REVIEW OF SCIENTIFIC INSTRUMENTS 85, 104102 (2014)
Thermal desorption spectroscopy from the surfaces of metal-oxide-semiconductor nanostructures Jan Philipp Meyburg, Ievgen I. Nedrygailov, Eckart Hasselbrink, and Detlef Diesinga) Fakultät für Chemie, Universität Duisburg-Essen, D-45117 Essen, Germany
(Received 22 January 2014; accepted 20 September 2014; published online 10 October 2014) An experimental setup, which combines direct heating and temperature measurement of metal nanofilms allowing temperature programmed desorption experiments is described. This setup enables the simultaneous monitoring of the thermal desorption flux from the surface of chemi-electric devices and detection of chemically induced hot charge carriers under UHV conditions. This method is demonstrated for the case of water desorption from a Pt/SiO2 -n-Si metal-oxide-semiconductor nanostructure. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4896979] I. INTRODUCTION
Different varieties of thin film metal heterostructures, such as metal-semiconductor (MS),1–6 metal-oxidesemiconductor (MOS),7 and metal-insulator-metal (MIM)8, 9 nanostructures have attracted significant interest in recent years as sensors to study energy dissipation pathways in surface chemistry. For simplicity, all of the above mentioned nanostructures will hereinafter be referred to as chemielectrical devices. These allow the direct detection and quantification of hot charge carriers created in the top metal nanofilm in the course of a catalysed chemical reaction, e.g., the recombination of atomic hydrogen1, 8, 9 or the oxidation of carbon monoxide2, 10 and hydrogen.3, 4, 11 Comparison of the amount of detected hot charge carriers to the number of chemical events on the surface of the chemi-electrical devices can in principle provide an answer to the long-standing question regarding the role of transient electronic excitations of the substrate in the pathways of energy dissipation;12–14 a question, the answer to which is fundamental for the development of a predictive theory of surface chemistry on a molecular level. Thermal desorption spectroscopy (TDS) is a versatile and prolific tool for studying the kinetics of desorption processes from surfaces, also including the associative reactions which are the final step in catalysis.15–17 Moreover, as a method of spectrometry it also allows to gain insights into the composition of the adsorbate layer. And, the spectra provide some information about the structure of the substrate surface. However, the complex structure and the larger size of chemielectrical devices typically hampers the use of TDS. Moreover, one would really like to be able to record a spectrum while monitoring the flux of hot carriers at the same time. As a rule, the top electrode of the chemi-electrical devices consists of a polycrystalline, nanoscopic film of a metal (in most cases it is silver, platinum or palladium) and acts both as a catalyst for a chemical reaction, and a terminal for the electric circuit, which allows to measure the current of hot charge carriers created by the chemical reaction. Both physical and chemical properties of the top metal electrode may a) Author to whom correspondence should be addressed. Electronic mail:
[email protected] 0034-6748/2014/85(10)/104102/5/$30.00
vary from one type of chemi-electric device to another due to different methods of preparation, to properties of the support, and the geometry of the electrode etc. Therefore, there is a need to combine the chemi-electrical devices with the standard techniques of surface science, which enables an in situ characterization of their properties and the state of their surface as well as monitoring the rate of a surface chemical reaction simultaneously with the detection of hot charge carriers. In this article, we discuss ways which allow to use TDS despite the complexity of the devices. We report that temperature programmed desorption (TPD) experiments become feasible by combining direct, resistive heating of the metal film with the temperature measurement18 in a way which does not hinder the detection of chemically excited carriers. Direct resistive heating and temperature programmed desorption are combined for the first time in one setup. We demonstrate this combination of methods by studying the thermal desorption of water molecules from the surface of a Pt/SiO2 -n-Si nanostructure.
II. EXPERIMENT
All measurements were carried out using a custom UHV chamber with a base pressure better than 5 × 10−10 mbar. The Pt/SiO2 -n-Si nanostructures were prepared on the polished side of a Si(111) substrate (n-type, 7.5 cm), 20×10×0.5 mm3 in size, cut from a commercial wafer. First, the substrate was etched in HF (6%) for 5 min, rinsed in Milli-Q water, and dried in a steam of high purity nitrogen. Then, a 100 nm silicon oxide layer was grown by oxidation in air at 1200 K for 1 h. Onto this surface, a hPt = 5 nm thick 0.9999 pure Pt film was prepared by electron-beam physical vapor deposition. The thickness of the Pt film was monitored during the deposition process using a quartz crystal microbalance. Finally, a large area low resistance contact was made at the back side of the Si substrate by thermal infusion of pure indium. A cross-sectional view of the fabricated Pt/SiO2 -n-Si nanostructure is depicted in Fig. 1. Two silver contact pads were located at opposite ends of the Pt film allowing us to pass a current through it. In order to obtain a reference temperature of the Pt/SiO2 -n-Si nanostructure, a Pt1000 RTD sensor, 2.3×2.1×0.9 mm3 in size, was mounted at its back side. The
85, 104102-1
© 2014 AIP Publishing LLC
Meyburg et al.
Rev. Sci. Instrum. 85, 104102 (2014)
itoring the desorption flux. For simplicity we made use of the H2 O content in the background to carry out this study. Doses are given in units of L (1 L = 10−6 Torr s) for which we derived the H2 O partial pressure from reading the pressure meter, which may overestimate the actual value. Finally, an ammeter connected to the top and back electrodes of the Pt/SiO2 -n-Si nanostructure through the switch S1 was used to measure the device current induced by a surface chemical reaction or an applied voltage bias, if a voltage source (not shown in Fig. 1) is connected in series to the ammeter. The current-voltage characteristics recorded from the Pt/SiO2 -n-Si nanostructure before and after TPD experiments were identical. Thus, we conclude, that the electronic properties of the nanostructure are not altered by heating of the top electrode to the extent necessary for H2 O TDS.
QMS
UHV chamber
Pt n-Si
SiO2
RTD
In A
S1
V A
S2
variable power supply
III. RESULTS AND DISCUSSION
FIG. 1. Schematic of the experimental setup, which combines direct heating and temperature measurement using Pt/SiO2 -n-Si nanostructures with the TPD technique. The temperature of the device is monitored by a Pt1000 RTD sensor.
device is glued onto a glass slide which is fixed onto a copper plate that is connected by a copper lead to a helium coldfinger. The device holder is also in thermal contact with the wires contacting the Pt film. The two silver contact pads were connected to a variable power supply. In the first step, the temperature dependence of the resistivity of the Pt film was calibrated. A constant voltage of 100 mV was applied through the switch S2 , see Fig. 1, and the nano film resistance was measured over the temperature range from 60 to 300 K. For this purpose, the Pt/SiO2 n-Si nanostructure was cooled at a very small rate of about 0.01 K/s, making use of the connection to the helium coldfinger, starting from room temperature. It was found, that the resistance of the Pt nanofilm can be well described by a linear dependence RPt = R0 + αT on the temperature T, where R0 = 6.96 and α = 0.039 /K are constants. Establishing the temperature dependence of the resistance of the Pt film allowed us to use further on the latter as a temperature sensor and a heater. To record thermal desorption spectra, the device was cooled to T0 = 80 K and exposed to H2 O. In the next step, the switch S2 was closed and the Pt film was resistively heated by an electric current controlled by the power supply. The heating power was varied by changing the voltage in the range of 0–10 V, which allowed us to dissipate a power of up to 9 W for which the heating rate reached 100 K/s. Simultaneously with the direct heating, the temperature reading was obtained by monitoring the current and using the inverted temperature dependence of the Pt film resistance U/I − R0 . T = α
Typical thermal desorption spectra of H2 O from the polycrystalline surface of the Pt/SiO2 -n-Si nanostructure are depicted in Fig. 2 for different values of exposure in the range of 0.7–8.0 L. These curves are observed, when the Pt nano film is resistively heated from 80 to 250 K in which case the heating rate reached 11 K/s. The heating rate is not constant in our experiments as we apply a constant power. However, it could easily be made so by adding a temperature controller. On the other hand, in view of the device current measurements intended it may be advantageous to work with a constant voltage and to accept the non-constant rate. For exposures smaller than 1 L, the TD spectrum consists of a single broad peak with a maximum at about 200 K. With increasing exposure this peak narrows and shifts toward lower temperatures. So, for instance, for the largest exposure used in our experiments (8.0 L) the maximum is observed at 170 K. These results are similar to previously published TD spectra for H2 O from different faces of Pt19–22 with respect to the position of the maximum and its shift with coverage. The areas under the curves shown in Fig. 2 are proportional to the number of H2 O molecules desorbed, as they are
1.2
8L
QMS signal [arb. unit]
104102-2
0.9
4L
0.6
2L
0.3
1.4 L 0.7 L
(1)
Following Ref. 18 this method will be called direct heating and temperature measurement. A quadrupole mass spectrometer (QMS) (MKS, Spectra Satellite LM 61) was used for residual gas analysis and mon-
0.0 90
120
150
180
210
240
Temperature [K]
FIG. 2. TD spectra of H2 O from a Pt/SiO2 -n-Si nanostructure for different exposures in the range of 0.7–8.0 L. The heating rate was 11 K/s.
104102-3
Meyburg et al.
Rev. Sci. Instrum. 85, 104102 (2014)
dV
1.5
NH2O [arb. unit]
QMS signal [arb. unit]
1.0
3.4 W
2.6 W
Pt
L
27
4.1 W
w
24 21
SiO2
hPt
18
n-Si hSi
15 2.0
2.5
3.0
3.5
SiO2
4.0
dx
Heating Power [W]
0.5
x FIG. 4. Illustration of the differential volume element considered in the onedimensional heat conduction model.
0.0
0
4
12
8
16
Time [s]
FIG. 3. TD spectra from the Pt/SiO2 -n-Si nanostructure exposed to 0.7 L of H2 O for several values of heating power. The heating rates were 100, 67, and 50 K/s. Inset: Amount of desorbed H2 O (area under the QMS signal curves) as a function of the heating power. The solid line depicts the calculated amount of desorbed H2 O by the model presented.
plotted versus time. Thus, the amount of desorbed H2 O can be found by integration of the QMS signal over the time t NH O = K1 I (t)dt, (2) 2
0
where K1 is a constant. The full application of the TPD technique may require heating with different rates.15 In our approach, this can be achieved by varying the electric power dissipated in the Pt film, i.e., by applying different voltages or electric currents. However, the measured value for NH O is expected not to de2 pend on the heating power dissipated in the Pt nano film as long as the pumping speed of the UHV system is sufficiently large. Thus, one would expect, that the area under the thermal desorption curves should be independent of the heating power used as long as the amount of adsorbed H2 O is kept constant. However, we experimentally observe that the area is increasing with increasing electric power (Fig. 3). This effect may be understood when taking into account the temperature profile in the Pt film, when the latter is heated. The problem is that the leads at the ends of the Pt film are such large in heat capacity that they will effectively hold the film at the idle temperature T0 at the contact points. This is a general problem when contacting chemielectrical devices at elevated temperatures. In this work, attention is paid to the result of the contact induced temperature gradient rather than getting information from a small part of the nano film in its midst, for example, by using a Feulner cup23 or other techniques as stagnation chambers.24 In order to quantitatively model the situation of electrical contact induced temperature gradients, we divide the film into finite volume elements dV = (hPt + hSi )wdx ≈ hSi w dx, where hPt is the thickness of the Pt film, hSi the one of the SiO2 /Si/SiO2 substrate, w the width, and dx the length of the differential volume element (see Fig. 4.) For simplicity, we will use the values for bulk Si for the prop-
erties of the SiO2 /Si/SiO2 substrate. At steady state (dT/dt = 0), the one-dimensional heat conduction along the x-axis is accounted for by25 1 d dT (x) + qJoule + qrad − qcooling = 0, (3) dx dx κ where κ = κ Pt + κ Si is the thermal conductivity of the differential volume element, with κ Pt the thermal conductivity of Pt film and κ Si the thermal conductivity of Si substrate, qJoule the power density (W cm−3 ) of the Joule heating in the Pt film, qrad the power density of heating due to radiative exchange between the Pt film and walls of the vacuum chamber, and qcooling the power density of the cooling caused by conduction to the helium coldfinger. Note, that Eq. (3) implies that the heat is generated in the Pt nano film only, but transported through the whole Pt/SiO2 -n-Si nanostructure. The Joule heating caused by a constant electric current flowing through the Pt film can be described as qJoule =
d 2 dQJoule I 2 ρe = I R = , dV dV hPt hSi w 2
(4)
where I is the current and R = ρe dx/ hPt w the electric resistance of the Pt film whose resistivity is ρ e . The radiative heat exchange between the Pt film and the walls of the vacuum chamber is described by the Stefan-Boltzmann law qrad =
1 dQrad = εσ (T 4 − T (x)4 ), dV hSi B wall
(5)
where ε is the emissivity of the Pt film, σ B the StefanBoltzmann constant, and Twall the temperature of the chamber walls. Prior to heating the device, the cooling caused by the connection though the device holder and the probe wires to the coldfinger is balanced by radiative heating of the surface exposed to the walls of the vacuum chamber, i.e., qrad − qcooling = 0.
(6)
Therefore, a steady-state at the initial temperature T0 is established. In order to find an analytical solution to Eq. (3), we use the following simplifications: (i) that the condition described by Eq. (6) is valid at all temperatures during the experiment, and (ii) that the values for ρ e and κ are independent of temperature. Assumption (i) is justified as the device is in a quasi steady-state also during the heating ramp as a constant temperature profile would establish itself in about 0.1 s.
Meyburg et al.
104102-4
Rev. Sci. Instrum. 85, 104102 (2014)
550
0.43 A 0.40 A 0.38 A 0.36 A
Experimental Linear fit
20 350
250
NH2O [arb. unit]
Temperature [K]
450
26
Tdes xmax
150
14
6
50 0.0
0.5
1.0
1.5
0 0.0
2.0
x [cm]
0.2
1.1
1.3
1.5
2
Area [cm ]
With these simplifications Eq. (3) can be rewritten as 1 I 2 ρe d dT (x) , (7) =− dx dx κ hPt hSi w 2 which taking into account the boundary conditions (T(0) = T(L) = T0 ) yields T (x) = T0 +
I 2 ρe (L − x) x. 2κhPt hSi w 2
(8)
Figure 5 shows temperature profiles across the Pt/SiO2 -nSi nanostructure, calculated using Eq. (8) for several values of the electric current flowing through the Pt film. All parameters used for the calculation are listed in Table I. As can be seen in Fig. 5, the temperature profiles are parabolic in shape with only the central section heated to a temperature high enough to desorb H2 O. Here, Tdes = 200 K was chosen as the limiting temperature that must at least be reached in order to observe significant desorption from a surface element arguing that this is about 20 K below the temperature at which we observe the peak in the desorption spectrum. Thus, the area from which H2 O will desorb is given by the product of the width of the Pt/SiO2 -n-Si nanostructure and the length of the section in which the temperature will exceed Tdes , i.e., Amax = wxmax . At the same time, there are TABLE I. The parameters used in the calculation of the temperature profiles along the Pt/SiO2 -n-Si nanostructure.
FIG. 6. Amount of desorbed H2 O (area under the QMS signal curves in Fig. 3) as a function of Amax = wxmax , the area of the Pt nano film with the temperature T > Tdes . Here, xmax is the length of the region, heated to a temperature higher than Tdes , w is the width of the Pt/SiO2 -n-Si nanostructure.
sections at the edges, where the temperature remains lower than the temperature required to desorb H2 O at an appreciable rate. Moreover, the area heated to T > Tdes increases with increasing heating power, which in principle could explain the dependence of the amount of desorbed H2 O on the heating power noted, when discussing Fig. 3. To lend further credit to this hypothesis, we plot the amount of desorbed H2 O, which is determined from the area under the QMS signal curves (Fig. 3), as a function of Amax derived from the calculation for various heating powers. As shown in Fig. 6, despite the fact that the values for Amax were calculated using a crude model, the amount of desorbed H2 O scales linearly with this quantity. In other words, the NH O can 2 be represented by NH
2
O
= k2 Amax ,
(9)
1.0
5V 4V 3V 2V 1V
0.8
Thermocurrent [mA]
FIG. 5. Temperature profiles across the surface of the Pt/SiO2 -n-Si nanostructure for different currents flowing through the Pt film calculated using Eq. (8). Tdes is the temperature that must be reached in order for a significant amount of H2 O to desorb, xmax is the length of the region, heated to a temperature above Tmax .
0.6
0.4
0.2
Symbol κ Pt κ Si ρe hPt hSi Twall T0
Physical parameter
Value
Thermal conductivity of Pt Thermal conductivity of Si Electric conductivity of Pt Thickness of Pt nano film Thickness of SiO2 /Si/SiO2 layer Temperature of chamber walls Initial temperature
7.2 × 10−1 W cm−1 K−1 6.1 W cm−1 K−1 2.5 × 10−6 cm−1 5.0 × 10−7 cm 5.0 × 10−2 cm 300 K 80 K
0.0 0
120
240
360
480
600
720
Time [s]
FIG. 7. Observed device currents when various heating voltages are applied across the Pt film. In this case, the current is thermoelectric in nature (not chemically induced) arising from the vertical temperature gradient across the Si substrate.
104102-5
Meyburg et al.
where k2 is the slope of the line, shown in Fig. 6. As shown in the inset of Fig. 3 the calculated values for NH O are in fair 2 agreement with the experimental findings. In order to demonstrate that in this setup heating does not conflict with recording the current between the front Pt electrode and the device back electrode, we have monitored the latter during several heating experiments (Fig. 7). No spurious currents are observed. The monitored current is identified as thermoelectric current due to the temperature gradient across the Si substrate.18 This current follows in time the evolution of the device surface temperature, whereas the heating current through the Pt film has a totally different time dependence. Hence, the setup presented here enables the simultaneous monitoring of the thermal desorption flux and the detection of chemically induced hot charge carriers. IV. CONCLUSIONS
In this article, we have reported a way to perform reproducibly temperature programmed desorption experiments from the surface of chemoelectronic devices. Resistive heating and monitoring the temperature are achieved by passing a current directly through the nanoscopic surface metal film and monitoring its resistance. The temperature is then inferred based on the resistivity calibrated beforehand. To demonstrate this technique, we studied the thermal desorption of water molecules from the surface of the Pt/SiO2 -n-Si nanostructures. Due to the small heat capacity of the device and in particular the metal film in comparison to the leads, the temperature across the surface is not uniform. A simple model can account for that. Using a parabolic profile for the temperature across the nanofilm allows the analysis of the TPD spectra from the inhomogeneously hot nanofilm surface. This non-uniform surface temperature leads to some broadening of the TD spectra, which is fortunately not too severe due to the exponential dependence of the desorption rate on temperature, such that it will not hinder the application of this proven technique. A surface characterisation of chemi-electrical devices becomes thus possible especially for getting desorption characteristics from the whole nano film area.
Rev. Sci. Instrum. 85, 104102 (2014)
ACKNOWLEDGMENTS
This study has been supported by the Deutsche Forschungsgemeinschaft (DFG) through the collaborative research center SFB 616 – Energy Dissipation at Surfaces. I.I.N. thanks the German Academic Exchange Service (DAAD) for the Ph.D. scholarship. J.P.M. thanks Evonik Industries AG and the German Federal Ministry of Education and Research (BMBF) for the Germany Scholarship. 1 H.
Nienhaus, H. Bergh, B. Gergen, A. Majumdar, W. Weinberg, and E. McFarland, Phys. Rev. Lett. 82, 446 (1999). 2 J. Park, H. Lee, J. Renzas, Y. Zhang, and G. Somorjai, Nano Lett. 8, 2388 (2008). 3 A. Hervier, J. R. Renzas, J. Y. Park, and G. Somorjai, Nano Lett. 9, 3930 (2009). 4 E. G. Karpov and I. I. Nedrygailov, Phys. Rev. B 81, 205443 (2010). 5 S. K. Dasari, M. A. Hashemian, J. Mohan, and E. G. Karpov, Chem. Phys. Lett. 553, 47 (2012). 6 E. G. Karpov, M. A. Hashemian, and S. K. Dasari, J. Phys. Chem. C 117, 15632 (2013). 7 X. Liu, B. Cuenya, and E. McFarland, Sens. Actuators B 99, 556 (2004). 8 B. Mildner, E. Hasselbrink, and D. Diesing, Chem. Phys. Lett. 432, 133 (2006). 9 B. Schindler, D. Diesing, and E. Hasselbrink, J. Phys. Chem. C 117, 6337 (2013). 10 X. Z. Ji, A. Zuppero, J. Gidwani, and G. Somorjai, Nano Lett. 5, 753 (2005). 11 I. I. Nedrygailov and E. G. Karpov, Sens. Actuators B 148, 388 (2010). 12 H. Nienhaus, Surf. Sci. Rep. 45, 1 (2002). 13 E. Hasselbrink, Surf. Sci. 603, 1564 (2009). 14 A. Wodtke, J. Tully, and D. Auerbach, Int. Rev. Phys. Chem. 23, 513 (2004). 15 P. A. Redhead, Vacuum 12, 203 (1962). 16 D. Menzel, Interaction on Metal Surfaces (Springer, Heidelberg, 1975), p. 110. 17 A. M. de Jong and J. W. Niemantsverdriet, Surf. Sci. 233, 355 (1990). 18 I. I. Nedrygailov, E. Hasselbrink, D. Diesing, S. K. Dasari, M. A. Hashemian, and E. G. Karpov, J. Vac. Sci. Technol. A 30, 030601 (2012). 19 G. B. Fisher and J. L. Gland, Surf. Sci. 94, 446 (1980). 20 H. Ibach and S. Lehwald, Surf. Sci. 91, 187 (1980). 21 M. L. Grecea, E. H. G. Backus, B. Riedmüller, A. Eichler, A. W. Kleyn, and M. Bonn, J. Phys. Chem. B 108, 12575 (2004). 22 G. B. Fisher and B. A. Sexton, Phys. Rev. Lett. 44, 683 (1980). 23 P. Feulner and D. Menzel, J. Vac. Sci. Technol. 17, 662 (1980). 24 D. A. Hoffman and J. B. Hudson, Surf. Sci. 180, 77 (1987). 25 M. M. Rathore and R. A. Kapuno, Engineering Heat Transport, 2nd ed. (Jones and Bartlett Learning, 2010), p. 187.
Review of Scientific Instruments is copyrighted by the American Institute of Physics (AIP). Redistribution of journal material is subject to the AIP online journal license and/or AIP copyright. For more information, see http://ojps.aip.org/rsio/rsicr.jsp
Thermal desorption spectroscopy from the surfaces of metal-oxide-semiconductor nanostructures Jan Philipp Meyburg, Ievgen I. Nedrygailov, Eckart Hasselbrink, and Detlef Diesinga) Fakultät für Chemie, Universität Duisburg-Essen, D-45117 Essen, Germany
(Received 22 January 2014; accepted 20 September 2014; published online 10 October 2014) An experimental setup, which combines direct heating and temperature measurement of metal nanofilms allowing temperature programmed desorption experiments is described. This setup enables the simultaneous monitoring of the thermal desorption flux from the surface of chemi-electric devices and detection of chemically induced hot charge carriers under UHV conditions. This method is demonstrated for the case of water desorption from a Pt/SiO2 -n-Si metal-oxide-semiconductor nanostructure. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4896979] I. INTRODUCTION
Different varieties of thin film metal heterostructures, such as metal-semiconductor (MS),1–6 metal-oxidesemiconductor (MOS),7 and metal-insulator-metal (MIM)8, 9 nanostructures have attracted significant interest in recent years as sensors to study energy dissipation pathways in surface chemistry. For simplicity, all of the above mentioned nanostructures will hereinafter be referred to as chemielectrical devices. These allow the direct detection and quantification of hot charge carriers created in the top metal nanofilm in the course of a catalysed chemical reaction, e.g., the recombination of atomic hydrogen1, 8, 9 or the oxidation of carbon monoxide2, 10 and hydrogen.3, 4, 11 Comparison of the amount of detected hot charge carriers to the number of chemical events on the surface of the chemi-electrical devices can in principle provide an answer to the long-standing question regarding the role of transient electronic excitations of the substrate in the pathways of energy dissipation;12–14 a question, the answer to which is fundamental for the development of a predictive theory of surface chemistry on a molecular level. Thermal desorption spectroscopy (TDS) is a versatile and prolific tool for studying the kinetics of desorption processes from surfaces, also including the associative reactions which are the final step in catalysis.15–17 Moreover, as a method of spectrometry it also allows to gain insights into the composition of the adsorbate layer. And, the spectra provide some information about the structure of the substrate surface. However, the complex structure and the larger size of chemielectrical devices typically hampers the use of TDS. Moreover, one would really like to be able to record a spectrum while monitoring the flux of hot carriers at the same time. As a rule, the top electrode of the chemi-electrical devices consists of a polycrystalline, nanoscopic film of a metal (in most cases it is silver, platinum or palladium) and acts both as a catalyst for a chemical reaction, and a terminal for the electric circuit, which allows to measure the current of hot charge carriers created by the chemical reaction. Both physical and chemical properties of the top metal electrode may a) Author to whom correspondence should be addressed. Electronic mail:
[email protected] 0034-6748/2014/85(10)/104102/5/$30.00
vary from one type of chemi-electric device to another due to different methods of preparation, to properties of the support, and the geometry of the electrode etc. Therefore, there is a need to combine the chemi-electrical devices with the standard techniques of surface science, which enables an in situ characterization of their properties and the state of their surface as well as monitoring the rate of a surface chemical reaction simultaneously with the detection of hot charge carriers. In this article, we discuss ways which allow to use TDS despite the complexity of the devices. We report that temperature programmed desorption (TPD) experiments become feasible by combining direct, resistive heating of the metal film with the temperature measurement18 in a way which does not hinder the detection of chemically excited carriers. Direct resistive heating and temperature programmed desorption are combined for the first time in one setup. We demonstrate this combination of methods by studying the thermal desorption of water molecules from the surface of a Pt/SiO2 -n-Si nanostructure.
II. EXPERIMENT
All measurements were carried out using a custom UHV chamber with a base pressure better than 5 × 10−10 mbar. The Pt/SiO2 -n-Si nanostructures were prepared on the polished side of a Si(111) substrate (n-type, 7.5 cm), 20×10×0.5 mm3 in size, cut from a commercial wafer. First, the substrate was etched in HF (6%) for 5 min, rinsed in Milli-Q water, and dried in a steam of high purity nitrogen. Then, a 100 nm silicon oxide layer was grown by oxidation in air at 1200 K for 1 h. Onto this surface, a hPt = 5 nm thick 0.9999 pure Pt film was prepared by electron-beam physical vapor deposition. The thickness of the Pt film was monitored during the deposition process using a quartz crystal microbalance. Finally, a large area low resistance contact was made at the back side of the Si substrate by thermal infusion of pure indium. A cross-sectional view of the fabricated Pt/SiO2 -n-Si nanostructure is depicted in Fig. 1. Two silver contact pads were located at opposite ends of the Pt film allowing us to pass a current through it. In order to obtain a reference temperature of the Pt/SiO2 -n-Si nanostructure, a Pt1000 RTD sensor, 2.3×2.1×0.9 mm3 in size, was mounted at its back side. The
85, 104102-1
© 2014 AIP Publishing LLC
Meyburg et al.
Rev. Sci. Instrum. 85, 104102 (2014)
itoring the desorption flux. For simplicity we made use of the H2 O content in the background to carry out this study. Doses are given in units of L (1 L = 10−6 Torr s) for which we derived the H2 O partial pressure from reading the pressure meter, which may overestimate the actual value. Finally, an ammeter connected to the top and back electrodes of the Pt/SiO2 -n-Si nanostructure through the switch S1 was used to measure the device current induced by a surface chemical reaction or an applied voltage bias, if a voltage source (not shown in Fig. 1) is connected in series to the ammeter. The current-voltage characteristics recorded from the Pt/SiO2 -n-Si nanostructure before and after TPD experiments were identical. Thus, we conclude, that the electronic properties of the nanostructure are not altered by heating of the top electrode to the extent necessary for H2 O TDS.
QMS
UHV chamber
Pt n-Si
SiO2
RTD
In A
S1
V A
S2
variable power supply
III. RESULTS AND DISCUSSION
FIG. 1. Schematic of the experimental setup, which combines direct heating and temperature measurement using Pt/SiO2 -n-Si nanostructures with the TPD technique. The temperature of the device is monitored by a Pt1000 RTD sensor.
device is glued onto a glass slide which is fixed onto a copper plate that is connected by a copper lead to a helium coldfinger. The device holder is also in thermal contact with the wires contacting the Pt film. The two silver contact pads were connected to a variable power supply. In the first step, the temperature dependence of the resistivity of the Pt film was calibrated. A constant voltage of 100 mV was applied through the switch S2 , see Fig. 1, and the nano film resistance was measured over the temperature range from 60 to 300 K. For this purpose, the Pt/SiO2 n-Si nanostructure was cooled at a very small rate of about 0.01 K/s, making use of the connection to the helium coldfinger, starting from room temperature. It was found, that the resistance of the Pt nanofilm can be well described by a linear dependence RPt = R0 + αT on the temperature T, where R0 = 6.96 and α = 0.039 /K are constants. Establishing the temperature dependence of the resistance of the Pt film allowed us to use further on the latter as a temperature sensor and a heater. To record thermal desorption spectra, the device was cooled to T0 = 80 K and exposed to H2 O. In the next step, the switch S2 was closed and the Pt film was resistively heated by an electric current controlled by the power supply. The heating power was varied by changing the voltage in the range of 0–10 V, which allowed us to dissipate a power of up to 9 W for which the heating rate reached 100 K/s. Simultaneously with the direct heating, the temperature reading was obtained by monitoring the current and using the inverted temperature dependence of the Pt film resistance U/I − R0 . T = α
Typical thermal desorption spectra of H2 O from the polycrystalline surface of the Pt/SiO2 -n-Si nanostructure are depicted in Fig. 2 for different values of exposure in the range of 0.7–8.0 L. These curves are observed, when the Pt nano film is resistively heated from 80 to 250 K in which case the heating rate reached 11 K/s. The heating rate is not constant in our experiments as we apply a constant power. However, it could easily be made so by adding a temperature controller. On the other hand, in view of the device current measurements intended it may be advantageous to work with a constant voltage and to accept the non-constant rate. For exposures smaller than 1 L, the TD spectrum consists of a single broad peak with a maximum at about 200 K. With increasing exposure this peak narrows and shifts toward lower temperatures. So, for instance, for the largest exposure used in our experiments (8.0 L) the maximum is observed at 170 K. These results are similar to previously published TD spectra for H2 O from different faces of Pt19–22 with respect to the position of the maximum and its shift with coverage. The areas under the curves shown in Fig. 2 are proportional to the number of H2 O molecules desorbed, as they are
1.2
8L
QMS signal [arb. unit]
104102-2
0.9
4L
0.6
2L
0.3
1.4 L 0.7 L
(1)
Following Ref. 18 this method will be called direct heating and temperature measurement. A quadrupole mass spectrometer (QMS) (MKS, Spectra Satellite LM 61) was used for residual gas analysis and mon-
0.0 90
120
150
180
210
240
Temperature [K]
FIG. 2. TD spectra of H2 O from a Pt/SiO2 -n-Si nanostructure for different exposures in the range of 0.7–8.0 L. The heating rate was 11 K/s.
104102-3
Meyburg et al.
Rev. Sci. Instrum. 85, 104102 (2014)
dV
1.5
NH2O [arb. unit]
QMS signal [arb. unit]
1.0
3.4 W
2.6 W
Pt
L
27
4.1 W
w
24 21
SiO2
hPt
18
n-Si hSi
15 2.0
2.5
3.0
3.5
SiO2
4.0
dx
Heating Power [W]
0.5
x FIG. 4. Illustration of the differential volume element considered in the onedimensional heat conduction model.
0.0
0
4
12
8
16
Time [s]
FIG. 3. TD spectra from the Pt/SiO2 -n-Si nanostructure exposed to 0.7 L of H2 O for several values of heating power. The heating rates were 100, 67, and 50 K/s. Inset: Amount of desorbed H2 O (area under the QMS signal curves) as a function of the heating power. The solid line depicts the calculated amount of desorbed H2 O by the model presented.
plotted versus time. Thus, the amount of desorbed H2 O can be found by integration of the QMS signal over the time t NH O = K1 I (t)dt, (2) 2
0
where K1 is a constant. The full application of the TPD technique may require heating with different rates.15 In our approach, this can be achieved by varying the electric power dissipated in the Pt film, i.e., by applying different voltages or electric currents. However, the measured value for NH O is expected not to de2 pend on the heating power dissipated in the Pt nano film as long as the pumping speed of the UHV system is sufficiently large. Thus, one would expect, that the area under the thermal desorption curves should be independent of the heating power used as long as the amount of adsorbed H2 O is kept constant. However, we experimentally observe that the area is increasing with increasing electric power (Fig. 3). This effect may be understood when taking into account the temperature profile in the Pt film, when the latter is heated. The problem is that the leads at the ends of the Pt film are such large in heat capacity that they will effectively hold the film at the idle temperature T0 at the contact points. This is a general problem when contacting chemielectrical devices at elevated temperatures. In this work, attention is paid to the result of the contact induced temperature gradient rather than getting information from a small part of the nano film in its midst, for example, by using a Feulner cup23 or other techniques as stagnation chambers.24 In order to quantitatively model the situation of electrical contact induced temperature gradients, we divide the film into finite volume elements dV = (hPt + hSi )wdx ≈ hSi w dx, where hPt is the thickness of the Pt film, hSi the one of the SiO2 /Si/SiO2 substrate, w the width, and dx the length of the differential volume element (see Fig. 4.) For simplicity, we will use the values for bulk Si for the prop-
erties of the SiO2 /Si/SiO2 substrate. At steady state (dT/dt = 0), the one-dimensional heat conduction along the x-axis is accounted for by25 1 d dT (x) + qJoule + qrad − qcooling = 0, (3) dx dx κ where κ = κ Pt + κ Si is the thermal conductivity of the differential volume element, with κ Pt the thermal conductivity of Pt film and κ Si the thermal conductivity of Si substrate, qJoule the power density (W cm−3 ) of the Joule heating in the Pt film, qrad the power density of heating due to radiative exchange between the Pt film and walls of the vacuum chamber, and qcooling the power density of the cooling caused by conduction to the helium coldfinger. Note, that Eq. (3) implies that the heat is generated in the Pt nano film only, but transported through the whole Pt/SiO2 -n-Si nanostructure. The Joule heating caused by a constant electric current flowing through the Pt film can be described as qJoule =
d 2 dQJoule I 2 ρe = I R = , dV dV hPt hSi w 2
(4)
where I is the current and R = ρe dx/ hPt w the electric resistance of the Pt film whose resistivity is ρ e . The radiative heat exchange between the Pt film and the walls of the vacuum chamber is described by the Stefan-Boltzmann law qrad =
1 dQrad = εσ (T 4 − T (x)4 ), dV hSi B wall
(5)
where ε is the emissivity of the Pt film, σ B the StefanBoltzmann constant, and Twall the temperature of the chamber walls. Prior to heating the device, the cooling caused by the connection though the device holder and the probe wires to the coldfinger is balanced by radiative heating of the surface exposed to the walls of the vacuum chamber, i.e., qrad − qcooling = 0.
(6)
Therefore, a steady-state at the initial temperature T0 is established. In order to find an analytical solution to Eq. (3), we use the following simplifications: (i) that the condition described by Eq. (6) is valid at all temperatures during the experiment, and (ii) that the values for ρ e and κ are independent of temperature. Assumption (i) is justified as the device is in a quasi steady-state also during the heating ramp as a constant temperature profile would establish itself in about 0.1 s.
Meyburg et al.
104102-4
Rev. Sci. Instrum. 85, 104102 (2014)
550
0.43 A 0.40 A 0.38 A 0.36 A
Experimental Linear fit
20 350
250
NH2O [arb. unit]
Temperature [K]
450
26
Tdes xmax
150
14
6
50 0.0
0.5
1.0
1.5
0 0.0
2.0
x [cm]
0.2
1.1
1.3
1.5
2
Area [cm ]
With these simplifications Eq. (3) can be rewritten as 1 I 2 ρe d dT (x) , (7) =− dx dx κ hPt hSi w 2 which taking into account the boundary conditions (T(0) = T(L) = T0 ) yields T (x) = T0 +
I 2 ρe (L − x) x. 2κhPt hSi w 2
(8)
Figure 5 shows temperature profiles across the Pt/SiO2 -nSi nanostructure, calculated using Eq. (8) for several values of the electric current flowing through the Pt film. All parameters used for the calculation are listed in Table I. As can be seen in Fig. 5, the temperature profiles are parabolic in shape with only the central section heated to a temperature high enough to desorb H2 O. Here, Tdes = 200 K was chosen as the limiting temperature that must at least be reached in order to observe significant desorption from a surface element arguing that this is about 20 K below the temperature at which we observe the peak in the desorption spectrum. Thus, the area from which H2 O will desorb is given by the product of the width of the Pt/SiO2 -n-Si nanostructure and the length of the section in which the temperature will exceed Tdes , i.e., Amax = wxmax . At the same time, there are TABLE I. The parameters used in the calculation of the temperature profiles along the Pt/SiO2 -n-Si nanostructure.
FIG. 6. Amount of desorbed H2 O (area under the QMS signal curves in Fig. 3) as a function of Amax = wxmax , the area of the Pt nano film with the temperature T > Tdes . Here, xmax is the length of the region, heated to a temperature higher than Tdes , w is the width of the Pt/SiO2 -n-Si nanostructure.
sections at the edges, where the temperature remains lower than the temperature required to desorb H2 O at an appreciable rate. Moreover, the area heated to T > Tdes increases with increasing heating power, which in principle could explain the dependence of the amount of desorbed H2 O on the heating power noted, when discussing Fig. 3. To lend further credit to this hypothesis, we plot the amount of desorbed H2 O, which is determined from the area under the QMS signal curves (Fig. 3), as a function of Amax derived from the calculation for various heating powers. As shown in Fig. 6, despite the fact that the values for Amax were calculated using a crude model, the amount of desorbed H2 O scales linearly with this quantity. In other words, the NH O can 2 be represented by NH
2
O
= k2 Amax ,
(9)
1.0
5V 4V 3V 2V 1V
0.8
Thermocurrent [mA]
FIG. 5. Temperature profiles across the surface of the Pt/SiO2 -n-Si nanostructure for different currents flowing through the Pt film calculated using Eq. (8). Tdes is the temperature that must be reached in order for a significant amount of H2 O to desorb, xmax is the length of the region, heated to a temperature above Tmax .
0.6
0.4
0.2
Symbol κ Pt κ Si ρe hPt hSi Twall T0
Physical parameter
Value
Thermal conductivity of Pt Thermal conductivity of Si Electric conductivity of Pt Thickness of Pt nano film Thickness of SiO2 /Si/SiO2 layer Temperature of chamber walls Initial temperature
7.2 × 10−1 W cm−1 K−1 6.1 W cm−1 K−1 2.5 × 10−6 cm−1 5.0 × 10−7 cm 5.0 × 10−2 cm 300 K 80 K
0.0 0
120
240
360
480
600
720
Time [s]
FIG. 7. Observed device currents when various heating voltages are applied across the Pt film. In this case, the current is thermoelectric in nature (not chemically induced) arising from the vertical temperature gradient across the Si substrate.
104102-5
Meyburg et al.
where k2 is the slope of the line, shown in Fig. 6. As shown in the inset of Fig. 3 the calculated values for NH O are in fair 2 agreement with the experimental findings. In order to demonstrate that in this setup heating does not conflict with recording the current between the front Pt electrode and the device back electrode, we have monitored the latter during several heating experiments (Fig. 7). No spurious currents are observed. The monitored current is identified as thermoelectric current due to the temperature gradient across the Si substrate.18 This current follows in time the evolution of the device surface temperature, whereas the heating current through the Pt film has a totally different time dependence. Hence, the setup presented here enables the simultaneous monitoring of the thermal desorption flux and the detection of chemically induced hot charge carriers. IV. CONCLUSIONS
In this article, we have reported a way to perform reproducibly temperature programmed desorption experiments from the surface of chemoelectronic devices. Resistive heating and monitoring the temperature are achieved by passing a current directly through the nanoscopic surface metal film and monitoring its resistance. The temperature is then inferred based on the resistivity calibrated beforehand. To demonstrate this technique, we studied the thermal desorption of water molecules from the surface of the Pt/SiO2 -n-Si nanostructures. Due to the small heat capacity of the device and in particular the metal film in comparison to the leads, the temperature across the surface is not uniform. A simple model can account for that. Using a parabolic profile for the temperature across the nanofilm allows the analysis of the TPD spectra from the inhomogeneously hot nanofilm surface. This non-uniform surface temperature leads to some broadening of the TD spectra, which is fortunately not too severe due to the exponential dependence of the desorption rate on temperature, such that it will not hinder the application of this proven technique. A surface characterisation of chemi-electrical devices becomes thus possible especially for getting desorption characteristics from the whole nano film area.
Rev. Sci. Instrum. 85, 104102 (2014)
ACKNOWLEDGMENTS
This study has been supported by the Deutsche Forschungsgemeinschaft (DFG) through the collaborative research center SFB 616 – Energy Dissipation at Surfaces. I.I.N. thanks the German Academic Exchange Service (DAAD) for the Ph.D. scholarship. J.P.M. thanks Evonik Industries AG and the German Federal Ministry of Education and Research (BMBF) for the Germany Scholarship. 1 H.
Nienhaus, H. Bergh, B. Gergen, A. Majumdar, W. Weinberg, and E. McFarland, Phys. Rev. Lett. 82, 446 (1999). 2 J. Park, H. Lee, J. Renzas, Y. Zhang, and G. Somorjai, Nano Lett. 8, 2388 (2008). 3 A. Hervier, J. R. Renzas, J. Y. Park, and G. Somorjai, Nano Lett. 9, 3930 (2009). 4 E. G. Karpov and I. I. Nedrygailov, Phys. Rev. B 81, 205443 (2010). 5 S. K. Dasari, M. A. Hashemian, J. Mohan, and E. G. Karpov, Chem. Phys. Lett. 553, 47 (2012). 6 E. G. Karpov, M. A. Hashemian, and S. K. Dasari, J. Phys. Chem. C 117, 15632 (2013). 7 X. Liu, B. Cuenya, and E. McFarland, Sens. Actuators B 99, 556 (2004). 8 B. Mildner, E. Hasselbrink, and D. Diesing, Chem. Phys. Lett. 432, 133 (2006). 9 B. Schindler, D. Diesing, and E. Hasselbrink, J. Phys. Chem. C 117, 6337 (2013). 10 X. Z. Ji, A. Zuppero, J. Gidwani, and G. Somorjai, Nano Lett. 5, 753 (2005). 11 I. I. Nedrygailov and E. G. Karpov, Sens. Actuators B 148, 388 (2010). 12 H. Nienhaus, Surf. Sci. Rep. 45, 1 (2002). 13 E. Hasselbrink, Surf. Sci. 603, 1564 (2009). 14 A. Wodtke, J. Tully, and D. Auerbach, Int. Rev. Phys. Chem. 23, 513 (2004). 15 P. A. Redhead, Vacuum 12, 203 (1962). 16 D. Menzel, Interaction on Metal Surfaces (Springer, Heidelberg, 1975), p. 110. 17 A. M. de Jong and J. W. Niemantsverdriet, Surf. Sci. 233, 355 (1990). 18 I. I. Nedrygailov, E. Hasselbrink, D. Diesing, S. K. Dasari, M. A. Hashemian, and E. G. Karpov, J. Vac. Sci. Technol. A 30, 030601 (2012). 19 G. B. Fisher and J. L. Gland, Surf. Sci. 94, 446 (1980). 20 H. Ibach and S. Lehwald, Surf. Sci. 91, 187 (1980). 21 M. L. Grecea, E. H. G. Backus, B. Riedmüller, A. Eichler, A. W. Kleyn, and M. Bonn, J. Phys. Chem. B 108, 12575 (2004). 22 G. B. Fisher and B. A. Sexton, Phys. Rev. Lett. 44, 683 (1980). 23 P. Feulner and D. Menzel, J. Vac. Sci. Technol. 17, 662 (1980). 24 D. A. Hoffman and J. B. Hudson, Surf. Sci. 180, 77 (1987). 25 M. M. Rathore and R. A. Kapuno, Engineering Heat Transport, 2nd ed. (Jones and Bartlett Learning, 2010), p. 187.
Review of Scientific Instruments is copyrighted by the American Institute of Physics (AIP). Redistribution of journal material is subject to the AIP online journal license and/or AIP copyright. For more information, see http://ojps.aip.org/rsio/rsicr.jsp
