Structural properties of Ni/Co/Pt(111) and

Chapter 5

Structural properties of Ni/Co/Pt(111) and

Co/Ni/Pt(111)

5.1 Growth modes of d^{N i} Ni/ 1 ML Co/Pt(111) and d^Co Co/1 ML Ni/Pt(111)

The growth modes of Co films and Ni films have been studied to be at least 3 ML in a layer-by-layer growth mode on Pt(111) surface [62, 31, 5]. Ni ultrathin films were then deposited onto the 1 ML Co/Pt(111) surface, alternatively, Co ultrathin films were then deposited onto the 1 ML Ni/Pt(111) surface. The deposition rate of Ni films and Co films were calibrated by Auger uptake curve and the oscillation in intensity of LEED specular beam during deposition. Because of the Auger sensitivity and the mean free path of Auger electrons, the Co 53 eV, Ni 848 eV and Pt 237 eV were chosen for the measurements of the Auger uptake curves. During the measurements of LEED (0,0) intensity, the kinetic energy of the incident electron beam was 60 eV. At this energy level, LEED should be very surface sensitive. The phase in our experimental conditions is ϕ(0, 0) = h K_⊥ = 2hKcosθ ≈ 5π for 60 eV electron energy and incident angle θ = 5^◦, where K_⊥ is the momentum transferred perpendicular to the surface, and h_{N i} = 0.203 nm and h_Co = 2.04 nm are the step height of the adsorbed Ni adatom and Co adatom, respectively. It is out-of-phase condition of diffraction (h K_⊥ = nπ, n = 1, 3, 5...) which is more sensitive to surface step density on the (0,0) beam [39]. The

49

Figure 5.1: The growth curves of Auger signals and LEED (0,0) beam intensity were used for observing the growth of (a) d_{N i} Ni/1 ML Co/Pt(111) and (b) d_Co Co/1 ML Ni/Pt(111). In graph (a), both the kinks on the straight lines of AES uptake curves and the local maximums of LEED(0,0) bean intensity are at 8 minutes and 16 minutes. They are at 22 minutes and 44 minutes in graph (b).

5.1. Growth modes of d_{N i} Ni/ 1 ML Co/Pt(111) and d_Co Co/1 ML Ni/Pt(111) 51

Auger uptake curves and LEED (0,0) beam intensities of d_{N i} Ni/1 ML Co/Pt(111) and d_CoCo/1 ML Ni/Pt(111) are shown in Fig. 5.1(a) and (b), respectively. In Fig. 5.1(a), the first and second kinks of the straight line of Ni 848 eV Auger uptake curve is correspond to the first and second complete layer of Ni at deposition time 8 and 16 minutes, respectively.

The normalized AES signals I(t)/I(0) of Co 53 eV and Pt 237 eV versus deposition time t are also shown in Fig. 5.1(a). I(t) is the AES intensity at t minutes. It is easy to see that both growth curves are linear in the time intervals from 0 to 8 minutes and 8 to 16 minutes. These linear relations of Auger uptake curves in Fig. 5.1(a) can be associated to the growth mode of at least 2-ML in layer-by-layer growth.[31] The evolution of the intensity of LEED specular beam during deposition also reveals two peaks in the oscillation curve at 8 and 16 minutes, same result as that of AES in Fig. 5.1(a). The LEED intensity diminishes to the background after the growth of two atomic layers. Thus, we conclude that the growth mode of Ni on 1 ML Co/Pt(111) is at least 2 ML in layer-by-layer growth before turning into 3-D island growth, i.e. it follows the Stranski-Krastanov growth mode.

It is interesting that the growth mode of Ni/1 ML Co/Pt(111) is identical to Ni/Pt(111) where the growth mode is a layer-by-layer, at least for first two monolayers [91].

In Fig. 5.1(b), the first and second kinks of the straight line of Co 53 eV Auger uptake curve are correspond to the first and second complete layer of Co at deposition time 22 minutes and 44 minutes, respectively. The normalized AES signals I(t)/I(0) of Ni 848 eV and Pt 237 eV versus deposition time t are also shown in Fig. 5.1(b). It is easy to see that both growth curves are linear in the time intervals from 0 to 22 minutes and 22 to 44 minutes. These linear relations of Auger uptake curves in Fig. 5.1(b) can be associated to the growth mode of at least 2-ML in layer-by-layer growth [31]. The evolution of the intensity of LEED specular beam during deposition also reveals two peaks in the oscillation curve at 22 minutes and 44 minutes, same result as that of AES in Fig. 5.1(b).

Thus, we conclude that the growth mode of Co on 1 ML Ni/Pt(111) is at least 2 ML in layer-by-layer growth before turning into 3-D island growth. It is interesting that the growth mode of d_{N i} Ni/1 ML Co/Pt(111) is identical to d_{N i}Ni/Pt(111) [5], alternatively, the growth mode of d_CoCo/1 ML Ni/Pt(111) is identical to d_CoCo/Pt(111) [62, 31]. The growth modes of both d_{N i}Ni/1 ML Co/Pt(111) and d_Co Co/1 ML Ni/Pt(111) follow the Stranski-Krastanov growth mode where the growth mode is a layer-by-layer, at least for first two monolayers.

Two factors are important for the growth mode of a thin film on a substrate: lattice

mismatch and surface free energy. The lattice mismatch between Ni and Co is only 0.7%.

From this point of view, both the two systems favor layer-by-layer growth. The surface free energy of Co and Ni are γ_Co = 2.709 J/m², γ_{N i} = 2.364 J/m² [41]. For the d_{N i} Ni/1 ML Co/Pt(111) system, the difference ∆γ = γ_{N i}+ γ_i− γCo is negative (usually the interface energy γ_i is very small and can be neglected). This indicates that the d_{N i} Ni/1 ML Co/Pt(111) system favors layer-by-layer growth. Thus, the initial growth is layer-by- layer of ultrathin Ni films on 1 ML Co/Pt(111) is reasonable. Alternatively, the difference

∆γ = γ_Co+ γ_i− γN i of ultrathin Co films on 1 ML Ni/Pt(111) system is positive, that means the system favors 3-D island growth. Due to the competition of surface free energy and lattice mismatch, the ultrathin Co films turn to 3-D island growth after growing 2 ML Co on Pt(111) with 1 ML Ni as buffer layer. These growth curves were also used to calibrate the thickness of Co films and Ni films in the experiments.

Figure 5.2: The plots of Auger intensities versus sample temperature of (a) 1 ML Ni/1 ML Co/Pt(111) and (b) 1 ML Co/1 ML Ni/Pt(111). Each data was detected after annealing at its respondent sample temperature for 20 minutes. The significant changes of both the two systems are at 420 K and 580 K.

5.2. AES studies in alloy formation 53

5.2 AES studies in alloy formation

Auger electron spectroscopy is a useful technique to study the diffusing dynamic during the heating process. We use AES to investigate the changes of the surface compositions during the annealing processes of 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111).

The heating rate was 15^◦C per minute during the heating processes. Each temperature of measuring AES data was held for 20 minutes in the annealing. We have carefully measured; the annealing for 20 minutes is enough for the sample to reach the thermal equilibrium. Owing to the diffusion signal comes from the top surface region; the surface sensitivity must be taking into consideration when choosing the remarkable Auger signals.

The inelastic mean free paths of Co 53 eV and Ni 102 eV Auger signals are 3.9 A and 4.6 A, respectively. These Auger signals are more surface-sensitive than 11.4 A of Co 656 eV and 13.2 A of Ni 848 eV signals. Fig. 5.2(a) and (b) shows the evolution of Ni 848 eV, Ni 102 eV, Co 656 eV, Co 53 eV, and Pt 237 eV AES intensities versus sample temperature (T_s) for 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111), respectively. In 300-420 K, all signals of Fig. 5.2(a) and (b) do not have significant changes until the temperature is higher than 420 K. For the 1 ML Ni/1 ML Co/Pt(111) system, the Co 53 eV signal increases while Ni 102 eV signal decreases when 420 K < T_s < 480 K, but Pt 237 eV signal does not vary within this temperature region. These results indicate that the mixing of Ni and Co layers occurs in the upper interface, without diffusing into the bulk of Pt. As the temperature increases, the Co and Ni signals reach equilibrium when 480 K < T_s < 580 K. The Pt signal begins to increases significantly when T_s > 580 K.

Beyond 580 K, Co 53 eV and Ni 102 eV signals drop rapidly. These drops correspond to the beginning of the alloying of the Ni-Co alloy with Pt. Owing to the sensitivity of Auger electron; we did not observe any significant changes for the Co 656 eV and Ni 848 eV Auger signals between 420 K and 580 K. As the temperature exceeds 650 K, the drops of Ni 848 eV and Co 656 eV signals indicate that the bulk diffusions for Co and Ni occur because high-energy Auger electrons are coming from the deep layers. According to the recent studies about the alloy formation of Co-Pt [31] and Ni-Pt [5] interfaces, these drops can be confirmed as the alloy formation of Co-Pt and Ni-Pt. Of course the formation of Ni-Co-Pt alloy is possible near the interface of the overlayers and the substrate. For further annealing to T_s > 800 K, the Co 53 eV and Ni 102 eV signals drop to zero and the Pt 237 eV almost reaches a saturated value. This indicates that the Pt atoms occupied

almost all surface sites at T_s > 800 K.

For the 1 ML Co/1 ML Ni/Pt(111) system in Fig. 5.2(b), the Ni 102 eV signal increases while Co 53 eV signal decreases when 420 K < T_s < 480 K, but Pt 237 eV signal does not vary within this temperature region. These results indicate that the mixing of Ni and Co layers occurs in the upper interface, without diffusing into the bulk of Pt. The Pt signal begins to increases significantly when T_s > 580 K. Beyond 580 K, Co 53 eV and Ni 102 eV signals drop rapidly. These drops correspond to the beginning of the alloying of the Ni-Co alloy with Pt. As the temperature exceeds 650 K, the drops of Ni 848 eV and Co 656 eV signals indicate that the bulk diffusions for Co and Ni occur.

For further annealing to T_s > 800 K, the Co 53 eV and Ni 102 eV signals drop to zero and the Pt 237 eV almost reaches a saturated value. This indicates that the Pt atoms occupied almost all surface sites at T_s > 800 K. The diffusion processes of both 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) systems can be divided into two main parts: first is the Ni and Co atoms intermixed at 420 K-580 K; second is the Ni-Co intermixing layer diffusing into the Pt substrate when the sample temperature is higher than 580 K.

Fig. 5.3(a) and (b) shows the temperature dependence of Ni 102 eV, Co 53 eV and Pt 237 eV Auger signals with x ML of Ni on 1 ML Co/Pt(111) and x ML of Co on 1 ML Ni/Pt(111), respectively, where x = 1, 2, and 3. In Fig. 5.3(a), except for the slight increase in the Ni 102 eV Auger signal for the coverage of 3 ML, no Ni 102 eV Auger signal changes significantly at T_s < 420 K. The starting temperature of the intermixing (T_mix) of Ni-Co is 420 K for d_{N i} Ni/1 ML Co/Pt(111) system, and is independent of Ni thickness. This result is consistent with the Co 53 eV Auger signal in Fig. 5.3(a). Since the sensitivity of the energy analyzer of Auger electrons for Co 53 eV is about seven times that of Ni 102 eV Auger electrons, the change in intensity of the Co Auger signal is much more remarkable than that of Ni during the bulk diffusion. The slight increase of Ni 102 eV Auger signal in the range 300 K < T_s< 420 K for 3 ML Ni thin film may be due to the smoothing effect of the annealing process. Only 2 ML of Ni coverage is in the layer-by- layer growth, while the third atomic layer changes to island growth. Therefore, the island surface becomes smoother after the temperature increases. However, T_bulk, the starting temperature of Ni-Co intermixing layer diffusing into the Pt substrate, is dependent on Ni thickness. The more the Ni coverage encompasses, the higher the starting temperatures of Co-Pt, Ni-Pt and Ni-Co-Pt become. The temperature T_bulk increases from 580 K to 625

5.2. AES studies in alloy formation 55

Figure 5.3: Evolutions of the Auger intensities of Co 53 eV, Ni 102 eV and Pt 237 eV as functions of sample temperature for different overlayer coverage. (a) d_{N i} Ni/1 ML Co/Pt(111); (b) d_Co Co/1 ML Ni/Pt(111).

K when the coverage of Ni increases from 1 ML to 3 ML (T_bulk = 580 K, 605 K and 625 K for 1 ML, 2 ML, and 3 ML of Ni, respectively). These values of T_bulk of Ni 102 eV, Co 53 eV and Pt 237 eV are the same in Fig. 5.3(a), indicating that Ni and Co diffuse into Pt bulk together. This interesting result may be due to the attractive interactions between Co and Ni.

In Fig. 5.3(b), the Co 53 eV Auger intensities starts to decrease when the sample temperatures are 420 K, 495 K, and 600 K with respect to the Co coverage equals to 1 ML, 2 ML, and 3 ML. Each Ni 102 eV Auger intensity curve in Fig. 5.3(b) has two

Figure 5.4: (a) The evolution of intermixing temperatures of d_Co Co/1 ML Ni/Pt(111) and d_{N i} Ni/1 ML Co/Pt(111) versus Co overlayer coverage and Ni overlayer coverage, respectively. (b) The evolutions of T_bulk of d_Co Co/1 ML Ni/Pt(111) and d_{N i} Ni/1 ML Co/Pt(111) systems versus Co overlayer coverage and Ni overlayer coverage, respectively.

Total coverage is the total coverage including overlayer and 1 ML buffer layer. (c) The evolution of T_bulk of d_Co Co/Pt(111)[31] and d_{N i} Ni/Pt(111) [5] versus Co coverage and Ni coverage, respectively.

5.2. AES studies in alloy formation 57

significant temperatures, one is starting temperature of the increase of Ni102 eV Auger signal and the other is the starting temperature of the decrease of Ni102 eV Auger signal.

The starting temperatures of the increase are 420 K, 490 K, and 600 K with respect to the Co coverage equals to 1 ML, 2 ML, and 3 ML. Alternately, the starting temperatures of the decrease for different Co coverage (1 ML, 2 ML, and 3 ML) are 580 K, 610 K, and 700 K, respectively. The three Pt 237 eV Auger intensity curves of Fig. 5.3 increased after annealing at high sample temperatures. The starting temperatures of Pt 237 eV Auger intensity increase for different Co coverage (1 ML, 2 ML, and 3 ML) are 580 K, 610 K, and 700 K, respectively. By comparing the evolution of Co 53 eV, Ni 102 eV, and Pt 237 eV Auger intensities versus sample temperature, we can conclude that the diffusion process of d_Co Co/1 ML Ni/Pt(111) is similar to that of d_{N i} Ni/1 ML Co/Pt(111). T_mix are 420 K, 490 K, and 600 K with respect to 1 ML, 2 ML, and 3 ML Co coverage. T_bulk are 580 K, 610 K, and 700 K with respect to 1 ML, 2 ML, and 3 ML Co coverage. The more the Co atoms deposited on 1 ML Ni/Pt(111), the higher the starting temperatures of the alloy formation become.

Figure 5.4(a) shows the overlayer coverage dependences of intermixing temperatures of d_Co Co/1 ML Ni/Pt(111) and d_{N i} Ni/1 ML Co/Pt(111) systems. T_mix keeps at 420 K and does not depend on the Ni overlayer coverage for the d_{N i} Ni/1 ML Co/Pt(111) system, while it is increases almost linearly with the Co overlayer coverage for d_Co Co/1 ML Ni/Pt(111) system. Fig. 5.4(b) shows the starting temperature of Ni-Co intermixing layer diffusing into the Pt substrate (T_bulk) with respect to different overlayer coverage and total coverage (the total coverage including overlayer and 1 ML buffer layer). T_bulk of d_CoCo/1 ML Ni/Pt(111) system increases rapidly with the Co coverage, while it increases slowly with Ni coverage in d_{N i} Ni/1 ML Co/Pt(111) system. Comparing Fig. 5.4(a) and (b), the attractive force between Co layer seems to be higher than that between Ni layer.

Hence, the increasing rates of both T_mix and T_bulk of d_Co Co/1 ML Ni/Pt(111) system are higher than those of d_{N i} Ni/1 ML Co/Pt(111).

The temperatures of alloy formations of d_Co Co/Pt(111) [31] and d_{N i} Ni/Pt(111) [5]

were plotted in 5.4(c) for confirming the deduction further. T_bulk of d_Co Co/Pt(111) is about 350 K when d_Co < 1 ML, after which it increases almost linearly from 350 K to 750 K when 1 ML ≤ dCo ≤ 4 ML. The Tbulk of d_{N i} Ni/Pt(111) increases slowly from 450 K to 500 K when d_{N i} increases from 0.8 ML to 2 ML. The T_bulk of d_{N i} Ni/Pt(111) increases gradually from 500 K to 600 K when the Ni coverage increases from 2 ML to

2.3 ML. This jump of T_bulk is corresponding to the 3-D island growth of Ni films on Pt(111) [5]. Not like the behavior of d_{N i}Ni/Pt(111), the T_bulk of d_Co Co/Pt(111) system depends on the Co coverage rather than on the growth mode. From the above discussion, one can conclude that the attractive force between the Co layers is higher than that of Ni layers, and the change of T_bulk of d_{N i} Ni/Pt(111) is sensitive to the growth mode.

The surface free energies of Ni, Co and Pt are γ_{N i} = 2.364 J/m², γ_Co = 2.709 J/m², and γ_{P t} = 2.691 J/m² [41]. Ni has minimum surface free energy, Thus, the Ni film on the topmost surface is relatively stable, and T_bulk of 1 ML Ni/Pt(111) is about 100 K higher than that of 1 ML Co/Pt(111) is reasonable. Comparing Fig. 5.4(a) and (b)with Fig. 5.4(c), one can conclude that the higher attractive force between Co layers is the main reason that causes the T_mixand T_bulk of d_Co Co/1 ML Ni/Pt(111) to be higher than those of d_{N i} Ni/1 ML Co/Pt(111). Besides, T_bulk of both Co-Pt for 2 ML Co/Pt(111) [31] and Ni-Pt for 2 ML Ni/Pt(111) [5] are 500 K. It is interesting that the temperatures of the alloy formation of Ni-Pt and Co-Pt for 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) are 80 K higher than that of 2 ML Co/Pt(111) and 2 ML Ni/Pt(111). By comparing the four systems, we can conclude that the strength of Ni-Co bond is stronger than that of the Ni-Ni bond and the Co-Co bond.

5.3 LEED studies in alloy formation

The evolution of LEED patterns of 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) were shown in Fig. 5.5(a)-(d) and Fig. 5.5(e)-(h), respectively. Each pattern was recorded at room temperature after annealing at the high temperature for 20 minutes.

In spite of graph (d) and graph (h), the kinetic energy of the incident electron beam is 65 eV. It is 75 eV for graph (d) and (h). The schematic diagram of diffraction pattern for each temperature range is also given in the middle column of Fig. 5.5. When the annealing temperature in between 300 K and 580 K, the LEED patterns (a) and (e) show a six-fold fine structure developed surrounding each integer spot of the structure.

When N atoms (N = N_Co+ N_{N i} ) arrange on (N − 1) atoms of Pt, the surface is not flat but becomes corrugated as shown in Fig. 5.6(a). This incommensurate structure causes some fine satellites of LEED pattern. The satellites of LEED pattern indicate an incoherent epitaxy of Ni and Co overlayers on the Pt(111) surface. The top view of this incommensurate structure in real space is shown in Fig. 5.6(b). The lattice mismatch is

5.3. LEED studies in alloy formation 59

Figure 5.5: The LEED patterns of 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111). Each pattern was detected at room temperature after anneal for 20 minutes.

In spite of graph (d) and graph (h), the kinetic energy of the incident electron beam is 65 eV. It is 75 eV for graph (d) and (h). The schematic diagram of the LEED pattern for each annealing temperature range is also plotted. The circles around the main spots in the second scheme (600 K-630 K) mean that the main spots are blurred.

Figure 5.6: When N atoms (N = N_Co+ N_{N i}) arrange on (N−1) atoms of Pt, the surface is not flat but becomes corrugated. Graph(a) is the side view of this corrugated surface;

graph (b) is the top view of so called Mori´e structure [92].

very small (about 0.6%) between Co and Ni. This is why the same LEED pattern can be observed between these mirror systems. The lattice mismatch between Co and Pt is about 9.5%; while the lattice mismatch between Ni and Pt is about 10.2% [93, 94].

The six-fold satellites become blurred when the temperature is greater than 580 K as shown in Fig. 5.5(b) and (f). When the annealing temperature is between 650 K and 750 K, the LEED patterns have three-fold satellites surrounding each (1× 1) first order spot as shown in Fig. 5.5 (c) and (g). From the result of Fig 5.3, Co and Ni have been diffusing into Pt substrate at this temperature. This indicates that the surface structure changes from the Moir´e structure to a major fcc-like phase, which is coherent with the Pt substrate. The three-fold satellites of LEED diffracted spots become fuzzy and gradually disappear after annealing at temperatures higher than 750 K. After subsequent annealing at higher than 800 K, the three-fold symmetry (1× 1) main spots becomes sharp and bright. It is same as the LEED pattern of the well prepared Pt(111) surface. From the

5.3. LEED studies in alloy formation 61

result of Fig. 5.3, almost all Co and Ni atoms have diffused into Pt bulk after annealing at temperatures higher than 800 K. This is why only (1× 1) LEED pattern was observed at 800 K < T_a< 850 K.

Fig. 5.7 shows the evolution of LEED patterns versus annealing temperature T_a of 2 ML Co/1 ML Ni/Pt(111). Unlike the LEED pattern of 1 ML Co/1 ML Ni/Pt(111), no Mori´e structure was observed unless the annealing temperature is between 450 K and 550 K. A non-sharp six-fold satellites surrounding the sharp main spots pattern appeared as T_a exceeding 450 K. Although the Co films grow in a 2 ML layer-by-layer mode, the growth of the second deposited Co layer is not as smooth as the first deposited one. This results in a reductive tendency of the periodic arrangement of the Co atoms. Since the sharp Mori´e pattern is due to an incommensurate structure of a high periodical corrugated surface, the appearance of the non-sharp Mori´e pattern can be associated to a thermal

Figure 5.7: LEED patterns were measured at RT for 2 ML Co/1 ML Ni/Pt(111) after annealing at (a) 300 K, (b) 450 K, and (c) 700 K for about 20 minutes. The circles around the main spots in the third scheme (650 K-850 K) mean that the main spots are blurred.

Figure 5.8: The LEED I-V curve and E− n² plot of 1 ML Ni/1 ML Co/Pt(111) at room temperature.

smoothing effect after low temperature annealing. In Fig. 5.3(b), the Co 53 eV AES signal increased by a thermal smoothing effect before intermixing with Ni. Thus, the appearance of the non-sharp Mori´e pattern resulted from the smoothing effect is reasonable.

It is important to understand the structures of 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) after annealing at high temperature. From our recent study, the changes in magnetic properties is sensitive to the subsurface structure. The magnetic properties will be discussed latter. The interlayer distance d of subsurface region can be determined from the I−V profile of the LEED specular beam. From the Bragg diffraction condition 2dcosθ = nλ = nh/√

2m_eE, one can easily obtain the useful formula [95, 96]:

E = n²( h²

8m_ed²cos²θ) (5.1)

where θ is the angle between the directions of incident electron beam and surface normal, E is the kinetic energy of the incident electrons, m_e is the electron mass, and h is the Planck constant. The interlayer distance d can be calculated from the slope of the E− n² plot. Fig. 5.8 shows the results of the I− V profile and E − n² plot of 1 ML Ni/1 ML Co/Pt(111) at room temperature. The angle θ = 5.0^◦ was used in our experiment. The relation between E and n²is perfectly linear, and therefore the interlayer distance can be determined.

The interlayer distance d versus the annealing temperature of 1 ML Ni/1 ML Co/Pt(111) is shown in Fig. 5.9. Each I− V curve was taken at room temperature after annealing at each T_a for 20 minutes. In the range 300 K < T_a < 600 K, the interlayer distance

5.3. LEED studies in alloy formation 63

Figure 5.9: The interlayer distance of 1 ML Ni/1 ML Co/Pt(111) as a function of annealing temperature.

does not change. It is equal to 2.035 ˚A. Note that the interlayer distances of fcc(111) are d_{N i}= 2.03 ˚A, d_Co = 2.04 ˚A , and d_{P t} = 2.27 ˚A . The interlayer distance d increases rapidly when 600 K < T_a < 800 K. After T_a exceeds 800 K, it approaches the value of Pt(111) substrate. By comparing Fig. 5.9 with Fig. 5.2, we can find that the change of interlayer distance is consistent with the evolution of surface composition. The surface composition does not change within 300 K < T_a < 420 K, and thus, the interlayer dis- tance does not change in this temperature range. Fig. 5.2 shows that the intermixing process of Ni and Co occurs between 420 K and 580 K. Owing to the very small difference between the interlayer distances of Ni and Co, the constant interlayer distance is reason- able when 420 K < T_a< 580 K. The intermixing layers of Ni and Co start diffusing into the substrate when T_a is greater than 580 K, and so, the interlayer distance increases significantly. After further annealing (T_a > 800 K), Pt atoms occupy almost all sites of the surface. Thus, the interlayer distance approaches 2.27 ˚A, which is the value of Pt (111) interlayer.

As you know, the LEED pattern have three-fold satellites surrounding each (1× 1) first order spot between 650 K and 750 K. R Baudoing-Savois et al have observed a similar structure in Co/Pt(111) after high temperature annealing.[97] They indicated that this structure corresponds to Pt₅₅Co₄₅alloy. In our system, it may correspond to the structure of an ordered alloy of Ni_xCo_yPt_[1−(x+y)]. Once the interlayer distance d of the ordered alloy of Ni_xCo_yPt_[1−(x+y)] is determined, the concentrations of the constituent elements

can be calculated according to the Vegard’s law,[98]

d = x· dN i+ y· dCo+ [1− (x + y)] · dP t (5.2)

where x and y are the concentrations of Ni and Co, respectively, and d_{N i}, d_Co, and d_{P t}, are the bulk interlayer distances of Ni, Co, and Pt, respectively. The range of Pt concentrations can be estimated from equation 5.2 after getting the value of d = 2.16 ˚A from Fig. 5.9. At T_a= 700 K, the concentrations of Pt are from 52% to 54% with respect to Co-rich (48% of Co, 0% of Ni) and Ni-rich (46% of Ni, 0% of Co) phases. This finding is close to the R Baudoing-Savois’ result of Pt55Co45alloy[97]. Hence, we conjecture that the phase transitions of Fig. 5.5(c) to (d) and from Fig. 5.5(g) to (h) are related to the transition of Ni1−xCo_xPt surface alloy to Ni1−xCo_xPt3 surface alloy.

5.4 Subsurface concentration of Ni-Co-Pt alloy

Assuming the Ni-Co-Pt surface alloy is homogenous in the subsurface region, one can calculate the concentration of each constituent from the Pt 237-eV, Ni 848-eV, and Co 656-eV AES data by equation 5.3.[99]

C_x=

Ix

Sxdx

 _Iα

Sαdα

(5.3)

C_x, I_x, S_x and d_x are the concentration, Auger intensity, relative sensitivity, and scale factor of the element x, respectively. The calculated values of Pt concentration for 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) are shown in Fig. 5.10(a).

Each data point was measured at room temperature after annealing. When the annealing temperature is between 690 K and 750 K, the Pt concentrations of both samples are almost following the same curve. They has significant difference when the annealing temperature is between 750 K and 795 K. According to the LEED studies in sec. 5.3, this temperature range is corresponding to a transition of surface structure. Since the calculated value of Pt concentration is affected by the contribution of Pt substrate, the values as shown in Fig. 5.10(a) might higher than those that they really are. By comparing the calculated value at 700 K (Pt% ∼ 67%) with that we obtained by LEED I − V curve (Pt%∼ 52-54%), we estimate that those calculated values of Pt concentration are approximately 1.25 times the real values of Pt concentration. Thus we modify the values and re-plot them in Fig. 5.10(b). The Pt concentration is arranged from 58% to 66%

5.4. Subsurface concentration of Ni-Co-Pt alloy 65

Figure 5.10: Frame (a) is the Pt concentration as a function of annealing temperature for 1 ML Ni/1 Ml Co/Pt(111) and 1 Ml Co/1 ML Ni/Pt(111). Each data point was measured at room temperature after annealing. Frame (b) is the result after modifying the Pt%.

when the annealing temperature is between 750 K and 795 K. From the phase diagrams of Co_1−xPt_x and Ni_1−xPt_x as shown in Fig. 4.5 [77] and Fig. 4.6 [78], this range of Pt concentration is corresponding to a structural transition of an ordered L10 phase (CoPt alloy and NiPt alloy) to an ordered L1₂ phase (CoPt₃ alloy and NiPt₃ alloy). Thus, we conclude that the LEED patterns which have three-fold satellites surrounding each (1×1) first order spot as shown in Fig. 5.5 (c) and (g) are corresponding to (Ni_1−xCo_x)Pt surface alloy. They become fuzzy during the structral transition when the annealing temperature is between 750 K and 795 K. The three-fold symmetry (1× 1) main spots becomes sharp

and bright as shown in Fig. (d) and (h) are related to the (Ni_1−xCo_x)Pt₃ surface alloy.

5.5 UPS studies in alloy formation

The change of density of state is very sensitive to the surface composition, alloy formation, and surface structure. We used UPS to investigate the evolution of this change during the heating processes. A He I radiation (hν = 21.2eV) was used in this approach, it is

Figure 5.11: UPS spectra of (a) 1 ML Ni/1 ML Co/Pt(111) and (b)1 ML Co/1 ML Ni/Pt(111). The evolutions of electron kinetic energy located at Fermi edge E^{F ermi}_K of (a) and (b) versus sample temperature are plotted in (c) and (d), respectively.

5.5. UPS studies in alloy formation 67

useful in studying the valence band near the Fermi level. Since the energy of an electron on a conductor’s surface is affected by the surface’s properties, the work function may change whenever the surface changes. Owing to the analyzer in our lab is less sensitive to the energy region of secondary electrons, the work function is hard for us to measure.

A directed evidence of the change of work function is the electron kinetic energy located at Fermi edge E_K^{F ermi} at the UPS spectra. Thus, we discuss the change of E_K^{F ermi} , this helps us to understand the change of work function during the alloy formation processes.

The UPS results near the Fermi edge of 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) are shown in Fig. 5.11(a) and (b), respectively. In Fig. 5.11(b), the electron kinetic energy located at Fermi edge E_K^{F ermi} moves to the left (lower electron kinetic energy) when 420 K < T_s< 580 K, after which it moves back to the right (higher electron kinetic energy). The movement of E_K^{F ermi} of 1 ML Ni/1 ML Co/Pt(111) as shown in Fig. 5.11(a) is not obvious when 420 K < T_s< 580 K. Except a small decrease within 660 K and 700 K, it has a significant change when the sample temperature is higher than 700 K. The evolutions of E_K^{F ermi} versus sample temperature for 1 ML Ni/1 ML Co/Pt(111) and 1 ML Ni/1 ML Co/Pt(111) are shown in Fig. 5.11(c) and (d). For the 1 ML Co/1 ML Ni/Pt(111) sample as shown in Fig. 5.11(d), E_K^{F ermi} is about 16.63 eV when sample temperature is between 300 K and 420 K. It decreases to 16.53 eV when 420 K≤ Ts≤ 580 K, after which it increases to a saturated value of 17.74 eV at 700 K. It is interesting that the temperature range of the decrease of E_K^{F ermi}is corresponding to the intermixing process of Ni and Co atoms. The Ni atoms may favor lower E_K^{F ermi} causes the decrease of E_K^{F ermi} when Ni-Co intermixing process begins as shown in Fig.5.11(d). The E_K^{F ermi} starts to increase at a sample temperature of 580 K which is corresponding to the temperature of Ni-Co intermixing layer diffuses into the Pt substrate. The Pt atoms may prefer higher E_K^{F ermi} and this causes the increase. For the 1 ML Ni/1 ML Co/Pt(111) sample as shown in Fig. 5.11(c), the change of E_K^{F ermi} is not obviously when the Ni and Co intermixing process occurred. Except a small decrease at 600 K < T_s < 700 K, the E_K^{F ermi}increases to a saturated value of 17.74 eV at 735 K. The Pt-rich surface dominates the location of E_K^{F ermi} after certain high temperature annealing, thus, the E^{F ermi}_K of both 1 ML Co/1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111) has the same value after certain high temperature annealing.

In order to discuss the density distribution versus binding energy, we move the all the Fermi edges in Fig. 5.11 (a) and (b) to the binding energy of 0 eV. The results of 1 ML

Figure 5.12: The UPS spectra of (a)1 ML Ni/1 ML Co/Pt(111) and (b) 1 ML Co/1 ML Ni/Pt(111) after moving E_K^{F ermi} to zero. Frames (c) and (d) are the evolutions of 0.45 eV peak intensity for 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111), respectively.

Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) are shown in Fig. 5.12(a) and (b) respectively. A peak with binding energy of 4.25 eV in Fig. 5.12(b) becomes sharper when the sample temperature is around 410 K which is very near the temperature of the Ni-Co

5.5. UPS studies in alloy formation 69

intermixing process. The Ni atoms diffuse to the sample surface may favor a sharper peak located at binding energy of 4.25 eV, this is the possible reason that causes the peak to grow. The same peak in Fig. 5.12(a) is sharper in the as-grown spectrum, and it becomes broaden when T_s > 410 K. This is because of the Ni concentration of the topmost layer decreases when the intermixing process of Ni and Co begins. It is interesting that the peaks at 4.25 eV of both Fig. 5.12(a) and (b) become broaden when T_s > 580 K which is the temperature of the Ni-Co intermixing layer diffusing into the Pt substrate. A new peak located at binding energy of 4.00 eV appears in both Fig. 5.12(a) and (b) when T_s> 735 K, that means the surface prefers a 4.00 eV peak in the Pt rich region.

Not only the evolution of 4.25 eV peak but also the evolution of 0.45 eV peak is related to the diffusion processes. The peak with a binging energy of 0.45 eV is the nearest peak with respect to the Fermi edge. The evolutions of 0.45 eV peak height of 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) are shown in Fig. 5.12 (c) and (d), respectively. The heights of both the samples increase with the sample temperature when T_s < 420 K. They do not increase obviously when 420 K≤ Ts ≤ 580 K, that means the intermixing process of Ni and Co atoms does not affect the intensity of 0.45 eV peak. The heights of both Fig. 5.12 (c) and (d) decrease when 580 K≤ Ts≤ 630 K. It is interesting that the LEED patterns are fuzzy in this temperature range as shown in Fig. 5.5(b) and (f), thus, the surface structures of the fuzzy LEED patterns do not favor a sharp Fermi edge. Both the heights increases when T_s> 650K, they reach their own maximum values after higher temperature annealing. The sample temperatures for the heights to reach their own saturated value are 700 K and 735 K for 1 ML Co/1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111), respectively. It is interesting that the sample temperatures for the E_K^{F ermi} to reach 17.74 eV are 700 K and 735 K for 1 ML Co/1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111), respectively. Since the LEED patterns of these two systems can not be distinguished, the temperature difference of 30 K may result from the difference in the surface composition of these two samples. From the study of the surface composition in sce. 5.2, we know that the Ni concentration of 1 ML Ni/1 ML Co/Pt(111) is higher than that of 1 ML Co/1 ML Ni/Pt(111) in the subsurface region. Thus, we believe that the delay temperature of about 35 K is corresponding to the surface composition. It is worth noticing that the distribution of electronic density of stste of Fig. 5.12(a) and (b) within 1.00 eV and 2.50 eV are very different when 660 K < T_s < 775 K. The peaks of Fig. 5.12(b) are more obvious than those of Fig. 5.12(a) in this energy range.

The differences in surface composition and in electronic density of state may cause some differences in the magnetic properties between 1 ML Co/1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111) systems. The differences in magnetic properties between these two systems will be discussed in Chapter 6.

The UPS result near the Fermi edge of 2 ML Co/1 ML Ni/Pt(111) is shown in Fig. 5.13(a). The electron kinetic energy located at Fermi edge E^{F ermi}_K moves to the right when T_s> 600 K. The evolutions of E_K^{F ermi} versus sample temperature is shown in Fig. 5.13(b). The E^{F ermi}_K is about 16.66 eV when sample temperature is between 300 K and 630 K. It increases rapidly to 16.74 eV when T_s> 630 K. Since the Ni concentration of the topmost layer is very small after the intermixing process of Ni and Co occurs, no change in E_K^{F ermi} can be observed is reasonable. The rapid increase of E_K^{F ermi} can be understood by the diffusion process of Ni-Co intermixing layer into the Pt substrate.

The UPS spectra of 2 ML Co/1 ML Ni/Pt(111) after moving E_K^{F ermi} to zero is shown in Fig. 5.13(c). The peak with binding energy of 4.25 eV in Fig. 5.13(c) moves to 4.00 eV when the sample temperature is higher than 450 K, it moves back to 4.25 eV and forwards to 4.60 eV when 515 K < T_s < 600 K which is corresponding to the tempera- ture of Ni-Co intermixing process. Because of the Ni concentration is very small on the topmost layer, the peak is not as sharp as that of 1 ML Co/1 ML Ni/Pt(111). The peak nearest to Fermi edge is located at 0.40 eV. The evolutions of 0.40 eV peak height of 2 ML Co/1 ML Ni/Pt(111) is shown in Fig. 5.13(d). The heights does not change when 300 K < T_s < 450 K. Except a plateau at sample temperature between 515 K and 600 K, it decreases when 450 K < T_s < 515 K. The temperature range of this plateau is corresponding to the temperature range of Ni-Co intermixing process. Although the Pt atoms occupied almost all the surface sites after high temperature annealing, the peak height still not increases as the behavior found in Fig. 5.12(c) and (d). From the LEED studies in sec. 5.3, the structural transition of 2 ML Co/1 ML Ni/Pt(111) is not similar to those of 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111). Hence, we believe that the difference in surface structure between 2 ML Co/1 ML Ni/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) results in a different behavior of the peak height near the Fermi edge.

5.5. UPS studies in alloy formation 71

Figure 5.13: (a)UPS spectra of 2 ML Co/1 Ml Ni/Pt(111). (b)The evolution of E_K^{F ermi} versus sample temperature. (c)The UPS spectra after moving E_K^{F ermi} to zero. (d)The evolution of 0.40 eV peak intensity.

Summary

1. Because of the competitions between lattice mismatch and surface free energy, the growth modes at room temperature of both d_CoCo/ 1 ML Ni/Pt(111) and d_{N i}Ni/1 ML Co/Pt(111) are at least 2 ML in layer-by-layer growth before the 3-dimensional island growth begins.

2. For both 1 ML Ni/1 ML Co/Pt(111) and 1 ML Co/1 ML Ni/Pt(111) systems, the temperatures for the Ni-Co intermixing process (denoted as T_mix) and the Ni-Co intermixing layer diffusing into Pt substrate (denoted as T_bulk) are around 420 K and 580 K, respectively.

3. T_mix of d_Co Co/ 1 ML Ni/Pt(111) is coverage dependent, while that for d_{N i} Ni/1 ML Co/Pt(111) is not. By comparing these results with those of Co/Pt(111) and Ni/Pt(111), we believe that the higher attractive force between Co layers is the main reason that causes the T_mix of d_Co Co/1 ML Ni/Pt(111) to be higher than those of d_{N i} Ni/1 ML Co/Pt(111).

4. The interaction force between Co and Ni is higher than that of Co-Co and Ni-Ni, this cause the T_bulkfor both 1 ML Co/ 1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111) systems is higher than that of 2 ML Co/Pt(111) and 2 ML Ni/Pt(1111).

5. The LEED pattern which have three-fold satellites surrounding each (1× 1) first order spot only can be observed in 1 ML Co/ 1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111) systems. It is predicted to be corresponding to the (Ni_xCo1−x)Pt surface alloy though the I − V LEED and Vegard’s law analyzations.

6. By comparing the Pt concentration and the phase diagrams of Co1−xPt_xand Ni1−xPt_x, we believe that the (Ni_xCo_1−x)Pt₃ surface alloy dominates the surface structures of 1 ML Co/ 1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111) systems after annealing at a temperature higher than 800 K.

7. The movement of Fermi edge in the UP spectra is related to the diffusion process, which implies the changes of work function occurred by the annealing process. The electron kinetic energy located at Fermi edge E_K^{F ermi} does not change anymore when the surface become Pt-rich.

5.5. UPS studies in alloy formation 73

8. The differences in surface composition and in density of state may cause some dif- ferences in the magnetic properties between 1 ML Co/1 ML Ni/Pt(111) and 1 ML Ni/1 ML Co/Pt(111) systems.