4 THE TECHNIQUE OF THE EXPERIMENTS
4.4 Methods of heating and cooling silver-palladium films in vacuum
For heating and cooling the films in vacuum, we used universal vacuum station VUP-5M, which was designed to prepare samples for studying by an electron microscope and other instruments. The device consists of a vacuum chamber, an air pumping system, power supply and control unit.
The studied film was placed in a vacuum chamber, from where air was pumped out to a pressure of 0.0013 Pa. Then its gradual heating or cooling began. The film was heated using a tungsten spiral with an electrical resistance of 15 Ohms, which allowed reaching a temperature of 581 K. The experiments on cooling the film were carried out with liquid nitrogen using standard means of the VUP-5M device. The photocurrent was excited through the window of a vacuum chamber using a laser system operating at a wavelength of 1064 nm according to the longitudinal photocurrent excitation scheme.
4.5 CHANGING THE PALLADIUM OXIDE CONTENT IN THE SIL-VER-PALLADIUM FILMS
To smoothly change the PdO content in the fabricated films, we used a reduction process PdO + H → Pd + H2O (see, for example, [114]), i.e. converting the palladium oxide to metallic palladium. By using approach proposed in [115], the test film was
78
placed in an electrolytic cell with stainless steel anode and Ag/Pd film cathode. The cell was filled with a weak aqueous solution of sulfuric acid (Figure 4.11). The studied film was hydrogenated at a current density of 0.7 mA/cm2. After processing the film was removed from the electrolyte, washed with distilled water, and dried.
Then, the ohmic resistance R between the electrodes of the film, photocurrent, Raman spectra, and X-ray diffraction was measured. Further, the hydrogenation and measurement processes were repeated.
Figure 4.11. Scheme of electrochemical hydrogenation of Ag/Pd films (adapted from [109]).
4.6 CONCLUSIONS TO CHAPTER 4
The experimental equipment made it possible to irradiate the studied films with nanosecond and femtosecond laser pulses of known power in a wide spectral range. During the experiments, the polarization could be smoothly changed from linear to circular, i.e. the degree of ellipticity of radiation polarization could change.
The orientation of the polarization plane of linearly polarized radiation could also be changed. The goniometric device on which the film was mounted made it possible to orient the measuring electrodes in the necessary way, which made it possible to register longitudinal and transverse photocurrents. The goniometric device also allowed changing the angle of radiation incidence on the film.
Measuring instruments made it possible to determine the amplitude and polarity of the photocurrent pulses, as well as to record their shape, to measure the decay time, rise time, and duration. It was possible to measure the photocurrent at different temperatures of Ag/Pd films in vacuum, as well as with different PdO contents.
Thus, the experimental equipment made it possible to obtain the orientation and polarization dependences of all parameters of the photocurrent pulses. Therefore, the equipment available corresponds the goals and objectives and allows to get reliable results.
79
5 THE LONGITUDINAL PHOTOCURRENT IN SILVER-PALLADIUM FILMS
This chapter describes the longitudinal photocurrent (see Figure 5.1) generated in the Ag/Pd films under irradiation with nanosecond and femtosecond laser pulses.
We found that duration and decay time of the photoresponse significantly exceeds the duration and decay time of the excitation pulses and strongly depends on polarization. It is shown that the amplitude and polarity of the longitudinal photocurrent also substantially depend on the angle of incidence and polarization of the excitation beam.
Figure 5.1. Registration scheme of longitudinal photocurrent in Ag/Pd films. The electrodes are orientated perpendicularly to the plane of incidence xz. N is unit vector along z-axis, which is normal to film surface, α is angle of incidence. k is wave vector, E is electric field vector of the p-polarized laser beam, which is parallel to plane of incidence, ne is slow axis of wave plate. The polarization of the excitation beam is determined by the angle φ (or ϕ) between the slow axis of the quarter-wave plate (or half-wave plate) and the plane of incidence (adapted from [109]).
5.1 THE CONVERSION EFFICIENCY OF LIGHT TO LONGITUDINAL CURRENT
In Ref. [89] we showed that photoresponse of Ag/Pd films essentially depends on the composition of the paste and its annealing temperature. In particular, we have found that the photoresponse of the PR50-based films obtained at Tann = 833 K and Tann = 878K is much stronger than that of PR5-based films. Correspondingly, in this
80
work, we study the photoresponse in the PR50-based Ag/Pd films obtained only at Tann = 878 and 833 K. The polarization and angle dependences of photocurrent in the samples annealed at higher temperature were not investigated.
Figure 5.2 shows that the amplitude of the photocurrent pulse is a linear function of the incident laser energy at wavelengths of 532, 795 and 1064 nm. That is in order to describe photoresponse it is convenient to introduce the light-to-current conversion efficiency by using the following equation:
𝜂𝑥=𝑈𝑚/(𝑅𝑖𝑛·𝑊𝑖𝑛), (5.1)
where Um is the maximum voltage recorded by the oscilloscope, Rin is the input resistance of the oscilloscope (in our experiments, Rin = 50 Ohm), and Win is the energy of the incidence excitation pulse.
Introducing the conversion efficiency ηx allows one to compare the experimental results obtained at different pulse energies. Therefore, to represent the experimental data, the amplitude of the photocurrent will be represented through the conversion efficiency ηx hereinafter. It is worth noting that SPGE and PDE are second-order in the light field, i.e. they manifest themselves as photocurrents proportional the intensity of the excitation light. Therefore, the experiments results are in agreement with predictions made in the Chapter 3 for the SPGE and PDE mechanisms.
Figure 5.2. Dependences of the longitudinal photocurrent amplitude on the energy of the excitation pulses: a) λ = 532 nm, Φ = 45°, nanosecond pulses; b) λ = 1064 nm, p-polarization, nanosecond pulse; c) λ = 795 nm, p- and s-polarization, femtosecond pulse.
5.2 TEMPORAL EVOLUTION OF THE LONGITUDINAL PHOTO-CURRENT
The section is devoted to the study of the temporal profile of the longitudinal photocurrent. In Figure 5.3 the shapes of normalized temporal profiles of the longitudinal photocurrent generated in the Ag/Pd film under excitation with p- and s-polarized laser pulses are presented together with the shape of the laser pulse at λ = 1064 nm. It can be seen that the photoresponse lasts much longer than the excitation laser pulse.
It should be noted that excitation with a femtosecond pulse also results in nanosecond photocurrent pulse. Figure 5.4 shows the longitudinal photocurrent
81 pulse for s-polarized femtosecond excitation (τlas = 120 fs), the shape of the exciting laser pulse is not shown, because the pulse duration was many times less than the response time of the photosensor. In experiments we used oscilloscopes with bandwidth from 0 to 400 MHz at least. In other words the oscilloscope is capable to register electrical pulses with duration of 2.5 ns. To observe shape of exciting light pulses and to measure photoresponse we used same cables. Therefore, long duration, rise and decay time of photocurrent pulses can not be explained by the limited time resolution of experimental setup.
Figure 5.3. The shapes of the normalized pulses of the longitudinal photocurrent upon excitation by p- and s-polarized laser pulses, as well as the shape of the exciting pulses at λ = 1064 nm (adapted from [5]).
Figure 5.4. The pulse shape of a longitudinal photocurrent upon excitation of by s-polarized femtosecond pulses.
82
Figure 5.5. Pulse shapes of the longitudinal photocurrent upon excitation of p-polarized radiation at wavelengths a) λ = 1350 nm, b) λ = 1700 nm c) λ = 2000 nm d) λ = 2400 nm e) λ = 3000 nm f) λ = 3300 nm; and also of s-polarized excitation at g) λ = 1700 nm h) λ = 3300 nm (adapted from [48]).
83 Figures 5.3-5.5a show the typical monopolar photoresponse observed in the wavelength range of 532 - 1700 nm. However, in the wavelength range of 1700 - 4000 nm, the temporal profile of the photoresponse for the p-polarized excitation beam changes. Specifically, when the wavelength become longer than 1700 nm, the longitudinal photoresponse becomes bi-polar (see Figures 5.5b, c, d, e, f), i.e. the photocurrent pulse contains both positive and negative parts. Starting form λ = 1700 nm, the photocurrent pulse is bipolar with a sharp negative front and a long positive tail (Figure 5.5b). The longer the wavelength, the higher amplitude of the negative pulse, at λ = 3000 nm, a negative pulse dominates the photoresponse (Figure 5.5f). It should be noted that s-polarized excitation beam produces a monopolar photocurrent pulse in the entire wavelength range of 532-4000 nm (see Figures 5.5g, h). According to theory the PDE and SPGE have opposite signs, therefore, the longitudinal photocurrent is explained well in terms of the interplay of the PDE and SPGE.
It is worth noting that the description of the PDE in terms of the high fre-quency Hall effect can be found, for example, in [49]. In that paper, the PDE was studied in graphene under THz excitation. They showed in particular that in gra-phene, the direction of PDE photocurrent depends on type of charge carriers. How-ever, the authors of that paper emphasize that the developed semi-classical descrip-tion of the PDE in terms of the AC Hall effect is limited to the simplest band struc-tures and relatively low excitation frequencies. In materials having valence and conduction bands of considerably different topology, the directions of the PDE current in p- and n-doped materials are not necessarily opposite. The conclusion on complexity of the microscopic PDE mechanism is solids have been also made in recent paper [116].
The Ag/Pd films, which were studied in the Thesis, contain p-doped PdO nanocrystals. However, at the s-polarized excitation beam we observed positive photocurrent pulse instead of negative one predicted by the semiclassical AC Hall effect theory for the p-doped materials. The reason may be two-fold. First, the semiclassical description of the photoresponse is valid at ωtp < 1. This condition is satisfied in the THz and far infra-red spectral range, while in our experiments, we use visible and infra-red excitation. Second, the PdO belongs to the tetragonal crystal system (4/mmm point group) having rather complicated band structure, in which the PDE in p-doped material not necessarily results in the negative photocurrent pulse predicted the semiclassical description. Thus in our experimental conditions, the PDE can not be described in terms of high frequency Hall effect.
Since both PDE and SPGE are second-order nonlinear effects [7], the pulse shapes of PDE and SPGE photocurrents can be described by the expression:
𝐽𝑃𝐷𝐸,𝑆𝑃𝐺𝐸(𝑡) =𝑓𝑃𝐷𝐸,𝑆𝑃𝐺𝐸(𝛷,𝛼)� 𝐼(𝑡 − 𝜏)𝑒𝑥𝑝 �− 𝜏 𝑡𝑃𝐷𝐸,𝑆𝑃𝐺𝐸�
∞ 0
𝑑𝜏, (5.2)
84
where I(t) is the intensity of the excitation pulse, fPDE and fSPGE have opposite signs and describe the polarization and angular dependences of PDE and SPGE, tPDE and tSPGE are the response times of the corresponding mechanisms. Figure 5.6 shows photocurrent pulses excited by p-polarized radiation at wavelengths of 2000 and 2400 nm approximated by formula (5.2). It can be seen that the calculated pulse shapes are quite close to the obtained waveforms.
Figure 5.6. Temporal profiles of photocurrent pulses (purple curves) recorded by an oscilloscope at p-polarized excitation for a) λ = 2000 nm and b) λ = 2400 nm. The blue, red, and green curves show the results of approximation by expression (5.2) for the PDE, SPGE, and the total photocurrent, respectively (from [48]).
Figure 5.7. Dependences of a) rise time, b) duration, and c) decay time of photocurrent pulses on the polarization plane azimuth Φ at the incidence angle of 45°. The orientation of the polarization plane for various angles Φ is shown at the top (from [48]).
Moreover, the temporal characteristics of the photoresponse at the p- and s-polarized excitation also differ significantly. Figure 5.7 shows the dependences of τrise, τfw and τdec on the polarization azimuth of the excitation laser beam at
85 λ = 1064 nm. The obtained dependences of the time parameters are well described by the expression:
𝜏𝑚(𝛷) =𝜏𝑚(𝛷= 0)𝑐𝑜𝑠2𝛷+𝜏𝑚(𝛷= 90°)𝑠𝑖𝑛2𝛷, (5.3) where the index m labels rise, fw or dec. The maximum values of τrise, τfw and τdec are achieved for the p-polarized beam (𝛷= 0) and exceed those for the s-polarized beam (𝛷= 90°) by 1.2, 2.6, and 2.1 times, respectively.
Figure 5.8 illustrates why the τrise, τfw and τdec depend on the polarization of the excitation beam. We have already showed that the film photoresponse comprises PDE and SPGE photocurrents, which have opposite signs, different durations, and polarization-dependent amplitudes. From fitting photocurrent pulse profiles (see Figure 5.6) one can see that SPGE pulses is shorter than PDE one. If we subtract PDE (pulse 1) and SPGE (pulse 3) photocurrents, the duration of the resulting pulse 2 will depend on the their amplitudes. It is worth noting that since the SPGE contribution to the longitudinal photocurrent vanishes for the s-polarized beam (𝛷= 90°), 𝜏𝑚(𝛷= 90°) is determined only by the PDE. When the polarization azimuth of the excitation beam 𝛷< 90°, the addition of the negative SPGE current results in the elongation of the resulting pulse shown in Figure 5.8.
Figure 5.8. Elongation of the photocurrent pulse due to the subtraction of the PDE and SPGE pulses.
In all experiments, it was found that the duration of the photocurrent pulses significantly exceeds the duration of the exciting laser pulses. This can be explained as follows. The Ag/Pd films comprise metallic Ag-Pd and semiconductor PdO nanocrystallites, which can be represented as conductive nanoparticles separated by dielectric (glass) barriers. That is the nanocomposite can be seen as array of capacitors that accumulate charge. In addition, Schottky barriers can occur at the junctions of PdO and Ag-Pd. As a result, Ag/Pd films can be seen as a complex electrical circuit composed of a random resistor–capacitor circuits (RC circuits).
When a photocurrent is generated, it is partially short-circuited in parts of the film that are not involved in the photogeneration, the electric charge accumulates in the
86
"capacitors" of this random circuit. When photoexcitation ends, the "capacitors"
begin to discharge, part of this current is registered by the oscilloscope, the other part charges neighboring RC circuits. Consequently, charge and discharge processes will occur in the Ag/Pd film, which can continue even after the termination of photoexcitation. One may expect that this mechanism is responsible for the elongation of the PDE and SPGE photocurrent pulses in Ag/Pd films.
5.3 DEPENDENCE OF THE CONVERSION EFFICIENCY TO THE LONGITUDINAL PHOTOCURRENT ON INCIDENCE ANGLE FOR THE S- AND P-POLARIZED EXCITATION BEAM
In order to study the dependence of the conversion efficiency on the angle of incidence α in the range λ = 532 - 4000 nm, the orientation of the polarization azimuth of the excitation beam was controlled by a half-wave plate. Figure 5.9 shows typical conversion efficiency for the s- and p-polarized radiation measured at λ = 1064 nm as functions of the angle of incidence.
Figure 5.9. Dependences of conversion efficiency for the longitudinal photocurrent on the incidence angle for s- polarized (yellow circles) and p-polarized (green triangles) exitation beam at λ = 1064 nm. Red and blue solid lines show results of fitting with Eqs. (5.4) and (5.5). The figure is adapted from [48].
It can be seen from the figure that the conversion efficiency change sign for mirrored incidence, i.e. it is zero at α = 0. That is the conversion efficiency is an odd function of α, ηx(α)=-ηx(-α), and the photocurrent is observed only with an oblique incidence. This statement is correct for the entire studied spectral range. The conversion efficiencies are maximum at α ≈ ±50° for both s- and p-polarized beams.
It is noteworthy that the longitudinal photocurrent excited by s-polarized radiation is greater than the photocurrent caused by p-polarized radiation.
87 Since SPGE is forbidden for the s-polarized excitation beam, the photocurrent at 𝛷= 90° includes PDE only and angle dependence should be described by the equation (3.59):
𝜂𝑥 ∝ 𝐴𝑃𝐷𝐸(2) (𝛼) =𝜔 𝑐
4𝑚2𝑠𝑖𝑛 𝛼 𝑐𝑜𝑠 𝛼
�𝑐𝑜𝑠 𝛼+√𝑛2− 𝑠𝑖𝑛2𝛼 �2𝑅𝑒�𝛤𝑥𝑦𝑦𝑥�, (5.4) At p-polarization the both effects contribute to the photocurrent, which is determined by equations (3.32) and (3.58):
𝜂𝑥∝ 𝐽𝑥𝑃𝐷𝐸(𝛷= 0°) +𝐽𝑥𝑆𝑃𝐺𝐸(𝛷= 0°) =𝐴𝑆𝑃𝐺𝐸(𝛼) +𝐴𝑃𝐷𝐸(1) (𝛼) = In Figure 5.9, solid lines show the fitting with Eqs. (5.4) and (5.5) performed at 𝑛= 1.189 + 1.026𝑖,𝐺𝑥𝑥𝑧𝑧= (19.48−15.7𝑖)∙10−13 𝑚𝛺∙𝑉,Γ𝑥𝑥𝑦𝑦= (1.24 + 1.14𝑖)∙
∙10−5 𝑚𝛺∙𝑉,Γ𝑥𝑦𝑥𝑦= (1.91 + 4.8𝑖)∙10−5 𝑚𝛺∙𝑉,Γ𝑥𝑦𝑦𝑥= (0.71 + 2.33𝑖)∙10−5 𝑚𝛺∙𝑉. One can observe that these equations well describe the dependence of the conversion efficiencies on the incidence angle.
Thus, the obtained angular dependences unambiguously indicate that the polarity and amplitude of the photocurrent depend on the polarization and angle of incidence of the excitation beam. The dependences are well described in terms of SPGE and PDE. Since at a normal incidence of the exciting radiation, the photocurrent is zero and independent of the place of irradiation and the distribution of radiation over the Ag/Pd film, thermoelectric [117] and Dember [118] effects does not contribute to the photocurrent.
5.4 DEPENDENCE OF THE CONVERSION EFFICIENCY ON THE POLARIZATION PLANE AZIMUTH OF THE EXCITATION BEAM
This section presents the dependence of the conversion efficiency to the longitudinal photocurrent on the angle Φ of the polarization plane azimuth of the excitations beam in a wide spectral range. Figure 5.10 shows the dependences the conversion efficiencies on Φ at λ = 532, 1064, and 1550 nm for the nanosecond excitation pulses, as well as for λ = 795 nm at τlas = 120 fs. It can be seen that at all wavelengths, the conversion efficiency is minimum for the p-polarized excitation beam (Φ = 0 and Φ = 180°) and maximum for the s-polarized excitation beam (Φ = 90°). The experimental data in the entire wavelength range are well described by the following equation:
𝜂𝑥(𝛷) =𝜂𝑥(𝛷= 0)𝑐𝑜𝑠2(𝛷) +𝜂𝑥(𝛷= 90°)𝑠𝑖𝑛2(𝛷), (5.6) where 𝜂𝑥(𝛷= 0) and 𝜂𝑥(𝛷= 90°) can be defined from Tables 3.1, 3.2, 3.3 and 3.4
𝜂𝑥(𝛷= 0)∝ 𝐴𝑆𝑃𝐺𝐸+𝐴𝑃𝐷𝐸(1) , (5.7)
88
𝜂𝑥(𝛷= 90°)∝ 𝐴𝑃𝐷𝐸(2) . (5.8)
The parameters of fitting shown in Figure 5.10 is presented it table 5.1. It is worth noting that the fitting parameters for λ = 1064 nm are equal to the parameters obtained for the angle dependences.
Figure 5.10. Dependences of conversion efficiency for the longitudinal photocurrent on the orientation angle Φ of the polarization plane. Circles show experimental data. Red solid lines show results of fitting with Eqs. (5.6 - 5.8).
From Eqs. (5.7) and (5.8) one can see that at s-polarization only PDE is possible but at p-polarization both PDE and SPGE are allowed. The experiments showed that at s-polarization photocurrent pulse is positive and origins from PDE only.
While the bipolar pulse appears only at p-polarization, therefore it results of PDE and SPGE. To describe bipolar photoresponse, which was observed at the excitation wavelength longer than 1700 nm for the p-polarized pulse, we calculated conversion efficiencies for the positive (ηx,p,pos) and negative (ηx,p,neg) parts of the photocurrent pulse (see Figure 5.11a). The dependence of the conversion efficiency for the s-polarized excitation ηx,s(λ) is also shown. It is seen that with increasing wavelength, both ηx,s, and ηx,p,pos decrease because the longer the wavelength, the smaller the photon momentum is transferred to the electrons and the higher the reflectivity of the Ag/Pd film. Both factors result in a decrease in the PDE
89 photocurrent with increasing wavelength. However, from Figure 5.11b it can be seen that the ratio of the amplitudes of the negative and positive parts of the bipolar pulse μ = |ηx,p,neg|/ηx,p,pos is a nonmonotonic function of the wavelength. Specifically, the dependence μ(λ) has a maximum at λ = 3350 nm and sharply decreases in the range 3400 - 3800 nm. The negative part of the pulse disappears at λ = 4000 nm. In other words, a bipolar photocurrent pulse is excited by p-polarized radiation in the wavelength range of 1700 - 4000 nm.
Table 5.1 The fitting parameters for experimental data in Figure 5.10.
λ, nm 532 795 1064 1550
n 0.574+0.257i 1.261+0.397i 1.189+1.026i 0.731+0.325i
Gxxzz, 10-13 m/(ΩV) 0.51+0.049i 7.12+0.848i 19.478-15.7i 1.979-3.909i Γxxyy, 10-5 m/(ΩV) 0.0089+0.189i 0.474+0.979i 1.236+1.14i 0.667+2.277i Γxyxy, 10-5 m/(ΩV) 0.0095+0.207i 0.47+1.784i -1.911+4.8i 0.667+0.756i Γxyyx, 10-5 m/(ΩV) 0.274+0.204i 0.919+1.342i -0.719+2.33i 1.759+1.516i
Figure 5.11. Dependences of a) the conversion efficiency for 1 – s-polarized excitation, 2 – p-polarized excitation (positive part of the pulse), 3 – p-polarized excitation (negative part of the pulse) b) the ratio of the pulse negative part to the positive on the excitation wavelength (adapted from [48]).
The observed nonmonotonic dependence μ(λ) can be explained by the fact that SPGE in Ag/Pd films arises due to interband transitions in PdO nanocrystallites. In the bulk PdO, the band gap is in the range from 0.6 to 0.8 eV [119,120], which corresponds to the wavelength range of 1550–2067 nm; however, in the Ag/Pd films, the band gap also depends on the size of nanocrystallites and dielectric properties of the matrix [121–123]. In Ag/Pd films, the size of PdO nanocrystallites ranges from 28 nm to several hundred nanometers. As a result of this, the PdO
The observed nonmonotonic dependence μ(λ) can be explained by the fact that SPGE in Ag/Pd films arises due to interband transitions in PdO nanocrystallites. In the bulk PdO, the band gap is in the range from 0.6 to 0.8 eV [119,120], which corresponds to the wavelength range of 1550–2067 nm; however, in the Ag/Pd films, the band gap also depends on the size of nanocrystallites and dielectric properties of the matrix [121–123]. In Ag/Pd films, the size of PdO nanocrystallites ranges from 28 nm to several hundred nanometers. As a result of this, the PdO