Aug 21, 2009 - Developing Sample Holders for Measuring Shielding Effectiveness of .... the so called âair referenceâ method [6] with a model calculation background. .... ducting single-walled carbon nanotube films,â App. Phys. Lett., Vol.
PIERS Proceedings, Moscow, Russia, August 18–21, 2009
1530
Developing Sample Holders for Measuring Shielding Effectiveness of Thin Layers on Compound Semiconductor Substrates A. Feh´ er1 , Sz. Nagy1 , and I. Mojzes2 1
Department of Telecommunications, Sz´echenyi Istv´an University H-9026 Gy˝or, Egyetem t´er 1, Hungary 2 Department of Electronics Technology, Budapest University of Technology and Economics H-1111, Budapest, Goldmann Gy. t´er 3., Hungary
Abstract— Ohmic contacts can be formed, if thin metallic layers grown on compound semiconductor substrates are heat treated. The type of the arising contact depends on the composition of the materials and the growing and thermal treating circumstances. If the layers of are thin enough, fractal-like topology of the metallic clusters can also occur. The systems can be studied by scanning electron microscopy or atomic force microscopy and the images can be analyzed by various topology determining algorithms. These values can characterize the metal-semiconductor contact. Also, resistance and other electromagnetic properties are needed for a proper description. In the present work sample holders and measuring setups for determining the shielding effectiveness or the reflectivity of such small scale, thin, planar samples are tested. The measured data are compared to finite element model results using a commercial program package for electromagnetic calculations materials with well known electromagnetic properties. The irregular shape, the small size and the mechanical rigidity of the samples makes the measurements complicated. In the models high cylindrical symmetry is used with surprisingly good accordance. The frequency range is rather large, 9 kHz to 6 GHz, which is not covered by the current standards of shielding effectiveness measurements. The sample holders originate from those of the standards with much smaller size and specially modified flanges and fixing. The measured data is shape and size independent up to a certain level. 1. INTRODUCTION
Because of the increasing need for computational power and the continuous development of the manufacturers, the size of the elements on a semiconductor slice and the line thickness in these devices reaches the mezoscopic scale soon. In these nanometer scaled objects the quantummechanical effects tend to overwhelm the bulk properties, thus new characteristics arise. The new phenomena also appear in the contacts to these semiconductor devices. The Ohmic contacts are generated by thermal treating of thin metallic layers on the desired spots. Because of the small size of the devices to be contacted, the metallic layer has to be rather thin, and as the thickness of these layers decrease, the heat treating causes interesting behavior of the metallic film. Scanning electron microscopy (SEM) and atomic force microscopy (AFM) shows, that on compound semiconductor surfaces the 50 nm to 100 nm thick metallic films transform into an island-like structure, and the arisen islands have often fractal topology [1–3]. The structure of these systems can be characterized by their fractal dimension as well as by their localization type based on structural entropy calculations [4, 5]. Mostly, the DC resistivity of the resulting structure is measured for determining wether the contact is of Ohmic type. Since a lot of the circuits operate in the microwave domain, the radio frequency behavior should be analyzed, as well. 2. SHIELDING EFFECTIVENESS MEASUREMENTS
Shielding effectiveness and dielectric permittivity characterizes the microwave behavior of the material rather well. However, the standards and the usually applied methods are not suitable for the above mentioned samples. The standards for measuring shielding effectiveness ar based on two setups. For measurements according to IEEE-STD-299, a shielded enclosure with a proper sized window is needed, and the sample should be fastened into the window frame. Two antennas in both sides of the sample measure shielding effectiveness. The size of the necessary sample is about 1 m in diameter, which is impossible to be produced even from clean semiconductor. The other usual measurement setup consists of an enlarged coaxial line, cut in two halves and flanged, measured by a network analyzer. The sample is to be fasten between the two flanges of the sample holder, as it can be seen in Fig. 1.
Progress In Electromagnetics Research Symposium Proceedings, Moscow, Russia, August 18–21, 2009 1531 aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa aaaa
aaaaa aaaaa aaaaa
aaaaa aaaaa aaaaa
Figure 1: Sample holder with specimen for standard ASTM D4935-99. The device is air-filled, the required sample diameter is 133 mm the inner electrode diameter is 33 mm. The center rod is usually elastic for the safe contact.
(a)
(b)
Figure 2: (a) Reference and (b) specimen for standard ASTM D4935-99. The sample is a full circle with 133 mm diameter, while the reference covers only the electrodes, it consists of a circle of diameter 33 mm, and a ring.
The size of the sample is not very large, and the coaxial line could be scaled to suitable size, thus this setup seems to be more satisfactory for testing the compound semiconductor based thin metallic films. Variations of the scaled sample holder for special purposes, like testing nano-designed materials [6], shielding textiles [7, 8], or carbon nanotube films [9] appeared recently in the literature. The setup, however has some drawbacks. First, for the standard measurement a reference is also needed according to Fig. 2. The reference is a circle covering the inner rod of the sample holder and a ring for covering the flanges. If a substrate is necessary for the specimen, it has to be a full circle in case of the reference, too. In our case the first problem would be to produce circular substrate, since although semiconductor slices are produced in almost cylindrical shape, their size is not proper and they brake by their crystal planes, thus the substrates are usually slices of circles, squares or trapezoids. It is also complicated to grow ring-like structures of the metallic layer, moreover, it is not guaranteed that the reference and the sample will have the same fractal structure. This is why we have chosen the so called “air reference” method [6] with a model calculation background. The second problem is the elasticity of the inner rods. The surface of the semiconductor slices are very precisely plane, but they are rather thin and break easily, thus we have chosen to have non-elastic inner electrode, polished to the plane of the flanges. Our sample holder is not air- but Teflon-filled resulting in a high stability of the inner rod and a good planar surface for the sample to lie upon. Scratching of the layers could also be avoided by this setup. 3. CALCULATIONS
The shielding effectiveness is defined as µ SE = 20 lg
Eref Eload
¶ ,
(1)
where Eref is the electrical field strength with the reference inserted into the holder whereas Eload with the specimen. The sample holder consists of two Teflon-filled coaxial line pieces with rather large flanges on the contact surfaces. The other end of the device is a standard N-type mother connector. The
1532
PIERS Proceedings, Moscow, Russia, August 18–21, 2009
characteristic impedance of the coaxial line is Z0 =
60 D ln , ε0 d
(2)
with ε0 being the relative permittivity of the filling material (εTeflon = 2.02, D the diameter of the outer electrode and d that of the inner rod. In our case, the diameters are d = 2.04 mm, D = 11 mm, the characteristic impedance is 50 Ω. The sample holder can operate up to the frequency, where the higher order modes appear [10], i.e., until fc =
1 2c πε d + D
(3)
with c being the velocity of light. For the above described sample holder this cutoff frequency is ≈ 7.3 GHz. For simulation the RF toolbox of the finite element program package COMSOL Multiphysics were applied. The high cylindrical symmetry of the sample holder was used, resulting in the sample being cylindrical, too. As a test, a Silicon specimen was tested with relative dielectric constant εSi = 11.7 for the whole frequency band. In order to model the large flanges, a ring of air was put around the sample with proper metallic or inter-dielectrics boundary conditions. In Fig. 3 the electric and magnetic field norms are summarized for 1 GHz frequency with coaxial excitation of the device at its left end.
Figure 3: The cross-section color plot of the norm of the magnetic and electric fields in the sample and sample holder at 1 GHz frequency.
Figure 4: The parameters S11 , S21 simulated with the model in Fig. 3 with 1 mW exciting power, coaxial port.
Progress In Electromagnetics Research Symposium Proceedings, Moscow, Russia, August 18–21, 2009 1533
Clearly, the electromagnetic field concentrates at the center of the sample. Similar results appeared in all the studied frequencies, the energy density between the flanges was zero, or very small. This results in the suggestion, that the shape of the sample is indifferent as long as it covers the high energy density central region of the sample holder. The parameters S11 and S21 were calculated for the 1 GHz to 4 GHz region of frequency, and plotted in Fig. 4. S21 has an increasing saturating type of behavior as the frequency grows, whereas S11 decreases. 4. MEASUREMENTS
The measurements were carried out in the Radio Frequency Test Laboratory of the Sz´echenyi Istv´an University in Gy˝or, Hungary. A network analyzer by R&S was used with a frequency range 9 kHz to 6 GHz. The sample holder without load gave the parameter S11 , S21 , S12 and S22 results summarized in Fig. 5 It can be seen, that the parameters S21 and S12 start to deviate from their ideal 0 dB level at 4 GHz, much lower than the expected cutoff frequency. However, in the 9 kHz to 4 GHz domain, the sample holder behaves very well. The 300 µm Silicon slice specimens had irregular shapes, all covering the whole surface of the teflon filling and the inner rod, with varying coverage of the flanges. The measured data for one specific trapezoidal sample are summarized in Fig. 6. Similar, almost shape-independent results came from other shaped pieces. Clearly, resonances appear around 1 GHz and 4 GHz. The parameter S11 differs from the numerically modeled values, while that of the parameter S21 is rather similar to the calculated line. Considering, that the model is oversimplified, the second result is surprisingly good. The high reflectivity can be the result of the insufficient coupling between the sample holder and the measuring instrument. The square shape of the specimen holder can cause extra loss. 5
60
0
40
5 20
10 Sxy [dB]
Sxy [dB]
15 20
20
25 30
40
ref S11 ref S21 ref S12 ref S22
35 40 45
0
0
1
2
3 f [Hz]
4
S11 S21 S12 S22
60
5
6 9
x 10
Figure 5: The parameters S11 , S21 , S12 and S22 measured on the sample holder without load.
80
0
1
2
3 f [Hz]
4
5
6 9
x 10
Figure 6: The parameters S11 , S21 , S12 and S22 measured on the sample holder with a trapezoidal load.
5. CONCLUSION
A coaxial line based sample holder for thin, small, rigid, irregularly shaped planar samples is studied both by numerical simulations and by measurements. The numerical study showed, that due to the large flanges of the specimen holder the non central regions of the sample does not affect the main results, which was also proven by network analyzer measurements. Comparing numerical and measured data can result in shielding affectiveness and dielectric constant determination methods without reference sample. ACKNOWLEDGMENT
This work was supported by the Bolyai J´anos Research Fellowship of the Hungarian Academy of Sciences.
1534
PIERS Proceedings, Moscow, Russia, August 18–21, 2009
REFERENCES
1. D´avid, L., L. Dobos, B. Kov´acs, I. Mojzes, and B. P´ecz, “Fractal character of in situ heat treated metal-compound semiconductor contacts,” J. Mater. Sci: Mater. Electron, Vol. 17, No. 4, 321–324, 2006. 2. Mojzes, I., Cs. Dominkonics, G. Hars´anyi, Sz. Nagy, J. Pipek, and L. Dobos, “Heat treatment parameters effecting the fractal dimensions of AuGe metallization on GaAs,” Appl. Phys. Lett., Vol. 91, No. 7, article No. 073107, 2007. 3. Schuszter, M., L. Dobos, K. A. Nemcu, Sz. Nagy, and I. Mojzes, “Analysis of morphology changes of heat treated metallization of compound semiconductors by the fast wavelettransform based on B-spline,” J. Optelectron. Adv. Mat., Vol. 9, No. 7, 2241–2244, 2007. 4. Pipek, J. and I. Varga, “Universal classification scheme for the spatial localization properties of one-particle states in finite d-dimensional systems,” Phys. Rev. A, Vol. 46, 3148–3164, 1992. 5. Pipek, J. and I. Varga, “Scaling behavior of energy functionals of highly complex distributions,” Int. J. Quantum Chem., Vol. 70, 125–131, 1998. 6. Vasquez, H., L. Espinoza, K. Lozano, H. Foltz, and Sh. Yang, “Simple device for electromagnetic interference Shielding effectiveness measurement,” EMC IEEE EMC Society Newsletter , Vol. 220, 62–68, 2009. 7. Bula, K., J. Koprowska, and J. Janukiewicz, “Application of cathode sputtering for obtaining ultra-thin metallic coatings on textile products,” Fibers and Textiles in Eastern Europe, Vol. 14, No. 5, 75–79, 2006. 8. Wi¸eckowski, T. W. and J. M. Janukiewicz, “Methods for evaluating the shielding effectiveness of textiles,” Fibers and Textiles in Eastern Europe, Vol. 14, No. 5, 18–22, 2006. 9. Xu, H., S. M. Anlage, L. Hu, and G. Gruner, “Microwave shielding of transparent and conducting single-walled carbon nanotube films,” App. Phys. Lett., Vol. 90, article No. 183119, 2007. 10. ASTM Standard Designation D 4935-99, “Standard test method for measuring the electromagnetic shielding effectivenes of planar materials,” 1999.
