Charaterization of the dieletri properties of various fiberglass/epoxy omposite layups Marotte, Laurissa (University of Kansas); Arnold, Emily Center for Remote Sensing of Ie Sheets, University of Kansas Lawrene, KS laurissa@ku.edu Abstrat The Center for Remotes Sensing of Ie Sheets (CReSIS) utilizes airraft equipped with multiple radar antenna arrays to image and sound polar ie sheets. The strutures required to integrate these antenna arrays on to the airborne platform must onsider both strutural and RF performane. Fiberglass/epoxy omposites are ommon materials used for the onstrution of antenna radomes and other support strutures. Strutural properties of fiberglass/epoxy omposite materials an be tailored for speifi appliations, and the material s high strength-to-weight ratio make it ideal for airraft appliations. In addition, the non-ondutive, low-loss nature of fiberglass/epoxy omposites are neessary harateristis for materials used for antenna radomes and support strutures. This study proposes to investigate the effets of staking sequene (order and orientation of the individual plies in a omposite laminate) on the dieletri properties of fiberglass/epoxy omposite materials similar to those utilized by CReSIS. Key words Composite materials, dieletri onstant, staking sequene, non-isotropi dieletris permittivity), ϵ r, governs a lossless, non-ondutive, nonmagneti material s impedane via the relationship: ƞ = ƞ 0 ε r (Equation 1)[1] where ƞ 0 is the impedane of free spae (377 Ω). When an eletromagneti wave propagates through a material with a dieletri onstant greater than one, part of the signal is refleted and the rest is transmitted at eah interfae between the material and free spae. Figure 1 illustrates the behavior of an eletromagneti wave as it passes through a dieletri. Free spae Dieletri Free spae I. INTRODUCTION A. Bakground Composite materials Fiberglass/epoxy omposite materials onsist of glass fibers embedded in an epoxy resin. Strutural omponents omprised of omposite materials mainly arry applied loads in the diretion of the fibers. Beause strutural properties of omposite materials an be easily manipulated and beause they have a high strength-to-weight ratio, these materials are widely used in the design and fabriation of modern airraft. B. Staking sequene Composite materials onsist of many layers of zero/ninety degree plies and forty-five/negative forty-five degree plies. The order in whih these plies appear in the layup is alled the layup s staking sequene. Staking sequene is speifi to the strutural requirements of the omponent being built. C. Dieletri materials Beause fiberglass materials are used in the onstrution of antenna radomes/fairings developed by the Center for Remote Sensing of Ie Sheets (CReSIS) for remote sensing appliations, it is important to understand the effets of staking sequene on their dieletri properties. A material s dieletri onstant (or Figure 1: Behavior of light as it passes through a dieletri The total refleted field over the total inident field, alled the refletion oeffiient Γ, an be alulated via the equation: Γ = ƞ ƞ 0 ƞ+ƞ 0 (Equation 2)[1] A dieletri onstant whih mathes that of free spae will results in a refletion oeffiient of zero. The transmission oeffiient T is given by: T = 2ƞ ƞ+ƞ 0 (Equation 3)[1] In airborne remote sensing appliations, it is important to maximize the transmission oeffiient of an antenna s signal toward the objet of interest below. Hene, haraterizing the dieletri properties of fiberglass/epoxy materials whih appear on many CReSIS airraft will help researhers improve antenna performane in airborne remote sensing systems.
S21 (db) A. Time delay method II. METHODS One method for finding permittivity values of fiberglass panels utilized the time delay a transmit antenna s signal experienes as it passes through a dieletri. When a signal travels from one antenna to another in free spae, it travels at approximately 3 x 10 8 m/s, or the speed of light,, along the entire path. When a dieletri is plaed between the two antennas, the signal slows down momentarily as it passes through the material. The veloity at whih the signal propagates through a non-magneti dieletri is given by: V p = (Equation 4)[2] ε r Figure 2 illustrates the passage of eletromagneti energy from a transmit antenna to a reeive antenna through a dieletri material. Transmit antenna Free spae Material V p thikness Free spae Figure 2: Slowing of an EM wave as it passes through a dieletri If two transmission measurements are taken with a network analyzer one with a panel in between and one without the transmission of the signals an be plotted as a funtion of time. Beause the signal s veloity dereases as it passes through the dieletri, the maximum transmission of the signal traveling through the panel will our later than the maximum transmission of the signal traveling only in free spae. Figure 3 shows an example of two S21 plots in the time domain. Reeive antenna The differene between the times of maximum transmission of these two signals is the time delay, whih an be used to find the dieletri onstant of the material by subtrating the free spae measurement s propagation time from the dieletri measurement s propagation time and substituting into Equation 4: t dieletri t free spae = thikness V p thikness Δt = thikness ( ε r 1) (Equation 5) The time delay aused by a thin dieletri panel is extremely small. Beause the time step of the network analyzer is inversely proportional to the operating bandwidth of the transmit and reeive antennas, antennas with a very wide bandwidth must be used in order to detet the slight time delay aused by a thin panel. B. High Frequeny Struture Simulator method Another method used to find the dieletri onstant of the fiberglass panels was to ompare the shift in resonant frequeny of a mirostrip antenna mounted on a fiberglass substrate with omputed values generated by a parametri analysis tool in ANSYS High Frequeny Struture Simulator (HFSS). HFSS is a powerful modeling program designed to aurately predit the frequeny response of high-frequeny antennas suh as mirostrip antennas mounted on a substrate. [3] Figures 4 and 5 show an example of an antenna design and its alulated frequeny response in HFSS, respetively. Free spae only Δt Time (ns) With Panel Figure 3: S21 measurements of two antennas in free spae and with a dieletri Figure 4: Example of an antenna design in HFSS To test for permittivity frequeny dependeny, two mirostrip antennas were designed using HFSS one resonating at 4 GHz and another resonating at 6.7 GHz. To ensure that the results predited by HFSS would be aurate, eah antenna was first mounted on a Rohaell substrate with a dieletri onstant of 1.1 and ompared to refletion values predited by HFSS. [4] The antennas were then mounted onto fiberglass/epoxy substrates and their resonant frequenies were reorded. A parametri analysis tool in HFSS was used to generate frequeny response plots of the antennas on substrates with varying permittivity values. Permittivity values orresponding with the most similar
generated responses to the real, measured resonant frequenies of the antennas were reorded as representing the approximate permittivity values of the fiberglass. Figure 5: Example of a frequeny response alulation in HFSS A. Materials tested III. EXPERIMENTAL SETUP The staking sequene families tested in this experiment were,, and biaxial loth. These families vary in the number and sequene of 0/90-degree and 45/-45 plies that appear in the layup. All panels were omprised of 24 layers of pre-impregnated S-glass/epoxy material. After uring, all panels were approximately 0.25 inhes thik. Family 0/90 degree plies 45 degree plies The times at maximum transmission were extrated from the.s2p files using a simple MATLAB ode and then substituted into Equation 3 to find the dieletri onstant of eah fiberglass panel. C. HFSS method Two high-frequeny mirostrip antennas were designed using HFSS, one resonating at 6.9 GHz and another resonating at 4.0 GHz. Eah antenna s frequeny response was alulated for a substrate with a dieletri onstant of 1.1. The two antennas were then built by hand using adhesive opper tape as the ground plane and mirostrip. For eah antenna, a port was soldered to the ground plane and its feed was soldered to the mirostrip. The feed ran from the port to the antenna via a feed line that passed through the material. Figure 6 illustrates the dimensions and setup of all mirostrip antennas. Copper ground plane Fiberglass panel 10 m Port Feed line Copper mirostrip 20% 80% 50% 50% A/2 A/2 100% 0% B/2 The and families were hosen to be tested beause these two families are ommon in airraft omponent fabriation. [5] The family was hosen to be tested beause it allows investigation into the effet of 45-degree plies on material dieletri properties 4.5 m Copper mirostrip B/2 9 m B. Time delay method A two antennas (one quad-ridged horn and another dualridged horn) were plaed 3.5 feet apart in an anehoi antenna hamber. All wires that were to be used were onneted to Ports 1 and 2 of an Agilent network analyzer, whih was then alibrated with an eletroni alibration unit. The two antennas were then onneted to the transmission lines. The network analyzer was set to measure between 2 GHz and 18 GHz, and an inverse Fourier transform was subsequently applied to onvert the olleted data into the time domain. A free spae S(21) measurement was taken. Eah fiberglass panel was then arefully plaed, one at a time, halfway between the antennas, and their orresponding S(21) responses were reorded. Figure 6: Top: Cross-setion of mirostrip antenna setup Bottom: Dimensions of mirostrip on the substrate Figure 7 shows one of the test antennas onneted to a transmission line in the anehoi hamber. The ratio of the refleted energy to the inident energy at the port (denoted as S 11 ) was measured in the frequeny domain. The Rohaell dieletri measurements were ompared to HFSS s predited results for a substrate with a dieletri onstant of 1.1. Eah antenna type was then mounted onto eah fiberglass panel, making six test panels total.
Figure 7: Mirostrip antenna setup in anehoi hamber A parametri analysis was then performed in HFSS for eah antenna type, varying the substrate s dieletri onstant with eah iteration. The measured frequeny response of eah antenna was then ompared to the parametri analysis results and mathed to the losest alulated response from HFSS to pinpoint the dieletri onstant of eah substrate. 0-2 -4 Rohaell, 6.7 GHz Antenna 1-3 -5-7 -9 1 3 5 Rohaell Chek, 4 GHz Antenna Rohael HFSS E = 1.1 IV. RESULTS A. Time delay method Beause the transmit and reeive antennas were limited to a bandwidth of 16 GHz, the smallest measureable time delay using the network analyzer was 62.5 pioseonds. A time delay of 62.5 pioseonds using panels 0.25 inhes thik would indiate a permittivity value of 15.6. The permittivity of S-glass is 5.2 and the permittivity of epoxy is 3.6, therefore a dieletri onstant of 15.6 is unrealisti. [6] In order to detet a permittivity value lose to those of S-glass and epoxy, antennas operating at a bandwidth of over 30 GHz must be used with panels that are 0.25 inhes thik. Beause the antennas available operated at a narrower bandwidth than was neessary, the permittivity values alulated during a test run of this method were impreise; thiker panels were neessary in order to produe aurate results. Beause aess to omposite material was limited, this method was abandoned early in the investigation. B. HFSS Method The dimensions and alulated resonant frequeny of the mirostrip antennas are tabulated below. A (m) B (m) f res (GHz) 7.5 2 4.0 8 1 6.9 The auray of the HFSS simulations was onfirmed by the measurements with the Rohaell substrate. The resonant frequeny for eah antenna fell within 0.15 GHz of its predited value. Figures 8 and 9 show the frequeny response of the antennas mounted on Rohaell ompared to the free spaesubstrate antennas in HFSS. -8 0 2 Rohael HFSS E = 1.1 Figures 8 and 9: Calulated vs. measured responses of antennas mounted on Rohaell The resonant frequeny of the antennas mounted on fiberglass panels shifted downward with respet to measurements taken of antennas mounted on Rohaell. Figures 10 and 11 show the resonant frequeny shifts of the measurements taken with fiberglass substrates. 0-2 -4-8 0 2 4 6.9 GHz Antenna 6 Rohael
1-3 -5-7 -9 1 3 4 GHz Antenna 5 Figures 10 and 11: Downwards resonant frequeny shift of antennas mounted on fiberglass substrates For all substrates, the HFSS result that most losely mathed the measurements taken using the network analyzer was that of an antenna mounted on a substrate with a dieletri onstant of 3.4. Figures 12 and 13 ompare the measured frequeny responses of both antennas mounted on the various omposite layups as substrates. 0-2 -4-8 0 2 4 6 1-3 -5-7 -9 1 3 HFSS Simulation, 6.9 GHz Antenna HFSS Simulation, 4 GHz Antenna 5 2E+09 7E+09 Figures 12 and 13: Measured results of antennas mounted on fiberglass vs. alulated results of a substrate with a permittivity of 3.4 Beause a loss tangent of zero was assumed for all substrates in HFSS, disrepanies in the Q fator between measurement results and HFSS preditions may be attributed to losses in the material. FS HFSS, E = 3.4 HFSS, E = 3.4 V. ANALYSIS AND CONCLUSIONS To use the time delay method to find the dieletri onstant of a thin material sample, antennas operating at a very wide bandwidth must be used in order to aommodate the fine time resolution required to detet a small time delay, as in the ase of a material with a small dieletri onstant. Alternatively, the antenna operating at a bandwidth of 16 GHz would need to be used with very thik material samples. Beause aess to preimpregnated omposite material was limited and there were no antennas available with a wide enough bandwidth to measure a small time step, the time delay method did not produe aurate results. Using the HFSS method, it was possible to approximate the dieletri onstant of S-glass/epoxy omposite materials to be around 3.4. Beause the same approximate dieletri onstant fit well with the results of both antennas and for all staking sequenes, it an be onluded that permittivity does not heavily rely on staking sequene or frequeny in the tested frequeny range. Differenes in the Q fator of eah resonane from the projeted results in HFSS imply that fiberglass/epoxy omposite materials may be lossy. Figure 14 shows the HFSS analysis results for antenna 1 mounted on two substrates, both with a permittivity of 3.4 but two different loss tangents: Effet of Loss Tangent on Resonant Frequeny, E = 3.4 1 6 2E+09 4E+09 6E+09 Figure 14: Comparison of two loss tangent values for a given relative permittivity In this experiment, all measured permittivity values were ompared to HFSS-predited values with a loss tangent of zero. Further investigation into the dieletri properties of omposite materials should explore the effets of staking sequene and frequeny on the material s loss tangent as these materials may exhibit non-isotropi loss. ACKNOWLEDGMENT HFSS, tan = 0 HFSS, tan = 0.015 I would like to thank CReSIS and the National Siene Foundation for giving me an opportunity to ondut this researh. I would also like to thank Dr. Emily Arnold for mentoring me and providing me with valuable guidane throughout this projet.
My work on the time delay method was based on previous researh onduted by Dr. Stephen Yan. He also provided helpful advie throughout this projet regarding antennas and working in the anehoi hamber. REFERENCES 1. Demarest, K. (1997). Engineering Eletromagnetis. Upper Saddle River, New Jersey: Prentie Hall. 2. Farahmand, Farid. Introdution to Transmission Lines. Sonoma State University, Rohnert Park, CA. 2012. Leture. 3. ANSYS HFSS. (n.d.). Retrieved June 2015 from www.ansys.om 4. Dieletri Properties. (n.d.). Retrieved June 2015 from www.rohaell.om 5. Summary of Published Material Properties. (n.d.). Retrieved June 2015 from www.niar.wihita.edu 6. Kozakoff, Dennis. "Radome Dieletri Materials." Analysis of Radome-Enlosed Antennas. Norwood: Arteh House, 1997. 49. Print.