Pshield is a direct result of two factors: the magnitude of the E-field present on the shield and the size of the exposed tip apex. At VS = 2.8 V, the E-field is strong enough to alter the trajectory of secondary electrons emitted from the Cu sample and attract additional electrons from the sample to be measured at the shield. The effective control of the current with only small Vs indicates that ISXSTMshield is governed by slow secondary electrons and not by high-energy primary electrons. In addition, the areal portion of the tip apex exposed in a nanofabricated smart tip (see Fig. 2 inset) is very small in comparison with that of the shield. As a result in the far-field, the E-field may be strong enough to reduce the photo-ejected electron current measured at the tip.

For the same experimental set-up, spectra for the shield, tip, and sample channels were collected with the tip stabilized in the near-field (tunneling condition, ItunnelSTM ∼1 nA), above the Cu sample surface, with VB = −1 V. In the near-field, the Cu K-edge absorption peak measured at ISXSTMshield also shows a strong VS-dependence. The magnitude of Pshield, measured from ISXSTMshield, is plotted as a function of VS in Fig. 3(a). ISXSTMshield displays a different behavior in the near-field compared with far-field measurements. Here, the current measurement is noisier; however similar to the far-field measurements an increase in the Pshield is observed at a slightly higher VS ∼ 2.8 V. The elevated noise and slight shift in VS are likely due to the close proximity of the shield to the Cu surface in the near-field (∼40 nm), compared to the far-field (∼300 nm). In the near-field, secondary electrons ejected from the Cu sample will be collected at the shield more efficiently, than in the far-field.

FIG. 3.

(a)–(c) The Cu K-edge (8.979 keV) peak intensity measurements collected in the near-field, plotted for the shield, tip, and sample, respectively, as a function of VS-dependence (VB = −1 V, ItunnelSTM ∼1 nA).

FIG. 3.

(a)–(c) The Cu K-edge (8.979 keV) peak intensity measurements collected in the near-field, plotted for the shield, tip, and sample, respectively, as a function of VS-dependence (VB = −1 V, ItunnelSTM ∼1 nA).

Close modal

As discussed above when the tip is under the tunneling condition and under x-ray illumination, an additional x-ray excited tunneling current Iexcitedtunnel contribution is present, in-addition to an x-ray excited electron photocurrent Iexcitedejected contribution (Eq. (1)). In tunneling, this additional Iexcitedtunnel contribution alters the observed ISXSTMtip signal response to VS compared with the far-field (see Fig. 2(f)), and as a result, Ptip shows a symmetric linear correlation with the magnitude of the shield bias, |VS| [see Fig. 3(b)]. As VS is reduced from +5 V to 0 V, ISXSTMtip a linear decrease in the measured current is expected. However for VS < 0 V, ISXSTMtip shows a steady increase in the current signal measured in the tip channel mirroring the linear response for VS 0 V. The same symmetric linear correlation is also observed in Psample [Fig. 3(c)].

The linear correlations observed in ISXSTMtip and ISXSTMsample can be directly attributed to the bias (VS) placed on the shield and its close proximity (∼600 nm) to the actual tunneling gap [see Fig. 2(a) inset]. Ignoring Iexcitedejected for the moment in Eq. (1), under x-ray illumination of the tunneling gap and a stabilized tunneling condition (near-field), Iexcitedtunnel depends on the shape of the potential barrier in the junction and is proportional to the tunneling probability [≈T(Ex)], where Ex is the electron energy, along the direction in which electrons are tunneling.24–26 Studies by Sonntag et al. and Maruyama et al. have more directly attributed these effects, from the local E-fields, to charge redistribution (or local density of states, LDOS) at the surface of interest, leading to effectively small (∼meV) changes in the resultant surface chemical potential.27,28

For VS = 5 V, the resulting maximum local external electric field Es that is generated by the shield is approximately Es = 10 MV/m for a sample bias VB = −1 V. Consequently, Es can influence the barrier potential within the junction, and as a result, any change in VS may enhance or suppress the junction barrier potential. For a shield a few hundred nanometers away from the tunneling gap and a VS = 5 V, the maximum external potential change within the junction is less than 1.2 meV. As a result, the external E-field generated by VS on the shield affects the barrier potential within the tunnel gap and for a symmetric barrier, the measured current will show a linear response about VS = 0 V, as shown in Figs. 3(b) and 3(c).24 

Under the tunneling condition and under x-ray illumination, the exposed portion of a MIM tip apex (∼40 nm) is subjected to photo-current Iexcitedejected from the sample surface. A majority of the photo-current is detected at considerably larger distances from the source (sample) than at the tunnel junction, and can thus be considered predominantly far-field. Therefore, the Iexcitedejected contribution to ISXSTMtip is typically measured to be twice the magnitude of ISXSTMtip in the far-field (see Fig. 2(f)). This correlation is not observed in the near-field (Fig. 3(c)), indicating that the ISXSTMtip and ISXSTMsample chemical signals measured are dominated by tunneling.

A similar influence of Vs can also be observed for lower photon energies. In addition to hard x-ray measurements, spectroscopic chemical information was collected in the soft x-ray regime for ISXSTMtip, ISXSTMsample, and ISXSTMshield channels at beamline 4-ID-C. A ferromagnetic Co/Cu(111) thin film sample surface was used to collect spectra for left- and right-circularly polarized x-rays (LCP and RCP), respectively. The spectra were collected with the tip positioned in the near-field, (ItunnelSTM ∼1 nA, VB = −1 V). The spectroscopic chemical information collected for ISXSTMshield shows a clear Co L3 (778.1 eV) and L2 (793.2 eV) absorption signal in Fig. 4(a). As the VS was ramped from negative to positive, the intensity of ISXSTMshield absorption Co L2 and L3 absorption edges increases. This increased intensity observed in the L2 and L3 absorption edges demonstrates the effectiveness of the shield at repelling (attracting) photo-excited electrons from the Co sample surface at negative (positive) biases, indicating that, similar to hard x-rays, most of the electrons measured in the soft x-ray regime are low-energy, secondary electrons. Analogous to ISXSTMshield, Co L2 and L3 absorption edge signals were observed in ISXSTMtip and ISXSTMsample channels (not shown).

FIG. 4.

(a) Near-field soft x-ray measurements of Co L3 (778.1 eV) and L2 (793.2 eV) edge spectra taken from the shield at different VS, left-circularly polarized shown; (b) Co L3 and L2 spectra measured at the tip in the near-field, for left- and right-circularly polarized x-rays: in (a) and (b) VB = −1 V, ItunnelSTM ∼1 nA. (c) XMCD signal collected at the tip sample and shield for VS = 0 V; the tip, sample, and shield data are normalized and plotted on the same scale. (d)–(f) Near-field soft x-ray measurements of XMCD contrast measured at the shield, tip, and sample (Co L3 and L2 edge) as a function of VS-dependence [in (d)–(f), the smallest data point in each plot ISXSTMtip, ISXSTMsample, and ISXSTMshield, respectively, was re-scaled to an approximate value of “1” for the purposes of comparison of scale and clarity of the VS effect].

FIG. 4.

(a) Near-field soft x-ray measurements of Co L3 (778.1 eV) and L2 (793.2 eV) edge spectra taken from the shield at different VS, left-circularly polarized shown; (b) Co L3 and L2 spectra measured at the tip in the near-field, for left- and right-circularly polarized x-rays: in (a) and (b) VB = −1 V, ItunnelSTM ∼1 nA. (c) XMCD signal collected at the tip sample and shield for VS = 0 V; the tip, sample, and shield data are normalized and plotted on the same scale. (d)–(f) Near-field soft x-ray measurements of XMCD contrast measured at the shield, tip, and sample (Co L3 and L2 edge) as a function of VS-dependence [in (d)–(f), the smallest data point in each plot ISXSTMtip, ISXSTMsample, and ISXSTMshield, respectively, was re-scaled to an approximate value of “1” for the purposes of comparison of scale and clarity of the VS effect].

Close modal

Absorption edge spectra collected at the tip, ISXSTMtip, for both left- and right-circularly polarized x-rays are shown in Fig. 4(b). From these absorption spectra, the x-ray magnetic circular dichroic (XMCD) signal can be calculated, as the difference in absorption intensity measured between RCP and LCP spectra (IRCPILCP). The XMCD signals are shown for the ISXSTMtip, ISXSTMsample, and ISXSTMshield channels in Fig. 4(c). The XMCD signal for ISXSTMsample (blue) is inverted compared with ISXSTMtip (green) and ISXSTMshield (black) signals with the magnitude of the XMCD signal the strongest in ISXSTMtip and the weakest XMCD signal observed in ISXSTMshield. Analogous to hard x-ray measurements, which show the influence of shield bias, VS, on Ptip and Psample, a similar effect is observed in the soft x-ray regime for XMCD contrast. As shown in Figs. 4(d) and 4(e), the XMCD contrast signals (for the Co L3 absorption peak) in ISXSTMtip and ISXSTMshield can be controlled by VS. In ISXSTMshield, negligible XMCD contrast is observed at negative VS [see Fig. 4(d)], however as VS is stepped to a positive bias, the Co L3 XMCD contrast signal for ISXSTMshield shows a steady increase.

Analogous to the hard x-ray measurements this observation is expected, given the shield's relatively close proximity (∼40 nm) to the sample surface. For ISXSTMtip, the Co L3 XMCD contrast signal shows a decrease as VS is stepped from 5 V down to 0 V, however as VS is decreased further to negative voltages [see Fig. 4(e)], the XMCD signal in ISXSTMtip begins to rise again steadily. In ISXSTMsample, relatively little VS -dependence is observed in the Co L3 XMCD contrast signal [Fig. 4(f)]. This is likely due to the large footprint of the incident beam and dominant contribution of photo-ejected electrons to the measured ISXSTMsample for the Co L3 peak. For the Co L2 peak, XMCD contrast showed a weak VS-dependence in the tip, sample, and shield channels (not shown). However as evidenced in Fig. 4(e), the Co L3 XMCD contrast signal measured in the tip (ISXSTMtip) can be locally optimized by simply adjusting the magnitude of VS.

In summary, this study shows the utility of the outer conducting (Au/Ti) shield layer for advanced nano-fabricated MIM coaxial smart tips in SX-STM experiments. By placing a bias VS on the shield layer of MIM tip, we are able to generate an external E-field. The external E-fields produced by the shield play a dual role in SX-STM experiments: (1) the external E-field effectively reduces the x-ray excited photo-current seen by the exposed portion of the coaxial MIM tip and (2) the external E-field can modulate the intensity of the spectroscopic chemical and XMCD contrast signal (electronic current signal) measured at ISXSTMtip. Similar to previous studies, the modification of tunneling currents in SX-STM experiments by external E-fields can be attributed to local changes in the potential barrier and may also be the result of charge modifications (LDOS) within the tunneling gap. As a result, this study demonstrates the capacity of MIM tips to enhance the measured chemical signal in SX-STM experiments, and could dramatically improve our understanding of local chemistries and environments at the surface.

This work was funded by the Office of Science Early Career Research Program through the Division of Scientific User Facilities, Office of Basic Energy Sciences of the U.S. Department of Energy through Grant No. SC70705. Use of the Advanced Photon Source and the Center for Nanoscale Materials was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.

1.
D.
Serrate
 et al,
Nat. Nanotechnol.
5
,
350
(
2010
).
2.
J.
Bork
,
P.
Wahl
,
L.
Diekhöner
, and
K.
Kern
,
New J. Phys.
11
,
113051
(
2009
).
3.
K.
von Bergmann
 et al,
Phys. Rev. Lett.
96
,
167203
(
2006
).
4.
Y.
Yayon
 et al,
Phys. Rev. Lett.
99
,
067202
(
2007
).
5.
T.
Okuda
 et al,
Phys. Rev. Lett.
102
,
105503
(
2009
).
6.
N.
Pilet
 et al,
Nanotechnology
23
,
475708
(
2012
).
7.
A.
Saito
 et al,
J. Synchrotron Radiat.
13
,
216
(
2006
).
8.
M.
Cummings
 et al,
Ultramicroscopy
112
,
22
(
2012
).
9.
H. A.
Bechtel
,
Proc. Natl. Acad. Sci. U.S.A.
111
,
7191
(
2014
).
10.
S.
Suzuki
,
Bull. Chem. Soc. Jpn.
88
,
240
(
2015
).
11.
V.
Rose
 et al,
J. Appl. Phys.
111
,
07E304
(
2012
).
12.
N.
Shirato
 et al,
Nano Lett.
14
,
6499
(
2014
).
13.
A.
DiLullo
 et al,
J. Synchrotron Radiat.
23
,
574
(
2016
).
14.
V.
Rose
 et al,
Appl. Phys. Lett.
99
,
173102
(
2011
).
15.
K.
Akiyama
 et al,
Rev. Sci. Instrum.
76
,
083711
(
2005
).
16.
A.
Saito
 et al,
Surf. Sci.
601
,
5294
(
2007
).
17.
R. P.
Winarski
 et al,
J. Synchrotron Radiat.
19
,
1056
(
2012
).
18.
J. W.
Freeland
 et al,
Rev. Sci. Instrum.
73
,
1408
(
2002
).
19.
K.
Wang
 et al,
Rev. Sci. Instrum.
84
,
063704
(
2013
).
20.
R.
Hammerling
 et al,
Phys. Rev. B
66
,
052402
(
2002
).
21.
L.-C.
Duda
 et al,
Phys. Rev. B
50
,
16758
(
1994
).
22.
J.
Stohr
,
J. Magn. Magn. Mater.
200
,
470
(
1999
).
23.
V.
Rose
 et al,
Appl. Phys. Lett.
92
,
193510
(
2008
).
24.
J. G.
Simmons
,
J. Appl. Phys.
34
,
1793
(
1963
).
25.
M.
Depas
 et al,
Solid-State Electron.
38
,
1465
(
1995
).
26.
J. C.
Fisher
and
I.
Giaever
,
J. Appl. Phys.
32
,
172
(
1961
).
27.
A.
Sonntag
 et al,
Phys. Rev. Lett.
112
,
017204
(
2014
).
28.
T.
Maruyama
 et al,
Nat. Nanotechnol.
4
,
158
(
2009
).