Magnetic properties of polymer matrix composites with embedded ferrite particles

Polymeric composite materials offer advantages for many applications because of a combination of properties, which includes high specific mechanical strength and elastic modulus and corrosion resistance. However, the non-magnetic nature of these materials impedes the use of nondestructive evaluation (NDE) techniques using magnetic sensors. In this work, glass fiber-reinforced epoxy magnetic composites were produced with the addition of 10 wt.% of cobalt or barium ferrite particles. Circular plates with notches of 1, 5 and 10 mm in diameter were produced and characterized using magnetic flux leakage (MFL) technique. The effect of particle size on the magnetic properties of the composites was also investigated for the barium ferrite. The results indicated a good correlation between the measured magnetic signals and the presence of notches. Smaller average particle sizes hindered the identification of the smallest notch.


Introduction
Polymer composite materials offer a combination of properties that makes them advantageous for many applications.These properties include low density, high mechanical strength and elastic modulus and corrosion resistance.Due to these unique characteristics, these materials are widely used in applications such as aeronautical and aerospace structures, sporting goods and various industrial applications.
Polymer composites are increasingly recognized as an enabling technology to meet specific and ever-increasing demanding requirements of the oil and gas industry.
As corrosion poses major challenges for conventional steel pipes, the inherent corrosion resistance of polymer composites combined with their superior mechanical properties and low density has made them the natural choice of material [1][2][3][4].Currently, many types of fiber-reinforced polymer materials are used in oilfields.While fiberglass has been the typical reinforcement, the predominant resins are polyester, vinyl ester and epoxy [1].
Damage in composite structures are known to affect their mechanical properties and functionality.In some cases, damage progression may ultimately lead to structural failure with potential consequences that include damage to property or the environment and even loss of human life.While some surface damages may be easily detected during routine inspections, internal damages that are often undetectable by simple visual inspection can cause significant reductions in strength and stability of the structure.Therefore, approaches for the detection of internal damages in polymer composite materials are of great interest.The use of nondestructive evaluation (NDE) techniques allows assessing the structural integrity of the component without damaging the materials.Typical nondestructive testing (NDT) techniques currently used to examine polymer composite materials include ultrasound, thermography, shearography and x-ray computed tomography (XR -CT) [5][6][7].
Shearography has been used to assess damages, including the detection of delamination, with various types of excitation such as temperature, vacuum and internal pressure [6].Thermography is a highly sensitive technique to detect delamination, but there are limitations to measurements on thick laminates [6].X-ray computed tomography (X-ray CT) provides high-resolution images and full 3D reconstruction of the assessed part and can detect defects such as voids, cracks, inclusions, dry fibers and delamination.In this case, the specimens to be tested may be thicker than those evaluated by thermography due to low x-ray attenuation of polymer composites.
Ultrasound have been widely used for the inspection of metallic parts [7].In composites, ultrasonic inspection can detect damages such as delamination, providing reliable information on relatively thick parts.
Magnetic Flux Leakage (MFL) testing is a widely used, non-destructive testing (NDT) method for the detection of corrosion, pitting, cracks and wall losses in steel structures, including steel pipelines [8].However, this technique is not suitable for polymer composites because the material cannot be magnetized.Thus, pipelines made with conventional polymer composites do not allow structural integrity assessment through nondestructive techniques (NDT) using instrumented pigs (pipeline inspection gauge), which operate with magnetic sensors.This is a major limitation of these materials in this type of application.
The addition of magnetic particles to polymer matrix composites offers great potential for the inspection of these materials using NDT that are sensitive to changes in magnetic field.Ferrites are ceramic magnetic materials with spontaneous magnetic induction [9], which can be used in a wide variety of technological applications.These materials are therefore natural candidates to be added to polymer composites.
The aim of this work was to develop polymer composites with the addition of ferrites as magnetic markers to allow the application of damage detection techniques using magnetic flux leakage and to study the influence of particle size on the magnetic properties of the material.Three moles of ethylene glycol per mole of metal ions were used for the synthesis.The gel was dried in a muffle furnace (Marconi, model MA385) to the primary calcination at temperature of 350°C for 3 h with a heating rate of 3°C/min and static oxidizing atmosphere to eliminate the organic material.This resulted in an expanded resin (spongy and dark colored product).The product was then milled to form fine and homogeneous cobalt ferrite powders.Subsequently, the powders were calcined at 800°C for 6 h in a muffle furnace using a heating rate of 10°C/min.

Preparation and characterization of the ferrite particles
Commercially available barium hexaferrite produced by Fermag Ltda -Brazil was milled to reduce particle size using a high-energy attritor mill at room temperature.
The milling process was conducted using zirconia balls in wet medium (isopropyl alcohol) for 12 h and 20 h.The process was interrupted for 30 min, at every 3 h, to minimize heating.The different milling times were used to allow the evaluation of effect of particle size on the magnetic properties of the composites.
The crystalline structure of the ferrites was analyzed by X-ray diffraction (XRD) using a XRD -7000 Shimadzu diffractometer with CuKα radiation source.Crystallite sizes were obtained with Scherrer's equation [10].The Rietveld method was used to refine the XRD data [11] using the MAUD (Materials Analysis Using Diffraction) program [12].Ferrite particles were also observed by field emission scanning electron microscopy FEG SEM (FEI Inspect F50) operating at an acceleration voltage of 20 kV.
Based on the FEG SEM images, average particle size and size distribution were estimated using a minimum of 15 particles in a typical sample.Magnetic measurements were performed at room temperature using a vibrating sample magnetometer (VSM).

Fabrication and characterization of the composite plates
Circular plates of 250 mm in diameter (Figure 1) were manufactured by hand layup for magnetic characterization.The materials used were: fiberglass 0/90 biaxial knitted fabric type KT C1000 SE 1500 (975 g/m 2 ) from Owens Corning, epoxy resin type AIRSTONE 780E from Dow Chemical Company and ferrite particles.The epoxy resin was mixed with ferrite particles using a mix ratio in parts by weight (pbw) of 90:10 (resin:particles).Composite materials with magnetic particles fillers have been reported in the literature with particle contents up to 60 wt.% [13,14,15].In this work, the particle content was limited to 10 wt.% in order to maintain the low viscosity of the epoxy, which is an important advantage in thermoset polymer composite manufacturing.Higher particle content have also been reported to reduce some mechanical properties of polymer composites, such as tensile strength.The evaluation of mechanical properties was not included in the scope of this work.
A mechanical mixer (Marconi MA147) with speed of 1500 rpm was used to disperse the ferrite particles in the polymer matrix over a period of 15 min.Then, the fiberglass fabric was impregnated with epoxy resin combined with ferrite particles and three layers were laid-up at [0/45/0], respectively, on a flat tool previously prepared with mold release agent.This staking sequence results in laminates with fibers at 0°, 90°, 45°, and -45° which are commonly used in structural applications.
The laminate was then vacuum bagged and cured at about 25 °C.Circular notches were drilled on the composite plates using drill bits of 1, 5 and 10 mm, respectively, for subsequent magnetic characterization.The sizes of the notches considered in this work are of the same order of magnitude as those typically produced by corrosion found in both internal and external surfaces of metallic pipes carrying crude oil.Overall, four circular plates were produced according to the type of ferrite added: cobalt ferrite, barium ferrite "as received", barium ferrite milled for 12h and barium ferrite milled for 20h.Field emission scanning electron microscopy images of the composite plates with magnetic particles were obtained using an FEI Inspect F50 to evaluate particle dispersion.
Variations in magnetic field on the four polymer matrix composite plates were measured using a custom-made apparatus (Figure 2).The plates were first mounted on a stepper motor in which it was possible to control the speed and direction of spin.A magnet and a Hall sensor were kept fixed near the rotating plate to measure the vertical component of the magnetic field.The Hall sensor with three pins and a sensitivity of 5 mV was connected to a nanovoltmeter and a current source.The Hall sensor detected the magnetic field produced by the magnetized composite plate along with distortions produced in the field by the notches.There are at least two reasons for the measured variations in magnetic flux lines at the notches.First, as observed in stainless steel, the flux lines are more concentrated at the notches because of the reduction in thickness that, in turn, increases locally the magnetic field and saturates the local magnetization.Secondly, a smaller component of the magnetic field produced by the magnetic particles is detected at the notch area due to the increased sensor-to-surface distance.Data were acquired via a LabView-based computer controlled data acquisition system using a GPIB interface.In order to verify the repeatability of the measurements, the direction of rotation of the circular plates was reversed over the limits of each notch using the stepper motor controller.Symmetrical curves about the turning point indicate a clear correlation of the notch with the deformation of the magnetic field lines over the region.

Current source
Stepper motor

Circular plate
Hall sensor

Results and Discussion
Figure 3 shows the X-ray diffraction pattern for the cobalt ferrite after heat treatment at 800 °C for 6 h.Peaks related to the CoFe 2 O 4 phase confirm the cobalt structure (22-1086).However, the additional peaks observed suggest the presence of other phases on the cobalt ferrite structure.In this case, the peaks correspond to the following impurities: hematite (Fe 2 O 3 ) and cobalt oxide (Co 3 O 4 ), with JCPDS -ICDD file numbers of (87-1165) and (80-1335), respectively, as determined using the X'Pert software.Average particle size and crystallite size of ferrite particles are shown in Table 1.For CoFe 2 O 4 , the crystallite size determined was 256.53 nm.The literature reports crystallite size for cobalt ferrite [17] when synthesized by another method at temperature of 800°C/6h, obtaining crystallite size of 102.32 nm.This larger average crystallite size is due to the effect of diffusion, which occurs between ions at high temperatures, since this favors the coalescence of grains, forming larger crystals.
Although Scherrer equation is known to be applicable to crystallite sizes below 200 nm, the value measured is an indication that the crystallite size of the CoFe 2 O 4 produced is larger than that obtained by a different synthesis method published in the literature [17].
FEG SEM micrographs of the ferrites are shown in Figure 5.For the CoFe 2 O 4 sample, the material obtained is composed of coarse clusters and of finely divided particles (Figure 5 a).This can be attributed to the characteristic of the carbonized material, which has been first disaggregated in a mortar, yielding particles with a broad distribution of cluster sizes and that preserves the size distribution after calcination.
Thus, clusters with estimated sizes ranging from 10 μm to 2 μm associated with particles of sizes ranging from 1 μm to 300 nm were observed.The estimated average particle size in this case was 340 nm.
For the BaFe 12 O 19 sample before milling (Figure 5 b), the micrograph indicates that the original barium ferrite is composed by clusters of tightly bound smaller particles.These clusters were found to have an average size of about 13 μm consisting of particles of about 750 nm in size.These characteristics are related to the manufacturing process of these ferrites which were obtained by solid state reaction using high heat treatment temperatures, typically above 1200 °C.This process results in the formation of the desired phase, as well as powder sintering.
After the milling process, clusters of finer particle sizes are obtained, as shown in Figure 5 a) and b).The powder showed average particle size of about 670 nm and 400 nm, after milling for 12 h and 20 h, respectively.This suggests that the milling 13 process was effective in breaking the clusters and reducing the particle size of the original material.However, the presence of some particles with angular shape and average size of around 2.30 μm was verified.These particles are remaining fragments of the original unmilled particles.It was also observed that the milling process of barium ferrites resulted in the formation of hematite phase, as previously discussed and shown in Figure 4.The formation of this phase was explained by contamination of the milling system (polyamide 66 -Technyl ® ball and support) or broken bonds between Fe atoms during the milling process.The presence of hematite may have influenced the saturation magnetization measurements.
For BaFe 12 O 19 sample (before milling), the coercivity value was 0.25T.After the milling process for 12 h and 20 h, this value was reduced to approximately 0.19T.
This can be explained by the fact that a smaller coercivity implies in a smaller saturation field and easier magnetization of the material.This effect was confirmed by the characterization of these materials by MFL.respectively.In figure 7 c), which corresponds to the composite with BaFe 12 O 19 (milled for 20 h), it can be observed that the milling process favored not only the reduction of particle size but also their dispersion in the polymeric matrix.Uniform dispersion of the magnetic particles on the composite is essential for damage detection since areas without magnetic particles may produce a false indication of damage.The curves of variation in magnetic field presented in Figure 8 show no change in magnetic signal over the notch area for composites with the addition of barium ferrite milled for 12 to 20 h.Thus, the milling of magnetic particles did not favor the detection of smaller notches (1 mm).For notches of 5 and 10 mm (Figures 9 and 10), the beginning and end of the notches can be clearly identified by the arrows on all four polymer composite plates.
For all measurements (Figures 8 to 10) there was a greater variation in magnetic signal over the notches for composite plates with cobalt ferrite and barium ferrite without milling.This can be explained by the higher magnetization of these samples which results in higher equipotential lines and consequently greater magnetic flux leakage, thereby contributing to higher resolution in detecting the notch, as evidenced by the hysteresis curves of these materials (Figure 6).However, the curves presented in Figures 9 and 10 indicate that the smallest particle size allowed producing smoother magnetic curves with fewer discontinuities.This can also be related to a more homogeneous particle distribution within the polymer matrix.Moreover, the magnetic field curves of composites with barium ferrite of smaller particles showed peaks only over the regions of the notches, which did not occur for barium ferrites without milling or for cobalt ferrites, i.e., it allowed better signal-to-noise ratio.

Conclusions
In this study, fiberglass reinforced polymer matrix composite circular plates with the addition of ferrites (10 wt.%) were produced for magnetic characterization using magnetic flux leakage (MFL) technique.Barium ferrites were added to the composites with different particle sizes: without milling and milled in a high-energy mill for 12h and 20h.In addition, composite plates were manufactures with cobalt ferrites synthesized by the Pechini method and the results were compared with those obtained with barium ferrite.Circular notches of 1, 5 and 10 mm in diameter were introduced on the composite plates to investigate the potential use of the magnetic characterization technique for the identification of defects and damages in polymer composites.
According to the results obtained, changes in the magnetic signal of samples with 10 wt.% of cobalt ferrite were reproducible and characteristic for all notch sizes studied.For composites with barium ferrite, 5 and 10 mm notches were detected with all particle sizes analyzed, and the magnetic signal showed good correlation with the location and size of the notches.However, the notch of 1 mm was detected only for larger particle sizes (micrometric clusters formed by particles of approximately 0.75 μm).For composite plates with barium ferrite milled for 12 and 20 h, 1 mm notch did not produce a significant variation in the magnetic field, suggesting that the milling of particles did not favor the detection of smaller notches.Nevertheless, the smaller particle size allowed producing smoother magnetic curves with variations only over the notch regions and therefore a better signal-to-noise ratio.Moreover, smaller variation in the baseline is expected with narrower particle size distribution and better particle dispersion in the polymer matrix.The overall results indicate that the proposed technique has great potential for the detection of defects and damages in polymer matrix composites by using non-destructive magnetic testing.

Highlights
 The potential of magnetic flux leakage testing for NDE of polymer composites with embedded ferrite particles was assessed. Changes in magnetic signals were reproducible and showed good correlation with location and size of notches in the material. Smaller ferrite particles did not favor the detection of smaller notches, but resulted in better signal-to-noise ratio.

Figure 1
Figure 1 Polymer matrix composite plates with notches.Stickers on each side

Figure 2
Figure 2 Custom-made apparatus for the magnetic measurements.

4 Figure 3 X
Figure 3 X-ray diffraction pattern of the cobalt ferrite.

Figure 4 Figure 4 X
Figure 4 shows X-ray diffraction patterns of barium ferrites "as received" and

Figure 6
Figure 6 Graphic representation of the magnetization as a function of the magnetic field

Figure 7
Figure7shows FEG SEM micrographs of the polymer matrix composites with

Figure 7 FEGFigure 2 .
Figure 7 FEG SEM micrographs of the polymer matrix composite with the addition of

Figure 10
Figure 10 Magnetic field curves (mT) versus distance measured in composite plates

Table 1
Particle size and crystallite size of ferrite particles.
[16]rding to the data presented in Table1, the milling process was effective in reducing the particle size of BaFe 12 O 19 ferrites from an average of 750 nm to 670 nm, after 12 h, and to 400 nm after 20 h milling time.For CoFe 2 O 4 , the average estimated particle size was 340 nm.For BaFe 12 O 19 samples, it was observed that the milling process reduced the average crystallite size to the nanometer range, in agreement to results previously reported in literature[16].Therefore, there was a reduction from 181.76 nm (starting Magnetic field curves (mT) versus distance measured in composite plates with CoFe 2 O 4 , BaFe 12 O 19 (no milling), BaFe 12 O 19 (12 h) and BaFe 12 O 19 (20 h) at the 1mm notch.Arrows indicate the beginning and end of each notch.