Energy Separation in Various Ultrathin Film System

CHAPTER 3: Experimental Setup

4.1 Energy Separation Measurement

4.1.2 Energy Separation in Various Ultrathin Film System

Besides uniform thin film, we also tried to observe the work function difference in structurally distinct thin films. One of the most well-known system is Co/Cu(111) system. The Cu(111) surface can be divided into two stacking-dependent electronic regions, faulted anf unfaulted^[4-4], which are different in the surface energy as on Si(111) 7×7 surface. Pendersen have observed the specific triangular islands growth of Co on Cu(111)^[4-5]. Hence when Co atoms are deposited onto Cu(111), two structres of faulted and unfaulted islands appear. O. Pietzsch et al. have demonstrated that the dI/dV-V curves for the two types of islands are different in their energetic position of peaks as well as in intensity^[4-6]. It is because of the electronic difference in the two regions. While the unfaulted regions exhibit higher local density of states (0.35eV) then the faulted ones (0.29eV), we can also utilize I-V mapping to distinguish the two regions. Through the I-V measurement on Co/Cu(111) system, we can have the same result as that of Pietzsch et al., as shown in Fig. 4.16. A. L. Vázquez et al. have also found that the Co/Cu(111) surface mapping can be obtained at different biases^[4-7].

1.1 eV

0.4 eV

operated I-V measurement on the two islands. Since the faulted Co islands have the local density of states at 0.29eV and unfaulted at 0.35eV, we can easily identify the two Co islands on Cu(111), as shown in Fig. 4.5.

Fig. 4.6 (a) shows the spectra of Cu(111), faulted, and unfaulted Co islands. We can see that the Gundlach oscillation of the faulted and unfaulted islands are quite different, as shown in Fig. 4.6. It is because the electronic stacking structure of the two islands is different. Taking the high order peaks into account, it is obvious that the energy separations of higher orders kept constant in both spectra, 0.48eV for the unfailted Co/Cu(111) and 0.38eV for the faulted Co/Cu(111), shown in Fig. 4.6 (b). It shows that the constant energy separation is also valid in structually different ultrathin film system.

Moreover, it is worth noting that the energy separations in the two spectra is only different in 0.1eV. This result shows that the energy resolution of this technique is less than 0.1eV. In other words, even though the spectra of the faulted and unfaulted Co islands are slightly shifted to 0.1eV, this technique is able to recognize the two distinct performances. Therefore it is predictable that the error of the measurement can be below 0.02 eV, which is much better than that of detecting apparent barrier height with STM^[4-8]. The technique can be prevailing in many reconstructed ultrathin film system to recognize the different Gundlach oscillation for distinct structures of the reconstructed surface.

Fig. 4.5 (a) STM image of faulted and unfaulted Co islands on Cu(111). The image size is 60nm×60nm. (b) dI/dV-V spectra of faulted and unfaulted Co islands.

Fig. 4.6 (a) dZ/dV-V spectra of Cu(111) and Co islands. (b) Energy separation for the two islands.

(a)

(b)

f

Cu(111)

f

u

Since the energy separation occurs in ultrathin film system, it is interesting to find out whether this phenomenon can be seen in single atomic ultrathin film systems. Co was deposited on reconstructed Au(111) surface at 8K as shown in Fig. 4.7(a). Based on thermodynamics, Co atoms should tend to compile on the regions possessing higher surface energy, such as the bending of the ridges or the step edge. However, the distribution of Co atoms in Fig. 4.7 (a) is random and shows little preference on any specific sites. It is due to the poor diffusibility of these adsorbates at such a low temperature. After annealing at room temperature, Co atoms obviously aggregate to the bending sites and step edges, as shown in Fig. 4.13 (b). Compared to the growth on Au(111), Co/Cu(111) is considered as a heteroepitaxial system with a high degree of perfection^[4-9], illustrated in Fig. 4.5(a). It is considered as the little mismatch of 1.9%

for Co/Cu system that is much lower than the mismatch of 38.5% for Co/Au(111) system.

However, what we concern is the performance of the Gundlach oscillation of the single Co atoms on Au(111). We tried to zoom in the STM image to obtain a single Co atom, and the observed diameter is about 13Å^[4-10].

Fig. 4.7 Reconstructed Au(111) surface with Co deposition at 8K and RT respectively. Both image

(a) (b)

surface in the scanning size of 6nm×6nm. The full-width of maximum height (FWMH) of the spot in Fig. 4.8 (a) is also 13Å, which is close to the results obtained by V.

Madhavan et al.(Fig. 4.8(b)). The image observed by STM is obviously larger than the real diameter of Co atoms (2.51Å). It is believed that the interaction between Co atoms and Au(111) surface leads to the fluctuation of the electronic structure, which inspires the tunneling current around the Co atoms and makes them look larger. Regarding the bunches 13Ǻ in diameter are isolated Co atoms, we tried to operate Z-V measurement on one single Co atom and Au(111) substrate as shown in Fig. 4.9.

Fig. 4.8 (a) The STM topography of single Co atom on reconstructed Au(111) surface. Three distinct structures in this image are all marked by colored arrows. The image size is 6nm×6nm. (b) STM topography of isolated Co atoms on Au(111) surface^[4-10].

FCC Ridge

Co HCP

(a) (b)

we can see, the Gundlach oscillation of Co at low bias looks much rougher than that of Au(111). It is because that the data points on single Co atoms are deficient to obtain an average smooth curve. However, it is still clear that there exhibits a separation between the dZ/dV-V spectra of the Co atoms and Au(111) substrate. The separation can be recognized as 0.13eV especially at high bias (indicated by green rectangle in Fig, 4.9).

This result shows that there exhibits a work function difference between single Co atoms and the substrate.

Recalling the definition of work function, it is considered to be a performance of a collection. In other words, the work function of isolated atoms may be the same as that of the substrate due to their negligible contribution to the concrete electronic performance. Through the Z-V measurement we can understand that even single atoms can devote themsleves to the variation of work function and act as collective atoms.

From the above discussion, we found that work function difference is a general phenomenon in all thin film systems.

Fig. 4.9 dZ/dV-V spectrum at higher bias. Energy separations between Co and Au(111) become constant when the applied bias is close to 7.5V.

Constant separation=0.13eV

During the scanning process, the tip may change its shape due to the varied applied voltage. In Sec. 3.4.1, we have introduced the dependence of tip-sample electric field on the tip shape. When the tip is blunt, tunneling current is easier to be generated and electric field between the tip-sample gap is stronger than that within sharp tip condition. It is interesting to find out how the energy separations become when the tip condition changes. In this subject we operated the observation on single-Ag/Cu(111) system.

4.2.1 Interplay between Discontinuous Contrast and Energy Separation

Although the feedback was fixed during the Z-V measurement, we found that there may exhibits a discontinuous contrast in a STM image as shown in Fig. 4.10(a).

In Fig. 4.10(a), A is the bared Cu(111) substrate, B and C the single-layered Ag film.

However, it can be seen that the image is discontinuous between B and C regions, indicated by the yellow arrow in Fig. 4.10(a). Hence we tried to operate Z-V measurement on the bared Cu(111) substrate and both regions of Ag film, B and C.

Figure 4.10(b) shows the average dZ/dV-V spectra of the three regions indicated in Fig.

4.10(a). Even though region B and C are both obtained on Ag film, the peak positions and intensities are not identical. It is attributed to the abrupt change of the tip condition.

The tip condition at area A and B is unchanged because of the constant energy separation, 0.3eV. However, the energy separation between spectra at A and C is decreasing with the order for Gundlach oscillation, as shown in Fig. 4.10(c). This observation reveals that the work function measurement with Gundlach oscillation demands the highly stability of the tip condition.

However, as seen in Fig. 4.10(c), one can still estimate the work function by

considered as the image potential state. In other words, the energy separation of that order is the one close to the real value of work function difference. In accordance with this result, we tried to find out whether the energy separation of the first order next to the image potential state changes with the tip-sample electric field.

Fig. 4.10 (a) STM topography of Cu(111) with 1ML Ag film in size of 102nm×43nm showing the discontinuous contrast in the image. (b) The average spectra acquired in A, B and C indicated in (a). (c) Energy separation as the function of the order for the spectra in (b).

4.2.2 Tip-Sample Field Influence in Energy Separation

Since the energy separation changes with tip condition accompanied with the variation of electric field, we tried to find out the relationship between the field and the energy separation of order next to the image potential state in the spectra. The tip condition may be varied by changing applied voltage. We ascertain whether the

(a)

C

A

B C

(b) (c)

various peak numbers. The peak numbers may change by tuning the fixed tunneling current. We had the dZ/dV-V spectrum on Ag/Cu(111) system as shown in Fig. 4.11.

The relationship between the energy separation and the tip-sample electric field was demonstrated by Kubby et al. According to the quantum mechanism, energy difference of the standing wave states in the triangular potential is proportional to F^2/3, where F is the electric field in the tip-sample gap^[4-11]. From the discussion in Sec. 4.1, it is known that the energy separations of higher order are intended to be constant and referred to the work function difference. As mentioned in Sec. 4.2.1, we know that the energy separation of the first order is approximate to the real work function difference.

Hence we took the peak next to the image potential states, n=1, as the calculated object.

Kubby’s formula is known as follow.

where n is the order of the standing wave states, m the electronic mass,

h

^the reduced Planck constant. En is the energy of the peak n, and F is the corresponding tip-sample electric field. In this formula order 0 is the first standing wave states above the vacuum level, the image potential state. Regardless the constant, F is directly proportional to E^2/3. Here we calculate the energy variation between the first two peaks above the image potential state, △E, to understand the strength of the tip-sample electric field. We replaced F with (^△E)^2/3 to know the tendency of the energy separation with varied F, as shown in Fig. 4.11.

When the amount of peaks in the spectrum increases, ^△E reduces as well as the tip-sample electric field. The constant energy separation and △E of Cu(111) are indicated in the right upper corner, and the interplay between the constant energy separation and △E is illustrated in Fig. 4.11.

2 1/ 3

3 1

2 / 3

(△E)^3/2

difference between the first two peaks, △E (Fig. 4.12). It can be seen that with the increasing electric field, the energy separation of the image potential states varies abruptly with the electric field. It is reasonable because the image potential is supposed to change with the triangular potential well whose shape is strongly dependent on the electric field, i.e., the tip-sample gap. That is why the energy separation of image potential states is inconsistent with changing electric field.

However, the energy separations of first order nearly keep constant as 0.293 eV when the electric field is changed. Through the discussion in Sec. 4.2.1, the energy separation of the first order can be referred to separation of the vacuum levels of the substrate and film, i.e. the work function difference.

Fig. 4.12 Energy separation of the order 0 and 1 as the function of (△E)^3/2 for Ag/Cu(111) system.

△E is the energy difference between the order 1 and 2 of Cu(111). The solid lines are the smooth fitting to both data points.

0.293 eV

means as long as the work function of the substrate is known, that of the film can be known by the constant energy separation even the tip condition is uncertain as well as the tip-sample field. We certainly provide an efficient way to estimate the work function in the thin film system.

4.3 Surface Profiling with Variable Gundlach Oscillation

Since the slight difference in the array of topmost atoms may lead to the variation of electronic structures, it is possible to find out the connection by means of Gundlach oscillation measurement. Besides the estimation of work function, Gundlach oscillation can also reveal more information about the electronic structure of the surface such as the transmissivity and reflectivity of electrons. Furthermore, since the tip moves away from the sample surface during the Z-V measurement, we wonder whether there is any difference in the topography when the gap width between the tip and the sample is broadening.

4.3.1 Dependence of Gundlach Oscillation on Distinct Surface Structures

As mentioned in section 2.2, Moiŕe pattern exhibits on Ag/Cu(111) system. we have observed the surface of the superstructure 9×9Ag/Cu(111) by STS of dZ/dV-V spectra. The Moiŕe pattern of Ag film is divided into two regions: the hollow and the protrusive regions, indicated in Fig. 4.13(a). The average spectra taken at the two regions are shown in Fig. 4.13 (b). In this spectrum, there are five standing wave states and are all taken into our discussion. Local variation of the intensity difference is observed in these five peaks. Although the energy levels of the valley and the apexes locate at the same position, the intensities at the valley of the protrusive regions are

Fig. 4.13 (a) STM topography image of a Ag/Cu(111) surface. The image size is 30nm×32nm (b) The average dZ/dV-V spectra of the hollow and protrusive regions on Ag/Cu(111) surface.

A similar phenomenon has been discussed by W. B. Su et al. who observed the three structures on a reconstructed Au(111), as shown in Fig. 4.14. There exhibit higher apexes and lower valleys in the spectra of FCC and HCP regions. W. B. Su et al.

have demonstrated that the transmission background of the regions with weaker intensity (ridge regions on Au(111) surface) is higher than that with stronger intensity (FCC and HCP regions on Au(111) surface)^[4-12]. This means that tunneling current is tended to reflect in the FCC and HCP regions so that leads to a stronger resonance in the standing wave states. W. B. Su et al. have also shown that on a reconstructed Au(111) surface, the peak with higher transmission background has lower intensity at the apex, and vise versa. The phenomenon has also been observed by McMahon et al.^[4-13] and is really fitted to what we obtained in our system. It is believed that the intensity in the spectra is conserved so that a complementary phenomenon occurs. This result indicates that such complementation of intensity should be a general phenomenon.

protrusive

hollow

Fig. 4.14 (a) STM topography image of a reconstructed Au(111) surface (286Å×286Å) (b) The average dZ/dV-V spectra of FCC, HCP and ridge regions.

With this conclusion, the transmissivity of the protrusive regions on Ag film is supposed to be higher than that of the hollow regions. It implies that the transmissivity changes with different orientation of the reconstructed superstructure. Since at the energy of standing wave states, there are differences in intensity in distinct regions. We can have differentiated STM images to understand the contrast variation during the voltage ramping, i.e. Z-V measurement. Not only the surface topography, but also the comparative contrast can be known.

4.3.2 Surface Mapping with Increasing Gap Width

During the Z-V measurement, the tip-sample gap increases with the applied tip bias. Generally the broader the gap width is, the fuzzier the image and worse the resolution become. Therefore it inspires us to find out how the image will become when the displacement is far away from the surface. The displacement can be expected by the formula dV=FdZ, where V is the applied tip bias, F is the electric field, and Z is

FCC ridge HCP

should be as small as possible to obtain the gap width as broad as we could.

Figure 4.15(a) shows the STM topography image of a reconstructed Au(111) surface. The average dZ/dV-V spectrum in Fig. 4.15(b) is measured from Fig. 4.15(a), and implies the small electric field in the tip-sample gap. There are twelve peaks of Gundlach oscillation, but only the last ten were marked and discussed in our system.

Figure 4.15(c) indicated the corresponding bias at each apex and valley, and in Fig.

4.15(d) the corresponding displacement of the tip. Generally the gap width is within 10 Å during normal scanning. However, as we can see in Fig. 4.15(b) and (d), the farthest distance between the tip and the sample should be nearly 60Å. Since the tip can move from the sample surface with such a long distance, it is interesting to understand the STM image performance during the tip movement.

The STM image in Fig. 4.15(a) was differentiated at the biases where the apexes and valleys locate, and shown in Fig. 4.16 (a) and (b) respectively. It is apparent in Fig.

4.16(a) that the FCC regions are slightly brighter than the HCP regions, and the ridges are the darkest among the three structures. The high transmissivity implies low reflectivity as well as stronger surface resonance and brighter performance in the images. It reveals that the ridges have the highest transmissivity on reconstructed Au(111) surface, and FCC region has slightly higher transmissivity than HCP region.

This result is corresponding to the discussion in Sec. 4.3.1. The mappings at the energy of valleys in Fig. 4.16(b) have the same contrast as normal STM images, and Fig.

4.16(a) vice versa. It proves the assumption of the intensity complementary demonstrated by W. B. Su^[4-12].

Fig. 4.15 (a) STM topography image of a reconstructed Au(111) surface. The image size is 24nm×24nm. (b) dZ/dV-V spectrum acquired on reconstructed Au(111) surface. The first two states are believed as image potential states. (c) Corresponding bias at each apex and valley. (d) Corresponding displacement of the tip at each apex and valley.

Fig. 4.16 (a) Mappings at the apexes of each peak whose biases and displacement are indicated in Fig. 4.12 (c) and (d). The sizes of these mappings are 24nm×24nm. The contrast in these mappings is opposite to that in STM topography image. FCC regions are brighter than the other two structures.

a a ( (1 1. .5 55 5n nm m) ) b b ( (2 2. .2 20 0n nm m) )

f f ( (4 4. .2 27 7n nm m) )

c c ( (2 2. .7 79 9n nm m) ) d d ( (3 3. .2 28 8n nm m) ) e e ( (3 3. .8 82 2n nm m) )

h h ( (5 5. .1 15 5n nm m) ) g g ( (4 4. .7 70 0n nm m) )

j

j ( (5 5. .9 92 2n nm m) ) i

i ( (5 5. .5 57 7n nm m) )

(a)

Fig. 4.16 (b) Mappings at the valleys between all peaks, whose biases and displacement are indicated in Fig. 4.12(b) and (c). The sizes of these mappings are 24nm×24nm. The contrast in these mappings is same as that in STM topography image. The ridge regions are the brightest in the mappings.

f’ f ’ ( (4 4. .4 49 9n nm m) ) h’ h ’ ( (5 5. .3 34 4n nm m) ) e’ e ’ ( (4 4. .1 14 4n nm m) ) a’ a ’ ( (1 1. .8 85 5n nm m) ) b b’ ’ ( (2 2. .4 48 8n nm m) )

d’ d ’ ( (3 3. .5 50 0n nm m) ) c’ c ’ ( (3 3. .0 05 5n nm m) )

i’ i ’ ( (5 5. .7 75 5n nm m) )

g’ g ’ ( (4 4. .9 93 3n nm m) )

(b)

To understand the variation of resolution with increasing tip-sample gap, we measured the Full-Width of Highest Maximum (FWHM) for HCP regions, which are the narrowest among the three local structures in Fig. 4.16. Figure 4.16(a) and (b) show not only the mapping at various biases but also the images obtained with increasing gap width. The dependence of the resolution on the gap width is illustrated in Fig. 4.17. We found that the width of HCP regions varies little with the increasing tip-sample separation, and is averagely 0.1145nm in width. This result stands for a significant meaning that the minimum width we can observe maintains while the tip is moving away.

Fig. 4.17 Dependence of the possible resolution on the increasing tip-sample gap width.

1 2 3 4 5 6

0.06 0.08 0.10 0.12 0.14

0.16 ave. value= 0.1145 nm

width of HCP (nm)

displacement (nm)

CONCLUSION

By STM and STS operation, we can observe the Gundlach Oscillation on metal substrate and ultrathin film by Z-V measurement. Energy separations exhibit in the dZ/dV-V spectra at high energy. Whatever in uniform ultrathin (Ag/Cu(111)) and various (Co/Cu(111)) thin film system, constant energy separation at high voltage can be obviously seen. The constant energy separation in each thin film system is considered as the work function difference between the film and the substrate. This assumption is pretty close to the theoretical calculation and be proved by our model

在文檔中利用掃描穿隧顯微鏡觀察昆拉赫震盪及其應用 (頁 53-0)