THE ANALYSIS OF ADHESIVE BONDS USING ELECfROMAGNETIC ACOUSTIC TRANSDUCERS S.Dixon, C.Edwards, S.B.Palmer Dept of Physics University of Warwick Coventry CV 4 7 AL INTRODUCfION EMATs have been used in ultrasonic based research for over 20 years [1,2]. Improvements in permanent magnet technology and the required supporting electronic circuitry have made EMATs a realistic alternative to piezoelectric transducers in many applications. EMATs are limited to working on electrical conductors and are roughly three orders of magnitude less sensitive than PZT, so generally there has to be a special reason to use an EMAT in preference to a PZT. The scanning of an adhesively bonded plate is one of these areas where the EMAT should offer an improvement over tests done with conventional PZT transducers. The EMAT is a broadband device ( 5MHz bandwidth centred on 5MHz), giving it high spatial resolution in the direction of propagation, and is also a totally non-contact device, so no acoustic couplant is required. The most attractive feature of the EMAT used here in this work is that it generates radially polarised SH waves which should be more sensitive than longitudinal waves at normal incidence to a bonded interface. The design and construction of the EMAT will first be discussed and this will be followed by the presentation of various results obtained using the EMAT systems. EMATs have been used to probe bond integrity in the past [3,4], but have to the best of the authors knowledge not been used in a broadband, true pulse-echo mode. RADIALLY POLARISED SHEAR WAVE EMAT The basic EMAT construction is shown in figl. The EMAT spiral coil was 16mm in diameter, and etched into printed circuit board to provide reproducibility and durability. The coil impedance at around 5MHz was closely matched to the preamplifier input impedance, and the impedance of the coaxial cable used to connect Review of Prog ress in Quantitative Nondestructive Evaluation. Vol. 14 Edited by D.O. Thompson and D.E. Chimenti. Plenum Press. New York. 1995 1497
the transducer to the pulse/receiver circuitry. The magnetic field normal to the plane of the spiral coil was provided by a NdFeB permanent magnet. A steel backing cylinder was used to lower the demagnetising factor and increase the field at the surface of the magnet from 0.35T to O.5T. The magnet and coil was cased in a brass housing which also provided a shield from electromagnetic noise. The front of the EMAT needed some form of protection and electrical insulation from the conducting sample. In this case, a thin layer of plastic film was used (loomm thick), but in cases where the EMAT is to be used on hot samples (upto 5000 C with water cooling) both mica and ceramic discs have been used to protect the EMAT face. The EMAT can then be used as a detector or generator of radially polarised shear waves and in send-receive mode where the transducer is simultaneously used as both generator and detector. Another type of spiral coil EMAT has been used in a sendreceive mode which uses a generation coil concentric with a detection coil, in this instance the coils being would from O.2mm diameter wire. This design is more akin to the piezoelectric twin crystal probes, and can be directly used with a two channel flaw detector only requiring an external preamplifier. All of the EMATs described here are comparable in size to typical PZT transducers, and are compatible with existing technology. EXPERIMENTAL DETAILS Adhesive bonds were constructed from two dissimilar thickness aluminium plate adherents (approx 3.34mm and 4.83mm thick respectively), bonded by an aerospace epoxy adhesive. Bonds were analysed in both through transmission (IT) and sendreceive (SR) geometries. IT requires access to both sides of the bond, but is more sensitive than SR. SR on the other hand only requires access to one side of the bonded sample, but is not as sensitive as IT. earthing pin 16mrn active area spiral coil on PCB mild steel backing NdFeB permanent magnet brass housing Figure 1. The cross section of the radially polarised shear wave EMAT. The coil is a flat spiral 'pancake' type, etched onto PCB. 1498
The acoustic pulse generated by the EMAT described here were broadband. Thus, the ultrasonic signal had a short time duration. giving it a high spatial resolution in the direction of propagation. This allowed the direct resolution of echoes within the adhesive layer which is typically loollm thick. A SR waveform obtained from a bonded sample is shown in fig.2. A thicker adhesive layer than normal was used here (approx 250mm), in order to clearly separate the features of the pulse reflected form the near interface, and that which reverberated within the adhesive layer travelling an extra 2 adhesive thicknesses than the proceeding pulse. This is shown in fig.3. From this window, amplitude measurements and time between the principal pulse and the adhesive echo were recorded. The amplitude of the first pulse contains information about the reflection coefficient at the near adherent-adhesive interface. The bond reverberation amplitude contains information of the near interface reflection coefficient, adhesive bulk and the far interface reflection coefficient. Experiments were also performed on a 2 part epoxy adhesive curing between 2 aluminium plates to form an adhesive bond. Both amplitude and shear wave velocity measurements in SR and TT geometries were performed simultaneously on the curing sample. 1.2 -E ":::'-1.0 0) '0. 0.8 c.. ~ 0.6 0.4 4 '6 8 10 time (1lS)', Figure 2. Adhesive bond SR waveform. ",,, 1.2 1.1 ---- -E 1.0 '--' '" Q.) 0.9 '0 ::I.'.:: 0.8 c.. a 0.7 '" 0.6 0.5 "-:"'::-_--:-'':-_-:-'-:-_--:-'~_-'- L..:j 10 32 3.4 3~ 3B 4.0 time (IlS) Figure 3. pulse reflected from far interface and adhesive bond reverberation. 1499
RESULTS The pulse amplitude c-scans for a defective sample containing a void region and a sample containing a grease contamination on one adherent are shown in figs 4&5 respectively. The void would have easily been detected using a conventional immersion test or couplant requiring piezoelectric transducer type scan, but using the EMAT none of the practical problems associated with the need for a couplant were encountered. The grease contamination is shown where the defect is on the far adhesive-adherent interface and is clearly visible as a drop in signal amplitude. During the monitoring of the epoxy cure, the amplitude of the pulse reflected back from the near interface was recorded (SR data) and the amplitude of the first shear wave arrival in TT was also recorded (see figs 6&7). These amplitude measurements appear reasonable when compared to those obtained by other workers[ 4,5]. Using EMAT shear waveforms obtained from an aerospace adhesively bonded sample and a blank adherent the phase velocities in the viscoelastic [6] epoxy layer (fig.8) and in the aluminium plate (fig.9) were calculated as described by Sachse[7]. AMP. (arb) Figure 4. Amplitude c-scan of back reflected shear wave on voided sample. AMP. (arb) Figure 5. Amplitude c-scan of back reflected shear wave on silicon grease contaminated sample with the defect on the far interface. 1500
..,.. 18.5..-... of 18.0 :, =I : --- Q,I '3 17.5. l' "17.0 =I 'i 16.5 ~ ~ 16.0 o...,.. ::.. -... -~.,J-..:"-,/.. \le."",...."'y:....'.. ~, v~,..r 5 10 15 20 25 30 time (hours) Figure 6. Peak-peak amplitude of the reflected ultrasonic shear wave arrival from the near adherent-adhesive interface in send-receive. 25~--~----~--~----~--~----~ 15 10.!, ~.. -.~ ---' ~.~ /'.-J,;v.N"'. ~ 5,; i 1-' 0~---5~--~1~0--~15~--2~0----2~5----3LO~ time (bours) Figure 7. Peak-peak amplitude of the first ultrasonic shear wave arrival in through transmission. 1501
1200r-------r------,,_------r-----~ ~ 1150 E '-".c. g 1100 a:i >" Q,) CIl "" -a 1050......... 1000~----~~----~------~------~ 2 4 6 8 10 frequency (MHz) Figure 8. Phase velocity of shear waves in the epoxy layer calculated from broadband signals. 3130~---r--~,---~--~----~---.~,--. -'!!. E '-".c u 0 a:i > Q,) CIl..c:: "" c.. 3120 3110-3100,,, phase velocity from IT data phase velocity from SR data 3090 2 4 6 8 10 12 14 frequency (MHz) Figure 9. Phase velocity of shear waves in aluminium calculated from broadband signals. 1502
CONCLUSION It has been shown that the EMAT system described in this paper has been successfully implemented in the analysis of adhesively bonded aluminium plates. Broadband radially polarised shear waves have been generated in a non-contact regime, making scanning far easier compared to instances where acoustic couplant is required. The system has also been used to monitor the state of cure of a 2 component epoxy between 2 aluminium adherents, showing the expected behaviour, but once again without any of the problems or considerations that occur when using coupled ultrasonic transducers. REFERENCES 1. H. M. Frost, Electromagnetic-ultrasonic transducers: principles, practice and applications in: Physical Acoustics XIV (1979), (Ed. W. P. Mason & R. N. Thurston), Academic Press, London, p179 2. K. Kawashima, Quantitative calculation and measurement of longitudinal and transverse ultrasonic wave pulses in solids, IEEE Transactions on Sonics and Ultrasonics (1984) SU-31 (2), p83 3. D. A. Hutchins, M. D. C. Moles, G. S. Taylor and S. B. Palmer, Non-contact ultrasonic inspection of diffusion bonds in titanium, Ultrasonics (1991) 29, pp 294 4 F. He S.l. Rokhlin and L. Adler Application of SH and Lamb wave EMATs for evaluating adhesive joints, Rev. of Prog. in QNDE (1987), p911 5. S.l. Rokhlin, D.K. Lewis, K.F. Graff and L. Alder, Real-time study of frequency dependance of attenuation and velocity of ultrasonic waves during the curing reaction of epoxy resin, J. Acoust. Soc. Am. 79 (1986), p1786 6. W. Philippoff, Relaxations in polymer solutions, liquids, and gels in: Physical Acoustics lib, (Ed. W. P. Mason & R. N. Thurston), (1979), Academic Press, London, pp. 1-87 7 W. Sachse and Y. Poa, On the determination of phase and group velocitis of dispersive waves in solids, J Appl Phys. 49(8) (1978), p4320 1503