CHARACTERIZATION OF DIFFUSION IN EPOXY RESIN USING ATOMIC
FORCE MICROSCOPY ANALYSIS
Olivia G. Blank1, Lewis M. Cox2, David A. Miller3
Montana State University
Bozeman, MT, 59715
ABSTRACT
Composites are an essential part of manufacturing turbine blades for hydroelectric power
systems. These blades undergo extreme loads and must be able to operate for at least 20 years.
Composites offer several advantages over metals in these applications, including but not limited
to material cost, manufacturing cost, and a high strength-to-weight ratio. This study investigated
the impact of water saturation on the polymer matrix using atomic force microscopy (AFM). The
samples studied were manufactured using MGS RIMR 035c resin and conditioned at 50 °C in
distilled water until saturation was reached. Initial dimensions and mass of each sample were
recorded, and moisture uptake was observed over 30 days. AFM analysis revealed that the elastic
modulus of the neat resin decreased in areas affected by moisture diffusion, and as saturation was
reached, the elastic modulus further decreased and became uniform throughout the sample
thickness. The diffusion properties calculated in this study, as well as results from the AFM
analysis, can be used to aid and validate computer modeling efforts of composite systems in
marine environments.
Keywords: atomic force microscopy, diffusion, glass fiber composites, epoxy resin
Corresponding authors: David Miller
1. INTRODUCTION
The need for renewable sources of energy generation will continue to grow as environmental
impacts from coal and oil come to a head [1]. Solar, wind, and waterpower are the main
contenders in the clean energy sphere, and the research surrounding them is ever evolving. The
focus of this paper surrounds the impacts of environmental conditions on marine hydro-kinetic
structures (MHK), or structures that generate energy from the natural movement of water, such
as waves, tides, and currents, without the need to divert waterways or construct new dams [2].
These structures have a large amount of flexibility when it comes to design and manufacturing.
However, due to the harsh conditions of marine environments, there are limitations on materials
that can be used when manufacturing certain components.
A prominent material used in manufacturing MHK devices are glass fiber reinforced polymers.
Some benefits that these types of polymers have over conventional materials, such as metals, is
the corrosion resistance, increased strength to weight ratio, and design flexibility [3]. However,
composite systems are not without their challenges. The behavior of these materials is highly
variable depending on manufacturing processes, as well as glass fabric and resin selection. The
unpredictable behavior of composite materials is further exacerbated when environmental factors
must be considered, specifically moisture effects. Polymer matrix degradation from moisture,
paired with variable loading cases, is detrimental to the lifetime of MHK structures. Expanding
research is essential in developing predictive models for these devices [4].
Copyright 2024. Used by the Society of the Advancement of Material and Process Engineering with permission.
SAMPE Conference Proceedings. Long Beach, CA, May 20-23, 2024. Society for the Advancement of Material and Process
Engineering – North America.
(DOI will be added by SAMPE)
1.1 Moisture Effects on Polymers
When absorbed into a polymer matrix, water can potentially exist in a few different forms. The
water can exist in its molecular form, nested within the polymer microstructure. Water molecules
can also create hydrogen bonds with the polymer chains. Both of these molecular forms are
responsible for swelling in the polymer, as they increase the distance between the chains. Liquid
water, on the other hand, fills voids or holes that exist in the microstructure, and do not
contribute to swelling effects [5]. A graphic depicting these forms of moisture can be seen in
Figure 1. Liquid water can still influence mass uptake values when measuring absorption
properties, since it is trapped in the system, but not truly absorbed into it. This phenomenon
becomes more apparent as the polymer reaches saturation due to new voids forming as
hydrolysis within the polymer microstructure occurs.
Figure 1: Schematic from Jakes et al. 2019 [5] showing the different forms of water in a water plasticized
polymer. The three forms of water are hydrogen-bonded water, molecular water that has diffused
between polymer chains, and liquid water that fills holes.
1.1.1 Plasticization
Plasticization is a process which changes both the thermal and mechanical properties of a
polymer through the addition of a new compound, which increases mobility of polymer chains
within the system [6]. In this case, the compound that causes plasticization is the water
molecules that are diffusing into the polymer matrix.
This phenomenon is especially relevant to the stiffness of a polymer or composite system
containing a polymer matrix. The moisture that has diffused into the system degrades the
strength of the material, which relates directly to the degradation of the elastic modulus.
Predicting polymer degradation becomes more difficult when considering heterogeneous
materials such as composites. Composite materials contain reinforcement agents; therefore, the
uniformity of diffusion is no longer a valid assumption, especially with complex geometries.
Measuring the moisture with a complex composite part can be done easily on a bulk scale
through moisture mass uptake, but as growth continues with computer modeling efforts of
diffusion in composites, more precise methods may be needed.
1.2 Atomic Force Microscopy (AFM)
Atomic force microscopy (AFM, Cypher S, Oxford Instruments) is a useful tool for measuring
properties topography, adhesion, stiffness of a material’s surface, and many other properties [7].
The AFM uses a micromachined cantilever with a sharp probe tip (radius of curvature 10-20 nm)
to interrogate a surface with nanometer spatial resolution. The sub-nanometer deflections of the
cantilever as the probe tip rasters across the sample surface are measured with a detection laser
reflected from the top surface of the tip onto a photodiode array [7], [8]. A schematic of typical
AFM operation can be seen below in Figure 2.
Error! Reference source not found.
Figure 2: Schematic of AFM operation in static mode [8].
1.2.1 Force-Displacement Curves
The elasticity of a sample surface is measured using force curves. During force curve
measurement, the probe tip starts off the sample surface, approaches at a prescribed rate, presses
against the sample surface until a specified trigger is met, and then retracts. A force setpoint for
this process can range in magnitude from nanonewtons to micronewtons. The resulting force
curves are fit with a contact mechanics model to calculate the contact modulus of the sample,
analogous to a quasi-static Young’s modulus. We employ a standard Hertz model to fit the
curves, consistent with literature on glassy resins [8], [9], [10].
2. EXPERIMENTATION
2.1 Moisture Absorption
Samples for moisture uptake were manufactured by casting and curing MGS RIMR 035c epoxy
resin in a small metal tin at 40 °C for 24 hours, and then post-curing the samples at 70 °C for 4
hours. These samples were cut to approximately 25 mm x 25 mm, with a thickness ranging from
6-8 mm across the 10 samples. They were dried out in an oven at 50 °C for 7 days to ensure the
removal of all moisture absorbed into the polymer during manufacturing. ASTM D570
recommends only a 24-hour drying period; however, this was modified to a 7-day period for this
study. The mass of each sample was measured using a Cole-Parmer Symmetry PA-Series
Analytical Balance after allowing the moisture on the sample surface to fully dry (30 minutes).
The samples were weighed at 1-5, 10, 20, and 30 days following the procedure outlined in the
ASTM D570 standard [11]. Additional data points will be collected after the 30-day mark, but
for the purpose of this study, these data points will be sufficient in calculating the diffusion
coefficient.
2.2 AFM Sample Preparation
The samples used in this study were manufactured using the same epoxy resin system as the
moisture diffusion study. The epoxy was cast and cured at 40 °C for 24 hours, and then post-
cured at 70 °C for 4 hours. The cast plate of epoxy was cut into a thin strip (7.79 mm x 59.33
mm x 2.15 mm) and placed in distilled water heated to 50 °C. An elevated temperature was
chosen for sample soaking to increase the rate of diffusion into the epoxy system.
The resin sample was measured at 1-5, 10, 20, and 30 days of soaking. At each time of
measuring, approximately 2 mm of the coupon was cut using a clean razor blade. The blade was
pressed into the resin coupon using a vice, which was tightened until the blade cut through the
sample. The internal side of the cut section was hand-polished from 800 to 2400 grit paper while
maintaining a flat surface for measurement by the AFM.
The sample was then press-fit into a sample holder that held a control resin sample (unsoaked)
and the moisture conditioned sample. A sample preparation diagram can be seen in Figure 3,
along with a prepared AFM sample in Figure 4.
Internal surface
Control
Water conditioned
resin sample
Sample holder
Figure 3: AFM sample preparation diagram. All soaked samples were cut from the same resin coupon,
with the internal portion mounted upwards.
19 mm
Figure 4: AFM sample
2.2.1 AFM Probe Tip Selection and Calibration
Cantilevers (RTESPA-525, Bruker AFM Probes) with a sufficiently high stiffness were selected
to indent the glassy resin. The cantilevers were first calibrated using the Sader method,
approximating their shape as a simple rectangle. Potential error in this approximation was
accounted for with a second calibration step using the control samples as a reference. The
modulus of the control sample was established to be 3 GPa, consistent with the value of the 035c
resin as reported by the manufacturer, and small adjustments (within the error bounds of the
Sader method) to the cantilever calibration parameters were made to ensure agreement with this
value during force curves.
2.2.2 AFM Procedure
Cross-sectioned samples were mounted face up under the AFM and the probe was positioned
over the edge of the cross-section. Cross-section edges were located using tapping mode, then
imaged with force maps. Force maps (5 x 5 µm area, 16 x 16 grid) were collected first at the
edge, then shifted inward, while assuring some overlap between maps is present so that distance
from the edge can be tracked with progressive scans. The control sample was scanned first to
complete the second step of the calibration (Section 2.2.1), and then periodically re-scanned after
imaging treated samples to control for any potential tip wear.
3. RESULTS
3.1 Mass Uptake of MGS RIMR 035c Resin
Figure 5 shows the moisture uptake of the RIMR 035c epoxy resin, plotted with respect to the
square root of time. As the samples soaked for longer periods of time, the spread of the data
began to increase, as seen in the range bars on the plot. This phenomenon may be linked to the
development of cracks in the polymer, causing liquid water to become trapped during the
soaking process. During the analysis of data, it was assumed that all moisture gain was due to
molecular water diffused into the polymer microstructure. If liquid water was trapped and unable
to evaporate during the drying process, the additional mass would artificially increase the
measured mass uptake values.
Figure 5:Moisture uptake vs square root of time plot for RIMR 035c epoxy resin.
3.1.1 Diffusion Coefficient Calculation
A diffusion coefficient for RIMR 035c epoxy resin can be calculated from the data used to
generate the plot in Figure 5. The method used for calculating an effective coefficient at 50 °C
utilized Equation 3.1 from Shen et al. [12].
ℎ " 𝑀" − 𝑀# "
𝐷 = 𝜋$ ( $ ( [3.1]
4𝑀! √𝑡" − √𝑡#
∆𝑴
Each sample data set was linearly fit, and the slope, ∆√𝒕 , was used to calculate the diffusion
coefficient for each sample. The average height (h) of each sample was used in the calculations,
and 𝑴! was approximated to 2 % moisture uptake.
Table 1
Property MGS RIMR 035c
!
Diffusion Coefficient (𝑚𝑚 /𝑠) 1.80 E-06
Standard Deviation (𝑚𝑚! /𝑠) 1.37 E-07
1.70 E-06
Confidence Interval (𝑚𝑚! /𝑠)
1.90 E-06
T-Value 41.53
Degrees of Freedom 9
3.1.2 Fick’s Law: Depth of Moisture
In Equation 3.2, a solution to Fick’s Second Law is presented. The equation uses a constant
surface concentration of 100 %. As the samples at an initial concentration of 0 % are then fully
submerged into water, an approximation of moisture concentration for different depths can be
determined.
𝐶(𝑥, 𝑡) − 𝐶( 𝑥
= 1 − 𝑒𝑟𝑓 $ ( [3.2]
𝐶) − 𝐶( 2√𝐷𝑡
When these values were applied to Equation 3.2 with the diffusion coefficient listed in Table 1,
concentration values were calculated as seen in Table 2.
Table 2
Diffusion Depth (mm) Time (hours) Moisture Concentration (%)
0.001 96 99.93
0.001 480 (20d) 99.97
1.075 96 33.52
1.075 480 (20d) 66.65
At 20 days, the sample had not yet reached full saturation, as the center (1.075 mm depth) was at
66.65 % moisture concentration. The sample with 1 micron depth was close to 100 % at 96
hours, but there was still an increase in concentration from 96 hours to 20 days. The effect of
moisture concentration with respect to plasticization and modulus degradation is further
discussed in Section 3.2.
3.2 Elastic Moduli Degradation
3.2.1 Data Analysis
Modulus maps were progressively collected starting at the sample edge and moving inward.
Maps were stitched together by correlating topographic features in the overlapping scan areas.
Outlier values, defined as pixels which fall outside two standard deviations from the average, are
considered to be artifacts arising from sharp topographic features and removed from analysis.
The stitched together modulus maps for the 96-hour-soaked samples can be seen below in Figure
6. The map was overlayed onto the error channel from the AC scans taken to show where the
outlier data was located.
Figure 6: Stitched together FMAP of the 4 scans taken from the 96-hour-soaked sample. FMAP was
overlayed on the respective error channel (AC mode) for that sample to show areas where data was
rejected.
From there, the force map data was processed using MATLAB code that averaged each row of
moduli and incorporating the overlapping regions from the scanning offset. Figure 7 shows a
visual of how the data from each force map was offset and subsequently stitched together.
Figure 7: Force map showing overlapping scans. These overlapping regions were accounted for in the
averaging of line-by-line data.
3.2.2 96 Hours Soaked
Figure 8 shows the elastic moduli vs depth of measurement for the two edges measured. A
second order polynomial fit was applied to show the trend in the data as cross-sectional depth
increased. The dashed line on the plot represents the average control modulus measured during
this period (2.87 GPa).
Elastic Modulus of 96 hr Soaked Resin Sample
3.5
3
Modulus (GPa)
2.5
Average Edge Modulus
1.5
Poly2 Best Fit
control
1
0 2 4 6 8 10
Depth of Cross-section (microns)
Figure 8: 96 hour-soaked sample moduli vs depth into cross section
3.2.3 20 Days Soaked
Figure 9 presented below shows the moduli values of a 20-day-soaked sample. Similarly, to the
96 hours sample, the data was fit using a second order polynomial to show the trend in data as
cross-section depth increased. Edge 2 of Figure 9 has several data points that are cause for
scrutiny as they do not follow the expected trend. This could be a result of several factors, the
most likely one being topological features on the sample surface skewing force data. What is
most important with this plot is the increase in distance from the control sample (~2.97 GPa).
This gap widening shows the effect of plasticization of the polymer at the micro scale and paired
with the information calculated in Table 3, indicates that the time component of plasticization is
a major component in material degradation.
Elastic Modulus of 20 day Soaked Resin Sample
3.5
3
Modulus (GPa)
2.5
1.5 Average Edge Modulus
1
Poly2 Best Fit
control
0.5
0 2 4 6 8 10
Depth of Cross-section (microns)
Figure 9: Elastic modulus degradation of 20 day-soaked resin sample vs cross-sectional depth
4. CONCLUSIONS
As the soaking time of the polymer increased, the elastic modulus at the surface is seen
decreasing from ~3 GPa to ~1 GPa. The bulk effects of saturation in the resin can be seen in the
trendline getting further away from the control resin value, implying that the increase in moisture
concentration is causing the plasticization effects that degrade the polymers stiffness. Further
analysis of the additional measurement times is needed to show a true correlation but was unable
to be completed at this time due to time constraints.
4.1 Broader Impacts
The broader goal of this work was to show that, on a nanoscale, material degradation can not
only be measured, but has the potential to be applied further to computer modeling validation of
composites. Extensive work must still be done to develop a constitutive relationship between
moisture concentration and modulus mapping, but this study shows that the there is a tangible
trend in modulus degradation gradients in polymers.
4.2 Future Work
For the expansion of this work, additional trials must be conducted with the previously
developed procedure, with an increase in both number of edges measured as well as an increase
in number of scans from the sample edge. From there, concentration profiles would be plotted
with respect to experimentally determined modulus values and a resulting constitutive model will
be generated.
4.3 Limitation of Study
The limitations of this study lie within the development of the AFM procedure during the
measurement time frame. This led to inconsistency at the beginning, specifically with how the
edge of the sample was identified. Along with this, the exterior surface of the soaked samples
was not scanned, which would have greatly increased the confidence of the modulus value at the
edge being accurate.
5. REFERENCES
[1] P. A. Owusu and S. Asumadu-Sarkodie, “A review of renewable energy sources,
sustainability issues and climate change mitigation,” Cogent Eng., vol. 3, no. 1, p. 1167990, Dec.
2016, doi: 10.1080/23311916.2016.1167990.
[2] “Marine and Hydrokinetic Technology Development and Testing | Department of
Energy.” Accessed: Jan. 09, 2024. [Online]. Available:
https://www.energy.gov/eere/water/marine-and-hydrokinetic-technology-development-and-
testing
[3] F. Rubino, A. Nisticò, F. Tucci, and P. Carlone, “Marine Application of Fiber Reinforced
Composites: A Review,” J. Mar. Sci. Eng., vol. 8, no. 1, p. 26, Jan. 2020, doi:
10.3390/jmse8010026.
[4] P. Davies, G. Germain, B. Gaurier, A. Boisseau, and D. Perreux, “Evaluation of the
durability of composite tidal turbine blades,” Philos. Trans. R. Soc. Math. Phys. Eng. Sci., vol.
371, no. 1985, p. 20120187, Feb. 2013, doi: 10.1098/rsta.2012.0187.
[5] J. E. Jakes, C. G. Hunt, S. L. Zelinka, P. N. Ciesielski, and N. Z. Plaza, “Effects of
Moisture on Diffusion in Unmodified Wood Cell Walls: A Phenomenological Polymer Science
Approach,” Forests, vol. 10, no. 12, p. 1084, Nov. 2019, doi: 10.3390/f10121084.
[6] N. A. J. Platzer, Ed., Plasticization and Plasticizer Processes, vol. 48. in Advances in
Chemistry, vol. 48. WASHINGTON, D.C.: AMERICAN CHEMICAL SOCIETY, 1965. doi:
10.1021/ba-1965-0048.
[7] R. Hiesgen and J. Haiber, “MEASUREMENT METHODS | Structural Properties:
Atomic Force Microscopy,” in Encyclopedia of Electrochemical Power Sources, Elsevier, 2009,
pp. 696–717. doi: 10.1016/B978-044452745-5.00073-3.
[8] B. Voigtländer, Atomic Force Microscopy. in NanoScience and Technology. Cham:
Springer International Publishing, 2019. doi: 10.1007/978-3-030-13654-3.
[9] M. R. Vanlandingham et al., “Nanoscale Indentation of Polymer Systems Using the
Atomic Force Microscope,” J. Adhes., vol. 64, no. 1–4, pp. 31–59, Aug. 1997, doi:
10.1080/00218469708010531.
[10] A. C. Fischer-Cripps, Introduction to contact mechanics, 2nd ed. in Mechanical
engineering series. New York: Springer, 2007.
[11] ASTM D570-22, “Test Method for Water Absorption of Plastics,” ASTM International,
2022. doi: 10.1520/D0570-22.
[12] C.-H. Shen and G. S. Springer, “Moisture Absorption and Desorption of Composite
Materials,” J Compos. Mater., vol. 10, p. 20, Jan. 1976.
