Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Contents lists available at ScienceDirect
Ultramicroscopy journal homepage: www.elsevier.com/locate/ultramic
Role of the resistivity of insulating field emitters on the energy of field-ionised and field-evaporated atoms L. Arnoldi n, E.P. Silaeva, F. Vurpillot, B. Deconihout, E. Cadel, I. Blum, A. Vella GPM UMR 6634, Université de Rouen, Avenue de l'Université, 76801, BP12, 76801 Saint Etienne du Rouvray, France
art ic l e i nf o
a b s t r a c t
Article history: Received 20 August 2014 Received in revised form 3 November 2014 Accepted 23 November 2014
In order to improve the accuracy of laser atom probe analyses, it is important to understand all the physical processes induced by the combination of the high electrical field and the femtosecond laser beam during field evaporation. New information can be accessed from the energy of evaporated surface atoms or field-ionised atoms of an imaging gas. In order to study the ions energy, we combine La-APT and FIM analyses in a new experimental setup equipped with electrostatic lenses. We report measurements for semiconductors and oxides and we study the influence of the illumination conditions (laser power and wavelength), the evaporation rate, the sample geometry and the tip preparation processes. The results are discussed taking into account the resistive properties of nonmetallic samples and the photo-stimulated conductivity. This work clarifies the role of the laser and DC field in the energy deficit of field evaporated ions. & 2014 Elsevier B.V. All rights reserved.
Keywords: Atom probe tomography Field ion microscopy Field emitter Semiconductors Insulators Resistivity
1. Introduction Thanks to the use of fast-laser pulses, Laser assisted atom probe tomography (La-APT); 3D nanoanalysis is now possible for a wide range of materials. In addition to metals, it is now possible to analyse not only semiconductors but also ceramics and oxides [1– 7]. The quantitative analysis of this class of materials is challenging in many aspects. For instance, the ability to quantitatively measure composition reliably is particularly attractive when dealing with oxides that are able to accommodate high concentrations of point defects [8]. The analysis of highly resistive materials is also of fundamental importance in the field of microelectronics, including the challenging aspect of finding and analysing local interfaces between the structures of a functional device [9]. In the early years of APT, only nanometres scale oxide inclusions in metals or highly conductive oxides were studied [10–13]. The first published work on the atom probe analysis of bulk insulator was performed by Kellogg in the early eighties [14]. The analysis was found tedious and a thin metal coating was used to improve the electrical conduction. This requirement was thought essential to avoid the huge voltage drop that should be observed between the end apex of the sample and the base where the high voltage is applied. Thereafter, progresses in the analysis of oxides were slow, and the electrical resistivity was thought to be one of the main limitations of the n
Corresponding author. E-mail address: [email protected] (L. Arnoldi).
technique. Only recently, with the large spreading of modern laser assisted APT instruments all over the world, a blooming of studies in APT of oxides and ceramics was observed. Surprisingly, successful analyses of uncoated specimen were obtained for several bulk ceramics [15–16]. In addition near atomic scale resolution was demonstrated. These observations question our understanding of field evaporation for the case of poorly conductive materials. In a simple approach, in atom probe, the field emitter must drain the charge induced by pulsed field evaporation. The analysed bulk dielectric material is generally a micron size sample attached to a macroscopic metallic needle. It may then be considered as an electrical resistor which resistance can be considered to be very high especially at the cryogenic temperatures used (1010–1020 Ω). Kellogg predicted several kV of voltage drop for fused glass or Pyrex samples. Nevertheless, in recent experiments, voltage drops are rarely observed. Only in the paper of Chen et al. [17], a drop of only a few volts was measured in MgO samples. Chen et al. and after Tamura et al. [18] proposed a model based on accumulation of photo-induced holes changing the internal and external electric field around the tip apex. This paper shows the important role of the laser excitation in the field evaporation process. Some other authors pointed out the importance of surface states in the actual electrical properties of the sample. For instance, ab-initio calculations suggest that the band gap of crystalline alumina is reduced from 8 eV in the bulk to 2.5 eV at the surface, which makes the alumina surface more conductive and more absorbing in the visible domain of the laser excitation used (532 nm laser pulse/
http://dx.doi.org/10.1016/j.ultramic.2014.11.018 0304-3991/& 2014 Elsevier B.V. All rights reserved.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
2
2.3 eV or 355 nm/3.6 eV). The presence of Ga and the possible amorphization of the oxide surface by the gallium beam during sample preparation in the focused ion beam instrument may also contribute to a modification of the surface states and thus may create a charge-draining path, much like coating with a thin metal layer. The presence of the surface field may also change the physical properties of the material, as calculated by Kreuzer et al. [19]. In addition, the high surface to volume ratio of an APT sample may enhance the effects of any modification of the surfaces electrical properties. In this work, several experimental procedures were developed to accurately measure the in-situ electrical resistance of the APT sample. In order to measure the voltage drop with a high accuracy, electrostatic lenses are introduced in the chamber. The setup is then used as an ion energy analyser, which amplifies the effect of the ions energy on their trajectories and time of flight. This enables an accurate measurement of the voltage drop using the relationship: ΔE = neΔV , with ne the ion charge. With such an experimental setup we are able to investigate voltage variations at the tip apex ranging from a few volts to hundreds of volts. To amplify the observed voltage drop, the FIM mode was used in addition to the APT mode. The ion current can, indeed, be increased by several orders of magnitude due to the contribution of the ionisation of the image gas. The variation of the observed resistance of semiconductors and oxide tip samples were studied. We demonstrate that the laser illumination creates free carriers in the material, which significantly reduce its resistivity. After the laser pulse, these carriers recombine up to an average equilibrium value. In the case of FIM measurements, this average carrier density is monitored. The role of the specimen preparation is also addressed.
2. Theoretical consideration The ion energy deficit is induced by the voltage drop ΔV existing between the tip apex and its metallic base. Two different effects contribute to this voltage drop: the Ohmic effect and the Band Bending Effect [20]. Indeed, the expression of the drop can be written:
ΔV = V0 − Vtip = ΔVbb + ΔVOhm = ΔVbb + RI
(1)
where ΔVbb is the voltage drop induced by the change of band gap created by the accumulation of holes, I is the electronic current inside the tip (equal to the ionic current emitted from the tip) and R is the resistance of the tip. This resistance is given in a first approach by
R = L tip (μ c nc eStip )−1
(2)
where Stip = πrapex rbase [21] is the mean tip cross section, μc the carrier mobility and nc the carrier density [22]. In the case of semiconductors analysed by La-APT, both the ionic current and the resistance are low: the resistive effect is negligible and the voltage drop at the tip apex is smaller than 1 V. In FIM mode, the ionic current is higher than in La-APT mode (more than a factor 1000) so the contribution of this resistive effect can be observed even during the FIM analysis of semiconductors. In the case of oxides like MgO, the resistance is so high that the ohmic contribution can become the main contribution to the voltage drop.
3. Experimental setup and methods Two types of materials were analysed: silicon (Si) and MgO. Si tips were prepared by annular milling from micro-post [24]
of around 3 μm in radius and 40 μm in length. Two types of microposts were analysed: un-doped o 100 4 oriented micro-posts and p-doped (1018 cm 3) o1114 oriented micro-posts. MgO tips were prepared from bulk samples using the lift-out method [25,26]. All measurements were performed on a FlexTAP instrument (CAMECA) in APT mode or FIM mode. The main characteristic of the FlexTAP is the presence of three electrostatic lenses, between the tip and the detector, which can be polarised to accelerate of decelerate the emitted ions. This setup was developed in order to offer a flexible angle of view on the tip during the analysis: ranging from 4° to 30°. The other main goal is to improve the temporal resolution of the ToF spectrum. The relative accuracy of ToF measurements depends on the ratio of the ion ToF (t) over the time accuracy of the detector (Δt). The lenses allow to increase the value of ToF (t) and hence the accuracy of its measurement. The FlexTAP is coupled to a femto-second pulsed laser, with a repetition rate of 50 kHz in APT mode and 100 kHz in FIM mode, a pulse duration of 500 fs and a tuneable wavelength (1030 nm, 515 nm and 343 nm). The FIM is a projection microscope, which images the tip surface with an atomic resolution [23]. Noble gas atoms are introduced in the analysis chamber and are field-ionised close to protruding surface atoms, where the field is highest. After the ionisation, these ions are projected onto a grounded screen. The ionisation of the gas atoms by tunnelling effect requires a high electric field, around tens of volts per nanometre, which is of the same order than the field for field-evaporation in APT mode. Neon gas was used as imaging gas in the following experiments. The main advantages of FIM for the following experiments are that: 1. The ionic emission current I (and the electronic current inside the tip) is between 1000 and 10,000 times higher than in the case of APT. 2. FIM images can also be obtained without laser illumination, hence the voltage drops with and without laser illumination can be compared. This is not the case for APT analysis, where the laser power can be varied, but cannot be zero as laser pulses are required for TOF measurements. 3. Because the tip surface is not evaporated, the sample geometry remains constant during the experiment. The FIM detector is placed at 44.5 cm from the tip. An aperture is placed in front of the tip, and stops ions emitted at too large angles, that can have aberrant trajectories. The ion trajectories are schematically represented in Fig. 1 for fields of view of 30° and 15°. The voltages applied to the lenses for these two modes are given in Table 1. The 15° mode was used in the case of small voltage drops at the tip apex. Indeed, only one lens is polarized, hence, the relationship between the voltage drop and the shift of trajectories is easier to predict. Moreover, the magnification of this mode is larger than of the 30° mode, which causes the ions trajectories and TOFs to be more sensitive to small variations in Vtip. The main problem of this mode, however, is the high value of the potential applied to the decelerating lens: it is higher in the case of the 30° mode (1.005 against 0.982), inducing a high potential barrier on the ions trajectories. Therefore, if the energy deficit is important (higher than about 1.5%), the ions can be repelled and not reach the detector. The 30° mode, to the contrary, was chosen to measure energy deficits as high as a few percent. One advantage of this mode is its large field of view, which enables to collect more spatial information because the area imaged on the tip surface is 4 times larger. However, one drawback is that when the ions have high energy deficit, their trajectories can exhibit strong aberrations, as discussed later.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Fig. 1. Simulated trajectories of an ion emitted at a fixed (12°) angle at a voltage Vtip ¼ Vapplied (full lines) and at a voltage Vtip o Vapplied (dashed lines) for a field of view of 15° with 0.5% voltage deficit (a) and 30° with 2% deficit (b). Table 1 Voltages applied to the different lenses for the two modes used in this work: 15° and 30° fields of view.
FIM Lens 1 Voltage FIM Lens 2 Voltage
15°
30°
1.005 Vapplied 0
0.982 Vapplied 0.4 Vapplied
In the case of the 15° mode, when the laser is switched off, a change in magnification can be observed, (Fig. 2(a)). In order to find the corresponding voltage drop, the ions trajectories were simulated numerically for the geometry of the electrostatic setup of the FlexTAP. The simulation was repeated for different voltage drops in order to obtain a calibration curve (Fig. 2(b)). The relative shift in the ions impact position is a linear function of the relative voltage drop between 0% and 1%. In this configuration voltage drop of a few volts can be measured. When using the 30° FIM mode, the image of the MgO surface is very distorted and the detector is only partially hit, as reported in
3
Fig. 3(a). MgO terraces and crystallography are not visible even if a fourfold symmetry can still be distinguished. In order to explain such images, numerical simulations of the trajectories were performed for ions having an energy deficit of 3.75%. Fig. 4 reports the position of the impact of an ion on the detector, normalized to the detector radius, as a function of the emission angle. First, one can note that the function is not bijective: one position on the detector corresponds to more than one emission angle. A second point is the big difference of impact density: most of the trajectories hit one fourth of the detector while only a few others hit the rest of it. This behaviour explains why, in the experimental results, most ion impacts are observed in a circle smaller than the detector diameter. As shown in Fig. 3, the diameter of this circle increases when the laser is switched on. Again, using simulations of the ions trajectories, a calibration curve can be obtained, giving the diameter of this circle as a function of the voltage drop (Fig. 3b). In addition, as shown in Fig. 4, for ions emitted at an angle 420°, the impact position on the detector is very sensitive with the emission angle and the deficit value, hence a small variation of this angle or deficit will transform a bright spot into a large bright line on the detector, as observed in Fig. 3. An advantage of the measure at 30° mode is that a single value of deficit can be measured without comparing two images, unlike the case of the 15° mode. However, the accuracy of this mode is worse than that of the 15° mode (around 0.1% of the tip voltage). In the case of the APT analysis, the geometry of the FlexTAP chamber is the same as for FIM analysis (as reported in Fig. 1) except for the detector that is now placed 60 cm far from the tip. The presence of the three lenses allows to form a potential barrier on the ions path, thereby increasing their ToF. Hence, the expression of the time of flight becomes:
ToF = L‵
m 2neVapplied
where m is the ion mass, n its charge, Vapplied the applied voltage and L′ is a virtual flight length accounting for the effect of the lenses on the ToF. When only the first decelerating lens is polarized, the higher the lens voltage the larger the virtual flight length and the higher the sensitivity. If the lens voltage is too high, however, the ToF can become larger than the acquisition window of the instrument, which is 12 μs. In the following experiments, the best compromise was found to be a lens voltage equal to 1.02 Vapplied, when the ToF tends to infinity for a voltage of 1.04 Vapplied. Using this configuration, a small variation of voltage at the apex corresponds to important variation of ToF, as reported in Fig. 5.
Fig. 2. (a) FIM images of o 111 4 p-doped. (imaging gas 3 10 6 mbar of Ne at 20 K, 15° mode). The red circles show the positions of the bright spots with laser illumination and green circles refer to these positions without laser illumination. (b) Calculated FIM spots radial positions as function of the relative voltage drop. The radial positions are normalised by the radial positions measured in the presence of laser illumination. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
4
Relative position on the detector(%)
Fig. 3. FIM images of MgO at 20 K with 10 5 mbar of Ne and 30° mode on FlexTAP: (a) without laser illumination; (b) with 343 nm laser at I¼ 0.1 GW/cm². (c) Radius of the illuminated area (Ri) normalized to the detector radius (Rd) as function of the relative voltage drop.
100
4.1. Ion current
50 0 -30
-10
10
30
-50 -100 Emission angle of the ion (°)
Fig. 4. Position of the impact of an ion on the detector, normalized to the detector radius as a function of the emission angle of the ion for a voltage drop of 3.75%.
Fig. 5. The accuracy, defined as the variation of ToF corresponding to a change in the tip voltage of 1 V, as function of the ratio between the tip voltage and the first lens voltage when only the first lens is polarized.
Therefore, the energy deficit can easily be obtained from the ToF during the analysis. To do so, during the analysis, a small voltage variation of 50 V is applied to the tip and the shift in ToF is measured. This measured sensitivity is compared to the calibration curve given in Fig. 5 in order to obtain Vtip.
MgO tips were analysed in La-APT mode at constant voltage using the lens configuration described previously (only the first lens is polarized at 1.02 V0). After the measurement of the ToF shift for a change in voltage of 50 V, the tip continues to evaporate and its radius increases, hence the field at the apex slightly decreases. This decrease induces a reduction of the evaporation rate: i.e. a decrease in the ionic current. To increase the evaporation rate, we can increase the applied laser power to compensate for the decrease of the surface field. Five laser steps are reported in Fig. 6. Using the measured ToF shift (see Fig. 6(b)) and the calibration measurement, the voltage drop at the apex of the tip can be obtained as a function of the evaporation sequence. This voltage drop can then be correlated with the ion current variation, for each step of the laser power, as reported in Fig. 7 (left). We can note that for each value of laser power the relationship between the ionic current and the voltage drop is linear, which is consistent with a resistive effect: ΔV ¼RI. Moreover the slope changes with the laser power, which is consistent with the photostimulated conductivity as discuss in the next paragraph. From the slopes of each line reported in Fig. 7 (left), the value of the tip resistance can be obtained for each value of laser power, as represented in Fig. 7 (right). Using the FIM mode on the MgO tip, the value of the ion current cannot be measured directly with the instrument used in this study. However, the gas pressure can be increased, inducing an increase in the emission current, as reported in the literature even if the relationship between pressure and ion current is not linear [27]. An increase in the voltage drop was observed when increasing the gas pressure; this behaviour confirms the strong contribution of the resistive effect to the ion energy deficit, as already measured by La-APT experiment. (Fig. 8).
4. Experimental results
4.2. Laser intensity and wavelength
In order to study the origin of the ions energy deficit in FIM or APT, the energy deficit was studied as a function of different parameters: the ionic current, the illumination conditions (laser power and wavelength), the evaporation rate, the samples geometry and the tip preparation processes. Two types of materials were investigated: MgO because it can be analysed both in FIM mode and La-APT mode, thanks to its high resistivity; and Si because its electrical properties are well known. In the case of Si, the analyses have been performed only in FIM mode because its low resistivity (around 104 Ω m for Si with 1012 cm 3 impurity concentration against 1012–1020 Ω m for MgO) gives a negligible resistive effect (never observed during previous analysis contrary to MgO) in La-APT where the ionic current is 1000 times lower than in FIM.
From the previous results, it appears that the resistive effect is the main contribution to the ions energy deficit in FIM or APT mode. However, many experimental parameters can influence this voltage drop, by changing the resistance R of the sample. In order to study the influence of the laser intensity and wavelength on the energy deficit, an un-doped Si tip was analysed by FIM using the 15° mode and keeping the ion current constant. As shown in Fig. 9 (left), when increasing the intensity of the laser beam, the voltage drop decreases and it disappears at high laser intensity. The values of the laser intensity to cancel the energy deficit are lower (around 10 times less) than the value generally used to perform La-APT analysis. Moreover, longer wavelengths are more efficient in contrast with the wavelength behaviour reported in the case of La-APT analysis.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
5
Fig. 6. (Left) Ionic current emitted from an MgO tip analysed at 9.2 kV, at T ¼80 K on FlexTAP instrument with the decelerating lens ata potential of 1.02 V0. The laser power is increasing by steps from 0.1 mW to 0.75 mW. The laser wavelength is equal to 343 nm and the beam waist on the tip is equal to 60 μm. (Right) Time of Flight measured for Mg2 þ at 12 Da from this experiment as function of the number of atoms (the line shows the mean value).
4
The focused ion beam (FIB) milling process used for tip preparation and the tip shape are investigated in order to understand their possible effect on the energy deficit of field ionised or field evaporated ions. o111 4 p-doped Si tips, with identical shape (radius of 50 nm, cone angle of 5° and length of 10 μm) but different ion milling conditions, are analysed in FIM mode at 20 K. All the tips are milled with 30 kV-Ga ions. Then, a few tips are milled with 2 kV Ga ions in order to remove the most damaged region of the tip: this is known as the cleaning procedure. For short cleaning time (low flux) the tips are milled on a dozen of nanometres; for long cleaning time (high flux), on several tens of nanometres. As reported in Table 2, the longer the cleaning time, the smaller the voltage drop. All the analyses are done at constant ion current; hence the change in the voltage drop is due to a change in the resistance of the tip caused by the cleaning process. Concerning the geometry of the tip, as in the case of a cone or a cylinder, the tip resistance depends on its length, the longer the tip is and the higher the resistance of the whole tip will be. In the case of two o100 4 un-doped Si tips, having a length of 15 μm and 30 μm, the measured voltage drops were equal to 50 V (Fig. 10a)) and 200 V (Fig. 10b)), respectively. This huge change in the voltage drop, cannot be explain taking into account only the difference in length between the two tips (Eq. (2)). As reported in Fig. 10, the two tips have the same shape. However, the first one was milled
ΔV/V0 (%)
4.3. Samples geometry and tip preparation processes
3.9 3.8 3.7 3.6 3.5
0
10
-6
20
PNe (10 mbar)
30
Fig. 8. Evolution of the voltage drop as function of the gas pressure (Ne) in FIM, performed on MgO at 20 K.
on all its length of 15 μm while the second tip only on the first 10 μm with the same milling process. Hence, along the second tip, at the bottom of the milled surface, the native oxide should still be present, changing the resistivity of the tip. In fact, the resistivity of silicon oxide is more than 1010 times higher than the resistivity of crystalline silicon at 300 K [28]. Another interesting parameter is the apex radius. If the contribution of the near apex part is important, due to its small end surface, we should observe a variation of voltage drop when the tip radius changes. A o100 4 un-doped Si tip was analysed by FIM in the 15 °mode at 20 K and with a Ne pressure of 10 5 mbar.
Fig. 7. (Left) Voltage drop as a function of the ionic current from data reported in Fig. 6. (Right) Values of tip total resistance calculated from these data.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
6
15
1.E+11
UV
ΔV(V)
Excess Carrier Density (cm-3)
Green
10
IR
5
C
1.E+10
UV Green IR
1.E+09 1.E+08
0 0
0.025
0.05
0.075
0.1
I (GW/cm²)
0
0.01
0.02
0.03
I (GW/cm²)
Fig. 9. (Left) Voltage drop of a o 1004 un-doped. Si tip analysed in FIM mode at 15°. (Right) Excess carrier density generated by laser illumination calculated from the voltage drop. Measurements performed at 20 K with a Neon pressure of 10 5 mbar at 6.5 kV. Three wavelengths are used: IR (1030 nm), Green (515 nm) and UV (343 nm). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). Table 2 Voltage drop of o 1114 p-doped Si tips prepared with different cleaning processes. Performed with FIM at 20 K with 10 5 mbar of Neon, the measurement accuracy is around 1 V.
Voltage drop
No cleaning
Short cleaning
Long cleaning
12 V
5V
3V
Fig. 10. Two tips of o 1004 un-doped. Si analysed in FIM at 20 K with 10 5 mbar of Neon. The tip (a) was milled on its whole surface contrary to the (b) tip.
We perform two different analyses: before and after evaporation in La-APT, using the UV laser at a power of 0.02 mW. At the beginning, the voltage applied to the tip to evaporate 0.01 atom/ pulse was 6 kV, at the end 12 kV. Because the surface field conditions are kept constant all along the analysis, this increase of the voltage corresponds to an increase of the tip radius of a factor of 2. The voltage drops measured before and after the La-APT analysis are the same: 190 V.
5. Discussions 5.1. Laser intensity and wavelength In Fig. 7 (left) it is shown that the resistance of the MgO tip decreases of a factor of ten in the range of laser power
experimentally explored in La-APT. According to the trend of the curve, switching from no laser illumination to low laser illumination should decrease the resistance of several orders of magnitude. Moreover, in Fig. 9 (left), the voltage drop decreases as a function of the laser intensity because the laser illumination changes the tip resistance. Indeed, each laser pulse creates an excess of free carriers inside the tip through photon absorption [22]. As reported in Eq. (2), this increase in the carrier density (nc) corresponds to a decrease in resistivity. After the laser illumination, the excess carriers recombine. However, if the recombination time is of the same order of magnitude as the time separating two laser pulses, the density of charge carriers does not have enough time to go back to its original value before the next laser pulse. In this case, the excess carrier density present after the laser pulse should reach a stationary regime after several laser pulses. In any case, the higher the laser intensity, the higher the value of the carrier density after the laser pulse, as calculated from Eq. (2) and reported in Fig. 9(right). The voltage drop becomes negligible at a value of the carrier density of almost 1011 cm 3; it means that the Si needs a small excess of carrier density to lower this resistivity enough to cancel the voltage drop. It can also be observed that higher laser intensities are required to cancel the voltage drop in the case of shorter wavelengths. In order to understand this phenomenon, the optical absorption behaviour of the tip was studied as a function of the wavelength. Using a commercial software for Finite Difference in Time Domain (FDTD) numerical calculations by Lumerical, the absorption map of the Si tip can be obtained for the different laser wavelengths, as reported in Fig. 11 for UV and Green light. In the case of an UV laser, all the absorption occurs at the tip surface, and a high density of carriers is generated in a small volume. This high density induces a fast recombination rate at the surface (recombination time 1 ps): most of the carriers recombine before they can diffuse inside the tip volume, thus the resistivity decreases only on a very short time scale of 1 ps. In the case of green and IR wavelengths, the light is absorbed in all the volume of the tip, hence the local carrier density is lower and the
Fig. 11. Optical absorption map (W/m3) computed by FDTD method for a Silicon tip (50 nm radius and 5° shank angle), corresponds to the power absorption density for an incoming intensity of 1 W m 2. The laser beam comes from the bottom. Laser wavelength of 343 nm (left) and 515 nm (right).
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
7
recombination rate is slower (recombination time 1 μs): the resistivity is decreased for a longer time. As a conclusion, absorption in the bulk of the tip results in a lower average tip resistivity, hence, green and IR lights are more efficient in reducing the tip resistivity. 5.2. Samples geometry and tip preparation processes It was reported in the literature, especially in the case of silicon, that the surface and the volume conduction properties are slightly different [29]. The electronic current travels more easily near the surface because of its smaller resistivity. In the case of APT tips, due to their nanoscale dimension, the surface phenomenon can become dominant compared to volume phenomena. In fact, the tip resistance can be schematically represented as two resistances in parallel and the total resistance can be simply given by:
Rtip =
R surface R core R surface + R core
(3)
In our case, by taking into account the tip geometry from Fig. 10 (cone on a cylinder), we can estimate the contribution of the volume and the surface for the cone (15 μm length, 50 nm apex radius and 1 μm base radius). We find Rsurface 5.1011 Ω and Rvolume ¼20.1012 Ω. It means that most of the electronic current travels in near the surface and not through the volume of the tip. The surface resistance is higher than those reported in previous work for plane, crystalline and clean surface [30]. Considering the low temperature of the analysis (20 K) and the high density of the surface defects, an increase of the surface resistivity is expected Focused Ion Beam (FIB) sputtering process used for tip sample preparation induces damages such as surface amorphization, gallium implantation and lattice defects (vacancies, interstitials, dislocations). Moreover, the depth and concentration of the FIB-induced damages are dependant on the target material, ion acceleration energy, ion dose, and the ion incident angle. Using transmission electronic microscopy (TEM), Lee et al. [31] showed that milling APT samples using 30 kV Ga ions leaves a surface damage layer of 25 nm in thickness. One can simulate, using SRIM software [32], the effect of the implantation of Ga ions in Si at different accelerating voltages. At 30 kV, the Ga-projected range reaches 28 nm in Silicon sample and is only limited to 5 nm when the Ga-beam is accelerated at 2 kV. Tip sample preparation improvements using ion acceleration under 5 kV have also been demonstrated by Thompson et al. [15]. The FIB-induced damages increase the tip resistivity due to the partial amorphization of the sample. Indeed, amorphous silicon has a resistivity ten times higher than crystalline silicon. This means that the surface layer damaged by milling, has a higher resistance. Hence, the resistance of the whole tip increases also if the volume resistance remains constant. In the case of tips with identical geometry, the different values of the energy deficit, experimentally measured for different FIB cleaning process and reported in Table 2, can give information on the thickness of the amorphization region. This thickness is strongly dependant on the energy used in FIB [33] and can be simulated numerically: it is around 28 nm in the case of non-cleaned tip, or only 5 nm in the case of the long cleaning process with 2 keV Ga ions. Concerning the tip shape, as shown in Fig. 10, the tip can be considered to be made of a cone (the milled extremity) on a cylinder (the part that is not milled, in the case of the second tip, Fig. 10(b)). The total resistance of the tip can be schematically represented by two resistances in series (one for the cone and one for the cylinder). Moreover, each of these resistances can be described as two resistances in parallel (one for the core and one for
Fig. 12. Simple model for the tip resistance. The two regions of the tip have their core and surface contribution (oxide silicon in on case and oxide amorphous in the other case).
the surface layer):
⎛ R surf R core ⎞ ⎛ R surf R core ⎞ amoprh Si ⎟ ⎜ oxide Si ⎟ Rtip = ⎜⎜ + ⎜ surf core ⎟ core ⎟ surf ⎝ R amoprh + R Si ⎠cone ⎝ R oxide + R Si ⎠cylinder
(4)
In the case of the second tip, in the region that is not milled, the presence of oxide on the surface of silicon causes an inversion in the current behaviour. In the case of pure Si (or Si with an amorphous layer at the surface) the surface conduction was better and the current flows more easily in this area. In the case of oxide surfaces, the surface has a huge resistivity: the current should go through the volume of the tip. Hence, the main contribution to the voltage drop of the second tip is coming from the volume of cylinder (150 V, 75%of the total voltage drop of 200 V). Therefore, to study and predict the voltage drop, all the tip structure should be taken into account, not only its extremity corresponding to the milled part. Therefore, the change in the apex radius has an influence on the resistance of the near apex part of the tip, however, the contribution of the cone at the base of the tip and the cylinder part (if present) remain unchanged. Hence, the change of the tip radius doesn't influence the total resistance of the tip if the contribution to the tip resistance of the cone and the cylinder are dominant. (this is the case for long tips prepared from microposts or for samples with a highly resistive bottom layer). (Fig. 12). The sample preparation method can be modified in order to suppress this resistive effect. For example, in the case of an MgO tip, after the annular milling, a thin layer of W was deposited on one side of the tip using ion beam assisted deposition. The layer was not visible by SEM, and thus, should be only a few nanometres thick. The FIM image obtained on an MgO tip prepared with this method is presented in Fig. 13(a). No voltage drop was reported on this tip also without laser illumination. Fig. 13(b) shows the image of an MgO tip without the W deposition. In this case, the voltage drop is of 195 V and, even under laser illumination (0.92 GW/cm 2), a voltage drop of 50 V still remains. Metal deposition is an easy way to avoid the problems induced by the resistive effect. Moreover it shows that it is not necessaryto deposit a layer on the whole tip, which avoids evaporation of a horizontal interface during the La-APT analysis. A thin layer on one side is enough to create a short circuit between the base of the tips and the apex.
6. Conclusions The energy deficit of field-ionised or evaporated atoms from semiconductors and oxides is mainly due to a voltage drop induced by the electrical resistance of the tip. This ohmic effect can cause a voltage drop from a few volts to hundreds of volts, depending on the material, the tip shape, the sample preparation process and the illumination conditions. The contribution of the band-bending (8 V for MgO and 2 V for Si [22] without laser illumination) is proved to be negligible in all
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
8
Fig. 13. FIM image of (a) MgO tip with W deposit on one side of its shank, without laser illumination, (b) MgO tip prepared without metal deposition and under laser illumination (0.92 GW/cm 2). Experiments are performed at 20 K with 10 5 mbar of Ne.
the cases presented in this work. In fact, the voltage drop is proved to increase linearly as a function of the ionic current, as predicted by Ohm's law. The laser illumination decreases the value of the resistivity because it generates a high density of free carrier by photo-absorption process and this carrier excess leads to a reduced voltage drop and ion energy deficit. Moreover, it is shown that, even if light absorption is more efficient for a laser wavelength of 343 nm, the voltage drop is reduced more efficiently using IR light at 1030 nm. This can be explained by the differences in the absorption maps. In the case of UV light, the absorption is concentrated near the surface; hence a high density of carriers is generated in a small volume. Then, due to the fast recombination rate, the carriers have difficulties to reduce the resistance of the whole tip during the whole time between two laser pulses. The tip shape and the FIB method used to prepare the tip play also an important role in the variation of voltage drop. We show the influence of the surface state: a strong amorphization induced by FIB milling or, worse, an oxidation of the surface, increases the voltage drop significantly. We also demonstrate that the tip curvature radius at the apex has a negligible influence. Instead, at least in the case of silicon microposts (around 30 μm before milling), the tip resistivity is directly proportional to the tip length. In the case of silicon analysed in La-APT, the simple exposition to the pulsed laser during the analysis is able to totally cancel the voltage drop. However, in the case of oxides, the analysis parameters need to be chosen carefully, because the voltage drop decreases the time and spatial accuracy of La-APT analyses, even for a standard linear atom probe. Fortunately, we demonstrate that the deposition of a metallic layer of only a few nanometres on one side of the entire shank is enough to create a short circuit between the tip base and its apex, and cancel the voltage drop.
Acknowledgements We gratefully acknowledge the financial support from LabEx EMC3 (ASAP Project) and CAMECA Instruments Inc. (RING-APT Project).
References [1] T.F. Kelly, D.J. Larson, Annu. Rev. Mater. Res. 42 (2012) 1. [2] P.V. Liddicoat, X.-Z. Liao, Y. Zhao, Y. Zhu, M.Y. Murashkin, E.J. Lavernia, R. Z. Valiev, S.P. Ringer, Nat. Commun. 1 (2010) 63. [3] K. Biswas, J. He, I.D. Blum, C.-I. Wu, T.P. Hogan, D.N. Seidman, V.P. Dravid, M. G. Kanatzidis, Nature 489 (2012) 414. [4] Y. Chen, T. Ohkubo, M. Kodzuka, K. Morita, K. Hono, Scr. Mater. 61 (2009) 693. [5] L.M. Gordon, D. Joester, Nature 469 (2011) 194. [6] L. Rigutti, I. Blum, D. Shinde, D. Hernandez Maldonado, W. Lefebvre, J. Houard, F. Vurpillot, A. Vella, M. Tchernycheva, C. Durand, et al., Nano Lett. (2013). [7] K. Tedsree, T. Li, S. Jones, C.W.A. Chan, K.M.K. Yu, P.A. Bagot, E.A. Marquis, G. D. Smith, S.C.E. Tsang, Nat. Nanotechnol. 6 (2011) 302. [8] A. Janotti, C.G. Van de Walle, Phys. Rev. B 76 (2007) 165202. [9] Koji Inoue, Fumiko Yano, A. Nishida, T. Tsunomura, Takeshi Toyama, Yasuyoshi Nagai, Masayuki Hasegawa, Appl. Phys. Lett. 92 (103506) (2008) (103506-3). [10] D.A. Shashkov, D.A. Muller, D.N. Seidman, Acta Mater. 47 (1999) 15–16. [11] C. Kluthe, T. Al-Kassab, R. Kirchheim, Mater. Sci. Eng: A 1 (2002) 327. [12] C. Oberdorfer, P. Stender, C. Reinke, G. Schmitz, Microsc. Microanal. 13 (2007) 342–346. [13] K.R. Kuhlman, R.L. Martens, T.F. Kelly, N.D. Evans, M.K. Miller, Ultramicroscopy 89 (2001) 1–3. [14] G.L. Kellogg, J.Appl. Phys. 53 (1982) 6383–6386. [15] E.A. Marquis, N.A. Yahya, D.J. Larson, M.K. Miller, R.I. Todd, Mater. Today 13 (2010) 10. [16] Y.M. Chen, T. Ohkubo, M. Kodzuka, K. Morita, K. Hono, Ultramicroscopy 7 (2009) 61. [17] Y.M. Chen, T. Ohkubo, K. Hono, Ultramicroscopy 111 (2011) 562–566. [18] H. Tamura, M. Tsukada, K.P. McKenna, A.L. Shluger, T. Ohkubo, K. Hono, Phys. Rev. B 86 (2012) 195430. [19] E.P. Silaeva, M. Karahka, H.J. Kreuzer, Curr. Opin. Solid State Mater. Sci. 17 (2013) 211–216. [20] T.T. Tsong, Surf. Sci. 81 (1) (1979) 28–42. [21] M. Gilbert, F. Vurpillot, A. Vella, H. Bernas, B. Deconihout, Ultramicroscopy 107 (9) (2007) 767–772. [22] L. Arnoldi, E.P. Silaeva, A. Gaillard, F. Vurpillot, I. Blum, L. Rigutti, B. Deconihout, A. Vella, J.Appl. Phys. 115 (2014) p. 203705. [23] E. Müller, K. Bahadur, Phys. Rev. 102 (3) (1956) 624. [24] D.J. Larson, D.T. Foord, A.K. Petford-Long, H. Liew, M.G. Blamire, A. Cerezo, G.D. W. Smith, Ultramicroscopy 79 (Iss 1-4) (1999) 287–293. [25] K. Thompson, D. Lawrence, D.J. Larson, J.D. Olson, T.F. Kelly, B. Gorman, Ultramicroscopy 107 (Iss 2–3) (2007) 131–139. [26] T.T. Tsong, Atom-Probe Field Ion Microscopy: Field Ion Emission, and Surfaces and Interfaces at Atomic Resolution, Cambridge University Press, 2005. [27] E. Müller, Phys. Rev. 102 (1956) 618–624. [28] S.M. Sze, Physics of Semiconductor Devices, John Wiley and Sons, Inc, New York, 1981. [29] S. Hasegawa, X. Tong, S. Takeda, N. Sato, T. Nagao, Prog. Surf. Sci. 60 (1999). [30] S. Heike, S. Watanabe, Y. Wada, T. Hashizume, Phys. Rev. Lett. 81 (1998) 890. [31] J.H. Lee, B.H. Lee, Y.T. Kim, J.J. Kim, S.Y. Lee, K.P. Lee, C.G. Park, Micron 58 (2014) 32–37. [32] 〈http://www.srim.org/〉. [33] J.P. McCaffrey, M.W. Phaneuf, L.D. Madsen, Ultramicroscopy 87 (3) (2001) 97–104.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
Contents lists available at ScienceDirect
Ultramicroscopy journal homepage: www.elsevier.com/locate/ultramic
Role of the resistivity of insulating field emitters on the energy of field-ionised and field-evaporated atoms L. Arnoldi n, E.P. Silaeva, F. Vurpillot, B. Deconihout, E. Cadel, I. Blum, A. Vella GPM UMR 6634, Université de Rouen, Avenue de l'Université, 76801, BP12, 76801 Saint Etienne du Rouvray, France
art ic l e i nf o
a b s t r a c t
Article history: Received 20 August 2014 Received in revised form 3 November 2014 Accepted 23 November 2014
In order to improve the accuracy of laser atom probe analyses, it is important to understand all the physical processes induced by the combination of the high electrical field and the femtosecond laser beam during field evaporation. New information can be accessed from the energy of evaporated surface atoms or field-ionised atoms of an imaging gas. In order to study the ions energy, we combine La-APT and FIM analyses in a new experimental setup equipped with electrostatic lenses. We report measurements for semiconductors and oxides and we study the influence of the illumination conditions (laser power and wavelength), the evaporation rate, the sample geometry and the tip preparation processes. The results are discussed taking into account the resistive properties of nonmetallic samples and the photo-stimulated conductivity. This work clarifies the role of the laser and DC field in the energy deficit of field evaporated ions. & 2014 Elsevier B.V. All rights reserved.
Keywords: Atom probe tomography Field ion microscopy Field emitter Semiconductors Insulators Resistivity
1. Introduction Thanks to the use of fast-laser pulses, Laser assisted atom probe tomography (La-APT); 3D nanoanalysis is now possible for a wide range of materials. In addition to metals, it is now possible to analyse not only semiconductors but also ceramics and oxides [1– 7]. The quantitative analysis of this class of materials is challenging in many aspects. For instance, the ability to quantitatively measure composition reliably is particularly attractive when dealing with oxides that are able to accommodate high concentrations of point defects [8]. The analysis of highly resistive materials is also of fundamental importance in the field of microelectronics, including the challenging aspect of finding and analysing local interfaces between the structures of a functional device [9]. In the early years of APT, only nanometres scale oxide inclusions in metals or highly conductive oxides were studied [10–13]. The first published work on the atom probe analysis of bulk insulator was performed by Kellogg in the early eighties [14]. The analysis was found tedious and a thin metal coating was used to improve the electrical conduction. This requirement was thought essential to avoid the huge voltage drop that should be observed between the end apex of the sample and the base where the high voltage is applied. Thereafter, progresses in the analysis of oxides were slow, and the electrical resistivity was thought to be one of the main limitations of the n
Corresponding author. E-mail address: [email protected] (L. Arnoldi).
technique. Only recently, with the large spreading of modern laser assisted APT instruments all over the world, a blooming of studies in APT of oxides and ceramics was observed. Surprisingly, successful analyses of uncoated specimen were obtained for several bulk ceramics [15–16]. In addition near atomic scale resolution was demonstrated. These observations question our understanding of field evaporation for the case of poorly conductive materials. In a simple approach, in atom probe, the field emitter must drain the charge induced by pulsed field evaporation. The analysed bulk dielectric material is generally a micron size sample attached to a macroscopic metallic needle. It may then be considered as an electrical resistor which resistance can be considered to be very high especially at the cryogenic temperatures used (1010–1020 Ω). Kellogg predicted several kV of voltage drop for fused glass or Pyrex samples. Nevertheless, in recent experiments, voltage drops are rarely observed. Only in the paper of Chen et al. [17], a drop of only a few volts was measured in MgO samples. Chen et al. and after Tamura et al. [18] proposed a model based on accumulation of photo-induced holes changing the internal and external electric field around the tip apex. This paper shows the important role of the laser excitation in the field evaporation process. Some other authors pointed out the importance of surface states in the actual electrical properties of the sample. For instance, ab-initio calculations suggest that the band gap of crystalline alumina is reduced from 8 eV in the bulk to 2.5 eV at the surface, which makes the alumina surface more conductive and more absorbing in the visible domain of the laser excitation used (532 nm laser pulse/
http://dx.doi.org/10.1016/j.ultramic.2014.11.018 0304-3991/& 2014 Elsevier B.V. All rights reserved.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
2
2.3 eV or 355 nm/3.6 eV). The presence of Ga and the possible amorphization of the oxide surface by the gallium beam during sample preparation in the focused ion beam instrument may also contribute to a modification of the surface states and thus may create a charge-draining path, much like coating with a thin metal layer. The presence of the surface field may also change the physical properties of the material, as calculated by Kreuzer et al. [19]. In addition, the high surface to volume ratio of an APT sample may enhance the effects of any modification of the surfaces electrical properties. In this work, several experimental procedures were developed to accurately measure the in-situ electrical resistance of the APT sample. In order to measure the voltage drop with a high accuracy, electrostatic lenses are introduced in the chamber. The setup is then used as an ion energy analyser, which amplifies the effect of the ions energy on their trajectories and time of flight. This enables an accurate measurement of the voltage drop using the relationship: ΔE = neΔV , with ne the ion charge. With such an experimental setup we are able to investigate voltage variations at the tip apex ranging from a few volts to hundreds of volts. To amplify the observed voltage drop, the FIM mode was used in addition to the APT mode. The ion current can, indeed, be increased by several orders of magnitude due to the contribution of the ionisation of the image gas. The variation of the observed resistance of semiconductors and oxide tip samples were studied. We demonstrate that the laser illumination creates free carriers in the material, which significantly reduce its resistivity. After the laser pulse, these carriers recombine up to an average equilibrium value. In the case of FIM measurements, this average carrier density is monitored. The role of the specimen preparation is also addressed.
2. Theoretical consideration The ion energy deficit is induced by the voltage drop ΔV existing between the tip apex and its metallic base. Two different effects contribute to this voltage drop: the Ohmic effect and the Band Bending Effect [20]. Indeed, the expression of the drop can be written:
ΔV = V0 − Vtip = ΔVbb + ΔVOhm = ΔVbb + RI
(1)
where ΔVbb is the voltage drop induced by the change of band gap created by the accumulation of holes, I is the electronic current inside the tip (equal to the ionic current emitted from the tip) and R is the resistance of the tip. This resistance is given in a first approach by
R = L tip (μ c nc eStip )−1
(2)
where Stip = πrapex rbase [21] is the mean tip cross section, μc the carrier mobility and nc the carrier density [22]. In the case of semiconductors analysed by La-APT, both the ionic current and the resistance are low: the resistive effect is negligible and the voltage drop at the tip apex is smaller than 1 V. In FIM mode, the ionic current is higher than in La-APT mode (more than a factor 1000) so the contribution of this resistive effect can be observed even during the FIM analysis of semiconductors. In the case of oxides like MgO, the resistance is so high that the ohmic contribution can become the main contribution to the voltage drop.
3. Experimental setup and methods Two types of materials were analysed: silicon (Si) and MgO. Si tips were prepared by annular milling from micro-post [24]
of around 3 μm in radius and 40 μm in length. Two types of microposts were analysed: un-doped o 100 4 oriented micro-posts and p-doped (1018 cm 3) o1114 oriented micro-posts. MgO tips were prepared from bulk samples using the lift-out method [25,26]. All measurements were performed on a FlexTAP instrument (CAMECA) in APT mode or FIM mode. The main characteristic of the FlexTAP is the presence of three electrostatic lenses, between the tip and the detector, which can be polarised to accelerate of decelerate the emitted ions. This setup was developed in order to offer a flexible angle of view on the tip during the analysis: ranging from 4° to 30°. The other main goal is to improve the temporal resolution of the ToF spectrum. The relative accuracy of ToF measurements depends on the ratio of the ion ToF (t) over the time accuracy of the detector (Δt). The lenses allow to increase the value of ToF (t) and hence the accuracy of its measurement. The FlexTAP is coupled to a femto-second pulsed laser, with a repetition rate of 50 kHz in APT mode and 100 kHz in FIM mode, a pulse duration of 500 fs and a tuneable wavelength (1030 nm, 515 nm and 343 nm). The FIM is a projection microscope, which images the tip surface with an atomic resolution [23]. Noble gas atoms are introduced in the analysis chamber and are field-ionised close to protruding surface atoms, where the field is highest. After the ionisation, these ions are projected onto a grounded screen. The ionisation of the gas atoms by tunnelling effect requires a high electric field, around tens of volts per nanometre, which is of the same order than the field for field-evaporation in APT mode. Neon gas was used as imaging gas in the following experiments. The main advantages of FIM for the following experiments are that: 1. The ionic emission current I (and the electronic current inside the tip) is between 1000 and 10,000 times higher than in the case of APT. 2. FIM images can also be obtained without laser illumination, hence the voltage drops with and without laser illumination can be compared. This is not the case for APT analysis, where the laser power can be varied, but cannot be zero as laser pulses are required for TOF measurements. 3. Because the tip surface is not evaporated, the sample geometry remains constant during the experiment. The FIM detector is placed at 44.5 cm from the tip. An aperture is placed in front of the tip, and stops ions emitted at too large angles, that can have aberrant trajectories. The ion trajectories are schematically represented in Fig. 1 for fields of view of 30° and 15°. The voltages applied to the lenses for these two modes are given in Table 1. The 15° mode was used in the case of small voltage drops at the tip apex. Indeed, only one lens is polarized, hence, the relationship between the voltage drop and the shift of trajectories is easier to predict. Moreover, the magnification of this mode is larger than of the 30° mode, which causes the ions trajectories and TOFs to be more sensitive to small variations in Vtip. The main problem of this mode, however, is the high value of the potential applied to the decelerating lens: it is higher in the case of the 30° mode (1.005 against 0.982), inducing a high potential barrier on the ions trajectories. Therefore, if the energy deficit is important (higher than about 1.5%), the ions can be repelled and not reach the detector. The 30° mode, to the contrary, was chosen to measure energy deficits as high as a few percent. One advantage of this mode is its large field of view, which enables to collect more spatial information because the area imaged on the tip surface is 4 times larger. However, one drawback is that when the ions have high energy deficit, their trajectories can exhibit strong aberrations, as discussed later.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Fig. 1. Simulated trajectories of an ion emitted at a fixed (12°) angle at a voltage Vtip ¼ Vapplied (full lines) and at a voltage Vtip o Vapplied (dashed lines) for a field of view of 15° with 0.5% voltage deficit (a) and 30° with 2% deficit (b). Table 1 Voltages applied to the different lenses for the two modes used in this work: 15° and 30° fields of view.
FIM Lens 1 Voltage FIM Lens 2 Voltage
15°
30°
1.005 Vapplied 0
0.982 Vapplied 0.4 Vapplied
In the case of the 15° mode, when the laser is switched off, a change in magnification can be observed, (Fig. 2(a)). In order to find the corresponding voltage drop, the ions trajectories were simulated numerically for the geometry of the electrostatic setup of the FlexTAP. The simulation was repeated for different voltage drops in order to obtain a calibration curve (Fig. 2(b)). The relative shift in the ions impact position is a linear function of the relative voltage drop between 0% and 1%. In this configuration voltage drop of a few volts can be measured. When using the 30° FIM mode, the image of the MgO surface is very distorted and the detector is only partially hit, as reported in
3
Fig. 3(a). MgO terraces and crystallography are not visible even if a fourfold symmetry can still be distinguished. In order to explain such images, numerical simulations of the trajectories were performed for ions having an energy deficit of 3.75%. Fig. 4 reports the position of the impact of an ion on the detector, normalized to the detector radius, as a function of the emission angle. First, one can note that the function is not bijective: one position on the detector corresponds to more than one emission angle. A second point is the big difference of impact density: most of the trajectories hit one fourth of the detector while only a few others hit the rest of it. This behaviour explains why, in the experimental results, most ion impacts are observed in a circle smaller than the detector diameter. As shown in Fig. 3, the diameter of this circle increases when the laser is switched on. Again, using simulations of the ions trajectories, a calibration curve can be obtained, giving the diameter of this circle as a function of the voltage drop (Fig. 3b). In addition, as shown in Fig. 4, for ions emitted at an angle 420°, the impact position on the detector is very sensitive with the emission angle and the deficit value, hence a small variation of this angle or deficit will transform a bright spot into a large bright line on the detector, as observed in Fig. 3. An advantage of the measure at 30° mode is that a single value of deficit can be measured without comparing two images, unlike the case of the 15° mode. However, the accuracy of this mode is worse than that of the 15° mode (around 0.1% of the tip voltage). In the case of the APT analysis, the geometry of the FlexTAP chamber is the same as for FIM analysis (as reported in Fig. 1) except for the detector that is now placed 60 cm far from the tip. The presence of the three lenses allows to form a potential barrier on the ions path, thereby increasing their ToF. Hence, the expression of the time of flight becomes:
ToF = L‵
m 2neVapplied
where m is the ion mass, n its charge, Vapplied the applied voltage and L′ is a virtual flight length accounting for the effect of the lenses on the ToF. When only the first decelerating lens is polarized, the higher the lens voltage the larger the virtual flight length and the higher the sensitivity. If the lens voltage is too high, however, the ToF can become larger than the acquisition window of the instrument, which is 12 μs. In the following experiments, the best compromise was found to be a lens voltage equal to 1.02 Vapplied, when the ToF tends to infinity for a voltage of 1.04 Vapplied. Using this configuration, a small variation of voltage at the apex corresponds to important variation of ToF, as reported in Fig. 5.
Fig. 2. (a) FIM images of o 111 4 p-doped. (imaging gas 3 10 6 mbar of Ne at 20 K, 15° mode). The red circles show the positions of the bright spots with laser illumination and green circles refer to these positions without laser illumination. (b) Calculated FIM spots radial positions as function of the relative voltage drop. The radial positions are normalised by the radial positions measured in the presence of laser illumination. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
4
Relative position on the detector(%)
Fig. 3. FIM images of MgO at 20 K with 10 5 mbar of Ne and 30° mode on FlexTAP: (a) without laser illumination; (b) with 343 nm laser at I¼ 0.1 GW/cm². (c) Radius of the illuminated area (Ri) normalized to the detector radius (Rd) as function of the relative voltage drop.
100
4.1. Ion current
50 0 -30
-10
10
30
-50 -100 Emission angle of the ion (°)
Fig. 4. Position of the impact of an ion on the detector, normalized to the detector radius as a function of the emission angle of the ion for a voltage drop of 3.75%.
Fig. 5. The accuracy, defined as the variation of ToF corresponding to a change in the tip voltage of 1 V, as function of the ratio between the tip voltage and the first lens voltage when only the first lens is polarized.
Therefore, the energy deficit can easily be obtained from the ToF during the analysis. To do so, during the analysis, a small voltage variation of 50 V is applied to the tip and the shift in ToF is measured. This measured sensitivity is compared to the calibration curve given in Fig. 5 in order to obtain Vtip.
MgO tips were analysed in La-APT mode at constant voltage using the lens configuration described previously (only the first lens is polarized at 1.02 V0). After the measurement of the ToF shift for a change in voltage of 50 V, the tip continues to evaporate and its radius increases, hence the field at the apex slightly decreases. This decrease induces a reduction of the evaporation rate: i.e. a decrease in the ionic current. To increase the evaporation rate, we can increase the applied laser power to compensate for the decrease of the surface field. Five laser steps are reported in Fig. 6. Using the measured ToF shift (see Fig. 6(b)) and the calibration measurement, the voltage drop at the apex of the tip can be obtained as a function of the evaporation sequence. This voltage drop can then be correlated with the ion current variation, for each step of the laser power, as reported in Fig. 7 (left). We can note that for each value of laser power the relationship between the ionic current and the voltage drop is linear, which is consistent with a resistive effect: ΔV ¼RI. Moreover the slope changes with the laser power, which is consistent with the photostimulated conductivity as discuss in the next paragraph. From the slopes of each line reported in Fig. 7 (left), the value of the tip resistance can be obtained for each value of laser power, as represented in Fig. 7 (right). Using the FIM mode on the MgO tip, the value of the ion current cannot be measured directly with the instrument used in this study. However, the gas pressure can be increased, inducing an increase in the emission current, as reported in the literature even if the relationship between pressure and ion current is not linear [27]. An increase in the voltage drop was observed when increasing the gas pressure; this behaviour confirms the strong contribution of the resistive effect to the ion energy deficit, as already measured by La-APT experiment. (Fig. 8).
4. Experimental results
4.2. Laser intensity and wavelength
In order to study the origin of the ions energy deficit in FIM or APT, the energy deficit was studied as a function of different parameters: the ionic current, the illumination conditions (laser power and wavelength), the evaporation rate, the samples geometry and the tip preparation processes. Two types of materials were investigated: MgO because it can be analysed both in FIM mode and La-APT mode, thanks to its high resistivity; and Si because its electrical properties are well known. In the case of Si, the analyses have been performed only in FIM mode because its low resistivity (around 104 Ω m for Si with 1012 cm 3 impurity concentration against 1012–1020 Ω m for MgO) gives a negligible resistive effect (never observed during previous analysis contrary to MgO) in La-APT where the ionic current is 1000 times lower than in FIM.
From the previous results, it appears that the resistive effect is the main contribution to the ions energy deficit in FIM or APT mode. However, many experimental parameters can influence this voltage drop, by changing the resistance R of the sample. In order to study the influence of the laser intensity and wavelength on the energy deficit, an un-doped Si tip was analysed by FIM using the 15° mode and keeping the ion current constant. As shown in Fig. 9 (left), when increasing the intensity of the laser beam, the voltage drop decreases and it disappears at high laser intensity. The values of the laser intensity to cancel the energy deficit are lower (around 10 times less) than the value generally used to perform La-APT analysis. Moreover, longer wavelengths are more efficient in contrast with the wavelength behaviour reported in the case of La-APT analysis.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
5
Fig. 6. (Left) Ionic current emitted from an MgO tip analysed at 9.2 kV, at T ¼80 K on FlexTAP instrument with the decelerating lens ata potential of 1.02 V0. The laser power is increasing by steps from 0.1 mW to 0.75 mW. The laser wavelength is equal to 343 nm and the beam waist on the tip is equal to 60 μm. (Right) Time of Flight measured for Mg2 þ at 12 Da from this experiment as function of the number of atoms (the line shows the mean value).
4
The focused ion beam (FIB) milling process used for tip preparation and the tip shape are investigated in order to understand their possible effect on the energy deficit of field ionised or field evaporated ions. o111 4 p-doped Si tips, with identical shape (radius of 50 nm, cone angle of 5° and length of 10 μm) but different ion milling conditions, are analysed in FIM mode at 20 K. All the tips are milled with 30 kV-Ga ions. Then, a few tips are milled with 2 kV Ga ions in order to remove the most damaged region of the tip: this is known as the cleaning procedure. For short cleaning time (low flux) the tips are milled on a dozen of nanometres; for long cleaning time (high flux), on several tens of nanometres. As reported in Table 2, the longer the cleaning time, the smaller the voltage drop. All the analyses are done at constant ion current; hence the change in the voltage drop is due to a change in the resistance of the tip caused by the cleaning process. Concerning the geometry of the tip, as in the case of a cone or a cylinder, the tip resistance depends on its length, the longer the tip is and the higher the resistance of the whole tip will be. In the case of two o100 4 un-doped Si tips, having a length of 15 μm and 30 μm, the measured voltage drops were equal to 50 V (Fig. 10a)) and 200 V (Fig. 10b)), respectively. This huge change in the voltage drop, cannot be explain taking into account only the difference in length between the two tips (Eq. (2)). As reported in Fig. 10, the two tips have the same shape. However, the first one was milled
ΔV/V0 (%)
4.3. Samples geometry and tip preparation processes
3.9 3.8 3.7 3.6 3.5
0
10
-6
20
PNe (10 mbar)
30
Fig. 8. Evolution of the voltage drop as function of the gas pressure (Ne) in FIM, performed on MgO at 20 K.
on all its length of 15 μm while the second tip only on the first 10 μm with the same milling process. Hence, along the second tip, at the bottom of the milled surface, the native oxide should still be present, changing the resistivity of the tip. In fact, the resistivity of silicon oxide is more than 1010 times higher than the resistivity of crystalline silicon at 300 K [28]. Another interesting parameter is the apex radius. If the contribution of the near apex part is important, due to its small end surface, we should observe a variation of voltage drop when the tip radius changes. A o100 4 un-doped Si tip was analysed by FIM in the 15 °mode at 20 K and with a Ne pressure of 10 5 mbar.
Fig. 7. (Left) Voltage drop as a function of the ionic current from data reported in Fig. 6. (Right) Values of tip total resistance calculated from these data.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
6
15
1.E+11
UV
ΔV(V)
Excess Carrier Density (cm-3)
Green
10
IR
5
C
1.E+10
UV Green IR
1.E+09 1.E+08
0 0
0.025
0.05
0.075
0.1
I (GW/cm²)
0
0.01
0.02
0.03
I (GW/cm²)
Fig. 9. (Left) Voltage drop of a o 1004 un-doped. Si tip analysed in FIM mode at 15°. (Right) Excess carrier density generated by laser illumination calculated from the voltage drop. Measurements performed at 20 K with a Neon pressure of 10 5 mbar at 6.5 kV. Three wavelengths are used: IR (1030 nm), Green (515 nm) and UV (343 nm). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). Table 2 Voltage drop of o 1114 p-doped Si tips prepared with different cleaning processes. Performed with FIM at 20 K with 10 5 mbar of Neon, the measurement accuracy is around 1 V.
Voltage drop
No cleaning
Short cleaning
Long cleaning
12 V
5V
3V
Fig. 10. Two tips of o 1004 un-doped. Si analysed in FIM at 20 K with 10 5 mbar of Neon. The tip (a) was milled on its whole surface contrary to the (b) tip.
We perform two different analyses: before and after evaporation in La-APT, using the UV laser at a power of 0.02 mW. At the beginning, the voltage applied to the tip to evaporate 0.01 atom/ pulse was 6 kV, at the end 12 kV. Because the surface field conditions are kept constant all along the analysis, this increase of the voltage corresponds to an increase of the tip radius of a factor of 2. The voltage drops measured before and after the La-APT analysis are the same: 190 V.
5. Discussions 5.1. Laser intensity and wavelength In Fig. 7 (left) it is shown that the resistance of the MgO tip decreases of a factor of ten in the range of laser power
experimentally explored in La-APT. According to the trend of the curve, switching from no laser illumination to low laser illumination should decrease the resistance of several orders of magnitude. Moreover, in Fig. 9 (left), the voltage drop decreases as a function of the laser intensity because the laser illumination changes the tip resistance. Indeed, each laser pulse creates an excess of free carriers inside the tip through photon absorption [22]. As reported in Eq. (2), this increase in the carrier density (nc) corresponds to a decrease in resistivity. After the laser illumination, the excess carriers recombine. However, if the recombination time is of the same order of magnitude as the time separating two laser pulses, the density of charge carriers does not have enough time to go back to its original value before the next laser pulse. In this case, the excess carrier density present after the laser pulse should reach a stationary regime after several laser pulses. In any case, the higher the laser intensity, the higher the value of the carrier density after the laser pulse, as calculated from Eq. (2) and reported in Fig. 9(right). The voltage drop becomes negligible at a value of the carrier density of almost 1011 cm 3; it means that the Si needs a small excess of carrier density to lower this resistivity enough to cancel the voltage drop. It can also be observed that higher laser intensities are required to cancel the voltage drop in the case of shorter wavelengths. In order to understand this phenomenon, the optical absorption behaviour of the tip was studied as a function of the wavelength. Using a commercial software for Finite Difference in Time Domain (FDTD) numerical calculations by Lumerical, the absorption map of the Si tip can be obtained for the different laser wavelengths, as reported in Fig. 11 for UV and Green light. In the case of an UV laser, all the absorption occurs at the tip surface, and a high density of carriers is generated in a small volume. This high density induces a fast recombination rate at the surface (recombination time 1 ps): most of the carriers recombine before they can diffuse inside the tip volume, thus the resistivity decreases only on a very short time scale of 1 ps. In the case of green and IR wavelengths, the light is absorbed in all the volume of the tip, hence the local carrier density is lower and the
Fig. 11. Optical absorption map (W/m3) computed by FDTD method for a Silicon tip (50 nm radius and 5° shank angle), corresponds to the power absorption density for an incoming intensity of 1 W m 2. The laser beam comes from the bottom. Laser wavelength of 343 nm (left) and 515 nm (right).
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
7
recombination rate is slower (recombination time 1 μs): the resistivity is decreased for a longer time. As a conclusion, absorption in the bulk of the tip results in a lower average tip resistivity, hence, green and IR lights are more efficient in reducing the tip resistivity. 5.2. Samples geometry and tip preparation processes It was reported in the literature, especially in the case of silicon, that the surface and the volume conduction properties are slightly different [29]. The electronic current travels more easily near the surface because of its smaller resistivity. In the case of APT tips, due to their nanoscale dimension, the surface phenomenon can become dominant compared to volume phenomena. In fact, the tip resistance can be schematically represented as two resistances in parallel and the total resistance can be simply given by:
Rtip =
R surface R core R surface + R core
(3)
In our case, by taking into account the tip geometry from Fig. 10 (cone on a cylinder), we can estimate the contribution of the volume and the surface for the cone (15 μm length, 50 nm apex radius and 1 μm base radius). We find Rsurface 5.1011 Ω and Rvolume ¼20.1012 Ω. It means that most of the electronic current travels in near the surface and not through the volume of the tip. The surface resistance is higher than those reported in previous work for plane, crystalline and clean surface [30]. Considering the low temperature of the analysis (20 K) and the high density of the surface defects, an increase of the surface resistivity is expected Focused Ion Beam (FIB) sputtering process used for tip sample preparation induces damages such as surface amorphization, gallium implantation and lattice defects (vacancies, interstitials, dislocations). Moreover, the depth and concentration of the FIB-induced damages are dependant on the target material, ion acceleration energy, ion dose, and the ion incident angle. Using transmission electronic microscopy (TEM), Lee et al. [31] showed that milling APT samples using 30 kV Ga ions leaves a surface damage layer of 25 nm in thickness. One can simulate, using SRIM software [32], the effect of the implantation of Ga ions in Si at different accelerating voltages. At 30 kV, the Ga-projected range reaches 28 nm in Silicon sample and is only limited to 5 nm when the Ga-beam is accelerated at 2 kV. Tip sample preparation improvements using ion acceleration under 5 kV have also been demonstrated by Thompson et al. [15]. The FIB-induced damages increase the tip resistivity due to the partial amorphization of the sample. Indeed, amorphous silicon has a resistivity ten times higher than crystalline silicon. This means that the surface layer damaged by milling, has a higher resistance. Hence, the resistance of the whole tip increases also if the volume resistance remains constant. In the case of tips with identical geometry, the different values of the energy deficit, experimentally measured for different FIB cleaning process and reported in Table 2, can give information on the thickness of the amorphization region. This thickness is strongly dependant on the energy used in FIB [33] and can be simulated numerically: it is around 28 nm in the case of non-cleaned tip, or only 5 nm in the case of the long cleaning process with 2 keV Ga ions. Concerning the tip shape, as shown in Fig. 10, the tip can be considered to be made of a cone (the milled extremity) on a cylinder (the part that is not milled, in the case of the second tip, Fig. 10(b)). The total resistance of the tip can be schematically represented by two resistances in series (one for the cone and one for the cylinder). Moreover, each of these resistances can be described as two resistances in parallel (one for the core and one for
Fig. 12. Simple model for the tip resistance. The two regions of the tip have their core and surface contribution (oxide silicon in on case and oxide amorphous in the other case).
the surface layer):
⎛ R surf R core ⎞ ⎛ R surf R core ⎞ amoprh Si ⎟ ⎜ oxide Si ⎟ Rtip = ⎜⎜ + ⎜ surf core ⎟ core ⎟ surf ⎝ R amoprh + R Si ⎠cone ⎝ R oxide + R Si ⎠cylinder
(4)
In the case of the second tip, in the region that is not milled, the presence of oxide on the surface of silicon causes an inversion in the current behaviour. In the case of pure Si (or Si with an amorphous layer at the surface) the surface conduction was better and the current flows more easily in this area. In the case of oxide surfaces, the surface has a huge resistivity: the current should go through the volume of the tip. Hence, the main contribution to the voltage drop of the second tip is coming from the volume of cylinder (150 V, 75%of the total voltage drop of 200 V). Therefore, to study and predict the voltage drop, all the tip structure should be taken into account, not only its extremity corresponding to the milled part. Therefore, the change in the apex radius has an influence on the resistance of the near apex part of the tip, however, the contribution of the cone at the base of the tip and the cylinder part (if present) remain unchanged. Hence, the change of the tip radius doesn't influence the total resistance of the tip if the contribution to the tip resistance of the cone and the cylinder are dominant. (this is the case for long tips prepared from microposts or for samples with a highly resistive bottom layer). (Fig. 12). The sample preparation method can be modified in order to suppress this resistive effect. For example, in the case of an MgO tip, after the annular milling, a thin layer of W was deposited on one side of the tip using ion beam assisted deposition. The layer was not visible by SEM, and thus, should be only a few nanometres thick. The FIM image obtained on an MgO tip prepared with this method is presented in Fig. 13(a). No voltage drop was reported on this tip also without laser illumination. Fig. 13(b) shows the image of an MgO tip without the W deposition. In this case, the voltage drop is of 195 V and, even under laser illumination (0.92 GW/cm 2), a voltage drop of 50 V still remains. Metal deposition is an easy way to avoid the problems induced by the resistive effect. Moreover it shows that it is not necessaryto deposit a layer on the whole tip, which avoids evaporation of a horizontal interface during the La-APT analysis. A thin layer on one side is enough to create a short circuit between the base of the tips and the apex.
6. Conclusions The energy deficit of field-ionised or evaporated atoms from semiconductors and oxides is mainly due to a voltage drop induced by the electrical resistance of the tip. This ohmic effect can cause a voltage drop from a few volts to hundreds of volts, depending on the material, the tip shape, the sample preparation process and the illumination conditions. The contribution of the band-bending (8 V for MgO and 2 V for Si [22] without laser illumination) is proved to be negligible in all
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
L. Arnoldi et al. / Ultramicroscopy ∎ (∎∎∎∎) ∎∎∎–∎∎∎
8
Fig. 13. FIM image of (a) MgO tip with W deposit on one side of its shank, without laser illumination, (b) MgO tip prepared without metal deposition and under laser illumination (0.92 GW/cm 2). Experiments are performed at 20 K with 10 5 mbar of Ne.
the cases presented in this work. In fact, the voltage drop is proved to increase linearly as a function of the ionic current, as predicted by Ohm's law. The laser illumination decreases the value of the resistivity because it generates a high density of free carrier by photo-absorption process and this carrier excess leads to a reduced voltage drop and ion energy deficit. Moreover, it is shown that, even if light absorption is more efficient for a laser wavelength of 343 nm, the voltage drop is reduced more efficiently using IR light at 1030 nm. This can be explained by the differences in the absorption maps. In the case of UV light, the absorption is concentrated near the surface; hence a high density of carriers is generated in a small volume. Then, due to the fast recombination rate, the carriers have difficulties to reduce the resistance of the whole tip during the whole time between two laser pulses. The tip shape and the FIB method used to prepare the tip play also an important role in the variation of voltage drop. We show the influence of the surface state: a strong amorphization induced by FIB milling or, worse, an oxidation of the surface, increases the voltage drop significantly. We also demonstrate that the tip curvature radius at the apex has a negligible influence. Instead, at least in the case of silicon microposts (around 30 μm before milling), the tip resistivity is directly proportional to the tip length. In the case of silicon analysed in La-APT, the simple exposition to the pulsed laser during the analysis is able to totally cancel the voltage drop. However, in the case of oxides, the analysis parameters need to be chosen carefully, because the voltage drop decreases the time and spatial accuracy of La-APT analyses, even for a standard linear atom probe. Fortunately, we demonstrate that the deposition of a metallic layer of only a few nanometres on one side of the entire shank is enough to create a short circuit between the tip base and its apex, and cancel the voltage drop.
Acknowledgements We gratefully acknowledge the financial support from LabEx EMC3 (ASAP Project) and CAMECA Instruments Inc. (RING-APT Project).
References [1] T.F. Kelly, D.J. Larson, Annu. Rev. Mater. Res. 42 (2012) 1. [2] P.V. Liddicoat, X.-Z. Liao, Y. Zhao, Y. Zhu, M.Y. Murashkin, E.J. Lavernia, R. Z. Valiev, S.P. Ringer, Nat. Commun. 1 (2010) 63. [3] K. Biswas, J. He, I.D. Blum, C.-I. Wu, T.P. Hogan, D.N. Seidman, V.P. Dravid, M. G. Kanatzidis, Nature 489 (2012) 414. [4] Y. Chen, T. Ohkubo, M. Kodzuka, K. Morita, K. Hono, Scr. Mater. 61 (2009) 693. [5] L.M. Gordon, D. Joester, Nature 469 (2011) 194. [6] L. Rigutti, I. Blum, D. Shinde, D. Hernandez Maldonado, W. Lefebvre, J. Houard, F. Vurpillot, A. Vella, M. Tchernycheva, C. Durand, et al., Nano Lett. (2013). [7] K. Tedsree, T. Li, S. Jones, C.W.A. Chan, K.M.K. Yu, P.A. Bagot, E.A. Marquis, G. D. Smith, S.C.E. Tsang, Nat. Nanotechnol. 6 (2011) 302. [8] A. Janotti, C.G. Van de Walle, Phys. Rev. B 76 (2007) 165202. [9] Koji Inoue, Fumiko Yano, A. Nishida, T. Tsunomura, Takeshi Toyama, Yasuyoshi Nagai, Masayuki Hasegawa, Appl. Phys. Lett. 92 (103506) (2008) (103506-3). [10] D.A. Shashkov, D.A. Muller, D.N. Seidman, Acta Mater. 47 (1999) 15–16. [11] C. Kluthe, T. Al-Kassab, R. Kirchheim, Mater. Sci. Eng: A 1 (2002) 327. [12] C. Oberdorfer, P. Stender, C. Reinke, G. Schmitz, Microsc. Microanal. 13 (2007) 342–346. [13] K.R. Kuhlman, R.L. Martens, T.F. Kelly, N.D. Evans, M.K. Miller, Ultramicroscopy 89 (2001) 1–3. [14] G.L. Kellogg, J.Appl. Phys. 53 (1982) 6383–6386. [15] E.A. Marquis, N.A. Yahya, D.J. Larson, M.K. Miller, R.I. Todd, Mater. Today 13 (2010) 10. [16] Y.M. Chen, T. Ohkubo, M. Kodzuka, K. Morita, K. Hono, Ultramicroscopy 7 (2009) 61. [17] Y.M. Chen, T. Ohkubo, K. Hono, Ultramicroscopy 111 (2011) 562–566. [18] H. Tamura, M. Tsukada, K.P. McKenna, A.L. Shluger, T. Ohkubo, K. Hono, Phys. Rev. B 86 (2012) 195430. [19] E.P. Silaeva, M. Karahka, H.J. Kreuzer, Curr. Opin. Solid State Mater. Sci. 17 (2013) 211–216. [20] T.T. Tsong, Surf. Sci. 81 (1) (1979) 28–42. [21] M. Gilbert, F. Vurpillot, A. Vella, H. Bernas, B. Deconihout, Ultramicroscopy 107 (9) (2007) 767–772. [22] L. Arnoldi, E.P. Silaeva, A. Gaillard, F. Vurpillot, I. Blum, L. Rigutti, B. Deconihout, A. Vella, J.Appl. Phys. 115 (2014) p. 203705. [23] E. Müller, K. Bahadur, Phys. Rev. 102 (3) (1956) 624. [24] D.J. Larson, D.T. Foord, A.K. Petford-Long, H. Liew, M.G. Blamire, A. Cerezo, G.D. W. Smith, Ultramicroscopy 79 (Iss 1-4) (1999) 287–293. [25] K. Thompson, D. Lawrence, D.J. Larson, J.D. Olson, T.F. Kelly, B. Gorman, Ultramicroscopy 107 (Iss 2–3) (2007) 131–139. [26] T.T. Tsong, Atom-Probe Field Ion Microscopy: Field Ion Emission, and Surfaces and Interfaces at Atomic Resolution, Cambridge University Press, 2005. [27] E. Müller, Phys. Rev. 102 (1956) 618–624. [28] S.M. Sze, Physics of Semiconductor Devices, John Wiley and Sons, Inc, New York, 1981. [29] S. Hasegawa, X. Tong, S. Takeda, N. Sato, T. Nagao, Prog. Surf. Sci. 60 (1999). [30] S. Heike, S. Watanabe, Y. Wada, T. Hashizume, Phys. Rev. Lett. 81 (1998) 890. [31] J.H. Lee, B.H. Lee, Y.T. Kim, J.J. Kim, S.Y. Lee, K.P. Lee, C.G. Park, Micron 58 (2014) 32–37. [32] 〈http://www.srim.org/〉. [33] J.P. McCaffrey, M.W. Phaneuf, L.D. Madsen, Ultramicroscopy 87 (3) (2001) 97–104.
Please cite this article as: L. Arnoldi, et al., Role of the resistivity of insulating field emitters on the energy of field-ionised and fieldevaporated atoms, Ultramicroscopy (2015), http://dx.doi.org/10.1016/j.ultramic.2014.11.018i
