Setting Parameters Influence on Accuracy and Stability ... - IEEE Xplore

890

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 4, APRIL 2016

Setting Parameters Influence on Accuracy and Stability of Near-Field Scanning Microwave Microscopy Platform Sijia Gu, Kamel Haddadi, Abdelhatif El Fellahi, and Tuami Lasri Abstract— In this paper, the impact of different setting parameters on the performance of a microwave microscope is evaluated. This study is performed on a homemade near-field scanning microwave microscope built up with a conventional vector network analyzer and a vertical coaxial evanescent probe. A broadband matching network based on an interferometric technique is inserted between the analyzer and the probe to enhance the measurement sensitivity/accuracy in the frequency range 2–20 GHz. The performance of the system in terms of measurement accuracy and stability is derived as a function of different setting parameters. Microwave surface imaging of subwavelength features is evaluated as a demonstration. This paper provides a guide for best practice in microwave microscopy. Index Terms— Accuracy, evanescent microwave probe, microwave imaging, near-field scanning microwave microscopy (NFSMM), nondestructive testing and evaluation (NDT&E), spatial resolution.

I. I NTRODUCTION ICROWAVE nondestructive testing and evaluation (NDT&E) methods have been widely used for the detection of defects and characterization of materials [1], [2]. As a result, methods using waveguide probes [3] and resonator structures [4], [5] are now well established. Recently, imaging techniques based on microwaves have demonstrated to be very efficient tools for the evaluation of structures [6], [7]. Today, among the NDT&E techniques, we also find nearfield scanning microwave microscopy (NFSMM) tools that have the advantage to achieve subwavelength spatial resolution [8]–[10]. Thanks to their high potential in a wide range of applications [11]–[14], industrial [15], [16] and metrology laboratories [17], [18] have developed near-field microwave microscopy platforms. In general, one can note that most of the works reported in the literature are based on the association of a vector network analyzer connected to a modified atomic force microscope [19] or a scanning tunneling microscope [20]. These works altogether demonstrate that microwave microscopy is a good candidate for addressing imaging applications with high spatial resolution.

M

Manuscript received August 3, 2015; revised October 26, 2015; accepted November 17, 2015. Date of publication February 12, 2016; date of current version March 8, 2016. This work was supported by the National Research Agency within the Program Equipex through the EXCELSIOR Project. The Associate Editor coordinating the review process was Dr. Samir Trabelsi. The authors are with the Centre National de la Recherche Scientifique, Institute of Electronics, Microelectronics and Nanotechnology, University of Lille 1, Villeneuve-d’Ascq 59652, France (e-mail: tuami.lasri@iemn. univ-lille1.fr). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIM.2015.2507699

Although these kinds of tools are now good candidates to perform numerous applications with high spatial resolution, some scientific issues are still to be overcome. In particular, the main difficulty of NFSMM lies in the measurement repeatability [9], [17], especially in micro and/or nano characterizations [21]. For this kind of experiment, typical measurement uncertainties around 10%–20% are reported [11], [21]. Even if there has been some repeatability analysis for a scanning probe microscope [22], [23], only a few works devoted to the microwave microscopes have been performed [24]. For this reason, there is an urgent need to address the repeatability issue of NFSMM and to identify the sources of the errors related to this new characterization mean. Indeed, experimentations require well-established measurement protocols to ensure the reproducibility of the measured data and to improve the microwave microscopy performance. For instance, comprehensive works focusing on the influence of setting parameters, such as standoff distance and scanning step size on the imaging quality, have been achieved [25]. In this paper, we put an emphasis on the impact of a large number of setting parameters (microwave source, impedance tuner, cables, x-y-z stage, and measurement duration) and analyze their contributions to the measurement repeatability and imaging quality. This is a very fine study that gives a good picture of the performance allowed by the developed platform. To that end, in Section II, a brief presentation of the NFSMM that incorporates a homemade broadband interferometric matching network to enhance the measurement sensitivity/accuracy is given. In Section III, the impact of the intermediate frequency bandwidth (IFBW) of the vector network analyzer (VNA) on the measurement repeatability is quantified in different scenarios. The drift of the NFSMM with time is also estimated for three different signal levels of the transmission coefficient. In Section IV, optimized key parameters, such as the IFBW and the transmission signal level, are defined to configure the NFSMM for 1-D scanning. The error associated with the scanner (x-y-z stage) is also successfully extracted. Furthermore, a 2-D microwave imaging application based on the optimized key parameters is proposed to evaluate the performance of this technique. II. M ICROWAVE M ICROSCOPY P LATFORM The architecture of the microwave microscope is shown in Fig. 1. Two arms of the hybrid coupler (reference and isolated ports) are connected to the VNA, whereas the direct and coupled channels are connected to the coaxial evanescent microwave probe (evanescent microwave probe (EMP), with

0018-9456 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

GU et al.: SETTING PARAMETERS INFLUENCE ON ACCURACY AND STABILITY OF NFSMM PLATFORM

Fig. 1.

891

Interferometry-based near-field microwave microscopy platform. Fig. 3. Variation of electric (o) and magnetic () fields as a function of the distance to the tip—2 GHz (Ansys/HFSS).

than the vacuum impedance. Fig. 3 presents the magnitude of the electric and magnetic fields as a function of the distance to the probe. We can notice that the electric field magnitude decays with the distance to the probe in all the distance range investigated. Therefore, to the benefit of the sensitivity of the probe, the material under investigation must be placed in this near-field (in the order of the tip apex size). III. I MPACT OF C ONFIGURATION PARAMETERS Fig. 2. 2-D distribution of (a) electric and (b) magnetic field magnitudes at 2 GHz (Ansys/HFSS). Probe apex size = 66 μm.

reflection coefficient EMP ) and to the impedance tuner (with reflection coefficient TUN ), respectively. By combining the reflected signals coming from the probe and the impedance tuner, respectively, it gives the possibility to tune the resulting signal S21 to a very low level at the desired frequency to benefit from the sensitivity of the VNA. To achieve a broad frequency range operation and high measurement accuracy, the interferometer is built up with a high-resolution programmable delay line (Colby Instruments PDL-200A Series), a motor-driven variable attenuator (ATM AF 074H-10-28), and an Anaren hybrid coupler. A Keysight PNA-X 5242A (10 MHz–26.5 GHz) is used to measure the transmission coefficient S21 . The device/material under test to be scanned is mounted on a x-y-z stage that offers a positioning accuracy of about ±0.1 μm. The microwave part of the microscope is kept fixed during measurement operation, whereas the sample under test is moved under the probe. More information on the instrumentation can be found in [26]–[28]. Before doing any measurement, electromagnetic simulations, by using a commercial software (Ansys/HFSS), are performed to study the distribution of the electromagnetic energy in the vicinity of the probe tip in free-space conditions (Fig. 2). Fig. 2(a) shows the distribution of the electric field at the apex of the probe. The theoretical lateral resolution is in the order of the apex size (i.e., 66 μm) that corresponds to about λ0 /2272 (where λ0 is the free-space wavelength at 2 GHz). Fig. 2(b) indicates that the magnetic field magnitude in the region of the apex is negligible. The ratio of the electric and magnetic fields magnitudes |E|/|H | is, therefore, much greater

After a preliminary repeatability investigation presented in [29], a complete study related to the influence of different setting parameters on the quality of measurement is proposed. The first part of the analysis is focused on the measurement precision. This latter is related to the deviation of repeated measurements from the mean value. Consequently, to investigate the performance of the technique, the mean and the standard deviation of the transmission coefficient are considered. The mean of the transmission coefficient is given by n 1 S21i S¯21 = n i=1

where n is the number of measurements. The standard deviation of the complex transmission coefficient is defined by  1 n 2 1  ∗ [(S21i − S21 ) × (S21i − S21 ) ] . Std(S21 ) = n−1 i=1

The relative standard deviation is given by Std(S21 )% = 100% ×

std(S21 ) |S21 |

.

The errors related to the electrical part of the system (VNA, impedance tuner, coupler, cables, and so on) and mechanical repeatability (x-y-z stage) impact the measurement precision. Concerning the contribution of the electrical part of the platform to the error budget, it is quite difficult to quantitatively analyze the influence of each component (impedance tuner, coupler, and cables) separately. Thus, we have considered this electrical contribution as a whole. Nevertheless, to lower

892

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 4, APRIL 2016

TABLE I S TANDARD D EVIATION OF THE T RANSMISSION C OEFFICIENT S21 AS A F UNCTION OF IFBW. |S21 | = −50 dB. F = 2 GHz. A CQUISITION T IME = 60 s. P0 = 0 dBm

Fig. 4. Measured transmission coefficient S21 as a function of the IFBW— P0 = 0 dBm (Keysight PNA-X 5242A).

this error source, the microwave part of system remains fixed during scanning and the environmental conditions are kept constant (in particular, the temperature). These two kinds of errors are studied thereafter. We begin with the influence of setting parameters such as the IFBW and the cancellation level of the transmission coefficient. Then, the mechanical repeatability is investigated. All the measurements are done at room temperature around 20 °C.

TABLE II S TANDARD D EVIATION OF THE T RANSMISSION C OEFFICIENT S21 AS A F UNCTION OF IFBW. |S21 | = −50 dB. F = 2 GHz. A CQUISITION T IME = 60 s. N UMBER OF P OINTS = 60. P0 = 0 dBm

A. Intermediate Frequency Bandwidth The receiver architecture of the conventional VNA is based on a tuned receiver that makes use of a local oscillator to mix the received microwave signal to a lower intermediate frequency (IF). The IF signal is bandpass filtered that narrows the receiver bandwidth and greatly improves sensitivity, dynamic range and precision. Analyzers use an analog-todigital converter and digital-signal processing to extract magnitude and phase information from the IF signal. As the measurement accuracy and sensitivity are mainly governed by the IF bandwidth of the VNA, the transmission coefficient noise floor is first determined as a function of this parameter. To that end, the source power of the VNA is set to P0 = 0 dBm and, both measurement ports of the VNA are connected to match loads. The measured transmission noise floor corresponds to the average of the measured transmission coefficient magnitude on the whole frequency band [10 MHz–26.5 GHz] (Fig. 4). From the graph in Fig. 4, it is clear that the noise floor can be significantly reduced by decreasing the IFBW. In this case, a minimum value around −110 dB for an IFBW of 1 Hz is obtained. Nevertheless, a low IFBW results in a longer acquisition time. Consequently, the measure is more sensitive to drift errors. Experiments are proposed in the following to quantify the impact of the IFBW on the measurement precision. The coupler arms connected, respectively, to the probe and the impedance tuner are equilibrated by the interferometric method proposed, so that the level of the transmission coefficient is set to a value around −50 dB at 2 GHz when the probe is in free space. The acquisition time is set to 60 s, and six values of IFBW between 1 and 1000 Hz are considered. Obviously, the number of measured samples increases with the IFBW (32 001 points for IFBW = 1000 Hz). We give in Table I the standard deviation obtained (third column) for the different cases. As the transmission coefficient

is a complex value, we have also given the standard deviation calculated for the magnitude and the phase shift separately. From these data, we show that the precision diminishes when the IFBW increases from 1 to 1000 Hz. In the next experiment, the acquisition time is still set to 60 s, whereas the number of points is now fixed to a constant value of 60 for each IFBW considered. The data obtained are summarized in Table II. We notice that for given acquisition time and number of points, the measurement precision falls with the IFBW. When the IFBW increases from 1 to 1000 Hz, the standard deviation is multiplied by a factor of 3, going from around 0.5% to 1.5%. We also retrieve when comparing Tables I and II that for given acquisition time and IFBW, the precision is obviously better when the number of points is higher. These experiments demonstrate the influence of the setting parameters of the VNA on the measurement precision. Therefore, to lower the impact of these parameters, a compromise between the IFBW, the number of points and the acquisition time has to be found. B. Signal Level of the Transmission Coefficient Magnitude In this section, the influence of the signal level of the transmission coefficient on the drift errors is under investigation. The EMP exhibits a very high impedance in comparison with the standard 50- impedance of the VNA. Consequently, if the

GU et al.: SETTING PARAMETERS INFLUENCE ON ACCURACY AND STABILITY OF NFSMM PLATFORM

893

TABLE III

TABLE IV

S TANDARD D EVIATION OF THE T RANSMISSION C OEFFICIENT S21 AS A F UNCTION OF |S21 |. F = 2 GHz. P0 = 0 dBm. IFBW = 100 Hz. A CQUISITION T IME = 5 min

S TANDARD D EVIATION FOR T HREE C ANCELLATION L EVELS OF T RANSMISSION C OEFFICIENT S21 . F = 2 GHz. P0 = 0 dBm. IFBW = 100 Hz. A CQUISITION T IME = 5 h

Fig. 5. Transmission coefficient magnitude (left) and phase shift (right) versus time of three signal levels for the transmission coefficient (around −30, −50, and −70 dB). F = 2 GHz. IFBW = 100 Hz. Number of points = 600.

EMP is connected directly to a measurement port of the VNA, the receiver becomes practically insensitive to the variations of the reflection coefficient. Therefore, the probe must be matched to the 50- input impedance of the VNA. In the proposed approach, a broadband matching network based on an interferometric technique is inserted between the analyzer and the probe to enhance the measurement sensitivity/accuracy in the frequency range 2–20 GHz [27]. Thus, the transmission coefficient S21 can be put to a very low level to benefit from the sensitivity of the VNA. As the measured signal is subject to drift errors, especially when very low signals are considered, the stability of the system as a function of acquisition time is quantified in the following for different signal levels of the transmission coefficient magnitude. The input source power of the VNA P0 and the IFBW are set, respectively, to 0 dBm and 100 Hz (that corresponds to a transmission coefficient noise floor around −90 dB according to (Fig. 4). In Table III, we present the standard deviations calculated for three different levels of the transmission coefficient S21 (about −30, −50, and −70 dB). A relatively short acquisition time is first considered (5 min). The measured data are acquired every 30 s. It is retrieved from Table III that the drift errors increase with a decrease in the signal level. A maximum deviation of 5.36% is found for the lowest magnitude level tested (−70 dB), which is acceptable for short-term measurements lasting a few minutes. After analyzing the drift errors for a short-time test, a measurement time of 5 h is investigated in Fig. 5. Indeed, one of the applications of the microwave microscopy platform proposed is to perform a 2-D imaging over an object under test, as detailed in Section IV. Because such process usually takes hours, the long-term stability must be carefully studied.

Fig. 6. Standard deviation of transmission coefficient as a function of acquisition time under three different magnitudes of the transmission coefficient S21 . F = 2 GHz. IFBW = 100 Hz. Number of points = 600.

Besides, 5 h is largely enough for measurement such as 2-D scanning. As for the short-time test, the measurement data are also acquired every 30 s. Fig. 5 clearly indicates that the measured data are impacted by drift errors. A higher signal level (−30 and −50 dB) leads obviously to a better stability of the system. For example, when a cancellation level around −70 dB is considered, the measurement stability study shows the variations of 15.5 dB and 29.4° for magnitude and phase shift, respectively, after 5 h. However, only slight variations of 0.1 dB and 1.5° for magnitude and phase shift, respectively, are obtained after 5 h when the cancellation level is around −30 dB. For a signal level in between (−50 dB), the variations observed in magnitude and phase shift are, respectively, 0.9 dB and 7.2°. From these data, the standard deviations of the measured transmission coefficient are derived for the three cases considered (Table IV). The standard deviation diminishes when the signal level of S21 increases, as found in Table III. When comparing Tables III and IV, it is evidenced that much more errors are found for all the transmission coefficients considered after a long-term test. Therefore, the acquisition time that considerably affects the test stability is an important data when imaging applications are aimed. After having an overall study of a 5 h-stability test, a further investigation is focused on deviation with acquisition time, as shown in Fig. 6. This paper provides the distribution of system’s drift errors versus acquisition time. As evidenced by Fig. 6, the deviation

894

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 4, APRIL 2016

first increases with acquisition time and then stabilizes for the three transmission coefficient levels considered. It is clearly visible for the −70 dB response that the rate of the rise of the drift error slows down with the acquisition time. This indicates a gradual downward trend of the deviation increase. It can also be observed in Fig. 6 that, for the same period, the test results turn out to be much more repeatable for higher magnitude levels (−30 and −50 dB) than that for the lowest one (−70 dB). In general, a low signal level (−70 dB) is supposed to be chosen to benefit of a larger dynamic range of the transmission coefficient. However, in that case, the standard deviation (Fig. 6) is quite high (33.2%). By contrast, for a higher signal level (−30 and −50 dB), the drift errors are much lower (0.15% and 2.3%) but at the expense of a smaller dynamic range. Thus, there is a hard compromise between the transmission coefficient level, the acquisition time, and the dynamic range. As a conclusion, given all the results obtained, we demonstrate that key parameters, such as IFBW, acquisition time, and signal level of the transmission coefficient, have a great impact on the measurement repeatability. A magnitude level of −50 dB is believed to be a good candidate for applications where a long-term measurement lasting for hours is needed. By contrast, for a short-term measurement typically lasting minutes, a level of −70 dB can be applied to benefit from the best dynamic range. IV. 1-D AND 2-D S CANNING A PPLICATIONS The optimized key parameters to perform a short-term and/or long-term test have been carefully discussed. It has been shown that the electrical components of the platform (VNA, coupler, impedance tuner, and so on) have a strong impact on the measurement reproducibility. In Section IV-A, we first present the influence of the x-y-z stage on the measurement precision by analyzing 1-D scanning tests. While, in Section IV-B, a 2-D microwave imaging application is investigated to quantify the platform performance.

Fig. 7. Repeatability tests of the motorized stage in (a) x-/y-directions and (b) z-direction. The displacement is performed in a step of 1000 μm.

At the probe initial position (10 μm as standoff distance), standard deviations are calculated as 5.20%, 5.44%, and 5.74% for x-, y-, and z-directions, respectively. For comparison, a close standard deviation in the order of 6% for z-direction was found in [29]. This uncertainty is acceptable for longterm applications such as 2-D surface imaging. To analyze the contribution of the errors brought by the mechanical functioning of the system, the results given in Fig. 6 can be used. The deviation evaluated for the conditions chosen (F = 2 GHz, IFBW = 100 Hz, |S21 | = −50 dB, and acquisition time = 2 h) is 2.3% when the stage is maintained fixed. Hence, the mechanical error can be extracted as 2.90%, 3.14%, and 3.44%, respectively, for the three axes by subtracting the electrical error part (2.3%) from the standard deviations mentioned above. In general, one can say that for equivalent equipment, the measurement errors (electrical error and mechanical error) can be split roughly in equal parts around 2% or 3%. It is worth noting that the impact of environmental conditions, such as temperature and humidity drifts, is integrated in this evaluation method of the errors occurring when using the platform. B. 2-D Nondestructive Microwave Imaging

A. 1-D Scanning Repeatability of x-y-z Stage The 3-D stage enables a large scanning area of 200 × 200 mm2 in x-/y-directions and 10 mm in z-direction with a positioning uncertainty in the order of 0.1 μm. However, as the measured data are obtained by stepping the scanning motors point by point, even such a small positioning error can lower the repeatability of one-line or 2-D scanning measurements. Therefore, the error brought by the mechanical part of the platform has to be evaluated. In the setup configuration, the key parameters mentioned above are fixed to 100 Hz, 2 h, and −50 dB, respectively, to run a long-term test. The EMP is positioned 10 μm over a metallic surface, as shown in Fig. 7. To test the reliability of the stage, the sample is moved to another position along x-/y-directions in a step of 1000 μm, which is much larger than the minimum step size of the stage (1 μm), and then moved back to the initial position [Fig. 7(a)]. This process is conducted ten times in 2 h. Similarly, in z-direction, a process of standoff distance jumping up to 1000 μm and falling back to 10 μm is repeated ten times during the same period [Fig. 7(b)].

The previous repeatability study has demonstrated excellent measurement reliability under optimized key setting parameters. Indeed, even for a long-term test, the relative deviations of the stage in x-/y-directions are only ∼5% after a 2-h test. Consequently, the good stability and versatility offered by the proposed NFSMM allow us to envisage a wide range of applications in different fields. In particular, we show that the platform is suitable for 2-D nondestructive imaging with subwavelength spatial resolution. As a demonstration, we have scanned a chip area of a widely available card (Fig. 8). A microphotograph of a metallic slot is also shown in Fig. 8. The 2-D image is completed by moving the x-y-z stage in a meandering track (sweep in one direction on y-axis followed by a step of 50 μm in x-direction and a sweep in the other direction on y-axis). Considering the increase in the drift errors with the measurement duration, we have kept for IFBW and the transmission coefficient level the values that have shown the best results in the case of long-term test (i.e., 100 Hz and −50 dB). The sweep speed of the evanescent probe is set to 600 pixels/min with a step size of 50 μm

GU et al.: SETTING PARAMETERS INFLUENCE ON ACCURACY AND STABILITY OF NFSMM PLATFORM

895

Fig. 8. Image of the chip area (11 × 8 mm2 ) of a common chip card. The chip region is metalized and the rest of the card is in plastic. The zoomed-in view slot line width stands for 220.2 μm (optical image).

Fig. 10. Averaged variation of transmission coefficient magnitude and phase shift calculated every 50 μm. (a) and (b) Mean in x-direction. (c) and (d) Mean in y-direction.

Fig. 9. 2-D images on chip area (5.5 × 8 mm2 ) of a card with a step of 50 μm. (a) and (b) Magnitude images with the same test conditions. (c) Subtraction between (a) and (b). (d) and (e) Phase-shift images with the same test conditions. (f) Subtraction between (d) and (e). Tip-sample separation = 10 μm. F = 2 GHz. IFBW = 100 Hz. Signal level of transmission coefficient = −50 dB.

that is smaller than the tip apex (66 μm). The scanning area corresponds to half of the chip surface (11×8 mm2 ), resulting in a scanning duration of 30 min. To assure the microscope operates in near-field region, the separation distance between the tip and the sample is set to 10 μm. To evaluate the accuracy and repeatability of the microscopy system, we investigate the difference between microwave images obtained by repeating the same scanning process several times (five times). We show in Fig. 9 the images [|S21 |, arg(S21)] that present the largest difference observed. First, it can be noticed that both the magnitude and the phase shift clearly return the image of the chip area. The maximum contrast for magnitude and phase shift is 8 dB [Fig. 9(a) and (b)] and 40° [Fig. 9(d) and (e)], respectively. As shown in [Figs. 9(a) and 9(d)], the slot width retrieved, equal to 200 μm, is close to the value obtained by the optical image [220.2 μm (Fig. 8)]. This width corresponds to about λ0 /660 at 2 GHz in free space. To better appreciate the difference between the 2-D scans, we have plotted the images obtained by subtracting pixel by pixel the results recorded for each case in terms of magnitude [Fig. 9(c)] and phase shift [Fig. 9(f)]. A maximum variation of 2 dB and 8°, respectively, for magnitude and phase shift is noted. Because of the very low transmission signal level, the variation is not uniform on the entire scanned surface. Therefore, we have collected the mean values for different positions materialized by the lines in Fig. 9(c) and (f). To have a global view of the distribution,

the mean is calculated on a line X X  (respectively, Y Y  ) every 50 μm. The influence of the setting parameters is analyzed by varying the values around the optimized ones (IFBW = 100 Hz and |S21 | = −50 dB). Two situations are considered, we change only one parameter, either IFBW or |S21 | (1st case: IFBW = 1000 Hz and |S21 | = −50 dB; 2nd case: IFBW = 100 Hz and |S21 | = −70 dB). The results are reported in Fig. 10. Fig. 10 clearly exhibits the drift level that affects the 2-D microwave imaging process. Because of the sweeping mechanism chosen (meander: continuous sweep along y-axis and stepped sweep along x-axis), the results are less noisy for the displacement on y-axis. We also confirm that the better results are obtained for the optimized conditions with fluctuations lower than 0.5 dB for the magnitude and lower than 3° for the phase shift. As highlighted by Fig. 10, if we deviate from these optimized setting parameters, the accuracy is strongly impacted, at least the errors are doubled. To perform these tests, only one parameter has been modified at once, and it is quite obvious that the performance will be even lowered if both are changed. These results altogether demonstrate the influence of the setting parameters on the platform accuracy. V. C ONCLUSION A scanning near-field microwave microscopy platform based on an interferometric technique is evaluated through the study of key configuration parameters. Setting parameters, such as IFBW, acquisition time and signal level of the transmission coefficient are investigated to determine the performance of the system in terms of precision and stability. Two error sources, electrical and mechanical errors, are quantitatively analyzed. In particular, mechanical error (in the order of 3%) and electrical error (in the order of 2%) are extracted from 1-D scanning measurements by considering the optimized IFBW and signal level of transmission coefficient. Moreover, the evaluation of a 2-D imaging technique, with subwavelength

896

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 65, NO. 4, APRIL 2016

spatial resolution, is performed. Indeed, this is the first study, to the best of our knowledge, where the error sources and their impacts on the stability and accuracy of an NFSMM platform are investigated. This paper provides very useful information to the NFSMM community, especially in light of the proliferation of inexpensive analyzers. R EFERENCES [1] S. Kharkovsky and R. Zoughi, “Microwave and millimeter wave nondestructive testing and evaluation—Overview and recent advances,” IEEE Instrum. Meas. Mag., vol. 10, no. 2, pp. 26–38, Apr. 2007. [2] M. T. Ghasr, B. Carroll, S. Kharkovsky, R. Austin, and R. Zoughi, “Millimeter-wave differential probe for nondestructive detection of corrosion precursor pitting,” IEEE Trans. Instrum. Meas., vol. 55, no. 5, pp. 1620–1627, Oct. 2006. [3] C. Huber, H. Abiri, S. I. Ganchev, and R. Zoughi, “Modeling of surface hairline-crack detection in metals under coatings using an open-ended rectangular waveguide,” IEEE Trans. Microw. Theory Techn., vol. 45, no. 11, pp. 2049–2057, Nov. 1997. [4] S. Trabelsi and S. O. Nelson, “Nondestructive sensing of physical properties of granular materials by microwave permittivity measurement,” IEEE Trans. Instrum. Meas., vol. 55, no. 3, pp. 953–963, Jun. 2006. [5] M. S. Boybay and O. M. Ramahi, “Non-destructive thickness measurement using quasi-static resonators,” IEEE Microw. Wireless Compon. Lett., vol. 23, no. 4, pp. 217–219, Apr. 2013. [6] M. Fallahpour, J. T. Case, M. T. Ghasr, and R. Zoughi, “Piecewise and Wiener filter-based SAR techniques for monostatic microwave imaging of layered structures,” IEEE Trans. Antennas Propag., vol. 62, no. 1, pp. 282–294, Jan. 2014. [7] M. A. Baumgartner, M. T. Ghasr, and R. Zoughi, “Wideband imaging array using orthogonally fed dual varactor-loaded elliptical slots,” IEEE Trans. Instrum. Meas., vol. 64, no. 3, pp. 740–749, Mar. 2015. [8] M. Tabib-Azar, N. S. Shoemaker, and S. Harris, “Non-destructive characterization of materials by evanescent microwaves,” Meas. Sci. Technol., vol. 4, no. 5, pp. 583–590, May 1993. [9] M. Tabib-Azar, D.-P. Su, A. Pohar, S. R. LeClair, and G. Ponchak, “0.4 μm spatial resolution with 1 GHz (λ = 30 cm) evanescent microwave probe,” Rev. Sci. Instrum., vol. 70, no. 3, pp. 1725–1729, 1999. [10] S. M. Anlage, V. V. Talanov, and A. R. Schwartz, “Principles of nearfield microwave microscopy,” in Scanning Probe Microscopy: Electrical and Electromechanical Phenomena at the Nanoscale, S. Kalinin and A. Gruverman, Eds. New York, NY, USA: Springer, Aug. 2007, pp. 215–253. [11] V. V. Talanov and A. R. Schwartz, “Near-field scanning microwave microscope for interline capacitance characterization of nanoelectronics interconnect,” IEEE Trans. Microw. Theory Techn., vol. 57, no. 5, pp. 1224–1229, May 2009. [12] S. Fabiani et al., “Broadband scanning microwave microscopy investigation of graphene,” in IEEE MTT-S Int. Microw. Symp. Dig., Jun. 2011, pp. 1–4. [13] K. Haddadi, D. Glay, and T. Lasri, “A 60 Ghz scanning near-field microscope with high spatial resolution sub-surface imaging,” IEEE Microw. Wireless Compon. Lett., vol. 21, no. 11, pp. 625–627, Nov. 2011. [14] M. F. Cordoba-Erazo and T. M. Weller, “Noncontact electrical characterization of printed resistors using microwave microscopy,” IEEE Trans. Instrum. Meas., vol. 64, no. 2, pp. 509–515, Feb. 2015. [15] Scanning Microwave Microscopy (SMM) Mode, Agilent Technol., Inc., Santa Clara, CA, USA, 2013. [16] Scanning Microwave Impedance Microscopy (sMIM), Oxford Instrum., Inc., Abingdon, U.K., 2014. [17] A. Imtiaz, T. M. Wallis, and P. Kabos, “Near-field scanning microwave microscopy: An emerging research tool for nanoscale metrology,” IEEE Microw. Mag., vol. 15, no. 1, pp. 52–64, Jan./Feb. 2014. [18] A. P. Gregory et al., “A near-field scanning microwave microscope for measurement of the permittivity and loss of high-loss materials,” in Proc. 84th ARFTG Microw. Meas. Conf. (ARFTG), Dec. 2014, pp. 1–8. [19] A. Karbassi et al., “Quantitative scanning near-field microwave microscopy for thin film dielectric constant measurement,” Rev. Sci. Instrum., vol. 79, no. 9, p. 094706, 2008.

[20] M. Farina, A. Di Donato, D. Mencarelli, G. Venanzoni, and A. Morini, “High resolution scanning microwave microscopy for applications in liquid environment,” IEEE Microw. Wireless Compon. Lett., vol. 22, no. 11, pp. 595–597, Nov. 2012. [21] H. P. Huber et al., “Calibrated nanoscale capacitance measurements using a scanning microwave microscope,” Rev. Sci. Instrum., vol. 81, no. 11, p. 113701, 2010. [22] H. Edwards, R. McGlothlin, and E. U, “Vertical metrology using scanning-probe microscopes: Imaging distortions and measurement repeatability,” J. Appl. Phys., vol. 83, no. 8, p. 3952, 1998. [23] Z. Xia, J. Wang, Y. Hou, and Q. Lu, “A high stability and repeatability electrochemical scanning tunneling microscope,” Rev. Sci. Instrum., vol. 85, no. 12, p. 125103, 2014. [24] Y.-C. Chen, H.-F. Cheng, G. Wang, X.-D. Xiang, C.-M. Lei, and I.-N. Lin, “Measurement of dielectric property by evanescent microwave microscope,” Jpn. J. Appl. Phys., vol. 41, no. 11B, pp. 7214–7217, Nov. 2002. [25] J. T. Case, M. T. Ghasr, and R. Zoughi, “Optimum two-dimensional uniform spatial sampling for microwave SAR-based NDE imaging systems,” IEEE Trans. Instrum. Meas., vol. 60, no. 12, pp. 3806–3815, Dec. 2011. [26] K. Haddadi and T. Lasri, “Interferometric technique for microwave measurement of high impedances,” in IEEE MTT-S Int. Microw. Symp. Dig., Jun. 2012, pp. 1–3. [27] K. Haddadi, H. Bakli, and T. Lasri, “Microwave liquid sensing based on interferometry and microscopy techniques,” IEEE Microw. Wireless Compon. Lett., vol. 22, no. 10, pp. 542–544, Oct. 2012. [28] S. Gu, K. Haddadi, and T. Lasri, “Near-field microwave microscopy for liquid characterization,” in Proc. 44th Eur. Microw. Conf. (EuMC), Rome, Italy, Oct. 2014, pp. 628–631. [29] S. Gu, K. Haddadi, A. El Fellahi, G. Dambrine, and T. Lasri, “Measurement accuracy and repeatability in near-field scanning microwave microscopy,” in Proc. IEEE Int. Instrum. Meas. Technol. Conf. (I2MTC), Pisa, Italy, May 2015, pp. 1735–1740.

Sijia Gu received the M.Sc. degree from the University of Lille 1, Villeneuve-d’Ascq, France, in 2013, where he is currently pursuing the Ph.D. degree with the Institute of Electronics, Microelectronics and Nanotechnology. His thesis work is mainly focused on the microwave characterization of micro- and nano-objects. His current research interests include the development of microwave instrumentation and related calibration techniques dedicated to the measurement of electrical properties of nanosized devices.

Kamel Haddadi received the M.Sc. and Ph.D. degrees from the University of Lille 1, Villeneuve-d’Ascq, France, in 2003 and 2007, respectively. He is currently an Associate Professor with the Institute of Electronics, Microelectronics and Nanotechnology, University of Lille 1. His current research interests include microwave and millimeterwave instrumentation, characterization and modeling of devices and materials, design of multiport RF systems for metrology and communications, and high-frequency characterization of nanometer devices.

GU et al.: SETTING PARAMETERS INFLUENCE ON ACCURACY AND STABILITY OF NFSMM PLATFORM

Abdelhatif El Fellahi received the M.Sc. and Ph.D. degrees from the University of Lille 1, Villeneuve d’Ascq, France, in 2009 and 2014, respectively. He currently holds a post-doctoral position with the Institute of Electronics, Microelectronics and Nanotechnology, University of Lille 1. His current research interests include the characterization of nanodevices for microwave frequencies for which there is not a tool for direct and accurate measurement.

897

Tuami Lasri has been involved in studies in the area of energy with the development of microgeneratorsbased on thermoelectric transduction. He is currently a Professor of Electronics and Electrical Engineering with the University of Lille 1, Villeneuve d’Ascq, France. His current research interests in the Institut d’Electronique, de Microélectronique et de Nanotechnologie, University of Lille 1, include the development of measurement techniques, the conception and realization of systems for microwave and millimeter-wave nondestructive evaluation purposes such as the characterization of nanodevices and electromagnetic wave interactions with different kind of materials (bulk, thin films, powders, etc.).