Solar Energy Materials & Solar Cells 95 (2011) 2037–2068

Contents lists available at ScienceDirect

Solar Energy Materials & Solar Cells

journal homepage: www.elsevier.com/locate/solmat

Durability of Fresnel lenses: A review specific to the concentrating

photovoltaic application
David C. Miller n,1, Sarah R. Kurtz
National Renewable Energy Laboratory (NREL), 1617 Cole Blvd., Golden, CO 80401, USA

a r t i c l e i n f o abstract

Article history: The durability of Fresnel lenses used in the concentrating photovoltaic (CPV) application is reviewed
Received 25 September 2010 from the literature. The examination here primarily concerns monolithic lenses constructed of
Accepted 19 January 2011 poly(methyl methacrylate) (PMMA), with supplemental examination of silicone-on-glass (SOG)
Available online 6 May 2011
composite lenses. For PMMA, the review includes the topics of: optical durability (loss of transmittance
Keywords: with age); discoloration (the wavelength-specific loss of transmittance); microcrazing and hazing;
Fresnel lens fracture and mechanical fatigue; physical aging, creep, shape change, buckling, and warping; and solid
PMMA erosion. Soiling, or the accumulation of particulate matter, is examined in the following contexts: its
Accelerated life testing magnitude of reduction in transmittance; variation with time, module tilt, and wavelength; the
Reliability
processes of adhesion and accumulation; particle size, distribution, composition, and morphology; and
Durability
its prevention. Photodegradation and thermal decomposition, mechanisms enabling aging, are
examined relative to the CPV-specific environment. Aspects specific to SOG lenses include: solarization
of the glass superstrate; corrosion of glass; delamination of the silicone/glass interface; change in focus
due to thermal misfit between the laminate layers; and the chemical stability of poly(dimethylsiloxane)
(PDMS). Recommendations for future research are provided, based on the most important and the least
explored topics.
& 2011 Published by Elsevier B.V.

1. Introduction spherical lens element would be prohibitive. As CPV modules are

Concentrating photovoltaic (CPV) technology uses relatively
sizable optical component(s) to focus optical flux onto a relatively
small photovoltaic (PV) cell [1]. The CPV application becomes
economically advantageous when the module cost is reduced
using optical components that are inexpensive relative to the PV
cell. To realize low levelized cost of electricity, the optical
components must provide good performance over the desired
service life of 30 years. Although the advancement of high-
efficiency PV cells (such as multijunction III–V technology) has
recently motivated interest in CPV, understanding related to the
durability of the optical components remains limited.
Refractive CPV technology typically uses a lens as the first
component of an optical system. Fresnel lenses consist of discrete
concentric prism elements patterned on a superstrate. Fig. 1
represents a typical Fresnel lens, which may be constructed
monolithically or out of separate layers of material. Since their
first application as a collimator in a lighthouse in 1822 [2], Fresnel
lenses have been used where the size, weight, and cost of a
n
Corresponding author. Tel.: +1 303 384 7855; fax: + 1 303 384 6790.
E-mail address: David.Miller@nrel.gov (D.C. Miller).
1
http:www.nrel.gov/pv/performance_reliability/
typically mounted on a tracker made to point toward the sun,
Fresnel lenses have traditionally been manufactured out of
poly(methyl methacrylate) (PMMA). Such lenses may be manu-
factured by hot-embossing, casting, extruding, laminating, com-
pression-molding, or injection-molding thermoplastic PMMA [4].
A heavier, but less researched, lens technology consists of acrylic
or silicone facets patterned on a glass superstrate [3,5].
Certain optical considerations are relevant to the discussion of
Fresnel lenses. For example, the prism side of the lens is typically
mounted facing the inside of a CPV module (as indicated in Fig. 1),
because the prism valleys are less likely to become contaminated
(‘‘soiled’’) with particulate matter, the smooth superstrate surface
may be more readily cleaned, and the transmission loss through
the lens is reduced [6,7]. Fresnel lenses may be designed to focus
flux at a spot or along a line. The concentration of flux that may be
realized in these geometries is represented in Eq. (1) [2,8], where
for system international (SI) units, Cmax represents the maximum
concentration of optical flux (unitless); n is the real component of
the refractive index (unitless); and ya and ca are the acceptance
angles (also known as the ‘‘half field of view’’ (HFOV)) of the
incident flux relative to the lens (radians):
n2 n
Cmax ¼ ) ð1Þ
sin½ya sin½ca  sin½ya 

0927-0248/$ - see front matter & 2011 Published by Elsevier B.V.

doi:10.1016/j.solmat.2011.01.031
2038 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
Fig. 1. Cross-section schematic that identifies the key features of a typical Fresnel
lens. The orientation relative to the sun and interior of the module is labeled.
Adapted from [3].
From Eq. (1), the Cmax that may be realized for imaging PMMA
optics (a solid lens in air, where n  1.003) is roughly 46,300  for
spot focus and 215  for line focus, for the angular radius of the
sun, i.e., 4.653 mrad [2]. In practice, a geometric concentration
(Cg) much less than the thermodynamic limit [9], e.g., Cg ¼5  –
1000  , is used to allow an optical tolerance, e.g., accommodation
of tracking error. Regarding lesser optical concentration, studies
have suggested that linear concentrators with Cg ¼2.1  [6] or
15.6  [10] can harness the direct and circumsolar region (the
angular radius of  50 mrad) without tracking the sun [8]. The
exact concentration limit for non-tracking CPV is debated, e.g.,
the maximum Cg of 2.3  in air is identified to be improved to
Cg ¼5.2  using a solid glass or polymeric optic [11].
Mechanical stiffness and the corresponding tracking-error
tolerance (as well as the flux concentration) can be improved,
whereas chromatic aberration and module thickness (lens-to-cell
separation distance) are reduced if a domed (rather than flat)
Fresnel lens geometry is used [2,6,7,12]. From analysis, the lens
comes to increasingly resemble an ideal Lambertian source if a
flat, roof, arch, or domed-shape lens is used [6,12].
The issue of chromatic aberration is important to CPV because
heterogeneous color distribution adversely affects the PV effi-
ciency [13,14]. In ‘‘ordinary dispersion,’’ the n of the lens material
decreases with l [15], making blue light focus further from the
lens than red light. The photovoltaic efficiency is then reduced
because the size of the image is increased (for all cells) and
because the spectrum varies with location on the cell (for multi-
junction cells). This heterogeneous color distribution on a multi-
junction cell causes a detrimental (current-limiting) spatial
variation in the photocurrent generation for each junction (unless
it can be redistributed laterally within the cell [16]).
Light loss (optical efficiency), chromatic aberration, tracking-
error tolerance, spot shape, and flux uniformity may be improved
using a non-imaging lens design, which must be analyzed using
ray-tracing software [2,10,18]. The same factors can also be
addressed using a secondary optic. A sharp-cornered square,
triangular, hexagonal, or rectangular light-pipe composed of
glass [17,19–22] or metal [23] can be used to homogenize optical
flux. A light pipe using a rounded cross-sectional geometry would
instead render the opposite flux distribution, i.e., with concentra-
tion increase from the edge to approach infinity at the cen-
ter [19,20]. Glass ‘‘kaleidoscope homogenizers’’ operate on the
principle of total internal reflection, which may render a mini-
mum-to-maximum irradiance ratio of 0.94 when the focused flux
is reflected twice (on average) within the optic [21]. In contrast,
the flux distribution may easily vary by one to three orders of
magnitude if no secondary optic is used [24,25]. The chromatic
and flux uniformity rendered by a secondary optic usually
improves the PV efficiency beyond the expense of the optical loss
associated with the secondary. Other examples of secondary
optics include lenses of a spheroidal [25] or sphere-plus-cone [25]
geometry, or reflectors including linear, elliptical, or hyperbolic
mirrors [6] situated about a linear concentrator.
The f-number, N (unitless), is a related concept. For imaging
optics, N relates the focal distance, f (m), to the beam radius, Rb
(m), i.e., N ¼f/(2Rb). In CPV, N specifies the compromise between
the thickness of the module (which decreases as N is decreased)
and the issues of chromatic aberration, tracking-error tolerance,
alignment tolerance, spot shape, and flux uniformity, which
become increased as N is decreased.
Ranges of values specific to Fresnel lenses used in CPV are
provided in Table 1. The values in Table 1 come from the historic
literature (originating in the 1970s), as well as literature describ-
ing contemporary products. The table is neither meant to guide
CPV module design, nor does it imply the most advanced
manufacturing capability; rather, the values are meant to provide
a figure of merit for reference, aiding discussion as durability
issues are examined. In Table 1, Ze refers to the overall optical
efficiency of the lens, including: reflections at the free surfaces,
losses at the tip radii, draft angle losses, and optical absorptance
(at wavelengths outside of specific absorption regions).
Key characteristics of materials including PMMA, poly(di-
methylsiloxane) (PDMS, moderately and highly cross-linked),
soda-lime glass, and aluminum (often used in CPV modules) are
provided in Table 2. The values in the table apply to unaged
material. Like Table 1, Table 2 provides representative values to
aid discussion. The values in the table (except for yc) were
obtained from manufacturer’s data [26] and were not indepen-
dently verified at the National Renewable Energy Laboratory
(NREL). The maximum strain, em, is given when the mode I critical
fracture toughness, KIC, was not available. Applicable formulations
of optical- or mirror-grade PMMA material may come from the
automotive, lighting, or skylight industries. Applicable formula-
tions of optical-grade PDMS material are shared with the aero-
space, electronic, and light-emitting diode industries (but
typically excluding the health, industrial, and automotive indus-
tries). PMMA and PDMS may be adhered to a glass superstrate
and patterned as a Fresnel lens. Applicable formulations of PDMS
are typically cured from a two-part system and are colloquially
known as ‘‘gels,’’ ‘‘elastomers,’’ or ‘‘encapsulants.’’ Polycarbonate
(PC) is a material sometimes suggested as an alternative to
PMMA. The significantly greater toughness prevents mechanical
fracture and fatigue in PC, but PC is less scratch resistant than
PMMA [27]. Although not explored here, potential disadvantages
of PC include its lesser spectral bandwidth [15], lesser optical
transmittance [15], greater optical dispersion (enhancing
Table 1
For Fig. 1, representative geometry and parameter values for a CPV Fresnel lens.
Parameter name Symbol Representative value
Facet thickness hf 0.1–0.5 mm
Superstrate thickness hs 2.0–5.4 mm
Total lens thickness ht 2.5–5.5 mm
Prism peak radius Rp 2–30 mm
Prism valley radius Rv 1–30 mm
Facet angle yf 0–1.6 rad
Draft angle yd 35–52 mrad
Prism width wp 0.01–1.0 mm
Lens width wl 2–30 cm
Acceptance angle (HFOV) ya 0–300 mrad
Geometric concentration Cg 5–1500
f-number N 0.5–1.5
Optical efficiency Ze 78–86%
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
Table 2
Representative characteristics for CPV lens and module materials.
Characteristic PMMA PDMS elastomer
E, elastic modulus (GPa) 3.3 0.003
n, Poisson’s ratio (unitless) 0.33 0.5
1
CTE, coefficient of thermal expansion (ppm K ) 70 300
kT, thermal conductivity (W m  1 K) 0.19 0.2
KIC, fracture toughness (MPa m0.5) 1.15 em r 110%
r, density (Kg m  3) 1190 1030
n, real refractive index (unitless) at l ¼ 589.5 nm 1.49 1.41
Aw, maximum water absorption (wt% ) 2.5 0.1
yc, contact angle (deg.) 63–69 96–110
chromatic aberration) [28,29], and its known vulnerability to
terrestrial solar-induced photooxidation [30–32]. In particular,
the degradation of bisphenol-A PC may be facilitated by a photo-
Fries rearrangement (initiated at shorter wavelengths, which can
reconfigure its structure based on the recombination of radicals),
as well as a separate photo-oxidative process (initiated at longer
wavelengths), as detailed in Refs. [33–36].
Possible failure modes, which will be categorized to aid their
discussion, include those related to: optical durability (reduced
transmittance, discoloration, and hazing); mechanical durability
(crazing, fracture, fatigue, delamination, reduced modulus,
embrittlement, physical aging (dimensional stability), sag, shape
setting, buckling, solid erosion, and wear (scratching)); soiling
(particulate adhesion and accumulation, embedded particulates,
and water-spotting); photodegradation (photolysis); and thermal
degradation (thermolysis). More specifically, fracture (cracking)
can arise from microcrack growth or impact, e.g., by rocks or hail.
Mechanical fatigue can arise from thermal cycling, humidity
cycling (hydro-expansion), wind flutter, or repeated snow-load.
Lens sag may arise from gravity, wind-load, thermal- or hydro-
expansion, and shrinkage. Photolysis and thermolysis are the
underlying mechanisms that may contribute to many failure
modes.
Some of the literature related to Fresnel lenses comes from
specific research efforts at certain institutions. For example,
numerous laboratory reports concerning CPV were generated at
Sandia National Laboratories (Albuquerque, NM) in the 1980s. In
addition to the Sandia (‘‘SAND’’) reports, the activities related to
specific grants are summarized in the annual project integration
meeting reports (‘‘PIMs’’). During the same time, the U.S. Depart-
ment of Energy subcontracted with McDonnell Douglass Astro-
nautics (MDC), often in conjunction with other facilities in
support of projects related to solar concentration. The Desert
Sunshine Exposure Testing (DSET), later Heraeus, and currently
Atlas Material Testing Technology outdoor facility in Phoenix, AZ,
finds frequent mention in the literature. Regarding the literature
related to CPV technology, a significant gap occurs between the
research conducted in the 1980s in response to the energy crises
of the 1970s and the current research motivated by the advance-
ment of highly efficient III–V PV cells. Numerous examinations of
the photodegradation and thermal degradation of PMMA (and
PDMS) occurred at academic laboratories. A notable specific effort
was conducted at the National Bureau of Standards (NBSIR), now
the National Institute of Standards and Technology (NIST), in
Gaithersburg, MD.
The goal of this document is to examine the existing research
literature applicable to the durability of Fresnel lenses used in the
CPV application. The topics of optical durability, mechanical
durability, soiling, photodegradation, and thermal degradation
will be examined. Key concepts, results, and conclusions from the
literature will be summarized. Key studies (including those not
2039
PDMS encapsulation SiO2 Al
0.08 74 69
0.4 0.18 0.33
220 9.5 23.6
0.16 1.1 154
em r 5% 0.75 29
1030 2520 2700
1.41 1.52 1.05
0.1 0 0
83–107 18–39 31–47
previously documented electronically) will be identified.
Although the review here applies primarily to refractive-enabled
CPV module designs, the results may sometimes be extended to
reflective CPV systems. While the survey here was originally
intended to apply to optics composed of PMMA, results obtained
from components using glass or acrylic front-surfaces some-
times proved relevant. Issues that may apply specifically to
the durability of silicone-on-glass Fresnel lenses are examined
separately.
2. Failure modes
Optical and mechanical failure modes, as well as the topic of
soiling, are examined in this section. Optical concerns include the
efficiency of transmittance and esthetic appearance. Mechanical
issues include: crazing, fracture, fatigue, elastic modulus, fracto-
graphy, physical aging, and solid erosion/wear. The mechanical
concerns include loss of physical integrity (including crack-
formation, crack-growth, and breaking of the lens), shape change
(coupled to the optical characteristics), and surface characteristics
(also coupled to the optical characteristics). The topic of adhesion
(e.g., between lens elements or component layers within a lens) is
not examined here in detail. Soiling, which is the contamination
of optical surfaces by particulate matter, is then explored in
detail. This includes: (1) magnitude of effect, (2) variation with
time, module inclination, and wavelength, (3) the processes of
adhesion and accumulation, (4) particle size, distribution, com-
position, and morphology, and (5) the prevention of soiling and
cleaning of the surface.
2.1. Optical durability
Reduced transmittance was observed for acrylic sheet or
lenses subject to extended deployment in the outdoor environ-
ment [28,38–49]. Furthermore, reduced transmittance (increased
absorptance) within the solar spectrum has been observed fol-
lowing artificial illumination within the solar spectrum [50–60]
and at wavelengths shorter than the solar spectrum [61–66]. The
studies most resembling the CPV application include those at
Sandia [28,38,39,41,45], Heraeus [43,44,46], and the Institut für
Solartechnik Prüfung Forschung (SPF) [48,49], which were con-
ducted outdoors for up to 36.5, 1, and 20 years, respectively.
Increased optical density (absorptance) [41,48,55,57,58,60,
61,64–67] distinguishes here between optical degradation occur-
ring within the PMMA material and that caused by other factors,
such as solid erosion or soiling of the surface. The reduction in
transmittance varies with depth, where the irradiated (front)
surface is the most damaged. Depth dependence was specifically
verified in Refs. [28,38,64] by sectioning and polishing the
surfaces of acrylic sheet.
2040 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068

Fig. 2. Measured optical transmittance for PMMA sheet specimens subject to outdoor exposure, shown relative to reference (unaged) specimens of the same material.
The normalized direct solar spectral irradiance (AM1.5 in ASTM G173) is also provided for reference. The data were obtained from a previous study [48,49].

Fig. 2 presents typical transmittance profiles for acrylic sheet Table 3

specimens, shown relative to the normalized direct solar spectral Representative degradation in optical transmittance of PMMA.
irradiance (AM1.5 in ASTM G173 [68]). The data come from a study
Reference Location Duration kT, typical Wavelength
in Switzerland [48,49], conducted in Rapperswil (suburban envir- (years) (%) range
onment at 0.42 km elevation) and Davos (rural forest environment
at 1.59 km elevation). All specimens were measured after being [28,38] Albuquerque 17.7 2.2–4.3 Avg., l ¼350–
cleaned with a soap solution followed by ethanol; aged PMMA 900 nm
[39] Albuquerque 22 0.5 Avg., l ¼350–
specimens of material #42 are shown relative to an unexposed
900 nm
reference specimen. The result for a different formulation (without [42] New River (AZ) 2.4 1.2–1.9 Solar weighted
an antireflection (AR) coating) is indicated as material #44. Speci- avg.
mens aged outdoors may demonstrate reduced transmittance at [43,44,46] Phoenix (AZ), 1 7 Solar weighted
Miami (FL) avg.
shorter wavelengths, manifested as a rounding of the otherwise
[48,49] Davos, Rapperswil 20 1.0–3.1 Solar weighted
distinct cut-on frequency (occurring at 390 nm in Fig. 2, prior to avg.
aging) [60]. For the aged specimens, no noticeable attenuation
occurs in the near-infrared wavelengths, aside from the absorp-
tance peaks typically found there [15]. The transmittance profile for
aged specimens varies with formulation of material, e.g., compare the ranges of wavelengths examined, whereas in other cases,
specimen #A44 (which originally had a greater spectral bandwidth) the researchers provided a solar weighted average (relative to the
to #A42. For polymers, the magnitude of degradation may vary AM1.5 global spectrum). Within wavelength ranges [28,38,39], the
with test location, owing to factors that include ultraviolet (UV) transmittance was reduced by as much as 70% for l ¼350–400 nm.
dose (at elevation), climate (temperature and humidity), and The reduction in direct transmittance (relevant to CPV systems
aerosol content (such as ozone) [69]. The relative influence of these using high optical concentration) in Refs. [39,45] may be 5–10  ,
factors was however not examined in the aforementioned studies. which is indicated in Table 3. For multijunction cells, a decrease
The profiles in Fig. 2 represent the range of effect on transmit- within a particular portion of the solar spectrum can induce a
tance for a variety of material formulations; application-suitable current-limiting condition at one junction, disproportionately
formulations of PMMA may resemble specimen #A42. For refer- decreasing power production for the entire cell [15].
ence, the range of wavelengths used in CPV extends from 300 to Although quick to come to discussion, studies specific to the
1120 or 300 to 1800 nm for crystalline silicon (c-Si) or III–V degradation of the optical transmittance of PMMA are few in
multijunction cell technologies, respectively [15]. Reduced trans- number. Furthermore, most were completed (or initiated) more
mittance near UV wavelengths renders a visual discoloration, than 20 years ago. In several cases, potentially meaningful studies
including a yellow [38,47,60,70] or brown [45,70] appearance. were not continued for a sufficient duration capable of discerning
Most transmittance studies were conducted using a spectro- age-specific trends or properly simulating field service life. The
photometer equipped with an integrating sphere. Direct trans- CPV industry would benefit from publications examining con-
mittance (measured without an integrating sphere and more temporary PMMA formulations, which might be used to more
relevant to the CPV application) was shown to be significantly confidently identify the effect of the environment on transmit-
more affected than hemispherical transmittance [39,45]. tance. Further, such studies should enable the prediction of
The decrease in hemispherical transmittance (kT) of PMMA optical service life in the field. The recent consolidation within
resulting from outdoor aging is further summarized in Table 3. the acrylic lens manufacturing industry has reduced the number
The table includes test locations near the city of Albuquerque (rural of candidate specimens, thereby making such study more feasible.
environment at 1.62 km elevation), New River (rural environment at
0.61 km elevation), Phoenix (rural environment at 0.34 km eleva-
tion), and Davos, which may prove more severe than some CPV 2.2. Mechanical durability
installation sites because of their elevation and/or hot desert
climate. The values in Table 3 represent the magnitude of effect 2.2.1. Crazing
obtained from suitable formulations of PMMA at application-specific Crazing may begin at the surface and come to extend through
wavelengths. In some cases, the data were taken as the average of all the bulk of PMMA [4,172]. Such crazes render a cloudy or
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
translucent appearance because they facilitate light scattering
(optical haze). No studies were identified that directly relate
(quantify) between such field-induced microscopic defects and
optical performance. In the outdoor environment, crazes typically
grow with time, which may be represented using environmental
stress cracking models [4,172]. Microscopic crazing, as well as
macroscopic mechanical failure, may be understood through
conventional fracture mechanics and fatigue representations.
2.2.2. Fracture
Micrometer-scale defects, originating during manufacturing or
external impact in the field, limit fracture strength [72]. ‘‘Fracture
failure’’ is defined as a sufficient crack (or load) that motivates the
physical breaking of the lens. In impact events, fracture is
frequently encountered at the edge of the lens because of the
mechanical constraint associated with mounting the lens to the
module. The flexural strength of acrylic sheet was reduced by as
much as 51% [28,38] and 33% [40] after 17.7 or 1 year(s) in the
outdoor environment, respectively. Decreases of 25% [59] and
57% [65] for the tensile strength of PMMA resulted from indoor
aging. Craze resistance (which relates to strength [172], and was
correlated against site temperature, humidity, solar irradiance,
specimen equilibrium water content, and weight loss) was gen-
erally decreased after 2 [73,393] and 4 [27] years outdoor
exposure in Australia; craze resistance was likewise decreased
after 5 years outdoor exposure in Germany [74] or 1 year in Saudi
Arabia [75]. Unfortunately, the morphology of the critical flaws
typically present in CPV lenses, which would allow for estimation
of strength, has not been examined. Excluding external defects,
the fracture of PMMA would be limited by internal microvoids,
estimated to be 35–60 mm in size [77]. In comparison, the area of
0.64 nm2 may be taken for the chain in cross-section when
estimating the theoretical maximum strength [76].
The characteristic of strength is governed by the intrinsic
property of fracture toughness (related to surface energy and
fracture energy), examined for PMMA in Refs. [76–81]. From pffiffiffiffiffiffiffi the
critical energy release rate [Gc (J m  2)], the KIC (KIC ¼ GE) for
PMMA was observed to decrease with molecular weight, from
0.90 to 0.087 MPa m0.5 [78,79], or 0.84 to 0.18 MPa m0.5 [80]. In
these studies, molecular weight, Mw, was controlled by chain
scission via g-radiation [78,79] or the polymerization proce-
dure [80], respectively. For adequate Mw, Gc is generally under-
stood to be about 1000 times greater than its theoretical value,
due to plasticity-aided crazing at the crack tip [76,79,82]. Crazing
occurs via molecular entanglement for PMMA above the critical
molecular weight, Mc, of 10,000–100,000 [76–81,83] (chains
exceeding about 90 monomers [81]). The KIC value of
1.15 MPa m0.5 in Table 2 applies to material where Mw bMc; it
may be applied regardless of the fracture mode [84]. Although
PMMA degrades without cross-linking [78], its toughness may be
improved by cross-links introduced during the initial polymeriza-
tion [83]. Commercial PMMA formulations may also use co-
polymers to enhance the yield characteristics and improve craze
resistance—often rendering decreased toughness and increased
modulus [81].
The absorption of water or solvents (which may plasticize
PMMA) has been shown to affect craze formation [27,73,85,
172,393] (reducing toughness) and cause microcracking and void
formation (facilitating crack initiation and thereby decreasing
toughness) [86]. Unfortunately, the absorption of water is the
greatest for high Mw PMMA [86]; therefore, mechanical degrada-
tion by hydro-expansion is expected to be most detrimental in
new lenses. Weathering of PMMA has been cited as being more
severe in hot-humid than hot-dry locations [27,44,47]. The con-
tribution of factors including temperature and irradiance is
2041
identified in addition to ambient humidity and absorbed moisture
in Refs. [27,73]. It is unclear from these studies whether the
microcracking of PMMA was the primary aging mechanism in the
humid locations. The chemistry (acid content) of rain may also be
a significant factor. For reference, PMMA manufacturers may
guarantee product durability (e.g., discoloration via yellowness
index) less stringently in hot-dry than in hot-humid locations [87].
Greater understanding of the role of water in degradation would
be helpful.
Impact-resistant formulations of PMMA may be used to
improve the resistance to fracture. Unfortunately, these impact-
resistant formulations aged more readily [43,44,46,88,89]. Inspec-
tion of weathered specimens revealed chalking at the surface, as
well as randomly oriented microcracks [89]. Outdoor weathering
for 10 years decreased the tensile strength and elongation at yield
by 36% and 65%, respectively (with a substantial reduction in
optical transmittance) [88,89]. The tensile strength was decreased
on average by 21% in an outdoor study [75], and 83% in a separate,
combined indoor and outdoor study [43,44,46]. Here, decreased
elongation at yield and decreased transmittance trends could also
be identified for some specimens.
2.2.3. Fatigue
Failure by fatigue refers to the growth of a micro-defect, where
a crack eventually facilitates the fracturing of the lens. Similar to
toughness, the fatigue behavior of PMMA has been stu-
died [90,91–97] independent of the CPV application. Following
initiation, crack growth typically follows the familiar log-linear
relationship with stress intensity proposed by Paris et al. [98]
(Eq. 2). In the equation, a represents the crack half-length (m);
N, the number of cycles (unitless); DK, the range of stress
intensity factor (MPa m0.5); and C, a material-specific constant
(m2 Pa  2 cycle  1). The values of 5.7 [91] and 2.5 [93] have been
reported for the coefficient, m. Because PMMA is a polymer, the
material constants vary with the frequency of loading, specimen
temperature, and humidity:
@a
¼ C DK m ð2Þ
@N
The results of a complete factorial experiment [93] are used to
extend Eq. (2) to account for the combined effects of frequency of
the applied load, mean stress intensity (Kmean), and DK at a
constant temperature. Similar to toughness, fatigue behavior
varies with Mw due to cooperative energy dissipation at chain
entanglements [90,95–97]. That is, qa/qN is significantly increased
as Mw is decreased. Fatigue life, however, continues to increase
with Mw even as KIC has stabilized above Mc [95,96]. Furthermore,
the waveform of the applied load, i.e., periodic profile of Kmax
and Kmin, may favor chain scission even during conditions of
greater chain entanglement, decreasing fatigue life as Mw is
increased [97]. In polymers, Mw typically correlates with the glass
transition temperature, Tg, which was decreased by 4 1C after 17.7
years [37] or 3.51C after 1 year [75] in the outdoor environment.
Similarly, Tg was decreased by 3 1C after 1570 h of indoor aging
(Xe lamp) [59]. Similar to toughness [80], the presence of a small
portion of high Mw content material within PMMA will greatly
enhance fatigue life [95]. Cyclic loading may enable stress-
induced hysteretic heating [91–93,95,96]; therefore, the max-
imum test frequency in the literature is limited to 1.7 Hz [92] and
4 Hz [93].
The stress motivating fracture and fatigue in PMMA may be
analyzed visually because the material is birefringent [4]. To
explain, color contours will be observed in cross-polarized light
in proportion to the stress within the material. Methods of
photoelasticity exist such that the color contours may be inter-
preted to allow a non-destructive evaluation of residual stress
2042 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
(found to evolve in [75]) or the real-time assessment of stress
present during an applied mechanical load [99,100]. The method
becomes ineffective for T470 1C because of structural relaxation
within PMMA [101].
2.2.4. Elastic modulus
Elastic modulus, also essential in maintaining the physical
dimensions of an optical component, readily affects the fracture
characteristics (fracture and fatigue). Modulus would not be
expected to readily vary below Tg, where polymer chains remain
spatially constrained. Modulus was observed to remain stable in
Refs. [78,79,102] for PMMA of different Mw, maintained in the
glassy state (occurring below 105 1C) [78]. To explain, the mini-
mum Mw that may be readily manufactured is greater than the
Mc, below which E would be significantly reduced [95]. E,
however, may decrease to 2.5 GPa when PMMA becomes satu-
rated with water [85].
2.2.5. Fractography
Fractography refers to the examination of the fractured sur-
face, by which the history of critical event(s) may be analyzed.
The fractured surfaces of PMMA are dominated by the traditional
mirror/hackle/mist morphology [103] above Mc [75,82,83,104
–106]. The branching of the incipient crack may be represented
using a cohesive zone model [106], where the variation between
bifurcation and crack velocity may be explored. The fractography
transitions to a featureless, smooth transparent surface as Mw is
decreased [104]. The fractured surfaces may demonstrate inter-
ference colors [76,77,79,104], due to a thin layer of structurally
modified (reoriented) PMMA. The fractography of fatigued PMMA
is examined in Refs. [88,90,91,97]. Striations, occurring within
each growth cycle, as well as discontinuous growth bands have
been observed. The latter are limited in size by the radius of the
plastic region facilitating crazing [93,94,97].
2.2.6. Physical aging
The densification (of polymers) over time, known as physical
aging, may adversely affect a Fresnel lens via: visco-elastic flow
(shape change), change in dynamic response (including embrit-
tlement), and change in opto-mechanical properties (n, affecting
optical performance). Studies of the physical aging of PMMA
include Refs. [107–124], whereas the general phenomena are
described in Refs. [125–134]. The terms ‘‘physical aging’’ and
‘‘structural relaxation’’ refer to the idea that polymers seldom
exist in thermodynamic equilibrium when cooled below Tg. At the
molecular level, the material instead exists in a metastable state,
favoring spontaneous evolution (consolidation) of excess free
volume. The elimination of heterogeneously distributed free
volume is not a process of material degradation, because the
process can be reset by heating above Tg [101]. The rate of
physical aging varies strongly with temperature. A variety of
models has been proposed to relate to the ‘‘retardation time
constant’’ associated with individual consolidation events. Physi-
cal aging also depends on the thermal history. For example, a
‘‘memory effect’’ (where volume can very suddenly increase or
decrease when the temperature is changed) is attributed to the
series of previously occurring events [117,122,131,132]. Exter-
nally applied stress (or strain) may also affect the free volume,
resulting in a volume greater or lesser than the equilibrium
volume [121,123]. Others have found that large stress (or strain)
can reduce or even ‘‘erase’’ the effect of temperature-induced
aging, ‘‘rejuvenating’’ the material [115,131,132]. The more com-
plicated interaction between temperature, stress, and free volume
will not be examined here because of the lack of agreement
concerning the behavior characteristics, boundaries of behavior
regimes, and explanation (modeling) of the aforementioned
phenomenon [131,132]. The behavior for PMMA is perhaps
more readily understood than other polymers because the
material is amorphous. Lastly, the relationship between tensile
failure (known as ‘‘creep rupture’’) and temperature has been
examined [115,118,131].
PMMA may have a free volume on the order of 2.5% [110].
Relative to other polymers, PMMA may have a wider range of
free-volume sites; however, the largest sites are typically
depleted rapidly by the aging process [132]. The a-relaxation
(where Ta ¼Tg) is the primary mechanism facilitating physical
aging. For atactic (typical) PMMA, where Ta ¼99 1C [127], the
modes of a-relaxation may correspond to the slipping, uncoiling,
and reorientation of the main chain [101]. The b-relaxation,
occurring at Tb ¼33 1C [127], is attributed to the ester (–COOCH3)
side group [53,64,101,108,117,124,127,129,135]. Although the
contributing structure within the side group is not agreed upon,
several studies provide good evidence that rotation facilitating b-
relaxation is coupled to the main chain [101,108,117,124,127].
The g-relaxation (occurring at Tg ¼  47 1C [127]) is not as widely
studied, but has been attributed to the rotation of the methyl (–
CH3) side group [64,108,127,135]. Below Tg, a spurious peak
observed at 82 1C is attributed to water absorption [108,117].
The temperatures of the relaxation modes are not rigidly defined,
because they may vary with factors including: additives (such as
antiplasticizers), absorbed water, Mw (and polydisparity), the
measurement technique, rate of testing, and specimen history.
Isotactic PMMA (where similar side groups are oriented on the
same side of the chain) has a lower Ta (50 1C) than syndiotactic
PMMA (Ta ¼122 1C, where similar groups are alternately
oriented), attributed to the higher chain mobility of the isotactic
geometry [136,137]. Regarding stereochemistry, isotactic PMMA
does not exhibit a b-relaxation [108]. Most commercial PMMA,
however, has both atactic and syndiotactic contents.
Polymers subject to an applied stress (‘‘creep test,’’ where
creep compliance, J (Pa  1), is proportional to log[t]) or an applied
strain (‘‘relaxation test,’’ where creep modulus, Ec (Pa), is propor-
tional to log[t]) may be represented according to the principle of
time–temperature superposition (TTS). A master curve, obtained
for a particular applied s or e vs. time, is first constructed
at a single reference temperature. Within a small strain limit
[131,132], the master curve will be self-similar to profiles
obtained at different temperatures. Therefore, a shift function
may be used to characterize the horizontal translation occurring
between master curves obtained at different temperatures. The
shift function quantifies the ‘‘time multiplier’’—the multiplicative
factor describing the relative increase in the duration required to
reach equilibrium at different temperatures. The use of a separate
shift function to compensate between different applied stress/
strain conditions using a horizontal and/or vertical translation
remains controversial [131,132]. The double-logarithmic depen-
dence for T (Eq. (3))—known as the Williams–Landel–Ferry (WLF)
representation—may be used to obtain the time multiplier. Eq. (3)
may also be used to relate the viscosity at a temperature and a
reference viscosity, such as the zero-shear viscosity, Z0:
 
Z C1 ðTTr Þ
log½at  ¼ log ¼ ð3Þ
Z0 C2 þTTr
New parameters include: at, the time multiplier (unitless);
Z, the viscosity (Pa s); C, the parameter-specific coefficients
(unitless or 1C); and Tr, the reference temperature (1C). Tr may
correspond to Ta or a fictive temperature, e.g., obtained from an
empirical fit. In addition to a specific relationship between
coefficients in Eq. (3), some instead advocate taking Tr as
Tb [138]. The values of C1 ¼16.7 1C [109] or 15.82 1C [124],
C2 ¼55 1C [109] or 52.20 1C [124] , and Tr ¼105 1C [109] or
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
98 1C [124] have been determined for PMMA. Examples of master
curves for PMMA may be found in Refs. [109,120,124,125].
Separately, a temperature-dependent activation energy may be
estimated from C1 and C2: Ea ¼  2.303(C1/C2)RT2 [124], where
R¼8.314 (J K  1 mol  1). The Tool–Narayanaswamy–Moynihan
model [139–141] is a popular phenomenological representation
that may be applied in finite-element analysis (FEA) to estimate
the consequences of physical aging for any materials exhibiting a
glass transition. To do so, the master curve is approximated as a
Prony series [142], with the shift function usually being specified
through a subroutine.
For polymers, Eq. (3) is generally limited to Ta oTo
Ta +100 1C [109,124]. To explain the limit at lower temperatures,
it becomes experimentally impractical to obtain data (equilibra-
tion times on the order of years for T oTa 20 1C); furthermore,
the contribution of the b-relaxation increases as temperature is
decreased. For the ‘‘Eyring regime’’ [143] (where ToTa,
i.e., To80 1C [109,114,118,131,138] for PMMA) or T4Ta +
100 1C [138,143], TTS no longer applies. Here, the time multiplier
may be estimated using an Arrhenius model:
Ea
log½at  ¼ þ ln½C3 
RT
As in the WLF regime, R represents the ideal gas constant. The
values of Ea ¼ 240 kJ mol  1 [109] and 214 kJ mol  1 [124] have
been reported for PMMA. Much of the literature focuses on
explaining the phenomena of physical aging from the perspective
of phenomenological or thermodynamically derived models, such
that the details (e.g., C1, C2, C3, Tr, and Ea) and their range of
application are not directly examined.
Physical aging may also occur in the absence of applied s or
e [127,129]. The isothermal volume relaxation rate (Eq. (5)) is
shown in Fig. 3.
1 @V
r¼
V @log½t
Parameters in Eq. (5) include r, the rate of relaxation (s  1); V,
the volume (m3); and t, the time (s). Fig. 3 identifies the behavior
for PMMA as well as polystyrene (PS), PC, and polyvinyl chloride
(PVC). Importantly, the profile for PMMA directly below Ta is
Fig. 3. Isothermal volume relaxation rate, measured as a function of temperature,
for several polymeric materials, including PMMA.
The figure adapted from Ref. [127] with permission.
2043
broad, with the peak rate occurring near room temperature. The
two peaks for PMMA (indicated with arrows in Fig. 3) are due to
the processes of g- and b-relaxations, respectively. In contrast, the
temperatures encountered at candidate PV installation sites
(examined in [144,145]) overlap the rb peak in Fig. 3. Volumetric
contraction might therefore be expected for a PMMA lens during
its service life in the CPV application environment. Regarding the
a-, b-, and g-relaxation processes, their effect is most prominent
in r [127,129] or dielectric analysis [135], rather than in volume
vs. temperature [127] or CTE vs. temperature [127] profiles.
The broad transition profile in Fig. 3 identifies significant
overlap between the a- and b-relaxation regimes, particularly
for PMMA [113,117,120,131,132]. To distinguish the a-relaxation,
the b-relaxation may be estimated using the ‘‘Cole-Cole’’ model
and then subtracted from the total relaxation [116]. Such distinc-
tion identifies that the b-relaxation dominates at room tempera-
ture [113,116,124]. Aging studies identify that the temperatures
for the b-relaxation regime remain unaffected, whereas the a-
relaxation regime may vary with the test rate [116,122,128]
(typically merging with the b-regime in aged specimens
[113,116,117,120]). The a- and b-regimes may also merge
ð4Þ when examined using high-frequency dynamic mechanical
analysis [124,126].
Physical aging is relevant to Fresnel lenses because it affects
the mechanical and optical characteristics, including: dimensional
stability (size), ‘‘Poisson’s ratio’’, yield stress, mechanical modu-
lus, mechanical dissipation (tan d), and refractive index. The
process of densification, shown in Fig. 3, implies that size will
decrease over time, which may affect optical alignment and may
induce mechanical stress. The contraction ratio (  ey/ex, analogous
to Poisson’s ratio) was determined to increase from 0.34,
approaching 0.5 for aged PMMA [101,110,113]. Yield stress was
identified to increase as a consequence of aging [27,121,123,
131,134]. This implies that PMMA will become more brittle over
ð5Þ time, i.e., KIC, fracture strength, and fatigue life are decreased. The
yield strength is lesser in tension than in compression for
PMMA [146]; therefore, less creep is observed for an applied
compressive s or e than a tensile load [110]. E was found to
increase because of physical aging [111,117]. Change in the shear
modulus was about twice that of the bulk modulus [124],
indicating that deviatoric (shear) stress affects PMMA more
significantly than dilation (volumetric) stress [113]. Regarding
stress, it was found to affect the b-relaxation behavior according
to the motion of (–OCH3) about C–O, whereas physical aging
affects the rotation of (–COOCH3) about C–C [101]. Mechanical
damping was found to decrease with aging [117]; the magnitude
of the tan d associated with the b-relaxation was particularly
reduced. The refractive index would be expected to increase with
density (aging) [147–149], and a change from n¼1.4907 to
1.4924 at l ¼514.5 nm was observed [149]. This would affect
the focal distance and the distribution (the uniformity of color
and intensity) of a Fresnel lens. Reheating above Ta will restart the
process of physical aging [393].
Regarding the effects of physical aging reported in the CPV
literature, a sag of 0.05000 (1.3 mm) across a 8.1600 (20.7 cm)
square lens element after 9 months of field use was reported in
Ref. [24], where the bowing and consequent reduction in module
efficiency was mitigated using annealing. In analysis, a 0.0800
(2 mm) sag at the center of a lens was predicted to render 0.500
(13 mm) lateral translation of the focused spot of a Cg ¼1000  ,
N  1.2 system [12]. Common optical variations including defocus,
stigmation, and coma could render intensity spikes, increased
chromatic aberration, or loss of light outside the cell aperture (all
compromising module efficiency). The temporary and long-term
stability of the lens shape (and prism elements) is therefore
important to CPV. Long-term dimensional stability is an issue
2044 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
with thermoplastic polymeric materials including PMMA [71,150],
even when the tooling and manufacturing processes have to be
designed to account for shrinkage [3,151]. Although the flow
properties for PMMA have not been specifically examined for the
CPV application, viscoelastic behavior is typically greatly reduced
below Tg. Mechanical buckling (caused by physical aging, CTE
misfit between components (Table 2) or hydro-expansion [152])
would be expected to severely distort lens components.
2.2.7. Solid erosion and wear
The tribological characteristics (including resistance to pitting
and abrasion) may also vary with photo- and thermal-history, as
well as temperature, but are not examined in the CPV-specific
literature. The surface roughness is inversely related to the optical
performance, and therefore the primary tribological concern.
Studies related to the general problem of ‘‘solid particle erosion’’
may be found in Ref. [153]; specific investigation of PMMA
includes Refs. [154–162]. Erosion of ceramics (such as glass) is
facilitated by the mechanisms of fracture [153,154] and deforma-
tion wear [158], both are the greatest at normal incidence.
Erosion of ductile materials (including metals and polymers) is
facilitated by the mechanisms of cutting wear (dominant parallel
to the surface) and deformation wear (favored perpendicular to
the surface) [153,154]. In particular, deformation wear dominates
during the impingement of spherical particles [157,158], whereas
cutting wear is increasingly significant for angular (sharp) parti-
cles [159]. Although the rate of erosion is the greatest at normal
incidence for ceramics, the fastest wear for ductile materials
typically occurs at an oblique angle. The maximum erosion rate
of PMMA has been identified as 301 or 701 [155], 251 [161], or
similar [154]. To explain, the angle providing the fastest erosion
depends on the velocity of the incident particulate matter. When
impacted by a continuous stream of particles, observed response
regimes include: an incubation period (characterized by accumu-
lation of particulate matter), a transition period, and a steady-
state state period (characterized by a constant rate of material
removal) [155–159,162]. PMMA was found to erode faster
than polycarbonate, polytetrafluoroethylene, and ultra-high-
molecular-weight polyethylene [157–159]. PMMA was further
distinguished from these other materials by its additional beha-
vior ‘‘regime’’, i.e., the rate of erosion continued to increase in the
case of impingement by spherical particles [157,158], whereas
erosion decreased in the case of angular particles [159]. Various
representative models exist to analyze the rate of erosion,
with key parameters including: the elastic modulus of the sub-
strate, hardness (substrate and particle), yield elongation
(strain), ultimate elongation (strain), angle of incidence, particle
velocity, particle density, and local temperature at the impact
site. Many of the key properties related to the substrate
can be determined using indentation. Impacting particles of
natural origin may be irregular in shape, have a dia-
metero100 mm, and be significantly harder than an acrylic
surface [153].
Following a continuous stream of incident particulate matter,
the morphology at the site of impact may demonstrate at least
four regions: (1) a central pit where PMMA was removed, (2) a
build-up consisting of non-uniform peaks and valleys, (3) a raised
area, possibly related to localized heating, and (4) a recessed outer
perimeter [157,158]. Additional observed features include a series
of longitudinal cracks, followed by transverse cracks, facilitating
spalling of the surface [156], as well as flakes of PMMA suggestive
of cutting wear [159]. A different magnitude of effect on optical
transmittance was seen between blue and red wavelengths,
depending on the test conditions [159]. Regarding more applica-
tion-specific situations, the impact of ice (hail) and resulting
damage morphology is modeled in Ref. [160]. Abrasion at the
microscopic level, including pitting and surface roughening, was
examined in Refs. [28,41,60,71]. The surface roughness resulting
from field deployment in a desert location and lens cleaning is
compared in Ref. [60]. The roughening of the surface eventually
makes PMMA more difficult to roughen than a pristine
surface [158].
Certain strategies may be used to reduce solid erosion includ-
ing installation height and surface coatings. Use on a tracker was
found to reduce erosion, particularly if the module is maintained
Z2 m above the ground [163]. To explain, the majority (90%) of
particulate matter is distributed within 1 m of the ground. It
remains to be proven, however, whether the reduced entrainment
observed in Ref. [163] applies to a variety of locations outside the
southwestern U.S. regarding surface coatings, where the use of a
relatively soft surface (such as a silicone resin) was found to
reduce the surface damage on PMMA [163]. Also, PMMA (a soft
surface layer) was found to improve the abrasion resistance of
glass (a hard substrate) in a laboratory test [164]. Presumably, the
result relates to greater compliance of silicone, mitigating the
momentum of impact. Soft coatings, however, are more prone to
accumulate particulate matter over time (Ref. [157] and Sec-
tion 2.3.8 below), which may prove more significant than solid
erosion in some locations.
2.2.8. Other concerns
Should the Mw of PMMA decrease with service life, the study of
its fracture and fatigue clearly demonstrates that PMMA would
become more brittle with time. Embrittlement may also occur as
the result of environmental conditions. For example, toughness in
polymers is known to readily vary with temperature, where a cold
ambient enhances brittleness [150,152]. The variation in tough-
ness with temperature would affect fracture, fatigue, and solid
erosion characteristics.
Delamination is suggested, but not examined directly, in
Refs. [4,24,71,151,165]. Problems with interfacial adhesion may
arise when layers of material are laminated together or when
singlet lenses are bonded into an optical paraquet assembly. In
addition to stress generated by the environment, residual stress
generated during the manufacture of each lens is of concern. Hot-
embossing and casting are manufacturing techniques recom-
mended in Ref. [4] to reduce residual stress.
2.2.9. Future study
In the literature, the issues related to the mechanical dur-
ability of Fresnel lenses in CPV are the least examined. The Fresnel
lens is expected to operate as an optical and mechanical compo-
nent, where mechanical requirements include impact resistance,
stiffness, shape permanence, and mechanical integrity/hermeti-
city. The lack of studies in the public literature may be due, in
part, to the desire of manufacturers to retain proprietary informa-
tion. Furthermore, early studies focused on verifying that ade-
quate optical efficiency could be obtained using a PMMA Fresnel
lens. In other cases, there may exist a fear of compromising the
marketability of PMMA, Fresnel lenses, and/or related surface
coatings. However, if the applicable issues are publicly examined,
then performance limitations can be evaluated and compensated
through the lens or module design. Embrittlement (loss of
toughness) is of great importance in locations prone to impact,
e.g., hail. Some information, such as the size of craze defects
(which limits fracture strength and fatigue life) may only be
definitively determined from long-term studies of fielded compo-
nents. Likewise, application-specific study of the dimensional
stability of the prism elements would be appropriately obtained
from fielded components. Lastly, there will likely always remain
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
certain characteristics, such as interfacial adhesion (which
depends on the base and bonding materials) that remain proprie-
tary because they are specific to a particular technology or
manufacturer.
2.3. Soiling
2.3.1. Definition and measurement
The accumulation of contamination on the optical compo-
nent(s) of a module is known as ‘‘soiling.’’ Soiling includes the
adhesion of particulate (upon), embedding of particulate (within),
and formation of residues—such as ‘‘water spotting’’—(upon) the
surface(s). Soiling reduces power production because the con-
tamination may absorb, diffuse, and/or reflect incident light. In
the worst case, soiling may compromise closed-loop control,
causing systems to lose track of the sun [166]. Although the topic
of soiling specific to CPV is not widely explored in the literature, a
broad base of research does exist within the fields of flat-panel
photovoltaic (FP-PV) modules and concentrating solar power
(CSP) that may be shared between the applications.
The optical characteristics of accumulated contamination are
most directly addressed when a component is examined using a
spectrophotometer, pyrheliometer, or pyranometer. Historically,
PV modules or arrays have also been used to infer the effects of
soiling, when compared to the local meteorological (precipitation)
history. Here, the preferred metric is electrical current (and not
power), because current directly relates to optical flux converted
at the cell. In module-based studies, the module is implicitly
assumed to be stable, with no other significant sources of
degradation. This assumption may hold true only over a short
duration (months) [167]. Loss in transmittance may be accentu-
ated in FP-PV module-based studies, where a single shaded cell
may trigger the reverse-bias condition [168,169].
2.3.2. Magnitude of effect
Table 4 summarizes the identified impact of soiling on optical
transmittance. For each study, the test location, type of environment,
type of specimens (both composition and configuration), test dura-
tion, reduction in transmittance, and ‘‘divergence’’ of measurement
are specified. The typical reduction in transmittance, kTtyp, is taken as
the steady-state asymptotic limit in figures or sustained average of
values in tables for the specimens most similar to PMMA sheet. The
maximum reduction in transmittance, kTmax, corresponds to the
greatest measured effect. The transmittance values indicate the
percentage change from the original condition, kT¼(Ti–Tf)/Ti, not
the difference between initial and final transmittances, Ti–Tf. In
studies where second-surface reflective specimens are used, the
results in Table 4 have been divided by 2 to account for the effect
of multiple passes through the material, i.e., incidence and
reflection. Table 4 renders the value of 976% (average (avg.)71
standard deviation (SD)) for kTtyp, whereas kTmax ranges from 3%
to 90%.
Table 4 suggests that the effects of internal degradation
mechanisms—such as corrosion or polymer degradation—may
be less significant than that associated with external soiling, as
concluded in Ref. [218]. For example, the ratio of the reduced
transmittance to the test duration in many cases implies a
degradation rate well in excess of 0.5–0.7% per year, the rate of
degradation identified for FP-PV modules [167]. To explain, many
of the studies of FP-PV module durability use a cleaning proce-
dure prior to performance (degradation) measurements and
therefore neglect the impact of soiling. Many of the studies in
Table 4 occur in desert locations, inherently prone to the accumula-
tion of contamination. Furthermore, because CPV systems use the
direct solar spectrum, they are inherently much more sensitive
2045
to optical scattering. For example, one module design was identified
as being 5  more sensitive than FP-PV [182]. If left unaddressed,
Table 4 suggests that soiling could easily compromise the added
benefit of enhanced cell efficiency, motivating use of CPV. To under-
stand the complex phenomenon of soiling and how it may be
addressed, the topic will now be explored in detail.
2.3.3. Variation with time
Fig. 4 demonstrates the effect of soiling on the measured
reflectance for metalized glass, PMMA, and polytetrafluoroethy-
lene (PTFE) specimens. The data are also typical of refractive
optics [171,180,183,185,190–194,196–199,205,206,210,213,215].
For prolonged exposure, an asymptotic decrease in optical per-
formance is observed with time (as opposed to a linear approx-
imation, which might be inferred from Table 4). This may be
modeled using Eq. (6) [210], where new parameters include W,
the weight of the deposited dust (g m  2); and d, the empirical
coefficients (unitless).
 
1T h i
¼ d1 erf d2 W d3 ð6Þ
Ti
The time constant of 1–2 months is reported for locations in
the desert southwestern U.S. [182,196]; however, the onset of the
asymptotic limit may be site-dependent. Although not considered
in Eq. (6), studies including Refs. [181,188,205] suggest that after
prolonged exposure, the transmittance cannot be restored to its
initial value, Ti. The permanent loss (on the order of a few
percent) may be due to tenacious contamination.
Fig. 4 also conveys the variability associated with synoptic
events. For example, reflectance was observed to decrease by 25%
( 13% loss of transmittance) in a single day, for light rain
occurring in the presence of wind-blown dust [177]. Although
heavy rainfall cleans well, the effectiveness of other natural
cleaning events may vary considerably. As in Fig. 4, light rain or
condensation events often degrade optical performance by
enabling sticking of particulate matter or the consolidation of
existing contamination [175–177,179,184,191,192,214,215]. Like-
wise, frost and snow are usually ineffective cleaning agents, except
in the case where accumulated snow may suddenly slide across
the optical surface in an abrasive manner [175,176]. Although
wind may erode deposited dust [219], its effects are often
negligible, except for particles of sufficient size, e.g., greater than
50 mm in diameter [191,192]. In one study, the rate of degradation
for a specimen that was never cleaned was found to be less
than that of identical, adjacent specimens that were cleaned
periodically (every 2, 6, or 12 days) [179]. This implies that
the effectivenesses of the deposition and natural removal pro-
cesses are decreased by the presence of accumulated contami-
nation.
Eventually, some level of contamination will become intoler-
able, mandating prescribed cleaning. Daily [198], weekly
[195,213], or monthly [197,199,215] cleaning schedules have
been suggested. In the case of performance-motivated cleaning,
inadequate frequency of performance monitoring may cause
some accumulation events to be overlooked [179].
2.3.4. Variation with module inclination
Several studies examine the effect of module tilt on the
accumulation of contamination [171,175,176,185–187,192,195,
197,198,203,206,207,210,211,213,215]. For specimens mounted
at a fixed angle, the effect of soiling is increased as the orientation
is changed from perpendicular (901) to parallel (01) to the horizon.
The difference is attributed to the effectiveness of natural cleaning
agents (precipitation, wind, and gravity), which may readily act on
the larger and more loosely adhered contamination [192]. Also,
precipitation cannot be readily shed from a module oriented at 01
2046 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068

Table 4
Summary of the examination of soiling in the literature.

Year Reference Location Environment Specimen(s) Specimen Duration kTtyp kTmax Measurement
(first surface) type (%) (%) type

1971 [170] N/A laboratory Glass mirror N/A 3 42 Hemispherical

1974 [171] Roorkie, India urban, continental Glass, PMMA sheet 30 days 14 63 Direct
1977 [173,178] Albuquerque, NM, USA rural, desert Glass, PMMA, mirror 5 weeks 5 23 Direct
Al
1978 [174] Albuquerque, NM, USA desert Glass mirror 1 month 4 13 Direct
1978 [175,176] CA, USA; NM, USA rural, desert Glass, PMMA mirror 1.5 years 7 14 Direct
1978 [177] Albuquerque, NM, USA desert Glass mirror 7 months 5 11 Direct
1979 [164] Hazardville, CT rural, temperate Glass sheet 6 weeks 2 7 Hemispherical
1979 [179] Albuquerque, NM, USA desert Glass mirror 1.5 years 4 8 Direct
1979 [180] Dhahran, Saudi Arabia desert Glass PV module 4 months 15 N/A Hemispherical
1979 [181] CA, USA; GA, USA; NM, rural, desert; rural Glass, PMMA, mirror 1–20 4 10 Direct
USA temperate Al months
1980 [182] AZ, CA, NM, USA desert Glass, PMMA mirror 1 year 5 6.5 Hemispherical
1980, 1983, [183,191,196] CA, USA desert Glass, PMMA sheet 1 year 9 19 Hemispherical
1986
1980 [185] CA, NM, NY, MA, WA, various, rural & urban Glass, PMMA, Sheet 150 days 8 21 Hemispherical
USA PDMS
1980 [184] Albuquerque, NM, USA rural, desert Glass, PMMA, mirror 10 7.5 12 Direct
Al months
1980 [186,187] Albuquerque, NM, USA desert Glass mirror 2 years 10 18 Direct
1981 [188] Crosbyton, TX, USA rural, arid Glass, PMMA, sheet, mirror 8 months 8 9.4 Hemispherical
Al
1981, 1982 [190,205] GA, HI, NM, NV, OH, OR, industrial: desert, Glass, PMMA, Sheet, mirror 1 year 8 50 Direct
TX, USA temperate, tropical Al
1983 [189] N/A laboratory Glass Mirror N/A 44.5 Direct
1984 [192] AK, WA, USA; Panama tropical, temperate Glass, PMMA, sheet 28 6 19 Hemispherical
PDMS months
1984 [193] Riyadh, Saudi Arabia rural, desert Glass PV module 21 22 33 Hemispherical
months
1985 [194] Riyadh, Saudi Arabia desert Glass CPV module 1 month 35 55 Direct
1985 [195] Safat, Kuwait desert Glass, PMMA sheet 36 days 15 80 Hemispherical
1987 [197] Safat, Kuwait desert Glass PV module 2 years 9 57 Hemispherical
1990 [198] Jodphur, India arid Glass, PMMA, sheet 1 year 4 62 Direct
PVC
1990 [199] Dhahran, Saudi Arabia desert Glass thermal-solar 1 year 7 65 Direct
module
1992 [200] N/A laboratory Glass sheet N/A 16 35 Hemispherical
1992 [201] Jodphur, India arid Glass PV module 1 year 8 30 Hemispherical
1993 [202] N/A laboratory Glass PV module N/A 90 Hemispherical
1995 [203] Morges, Switzerland urban Glass sheet 1 year 5 25 Hemispherical
1997 [204] Köln, Germany urban Glass PV module 5 years 4 24 Hemispherical
1997 [206] Phoenix, AZ, USA urban, desert Glass PV module 3 years 1 3 Hemispherical
1998 [207] N/A laboratory Glass sheet N/A 73 Hemispherical
1998 [208] Burgdorf, Switzerland urban Glass PV module 3 years 10 Hemispherical
1999 [209] N/A laboratory Glass PV module N/A 90 Hemispherical
2001 [210] Minia, Egypt rural, desert Glass sheet 1 month 15 26 Hemispherical
2001 [211] Madrid, Spain rural, continental Glass, PMMA CPV module 4 months 12 26 Direct
2003 [212] Abu Dhabi, UAE urban, desert Glass desalination 1 year 29 Direct
plant
2005 [213] Las Vegas, NV, USA urban, desert PMMA sheet 1 year 11 24 Direct
2006 [214] CA, USA urban, desert Glass PV site 1 year 6 27 Hemispherical
2007 [215] Southwest USA urban/desert Glass PV site 1 year 4 6 Hemispherical
2008 [48,49] Davos, CH; Raperswil, CH urban+ rural, temperate Glass, PMMA, sheet 20 years 4.5 14 Hemispherical
others
2009 [216] Las Vegas, NV, USA urban, desert PMMA sheet 1 month 10 12 Direct
2009 [217] Mountain View, CA urban Glass PV site 15 50 Hemispherical
months

(commonly used in installations on a flat roof to obtain the
maximum packing [therefore power production] for FP-PV mod-
ules); therefore, any dissolved sediment would remain trapped on
the module surface upon drying.
Studies of modules tracking the sun include Refs. [206,211].
Analysis examining the effect of soiling on concentrating
optics suggest they could be affected by an order of magnitude
relative beyond FP-PV [173,174]. Unless packaged with a front
glass [15], light concentrated with reflective optics must
pass through the contamination layer twice [186]. Reflective
optics are therefore more susceptible to soiling than refractive
optics [174].
Studies of soiling of modules mounted on a tracker indicate
that the accumulation of contamination is the least if the module
is stowed face down (1801) during the night. In this position, the
process of accumulation by sedimentation may be largely elimi-
nated [186], and the module radiates to the Earth rather than the
sky, cooling less rapidly-—preventing moisture condensa-
tion [175,176]. This storage position, however, does not take
advantage of natural cleaning events such as rain. If the weather
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
Fig. 4. Reflectance vs. time data for plate specimens examined in Albuquerque,
NM. Meteorological conditions are noted in the figure.
Adapted from Ref. [184] with permission.
can be predicted overnight or during adverse conditions, some
weather events may warrant prescribed orientation to exploit
natural cleaning. Orienting modules into the sun in the morning
or after precipitation will affect the dissipation of condensed
moisture, which has not been studied specific to soiling.
2.3.5. Variation with wavelength
The wavelength-dependent effect of soiling is mentioned in
Refs. [48,49,167,174,182,184,186,187,207], where the magnitude
of attenuation varies with wavelength, being more significant at
shorter wavelengths. Fundamentally, optical absorptance and
scattering may be analyzed using a Mie representation
[48,49,174,186,187,195,200,207]. Because the size of particulate
matter in the field (0.1 mmr|o50 mm [187,191,192,220]) is
similar to the wavelengths of the solar spectrum, a Mie model
is well-suited to the analysis of optical attenuation (as opposed to
the Rayleigh representation, applicable to smaller-sized matter).
The extinction coefficient, k (unitless), may vary by a factor of
5 with wavelength [187]; however, the wavelength dependence is
identical for both direct and hemispherical transmittances
[184,186,187]. In practice, the wavelength-dependent effect was
observed to evolve over time with the change in size distribution
of accumulated matter [186,187]. The use of a mirror in addition
to a Fresnel lens further complicates the wavelength dependence;
the effects are predicted for a silver, second-surface reflector in
Refs. [174,178,186,187].
2.3.6. Processes of adhesion and accumulation
The process of accumulation and mechanisms of adhesion for
soiling are examined in Refs. [174,181,183,185,187,191,192,
196,219,221,222]. Because of their mass, particles will settle out
of suspension in air when their velocity falls below the Stokes
settling velocity, analyzed in Refs. [174,221]. This implies that for
a Reynolds number o105 (laminar flow, occurring at low velocity
or near the leading edge of a surface) [174], sedimentation will
occur readily. Here, the rate of accumulation may approach the
terminal fall velocity [222]. Large surfaces, such as the front of a
refractive lens, will invoke turbulent flow. Turbulent flow exceed-
ing the deflation threshold may remove sediment, cleaning the
surface, as analyzed in Refs. [183,191,222]. The process of sedi-
mentation also dictates that the smallest particles will likely
2047
travel the greatest distance. The forces motivating a particle to
settle on a surface include: thermal gradient, electrical charge,
gravity, and localized turbulence (including Brownian motion for
small particles). The settling of airborne contamination is pre-
dominantly due to localized turbulence, which imparts an
approach velocity relative to the surface [222]. In cases where
the entrainment of contamination impinges directly on the sur-
face, deposition may instead occur as the result of kinetic impact,
in both laminar and turbulent flow regimes [174]. Dust and
water-soluble matter may also be transported by precipitation,
remaining behind as ‘‘water spots’’ after the surface has
dried [223]. The process of accumulation is believed to occur on
all materials (e.g., PMMA, glass, etc.) at a similar rate; surface
materials are therefore distinguished according to the rate of
removal of contamination [185].
Following its transport to the surface, contamination may
become subject to the mechanisms of adhesion, including gravity,
mechanical interference/friction, capillary condensation, electro-
static charge, charge double-layer (contact potential owing to
difference in the work functions), surface (van der Waals) forces,
hydrogen bonding, and chemical bonding. The attractive surface
forces, including those associated with London forces (permanent
and induced dipoles), are not distinguished here. Intimate con-
tact, facilitating the aforementioned mechanisms of adhesion, is
favored for a wet surface [222] because of its greater sticking
coefficient [182].
Moisture on the surface, including monolayers of water pre-
sent at elevated relative humidity, will significantly improve the
adhesion of contamination through the process of capillary
condensation. Capillary condensation will occur rapidly within
the first hour of contact, proceeding asymptotically within  24 h,
and becoming complete within a few humidity (e.g., moisture
condensation) cycles [174]. On a soft substrate surface, the sur-
face tension of the drying droplet may drive trapped particles into
the surface, forming craters [183]. The process of capillary
condensation is cited as the primary mechanism by which
adhesion facilitated by weak surface attraction is transformed
into much more robust chemical or physical bonds [183,221].
Most activity of the adhesive mechanisms will occur within a
period of 30–60 days [192,196].
The adhesion of particles is further explored in a three-layer
model (Fig. 5), proposed to represent accumulated contamina-
tion [183,185]. In Fig. 5, the lens and its surface are shown relative
to the three successive layers of accumulated matter, ‘‘A,’’ ‘‘B,’’ and
‘‘C.’’ Layer A consists of salts, organics, and cement-like matter,
which may accumulate from the environment and/or leach from
Fig. 5. Cross-sectional schematic (not to scale) demonstrating the three-layer
model for deposited contamination.
Adapted from [191].
2048 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
the surface of the lens [177,182,183,191,224]. The matter here
(which may be water-soluble) can move across the surface and
agglomerate in the presence of precipitation, condensation (dew),
or the monolayers of water found at high relative humidity
( 4  80% [183]). Layer A is tenaciously bound to the surface,
and typically may only be removed using abrasive scrubbing or
chemical etching. Layer B consists of particles that may range in
diameter (from sub-micrometer to micrometers), composition
(from organic to inorganic and/or polar to non-polar), and surface
energy (from high to low) with thickness. Particles here may be
loosely bound with intermediate matter. Layer B may be removed
by scrubbing. Layer C consists of particles typically of size greater
than in layer B, that are predominantly mechanically bound. The
layer is relatively inert, lacking the through-thickness energy
gradient present in layer B. Layer C is removed by adequate
precipitation. A temporary, fourth layer (not shown) consists of
even larger particles (diameter 450 mm). The sediment in layer D
is readily removed by wind.
One limitation of the three-layer model, shown in Fig. 5, is that
is has not been widely validated. The model originates within the
aerospace industry, based on the study of absorbed layers of
atmospheric gas on metallic surfaces [225]. The most similar
independent results occur in Ref. [181], where flakes of mineral
matter (composed of mica, quartz, or feldspar) were tenaciously
held in place by a cement-like layer (composed of carbonates,
salts, or sulfates). Here, the plate geometry of the mineral layer
and the limited thickness of the cement layer prohibited wetting
of the cement layer, preventing both materials from being
removed by washing/scrubbing techniques. The language and
specific details of the three-layer model were, however, refined
during the course of empirical study.
One of the great merits of the three-layer model, however, is its
consistency with many of the trends observed in the literature. For
example, the asymptotic accumulation of tenacious contamination in
30–60 days can be explained as the formation of the A and B
layers [185,191,192]. The formation of a tenacious film (composed
of carbonates, salts, and/or sulfates) in many desert sites is a very
significant aspect not simulated in artificial wind-tunnel experi-
ments [164,167,174,185,187,189,192,200,202,207,209,222,226,227].
The C layer is identified as being readily subject to synoptic events
(as in Fig. 4) or factors of influence (such as the inclination angle of
the module). Lastly, not all combinations of materials and site
environments were found to result in the successive formation of
accumulation resembling the A, B, and C layers [192]. In instances
where the A and/or B layers did not form, the subsequent accumula-
tion was more readily removed by natural cleaning processes.
The issue of uniformity of accumulation is examined in
Refs. [175,176,178,204,208,223]. The issue results not only from
the nature of the processes of accumulation and adhesion (which
are inherently heterogeneous) but also from the geometry of
installed PV modules. Five patterns of residual marks (distin-
guished by size, spacing, and streaking) were correlated with the
intensity of precipitation and wind [223]. In contrast, the fine dust
accumulated at ocean sites (composed of salt) can be more
uniformly distributed [226]. Regarding the module geometry,
the latitude-tilt angle used for FP-PV modules resulted in accu-
mulation at the bottom one-third of the front surface, due to
runoff of precipitation [204]. To explain, some module designs use
a frame (or clips) that protrudes above the edges (trapping runoff
containing dissolved sediment). This strip of accumulation ren-
dered a partial-shading condition (up to 24% power reduction),
confirmed in the measured current–voltage (I–V) profile of the
modules and models of the reverse-bias condition for the strings
of cells [204,208]. Heterogeneous accumulation can therefore
render the potentially hazardous hot-spot condition in FP-PV
modules [168,169], which is typically not an issue in CPV
modules—where each cell has a corresponding bypass diode. Flux
uniformity, which could be compromised by non-uniform soiling,
could reduce the operating efficiency of CPV cells.
2.3.7. Particle size-distribution, composition, and morphology
The size distribution of accumulated particles and those in the
atmosphere is examined in Refs. [167,185–187,219,223,228,229].
The size of accumulated particles was determined to range from
+¼0.02 to 190 mm, with 1–40 mm being common. The distribu-
tion of accumulated particles is modeled using a power law
(‘‘Junge’’) [186], logarithmic Gaussian plus power law [187], or
log-normal distributions [223]. Log-normal distribution is fre-
quently used in the atmospheric and aerosol sciences. The size
distribution may not be Gaussian (as is often assumed), because
the profile can be bimodal [187,229] or non-symmetric [167,187].
The distribution profile may also change over time as the result of
particle-particle attraction and agglomeration [187]. The size
distribution of particles present in the atmosphere varies with
the meteorological stability; stable conditions (and therefore finer
particles, less likely to deposit) tend to be present at night and
during cold weather [219]. Regarding the local topography of an
installation site, coarser particles generally deposit on the wind-
ward side of a hill, with little accumulation occurring beneath the
separation zone (formed at the peak and extending for some
distance along the leeward side) [222,227].
The composition and morphology of contamination is exam-
ined in Refs. [181,183,185,186,190–192,205,208,209,228,229]. In
these studies, materials were observed that included soil (dust,
sand, clay, and debris), salt, pollen, algae, lichens, metals (from
manufactured, natural, industrial, or transportation sources), soot,
paint, oil (organic or industrial), and plant matter (fallen leaves).
The deserts of the southwestern U.S. are known for their alkaline
chemistry, which may affect the availability, composition, and
adhesion of contaminants [190,205]. Less accumulation was
observed in specimens exposed in temperate locations, which
may be due to the acidic soil and/or the greater precipitation
occurring there [205]. Swellable clays (such as montmorillonite)
can develop a large contact area, increasing adhesion in their
hydrolyzed state. Carbonates, salts, and sulfates (including calcite,
gypsum, halite, and sylvite) are water soluble. This allows them to
be transported interstitially in their dissolved state, where they
may bind undissolved matter to the lens surface upon drying.
Calcium carbonate (present in both alkaline and acidic environ-
ments) is one of the principle ingredients of Portland cement
(identifying its adhesive capability). Less soluble matter found in
the natural environment includes non-swellable clay (such as
illite and kaolinite) and aluminum silicates (such as quartz,
feldspar, and mica). These materials may act as the ‘‘bricks’’
present in tenaciously bound contamination. Bird droppings are
a particularly problematic biological contamination, because
they are not readily removed by natural cleaning agents
[206,211]. Particulate morphology is specifically examined in
Refs. [181,185,186,208,229], where it includes spheroidal, oblate,
irregular, and plate shapes.
The accumulation, adhesion, and composition of contamina-
tion may demonstrate significant site-specific variation. For
example, the factors inherent to urban installation sites are
described in Refs. [182,190,204,205,208,228,229,230]. Early PV
studies describe the urban environment as being more severe
than rural locations. For example, the power yield (i.e., current) of
an urban site (New York City) over time is identified to become
half that of an agricultural site (Meade, Nebraska) [230]. In a
separate study, the transmittance of urban and rurally located
concentrators decreased by 60% and 15%, respectively [182].
Scattering (which may aid FP-PV by providing multiple passes
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
within the cell, but renders optical loss in a CPV system) is
identified as being 2  and 10  more significant than absorption
in urban and rural installation sites, respectively [182]. In addi-
tion to the reduced proportion of scattering relative to absorption,
the size distribution of particles (e.g., soot [185]) in an urban
environment may heterogeneously filter the wavelengths of the
solar spectrum, affecting the current-limiting condition in multi-
junction cells. For example, reduced throughput of blue wave-
lengths could result in a more pronounced power loss than
predicted from the complete spectrum, because the junctions
are connected in series [231]. Pollution control has evolved
significantly since the early studies, as enabled by catalytic
converter, scrubber, and other technologies. The factor of proxi-
mity [229], however, is still expected to be important. For
example, specimens located close to food or beverage processing
facilities are observed to accumulate surface films composed of
oily hydrocarbons and/or organics [185,190,205]. These films
were found to greatly mitigate soiling, presumably by preventing
the formation of the A layer (Fig. 5). The difference between urban
and rural application sites is used as an example here, but factors
including the meteorological micro-environments in or near
desert areas can also render significant site-specific variation.
The requirements for an environment to remain less soiled
include: minimal airborne organic vapors, either frequent con-
sequential rain or extended dry conditions (minimal dew and low
relative humidity), and few dew cycles or occurrences of high
relative humidity between rain periods [183,191,192].
In some countries, the concentration and composition of
particulate matter is publicly examined by specific task
groups [232] or agencies, such as the Environmental Protection
Agency (EPA) [233]. The EPA distinguishes between fine-sized
(+r2.5 mm, indicated as PM2.5) and intermediate-sized
(2.5 mmo+r10 mm, PM10) particulate matters. One reason for
this distinction is the anticipated range of travel: PM10 may
remain airborne for  150 m, whereas PM2.5 may travel much
farther. Including composition and origin, existing studies typi-
cally examine long-term, seasonal, and local trends. Extensive
location-specific data are available in some metropolitan areas.
The planning (including choice of site location), operation (solar
resource, including intensity and spectral effects), and mainte-
nance (including cleaning based on forecast information [234]) of
large-scale PV installation sites may readily benefit from available
studies or collaboration with existing agencies.
2.3.8. Prevention of soiling and cleaning of surfaces
Techniques related to the prevention and cleaning of accumu-
lated matter are examined in Refs. [164,173,174,176,178,181,
183,185,187,190–192,196,205]. A surface resistant to soiling is
identified as requiring the following qualities: hard, smooth,
electrically neutral, low surface energy, chemically clean (of
adhesive materials, water soluble salts, and group I elements),
and minimally diffusing/absorbing of water [183,191,192]. These
requirements have not been widely validated, and contradicting
requirements may occur for specific location types. The use of a
hard coating, such as PDMS [183,185,191,192,196], KORADs
acrylic [183,185], SiO2 thin films [235], ORMOCERs [236],
or plasma treatment and deposition [237], are advocated for
PMMA, which is substantially softer than glass. Regarding PDMS,
a hard (‘‘resin’’) formulation must be used, because the softer
elastomer and gel formulations were prone to significant soil-
ing [185,191,192,196]. The issue of surface smoothness, includ-
ing specific roughness requirements, is not examined directly
in the literature. The contrasting requirements of smoothness
(minimizing windward features) and hierarchal microscale sur-
face texture (proven in nature (e.g., Lotus leaf) to promote
2049
cleanliness [238,239]) suggest an optimization problem. The
soiling of a surface textured for antireflection as in moth’s
eye [236,240,241] or porous films added to create an index-
graded AR coating remains to be examined.
Surface coatings including transparent SnO2 [174,182] films
and antistatic coupling agents [164] proved somewhat helpful. In
theory, the use of applied potential and/or antistatic technology
becomes more applicable as the system voltage is increased,
which is the historical trend. Lenses may however be constructed
out of polymeric materials, prone to charge accumulation.
Popular approaches used to manipulate water for the purpose
of cleaning include ‘‘superhydrophilic’’ and ‘‘superhydrophobic’’
coatings [242]. A superhydrophilic surface would have a large
surface energy and the corresponding contact angle approaching
01. This encourages wetting during precipitation, enabling the
water to readily transport dirt. TiO2 is a hydrophilic material,
where its character as a semiconductor (Eg ¼3.2 eV [243])
encourages photocatalysis of organic matter (including oils,
bacteria, and viruses) [242]. That is, external irradiation (where
l o390 nm) enables TiO2 to automatically decompose organic
matter into product species, including water. A superhydrophobic
surface has a surface energy much less than that of water,
rendering a large contact angle (preferably yc 41601). To achieve
superhydrophobicity, synthetic fluorinated polymers (where
yc o1301) must also be patterned with a microscale surface
texture to ensure that water will roll to mimick the cleaning
action found in the Lotus plant [242]. The goal for superhydro-
phobic surfaces is to shed water immediately, preventing any
wetting to the surface. The rolling of droplets furthermore
facilitates cleaning of surface contamination. The appropriate
coating (hydrophilic or hydrophobic) for a particular environment
(based on frequency of precipitation) is not currently well-
understood.
Historically effective hydrophobic surface coatings include:
fluorinated silane (L-1668, 3M Corp.) [191,192], fluorinated
acrylic films such as Acrylars [191,192], perfluorodecanoic acid
(E-3820-103B, Dow Corning Inc.) coupled with silane (Z-6020,
Dow Corning Inc.) [191,192], PDMS (650 glass resin, Owens-
Illinois Inc.) [164], PMMA (AT-50, Rohm, and Hass Co.) combined
with polyethylene glycol ester (Flexol 40G0, Union Carbide
Co.) [164], hydroxyethyl cellulose coupled with titanate (Tyzor
TE, E. I. du Pont de Nemours and Co.) [164], silicone fluid (L-31,
Union Carbide Co.) coupled with stearic acid [181], and silicone
fluid (L-7001, Union Carbide Co.) coupled with stearic acid [181].
As suggested by their contact angle in Table 2, PDMS and PMMA
inherently have a low surface energy, rendering hydrophobicity
relative to glass. None of the aforementioned coatings or materi-
als are inherently superhdrophobic. The different surface energy
and chemistry of glass, PMMA, and PDMS suggests that these
different superstrates may require different surface functionaliza-
tions and/or coating materials. The materials investigated above
are by no means the only possible methods of reducing surface
energy. Surface functionalization is one of the core topics in the
recently popular field of nanoscience.
The tin-rich float side in float-manufactured glass is relevant to
surface coatings [164]. This side is usually oriented toward the
environment in FP-PV because of improved adhesion between
polymeric encapsulation and the tin-poor side of the glass. The
convention for CPV is not well defined; however, better adhesion
of hydrophobic coatings is anticipated for the tin-poor surface of
float glass [164]. Any coating (whether hydrophilic, hydrophobic,
or anti-static) will absorb some of the solar spectrum, reducing
the available optical flux and affecting its color balance. The cost
analysis for use of coatings (which may reduce soiling) vs. the
reduced power yield (due to the lesser optical transmittance of
the coating) is not examined here.
2050 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
Regarding cleanliness, it is important to remember that sub-
stances used in manufacturing (including organics or mold-
release agents) may remain as residuals near the surface. Glass
is prone to corrosion in the natural environment, including ionic
transport in its interior and aqueous transport of Na, Ca, Mg, and
other trace species at its surface [183,244]. The group I ions,
including sodium and potassium, create a hydrophilic salt film if
leached to the surface; therefore, the replacement of the group I
species with group II/III ions (yielding a hydrophobic surface film)
was recommended for glass or acrylic [183]. Ingress or absorption
of water in the field enables the leaching or addition of species at
the surface, facilitating the formation and/or adhesion of a
contamination layer.
Mitigation of soiling may also be achieved at the installation
site itself. The use of gravel, grass, and vegetation (including
microphytic biological crust in desert areas [226]) rather than
exposed soil can limit the transport of particulate matter at
its source. The same might be achieved by watering or chemi-
cally treating parched soil. Regarding the use of gravel, its
installation must prevent the risk of its transport and subsequent
impact during wind storms, as in Ref. [246]. Methods for urban
installation sites include the sealing of roofs and other mate-
rials that might outgas [190,205], and the filtering/directing of
exhaust [185,190,192,205].
Methods of actively cleaning or maintaining a clean surface
include: use of water (rinse, high-pressure spray, or mist),
detergent, chemical etchant, wiping/scrubbing, sonication, remo-
vable/replaceable coatings, and air flow (boundary layers, spoi-
lers, or turbulators). Regarding the use of water, its effectiveness
might be improved when the contaminated surface is already wet
with dew [191]. The effectiveness of water decreases after
accumulated contamination experiences its first humidity cycles
(within about 2 weeks) [174]. Detergents examined include: Jet-
X [173], CB-120 (McGean Chemical Co.) [173,176,181,190,205],
A69M (for acrylics, McGean Chemical Co.) [176,181], Lime Brite
(Vestal Laboratories Div.) [181,190,205], and Alconox (Alconox,
Inc.) [185,230]. Requirements for detergents include: reduced
surface tension, low cost, dispensability, and biodegradability.
Although chemical reaction is not forbidden, the effective clean-
ing solution CB-120 may not be practical in many locations,
because it contains 5% hydrofluoric acid [181] (which may be
harmful to the site environment or even legally prohibited). The
addition of a sheeting agent to a cleaning solution is recom-
mended to prevent spotting [181], particularly if hard water is
present. Cloth towels, paper tissues, sponges, and pumice were
used for mechanical scrubbing. Some studies avoided mechanical
cleaning to prevent damaging the surface [190,205]. One study
qualitatively documents the effects of scrubbing [164], which
might now be examined quantitatively using a variety of surface
characterization methods. Deionized or particulate-filtered water
was often used both before and after surfactants and scrubbing.
Additional techniques (including sonication [174,187]) are
described, but not used beyond laboratory study. The use of a
spoiler [174] to regulate turbulence and uniformity is likewise
examined only in a cursory manner. Spoilers, in addition to active
air currents [174] or streamlining [174], are proposed but not
examined in field tests. The use of a substitute, removable, and/or
replaceable surface [174,181] is likewise only proposed. Here,
material may resurface (fill in) damaged surfaces, replenish a
hydrophobic coating, or simply be removed and replaced for
cleaning.
Even with the aforementioned methods of prevention and
cleaning, a permanent loss of transmittance is often observed.
Refs. [171,176,181,184,185,187,191,192,195,198,205] suggest
that soiling reduces transmittance for PMMA relative to glass
surfaces. From the studies, at least a 2% and 4–8% permanently
retained loss should be expected for glass and PMMA, respec-
tively [181,187]. Refs. [48,49] instead identify PMMA remained
cleaner, because it could be better cleaned than glass. In contrast,
Ref. [185] indicate that glass remained cleaner, identifying that
PMMA was scratched by contact cleaning methods. Ref. [176]
convincingly argues that glass is more readily cleaned, because
PMMA is more hydrophobic. Separately, Refs. [190,205] advocate
the selection of glass or PMMA based on other criteria, e.g., the
likelihood of a reflective layer being corroded in certain environ-
ment types. Although several early studies favor glass relative to
PMMA, the phenomenon of soiling is so complex—depending on
precipitation, particulate composition, pH variation, and other
factors—that each material may have separate preferred locations
for use.
2.3.9. Future study
A reasonable collection of studies related to soiling comes
from disparate sources. Although some consistent themes and
conclusions may be drawn, the subject could benefit from wider
validation and further examination.
The literature lacks comparison of key representative sites
throughout the world. Examples of unexamined prospective
regions and countries include: Europe (e.g., Spain and Italy), north
and south Africa, the Middle East (e.g., Saudi Arabia, Kuwait),
India, west China, and Australia. Examples of environments
needing greater study include: tropical ocean (e.g., Florida), ocean
desert (e.g., the Baja California peninsula), and urban locations. In
all cases, the particles (size distribution, shape, and composition),
sticking coefficient of the surface, and optical properties (trans-
mittance) of accumulated layers should be examined. Additional
relevant information includes: mass accumulation rate, module
mounting (orientation, tilt angle, and cardinal direction), and
meteorological conditions at the site. The issues related to air-
borne and deposited matter (as well as temporal variation) are
also relevant. Natural soiling is a complex subject; therefore,
seemingly contradictory results may be expected, as dictated by
location-specific phenomena.
While expanding and comparing the database of known test
sites, several other concerns may be addressed, summarized here
in the following set of questions: How does soiling compare
between fixed-angle and tracking modules? Is the three-layer
morphology model in Fig. 5 valid for a variety of test sites over an
extended duration? How does a PMMA lens compare to a glass
lens—specifically, the evolution of damage to the first surface?
(there may be separate types of sites where either PMMA or glass
prove more advantageous). Are the contemporary prevention and
cleaning technologies effective after prolonged exposure in the
field? Can the accumulation of contamination be predicted via
modeling (or a historical meteorological baseline), which may be
used to prescribe cleaning? (The solar resource [compensated for
existing environmental conditions] could alternately be compared
to the actual power yield to prescribe cleaning). Would modeling
of the adhesion phenomena, such as capillary condensation,
reveal actionable insights?
3. Failure mechanisms
The following section focuses on the degradation mechanisms
specific to PMMA, i.e., the processes of photodegradation and
thermal degradation. Photodegradation refers to aging facilitated
by high-energy radiation (ultraviolet light), whereas thermal
degradation refers to aging facilitated by elevated temperature.
Regarding photodegradation, topics here include: background
information (figures of merit, details of specimen preparation,
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
additives, and the light source used), the enabling physiochemical
mechanism, monomer-mediated catalysis, comparison against
radiolysis, the influence of the atmosphere (test environment),
and relevant analysis methods (including optical and infrared
spectroscopy). Regarding thermal degradation, topics here
include: background information (figures of merit and the details
of specimen preparation), the enabling physiochemical mechan-
ism, separate factors of influence (ambient air pressure, content of
atmospheric O2, Mw, and stereochemistry), synergy between
photodegradation and thermal degradation, and monomer-
mediated catalysis. A broad variety of information specific to
PMMA exists in the literature; much of this work was conducted
at academic institutions. General resources describing photode-
gradation and thermal degradation include Refs. [247–249].
3.1. Photodegradation: background and related details
Much study related to the photodegradation of vinyl polymers,
including PMMA, was conducted in the 1950s to 1970s.
Table 5 summarizes that seminal and the more recent work.
Details in the table include: the irradiation wavelength(s), source
of illumination, atmosphere (or solution) present, test tempera-
ture, use of reprecipitation (when applicable), formulation addi-
tives (when examined), and the quantum yield of reaction. The
latter quantity is related to the dose of radiation through Eq. (7)
[250]:
Z lmax
D¼ E½l,t ð1e½ha½l ÞF½l@l ð7Þ
lmin
New parameters in Eq. (7) include: D, the dose (W m  2); E, the
spectral irradiance (W m  2 nm  1); h, the thickness (m); a, the
absorption coefficient (m  1); and F, the quantum yield (unitless).
The quantities E, a, and F are inherently wavelength
dependent. Table 5 provides multiple F values in cases where
multiple conditions (atmosphere, l, or T) were used.
Many of the specimens in Table 5 were obtained from purified
(washed and precipitated) monomers, which can be polymerized
by light or heat, either with or without an initiator. ‘‘Reprecipita-
tion’’ in Table 5 specifically refers to the refinement (removal of
additives or impurities) of PMMA after it has been polymerized. In
this way, trace components (including additives such as stabili-
zers), release agents, co-polymers, and residual monomer may be
removed. Regarding initiators, benzoyl peroxide [251,256], azo-
di-iso-butyronitrile [62,63,251,252,257,268], and lauroyl perox-
ide [254] were used. Rather than direct fabrication via polymer-
ization, sheet [67,253,263,270] or powder [54,55,57,58,269]
specimens were sometimes obtained from commercial vendors.
In some cases, additives were introduced into PMMA during
polymerization or reprecipitation to examine their performance
as photostabilizers. Many of these substances were found to be
effective based on their opacity at 254 nm [263]. That is, the
additives had no photochemical interaction with PMMA. How-
ever, pyrene [253] as well as the combination of benzotriazle (a
UV stabilizer) and hindered phenol (a radical scavenger) [266]
were found to protect PMMA, in addition to their optical-shield-
ing capability. Benzophenone (a matrix-softening plasticizer) was
also found to protect against photodegradation, attributed to
enhanced radical mobility (enhancing combination and scaven-
ging) [269]. Curiously, the same substance was identified as a
detrimental photosensitizer in a later study, where the spectral
bandwidth of damaging wavelengths was also increased by
benzophenone [58].
The UV absorbers and stabilizers used in commercial PMMA
products, which may readily affect their UV aging characteristics,
are not examined in the public literature specific to PMMA. Such
2051
details have instead remained proprietary. In general, some types
of stabilizers, such as hindered amine light stabilizers (HALS), are
self-regenerating, whereas others are not. The UV stabilization
system, however, need not be depleted to enable degradation.
Other additives (such as those used to facilitate processing during
manufacture) may catalyze material degradation. Furthermore,
comprehensive study of the effect of the common co-monomers
(methyl acrylate and ethyl acrylate, which may be included up to
5% [393]) was not found in the literature.
The majority of studies in the literature use a low-pressure Hg
lamp, for which most of the optical flux (at least 80–90%) occurs
at the wavelength of 253.7 nm [263]. Although this is within the
range of the extraterrestrial solar spectrum (i.e., AM0 [272]),
irradiation at this wavelength is not encountered at locations on
Earth (e.g., AM1.5 [68]). When using artificial illumination, one
should be wary because the degradation mechanism(s) and the
corresponding F may not be applicable at all wavelengths.
In Table 5, the mechanism of random chain scission is described
in most of the studies for l ¼254 nm. Another radiation peak at
184.9 nm may also be used in some low-pressure Hg lamps [66].
For greater internal pressure, the emission spectrum of a Hg lamp
falls within terrestrial solar spectrum, i.e., l Z280 nm. A xenon-
arc lamp (which closely emulates the terrestrial UV solar spec-
trum [250]) or a custom monochromatic instrument (using a
double-blazed grating) [54,55,57,58,269] are used in more recent
studies. These studies confirm that degradation, i.e., chain scission
at random locations along the polymer chain, can occur at
wavelengths within the terrestrial solar spectrum, with the upper
limit for damage being 300 r l r320 nm [55,57,58]. That is,
damage did not occur at 340, 400, or 500 nm. A different study
reports degradation at greater wavelengths (decrease in effect for
l ¼350, 436, and 546 nm, respectively) [52]. A distinction is made
between homolytic (direct) chain scission ([66], specifically for
300r l r320 nm in [55,57,58]) and that initiated at ester side
groups (b-scission) ([54,55,62,63,65,66,253,261,267,270], specifi-
cally for l o280 nm in [57]). The initiating site (main chain or
ester side group) is not specifically identified in [50,56,58,67,
252,254–256,258–260,263,264]. Because the initiating site is
contended, suggesting the need for additional research using
radiation within the terrestrial solar spectrum.
3.1.1. The process of photolysis
The dominant mechanism of photodegradation of PMMA is
summarized in Fig. 6. In many studies, initiation (aided by UV
light, hv) is attributed to the carbonyl structure (CQO) of the
ester side group (  CO2CH3) (arrow in Fig. 6(a)). This results in the
cleavage of the ester side group, rendering the methylene groups
(  CH2  ) vulnerable (arrows in Fig. 6(b)). The loss of the ester
group renders a greater free volume in PMMA (enabling physical
aging and creep), and also inhibits the b-relaxation mode
(decrease in tan db) [135,291]. Once freed, the ester group may
decompose into species including methanol or methyl formate
(Fig. 6(c)). In air, additional volatile product species include
hydrogen, carbon dioxide, and carbon monoxide [51,65,261].
The relative yield of volatile species is examined in
Refs. [51,52,62,252,257,264,268], whereas the quantum yield of
the volatiles and the radical species is identified in Ref. [268]. The
example in Fig. 6 illustrates an unsaturated termination at one
end of the severed chain (arrow in Fig. 6(c)). Prior to additional
chemical interaction eliminating this remaining free radical
(arrow in Fig. 6(c)), scission may also release methyl methacrylate
(monomer) from the unterminated end (compare the left end
in Fig. 6(c) with the active end in Fig. 7(b)).
Although cross-linking occurs during the photodegradation of
other polymeric materials, PMMA has been identified to degrade
 2052
Table 5
Summary of the examination of the photodegradation of PMMA in the literature.

Year Reference Reprecipitation Additives l Irradiation Irradiation source Atmosphere or solution T, F, quantum yield
(nm) temperature (scissions/quantum)
(1C)

1952 [251] 254 Hg lamp (low pressure) Vacuum 150–201 220

1961 [252] Tetrahydrofuran- 300–400 Hg lamp (medium pressure) Air and vacuum 23 0.017 0.044
methanol
1961 [253] Acetone Pyrene, p-terphenyl, xylene, benzene, Pb acetate, 254 Hg lamp (low pressure) Air 23 0.032
and Mn acetate
1961 [254] 254 Hg lamp (low pressure) Air 26 0.0023
1962 [50] Methylene chloride 254,300 Hg lamp (low pressure) and Air and nitrogen 20 0.012
biosol

D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
1963, [62,63] Tetrahydrofuran- 300–400 Hg lamp (medium pressure) Air and vacuum 23 0.016 0.035
1964 methanol
1964 [255] Tetrahydrofuran- 254 Hg lamp (low pressure) Methylene chloride, dioxane, 25 0.071 0.085 0.187
methanol and 4:1 CCl4:CH2Cl2
1965 [256] Pure, hydroquinone 254 Hg lamp (low pressure) Chloroethanol 25–155
1966 [257] Benzene-heptane Acrylaldehyde, methacrylaldehyde, and methyl 254 Hg lamp (low pressure) Benzene 30
acrylate
1967 [258] Methylene chloride 254 Hg lamp (low pressure) Air 23 0.002
1971 [259] 254 Hg lamp (low pressure) Air 160–200 0.054 0.52
1972 [260] Ethyl acetate- 254 Hg lamp (low pressure) Air 23
methanol
1973 [261] Poly(methyl acrylate) and n-butyl acrylate 254 Hg lamp (low pressure) Air 23
1973 [262] Methanol Methyl vinyl ketone 4 290 Arc lamp (high pressure) Air 25
1974 [263] Methylene chloride- p-terphenyl, pyrene, chrysene, and monomer 254 Hg lamp (low pressure) Air and helium 23 0.0079 0.0040
methanol
1977 [264] Choloroform - 200–600, flashed Xe lamp (unfiltered, Vacuum and nitrogen 30-Oct
methanol 4330 filtered)
1978, [51,52,265] Methylene chloride- 254, 300–400, Hg lamp (low, medium Air and vacuum 23 0.038 0.029
1979 methanol 436, 546 pressure) , Xe arc lamp
1979 [266] Sterically hindered phenol, benzotriazole, chain 4 280 Xe arc lamp Air 30–40
transfer agent, and methyl acrylate
1980 [267] Acetone Methyl vinyl ketone and methyl isopropenyl 254 Hg lamp (low pressure) Methyl acetate 23
ketone
1980 [268] Methanol 254, 266 Hg lamp (low pressure) and Vacuum  196 0.05
Nd-Yag laser
1982, 1992 [56,269] Acetone, methylene Benzophenone and 2-hydroxy-4- 300–400 Hg lamp (medium pressure) Air and vacuum 30,40
chloride-methanol methoxybenzophenone
1984 [53,270] Monomer 4 280 Xe arc lamp Air and vacuum 50, 85, 115,
125
1990 [54] 365 Hg lamp (high pressure) Air 140–180
1990 [55] Acetone-methanol 260, 280, 300 Custom Air 23 0.00021 0.00023 0.00041
1991 [67] 254, 4300 Air 23
1993 [57] Acetone-methanol 260, 280, 300 Custom Vacuum 23 0.000084 0.00011 0.00042
1995 [58] Acetone-methanol Benzophenone (photosensitizer) and 2-hydroxy-4- 260–400 Custom Air 23
methoxybenzophenone
1997 [271] Benzophenone 284 KrF laser Air 23
1999 [65] Tetrahydrofuran- Vinyltriethoxysilane 259 Hg lamp (high pressure) Air 23
methanol
1999 [66] Tetrahydrofuran, 185 Hg lamp (low pressure) Vacuum 30–130
chloroform-methanol
2001 [59] 4 280 Xe arc lamp Air 23
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
Fig. 6. Schematic representing the photodegradation of PMMA. In the sequence:
(a) initiation begins at an ester side group, whereupon (b) the resulting radical
(c) facilitates b-scission of the chain, while the cleaved side group decomposes.
Adapted from [261].
Fig. 7. Schematic representing the thermal decomposition of PMMA into MMA.
Shown here for initiation at a chain end, the sequence occurs when: (a) an
unsaturated bond becomes vulnerable at elevated temperature, T. After being
broken, (b) the resulting radical configuration will depolymerize the chain by
sequentially releasing monomer, unless (c) a stable end-group is formed.
Adapted from [51,310].
solely through photolysis. The dominance of the process of
photolysis is attributed to blocking by the methyl groups on the
main chain [273]. In the studies in Table 5 that examine Mw,
random photolysis is deduced from the constant rate of scission
2053
with time. The mechanisms of scission and cross-linking, however,
compete when additives [58] or co-polymers [257,260–262,274]
are present. In those studies, the incidence of photolysis was
always increased relative to pure PMMA homopolymer. Quantum
yield was decreased by  3  for a (2:1) syndiotactic:isotactic
PMMA mixture [258]. However, this enhancement of photostabil-
ity was not explored further. A general model for predicting
service life of PMMA components was developed in Ref. [275],
based on the F of photolysis and an assumed Poisson damage
distribution.
3.1.2. Monomer-enabled catalysis
The effect of monomer (methyl methacrylate (MMA)) on
photolysis is directly examined in Refs. [36,263,270,276]. MMA
typically remains as the principle impurity ( o5 wt% [277]) pre-
sent in commercial PMMA, particularly if polymerization does not
occur above Ta [276]. The monomer remains as a residual because
it was not polymerized during material fabrication. Additional
molecules of MMA may be liberated by photolytic chain scission
in the time duration required for stable chemistry to develop at
the severed chain ends (Fig. 6(c)). Unfortunately, its F identifies
monomer to be photoactive, resulting in the accumulation of free
radicals and consumption of –CH2– groups [263]. The formation
of a hydroperoxide from MMA, vulnerable to heat and/or light, is
identified in Ref. [36,266]. The diffusion of monomer and per-
oxide-forming species such as O2 is limited at room temperature
(limiting synergistic effects), but diffusion becomes greatly
enhanced as the temperature [270] or surface area of the speci-
men is increased. In contrast to MMA catalyzed degradation, an
earlier study attributed an increase in refractive index to the
photo-induced (l ¼325 nm) polymerization of residual monomer
(aided by residual initiator) [276].
3.1.3. Comparison with radiolysis
A substantial literature, including Refs. [50,135,253,273,
278–293,295], exists regarding the effect of alternative irradiation
on PMMA. Some of this work facilitates understanding of
advanced photolithography conducted using PMMA photoresists.
Sources include extraterrestrial radiation [64], 60Co (gamma rays)
[50,135,273,279–281,283,286,288–293], a Van de Graaf generator
(electron radiation) [253,282], an atomic pile (neutrons and g-
rays) [278,281], a cyclotron (proton radiation) [290], and an X-ray
synchrotron [287,295]. High-energy radiation is identified as
being 8 times more likely to render chain scission [253], with
notable differences in the damage yield for electrons and ions,
depending on the incident energy [287]. Refs. [284,285] explicitly
identify that the products, their yield, and propensity for chain
scission (vs. side-group cleavage) is different between g and UV
radiations. The accumulated energy per scission ranged from
5.7 to 83 eV [50,253,273,280,281,295], with 61 eV (F ¼0.0073
scissions/ion) being common for 60Co.
The following details specific to the literature of high energy
radiation-induced degradation are included here because of their
profound implications if they were identified for optical radiation.
Additional chemical species including CO, CO2, and propane
(formed in a vacuum) [135,253,280,285] and formaldehyde
(formed at ambient pressure) [281] are observed. These were
identified as molecular species cleaved from the main chain and
more greatly affected by the high energy radiation. Irradiated
PMMA was observed to bubble when heated [278–281,293] or
dissolved in solution [278]. The bubbles are identified as volatile
product species, trapped within the bulk PMMA at low tempera-
tures [281]. The observed bubbling implies that significant
porosity and mechanical stress are created during irradiation. In
some studies, radiation-induced discoloration (yellowing) is
2054 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
observed (or determined using a spectrophotometer) to lessen
over time [50,252,253,293]. The responsible color centers are
identified as free radicals, trapped within the bulk material until
they become quenched by diffused oxygen [288,293] or
light [286]. If so, the greater discoloration observed for neutron
vs. g radiation [278] suggests that radicals are formed in propor-
tion to the energy of radiation. Lastly, cleavage of C–H bonds in
the quaternary carbon (–CH2–) group (a Norrish type II pro-
cess [59,262,294]) is proposed to initiate abstraction of the ester
group in Ref. [283]. This explanation is consistent with the ratio of
ester abstraction to chain scission (  0.8:1) [281,283], as well as
the production of hydrogen identified in addition to that occur-
ring from the side groups [135,281,283,295]. In contrast, other
researchers identified the –COOCH3 anion, –CH–, _ and – COOCH _ 2
radical species at 77 K, with only the latter contributing to chain
scission (and remaining stable) at elevated tempera-
tures [286,289]. Whether or not the Norrish type II process
applies as a degradation mechanism (in addition to ester side-
group initiated scission) for any radiation source warrants
examination.
To summarize, although radiolysis may occur via the same
mechanism as photolysis, there are some disparate observations.
Fundamentally, radiolysis may invoke ionization reactions
(including alternate radical reaction processes, consumption of
additives/stabilizers, excess trapped radicals, and deteriorization
of byproducts) that are not generated by UV exciton events.
Further, the different figures of merit for photolysis (quantum
yield, F (the number of scissions per quantum of light), which
may be characterized according to wavelength via the action
spectrum [250]) and radiolysis (yield, Gr, the number of scissions
per 100 eV of total energy absorbed) make direct comparison
difficult. Regarding the development of PMMA Fresnel lens
technology, rather than attempting to simulate outdoor aging
using alternative radiation sources (which may prove infeasible),
it is advisable to work with concentrated light within the
terrestrial solar spectrum. It remains to be proven that optically
facilitated chain scission is the dominant degradation mechanism
affecting modern PMMA co-polymer materials used in CPV lenses.
3.1.4. Influence of ‘‘the atmosphere’’
Regarding the atmosphere, there is disagreement whether an
ambient or vacuum condition enhances photolysis. In the more
recent studies, where atmospheric conditions are directly com-
pared (Table 5), the ambient condition is often slightly (F r2  )
more damaging. Physical appearance, including increased disco-
loration, is also used to argue that O2 aids initiation of photo-
degradation [270]. Taking this further, the diffusion and
consumption of oxygen is considered in Ref. [53], where the
activation energy of photolysis (19 kcal/mol) is estimated. Per-
haps more significant is the influence of a solvent, where F may
be increased by more than an order of magnitude. Photodegrada-
tion of PMMA is explained to be regulated by a cage effect, where
the ordinary chain and periodic side-group structure motivate the
radicals generated by radiation to be confined and recombined,
preventing them from facilitating degradation. The cage effect is
reduced in solution [259,261] or at elevated tempera-
tures [275,53]. In both cases, neighboring polymer molecules
become separated, reducing the cage effect. The cage effect may
preclude many common solvents (including methanol and acet-
one) from use as cleaning agents.
3.1.5. Analysis tools
Photodegradation may be assessed by spectral analysis. Regarding
the optical spectrum, a peak at 224 nm occurs in proportion to the
monomer concentration [256,259]. This absorption region for MMA
extends to  320 nm, while the optical absorptance region for
commercial PMMA extends to 390 nm [60]. The absorption at
254 nm for monomer, however, is identified as being 200  that of
PMMA [263]. From Ref. [135], the participation of MMA is therefore
likely in Table 5, particularly in those studies using a low-pressure Hg
lamp. Note that the optical absorptance for MMA does not specifically
imply light induced degradation; furthermore, a quantum yield
typically applies at vulnerable wavelengths. Increased absorptance
at 285 nm or for 260 nmo l o280 nm is documented to result from
photo-aging in many studies [41,48,51,52,55,57,58,60,61,64–67]. This
phenomenon is attributed to the formation of a carbonyl
group [51,52,55] (or similar ketone [55] or aldehyde [55,257,260]
species containing a conjugated double bond [260]) on the chain or in
the ester side group [55]. The lack of precise consensus here may
indicate multiple chromophore species. Separate from the transient
discoloration mentioned in Refs. [50,252,253,293], a fluorescence was
observed at 375 nm [270] and other unspecified wavelengths [53].
The spectrum at longer wavelengths is examined in
Refs. [51,55,56,58,59,65–67,81,235,237,260,271,296–299] using
Fourier transform infrared spectroscopy (FTIR). Major and minor
peaks are tabulated in Refs. [59,296]. Changes in the spectrum
may include: a decrease at 2962 cm  1 (corresponding to stretch-
ing of the CH3 alkanes [300]), a decrease at 2926 cm  1 (stretching
of the –CH2– alkanes [300]), a decrease at 2872 cm  1 (stretching
of CH3 [300]), a decrease at 2855 cm  1 (stretching of
 CH2  [300]), a decrease with broadening of the peak at
1734 cm  1 (stretching of CQO in the ester side group), new
bands at 1615 and 1640 cm  1 (stretching of CQC in olefinic
unsaturation [51,66,271]), a decrease at 1465 cm  1 (stretching of
 CH2  coupled with O–CH3 [296]), a decrease at 1450 cm  1
(stretching of the O–CH3 in the ester group [59,235,296]), a
decrease at 1190 cm  1 (stretching of C–O [296] or C-O-C [56] in
the ester group), a decrease at 1150 cm  1 (stretching of C–O in
the ester group [271,296]), and a decrease at 753 cm  1 (rocking
of the –CH2– group coupled with skeletal deformation [296]).
Similar to a yellowness index for spectral data in the optical
range [301], a carbonyl index (CI) can be defined as the ratio of
the intensity at 1734 cm  1 to the intensity at 753 cm  1 [299] to
relate between the loss of the ester group and the reduction of the
population of –CH2–. Unfortunately, the prominent peaks for
PMMA all correspond to species lost during chain scission. That
is, there are no new peaks formed as a result of degradation
(except for the seldom reported peaks at 1615 and 1640 cm  1). In
contrast, distinct peaks occur for MMA at 1637, 1326, 1303, and
941 cm  1 [60], allowing monomer and PMMA to be distin-
guished. MMA, the product of depolymerization, is not produced
abundantly via photolysis. Furthermore, the monomer will outgas
at ambient temperatures, necessitating timely measurement or
cryogenic storage. Although FTIR is not a strong tool for monitor-
ing the photodegradation of PMMA, the results in the literature
for UV aging are consistent with Fig. 6, supporting the mechanism
of photolysis.
3.2. Thermal decomposition
The thermal decomposition of PMMA is more broadly studied
by academics than its optical degradation. Like the effect of high-
energy g-radiation on PMMA, the summary here is not exhaus-
tive. For example, the literature related to the topics of flamm-
ability or spot-specific (laser) heating (some of which are
identified in [303]) will not be examined.
3.2.1. Thermal decomposition: background and related details
Table 6 summarizes prominent studies examining the thermal
degradation and thermal oxidation of PMMA. Details in the table
Table 6
Summary of the examination of the thermal decomposition of PMMA, in the literature.

Year Reference Initiator Reprecipitation Additives Atmosphere Td End-initiated Scission initiated

(1C)
c1 Ea1 c2 Ea2 Z, zip
(s  1) (kJ/mol) (s  1) (kJ/mol) length (mers)

1947 [304] Benzoyl peroxide 1:4-diamiao anthraquinone Vacuum 160–220 130

1949 [305] Benzoyl peroxide Acetone-methanol Vacuum 170
1952 [251] Benzoyl peroxide Benzene UV light Vacuum 130–201 78 1250–7700
1953 [306] Fe++–H2O2 Benzene Vacuum 100 201 1000
1953 [307] Benzoyl peroxide MeOH Vacuum 240–325 138
1956 [308] 250 1000
1960 [309] Azo-di-iso-butyronitrile 330–463 126 251

D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
and lauryl mercaptan
1960 [310] Benzoyl peroxide Vacuum 193–270 1.0E+ 09 124 430–5810
1965 [311] Benzoyl peroxide Adams catalyst and acetic acid 250–400 9.0E+ 09 109 4.9E+ 16 207
1965 [256] Benzoyl peroxide Acetone-methanol UV light N2 54–159 37 14–312
1966 [312] Phenylmagnesium bromide, ammonia, and Li Acetone-methanol Vacuum 300–400 8.0E+ 21 280 460–900
1968 [313] Benzoyl peroxide Acetone-methanol 300–350 4.5E+ 10 107 3.2E+ 25 274
1968 [314] Azo-di-iso-butyronitrile Chloroform-methanol Various copolymers 140–450
1971 [259] UV light Air 160–200
1984 [315] 370–400 2.5E+ 16 210
1984 [270] UV light and monomer Air and vacuum 50–125
1984, 1985 [316,317] Free radical Monomer and release agent Air and N2 150–400 4.8E+ 15 174 2.1E+ 15 210 1270–1760
1985 [318] Free radical Monomer and release agent Air and N2 205–325 8.5E+ 13 64 8.5E+ 13 233 20–2620
1986 [319] Azo-di-iso-butyronitrile, benzoyl Benzene/chloroform-methanol Air and N2 130–450
peroxide, diphenylhexyllithium,
tert-butyl mercaptan, and Co-TPP
1987 [320] 410–485 159
1987 [277] n,methylene chloride-methanol Monomer Air and vacuum 50–125
1988 [321] Azo-di-iso-butyronitrile Petroleum ether Poly(acryloyl chloride) N2 200–400 113
1988 [323] 2.0E+ 16 260
1988 [322] Anionic Methanol Vacuum 340–380 1.0E+ 13 265 200–10,000
1989 [324] Anionic 300–410 390 200–3000
1989 [325] Azo-di-iso-butyronitrile and CoCTC Tetrahydrofuran He 125–410
1990 [54] Bis(4-benzoylphenyl)ether UV light Air 140–180 57 790–3200
1990 [326] Azo-di-iso-butyronitrile Methanol Cross-linking agents Air and N2 80–400 1.1E+ 11 148 4.2E+ 14 211
1991 [327] Methyl trimethylsilyl 2,2-dimethylketene He
1993 [137] t-C4H9MgBr, t-C4H9Li-R3Al, Hexane/methanol-benzene Methyl acrylate and Air and N2 125–410
and tert-butyl various copolymers
mercaptan
1993 [328] Benzoyl peroxide Trimethylolpropane triacrlyate N2 155–400 3.7E+ 11 168
and trimethylolpropane
trimethacrylate
1997 [329] Anionic Ar 380–460 2.0E+ 11 180
1997 [302] Propyl ester phosphazene N2 290–450 154
1999 [330] Free radical Air and N2 180–400 60 190
2001 [331,332] Air and Ar 340–420 150 210
2003 [333] Free radical Phosphorus Ar 150–390 4.5E+ 10 119 9.5E+ 15 199
2003 [334] Azo-di-iso-butyronitrile, benzoyl Methanol N2 180–410 6.0E+ 04 58 9.5E+ 08 170
peroxide, e-caprolactum, n-
dodecanethiol, and 3-mercaptoprionic
acid 2-mercaptoethanol
2004 [335] Air 200
2008 [336] Free radical Tinuvin 622 N2 290–325 165

2055
2056 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
include: the initiator used in polymerization (when specified), use
of reprecipitation (when applicable), formulation additives (when
examined), the atmosphere present, temperatures at which
decomposition was observed (Td), Arrhenius rate coefficients
(c1 and c2), activation energies (Ea1 and Ea2), and zip length (Z).
In Table 6, azo-di-iso-butyronitrile and benzoyl peroxide are
free-radical-based initiators, whereas bis(4-benzoylphenyl)ether,
diphenylhexyllithium (DHLi), bis(4-benzoylphenyl)ether, methyl
trimethylsilyl 2,2-dimethylketene, e-caprolactum and n-dodeca-
nethiol, 3-mercaptoprionic acid, and 2-mercaptoethanol are anio-
nic initiators. The chemicals tert-butyl mercaptan (t-BuSH), lauryl
mercaptan, n-dodecanethiol, cobalt(II) chelate (Co-CTC), and
cobalt(II) tetraphenylporphyn (Co-TPP) are chain transfer agents
that aid the process of generating polymeric chains from mono-
mer. The chemicals phenylmagnesium bromide, ammonia, Li,
tert-C4H9MgBr, and tert-C4H9Li-R3Al may be used to control the
stereochemistry (tacticity) during polymerization.
In Table 6, 1:4-diamiao anthraquinone, adams catalyst, and
acetic acid are used as radical scavengers or to stabilize unsatu-
rated bonds. The additives poly(acryloyl chloride), methyl acry-
late (MA), and propyl ester phosphazene are co-polymers. The
additives trimethylolpropane triacrylate and trimethylolpropane
trimethacrylate are cross-linking agents. Lastly, Tinuvin 622 is
a HALS.
The rate of decomposition may be modeled using an Arrhenius
relationship:
   
Ea1 Ea2
kr ¼ c1 exp or kr ¼ c2 exp ð8Þ
RT RT
Additional parameters in the equation include: kr, the rate of
reaction (i.e., monomer production) (s  1); c, the Arrhenius rate
coefficients (s  1); Ea, the activation energies (kJ mol  1); R, the
ideal gas constant; and T, the temperature (K). The c and Ea values
are given in Table 6 for the regimes of end initiation and random
scissions, i.e., subscripts  1 and 2, respectively. In Eq. (9), the
activation energy (typically identified as a first-order process for
the thermal decomposition of PMMA) reflects the processes of the
initiation reaction (-ir), chain depropagation (  dp), and termina-
tion ( t):
Ea ¼ Eir þ Edp Et
The order-of-magnitude estimates (obtained from the average
of the logarithm of the values in Table 6) for c1 and c2 are
3.8  1010 71.4  103 and 3.6  1015 76.4  103 s  1, whereas the
representative values (avg. 71 SD for Table 6) for Ea1 and Ea2 are
126740 and 226755 kJ/mol, respectively. The Ea values
in Table 6 are notably less than 326–348 kJ/mol, the energy
capable of facilitating C–C scission [318,324,328,331]. This imme-
diately suggests that other mechanisms facilitate the decomposi-
tion of PMMA (as in Fig. 8). Variation in Ea values is identified in
many studies. Ea is observed to depend on the chain length,
history of degradation, and population of vulnerable end-
groups [313]. Others found that additives such as co-poly-
mers [137,302,314], cross-linking agents [326,328], or stabili-
zers [336] can increase Ea and Td. This suggests that disparity in
Ea and Td relative to the values in Table 6 may be encountered
when commercial PMMA is not reprecipitated to remove
additives.
Several shortcomings related to the thermal-degradation mea-
surements can be identified from the literature. Transport issues,
such as the diffusion of monomer out of specimens, may affect
Ea [328]. For this reason, powder or thin-film specimens
(ho1 mm [325]) are used in academic studies to ensure adequate
transport at high temperature, providing ideal results, possibly
different from a Fresnel lenses of finite thickness (Table 2).
Separately, the test and measurement procedure (often conducted
Sample A
m, Rate of mass loss (∂[W/W°]/∂ [t]),{s-1}
0.0007
Sample B
Sample C
0.0005
0.0003
0.0001
0
-0.0001
0 110 140 170 200 230 260 290 320 350 380 410 440
T, Temperature {°C}
Fig. 8. TGA results for PMMA, with temperature ranges identified for the enabling
degradation mechanisms of: (a) ‘‘weak sites,’’ indicated as ‘‘regime I,’’
(b) unsaturated end-groups, ‘‘regime II’’, and (c) homolytic b-scission, ‘‘regime
III.’’ .
Adapted from Ref. [319] with permission.
at 20 1C min  1) for thermal decomposition is not standardized.
Importantly, specimen illumination (which is often precluded by
the thermal chamber in commercially available test instruments,
but may synergistically affect decomposition [54,251,256,259,
316,317]) is not specified in most studies. As another example,
the interval for the analysis of measurements may not be
specified, although data occurring at 0%, 40%, and 60% mass loss
are most frequently reported.
3.2.2. The process of depolymerization
The process of thermal degradation is represented in Fig. 7. As
shown for end-group initiation, elevated temperature renders
unsaturated bonds vulnerable (arrow in Fig. 7(a)). High tempera-
ture may also motivate random scission, as in Fig. 6. Regardless of
ð9Þ
the initiation mechanism, the thermal decomposition of PMMA
consists of the depolymerization of the chain into molecules of
MMA (Fig. 7(b)). Once the radical structure in Fig. 7(b) has
formed, depolymerization is autocatalytic, with the formation of
monomer occurring in rapid, successive ‘‘unzipping’’ events. The
average number of decomposition events can be estimated,
identified as the zip length in Table 6. If the kinetic length
(number of unzipping events) during decomposition exceeds the
length of the molecular chain, the entire chain will depropagate.
For longer chains, the process may become interrupted. For
reaction termination to occur, a more stable end-group must be
formed, such as the saturated configuration in Fig. 7(c). For
example, termination may occur as the result of interaction with
the surrounding environment, such as the reaction with an
external species (‘‘X’’ in Fig. 7(c)).
Multiple ingredients may facilitate the thermal decomposition
of PMMA, as identified in thermal volatilization analysis
(TVA) [274,314] and thermogravimetric analysis (TGA) [137,
316,317,319,325,326,328,330,334]. As summarized in Fig. 8, three
distinct peaks (regimes) were observed in Refs. [137,314,
319,325,326,328]. The first peak typically occurred in the range
1401oTo200 1C, as in Fig. 8 (regime I), with some mass loss
occurring at even lesser temperatures (Table 6). Degradation in
the aforementioned temperature range implies an Ea of 60 kJ/mol.
Initiation here is attributed to head-to-head linkages [137,
319,322,330], co-polymerized oxygen (peroxide links, generated
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
when PMMA is polymerized in air) [322,331], residual mono-
mer [277], unreacted initiator [277], and free radicals remaining
in additives [277]. The latter phenomena will be collectively
referred to here as ‘‘weak sites.’’ Regime I remains controversial,
because it is not always observed. The presence of head-to-head
linkages (proposed to result from a combination reaction occur-
ring during polymerization) is itself contested. To explain, weak
head-to-head linkages have been directly identified via nuclear
magnetic resonance (NMR) [137]. Others have been unable to
corroborate this observation and insist that the latter factors
(oxygen, residual monomer, and additives) instead contribute to
decomposition at low temperatures [332]. An adequate number
of independent investigators identify regime I; however, the
responsible mechanism(s) remains to be agreed upon.
A second peak (typically occurring from 2201oTo300 1C as
in Fig. 8(b)) is attributed to end-group initiation. When end-
termination is facilitated by a disproportionation reaction, an
equal population of saturated (Fig. 7(c)) and unsaturated
(Fig. 7(a)) end-groups will exist following polymerization facili-
tated by a free-radical initiator. The unsaturated (QC, ‘‘vinyl’’)
ends are clearly confirmed from the vinylidene (QCH2) peaks
observed at 5.4 and 6.14 ppm in NMR [137]. The thermal decom-
position of PMMA occurs here from attack (homolytic b-scis-
sion [333]) on the unsaturated terminus by external radical(s)
(Fig. 7). To clarify, decomposition does not occur from attack on
the radical fragment of the initiator species, which exists on the
opposite end of the chain (remaining as an artifact of polymer-
ization) [310]. In contrast to Fig. 8, a recent study [333] identifies
four mechanisms of thermal decomposition, with the first three
occurring from different damage modes for the unsaturated end-
groups. Because decomposition in regime II is initiated at the
chain ends and typically proceeds to completion (100% decom-
position), the Mw and its distribution do not change with time.
The third peak (typically occurred for 3101oTo400 1C as
in Fig. 8(c)) is attributed to initiation via random scission. If a
free-radical initiator was used during polymerization, typically
40–60% of the mass of PMMA will have already been lost (in
regime II) prior to the onset of random scission (regime III).
Anionic initiators or chain transfer agents may be used to avoid
generating unsaturated end-groups during polymerization. In
such cases, thermal decomposition will occur solely via homolytic
b-scission. Because decomposition in regime III is initiated by
chain cleavage, the Mw decreases with time.
Possible mechanisms enabling termination (breaking the auto-
catalytic process) include a transfer reaction between the depro-
pagating chain and a monomer molecule, or an interaction
between the depropagating chain and the fragment of initiation
(the process of ‘‘geminate recombination’’) [309]. Termination
renders the same results as radical disproportionation, i.e., termi-
nation into –C–X (stable end-group, Fig. 7(c)) or –CQCH
(unstable end-group, Fig. 7(a)) [310]. Although not carefully
distinguished throughout the literature, end-group-initiated
decomposition is identified as being a 2nd order process [311,313]
(where both diffusion and reaction must occur), whereas
endothermic chain scission [289] occurs as a pseudo 1st order
process (limited by mass depletion) [313]. Separately, numerical
methods have been employed to model the heat transfer [337]
and mechanisms [338] related to thermal decomposition. Rather
than advancing the understanding of thermal decomposition,
numerical analysis merely verifies existing observations related
to initiation and termination.
3.2.3. Factors of influence
Several influential factors emerge as the temperature present
during decomposition is increased. MMA boils at 101 1C [318],
2057
which may affect heat transfer (reducing the temperature) in
large specimens [320]. Decomposition may also be affected by the
ambient pressure [329]. Size and geometry in regime III is critical,
as reduced outward diffusion of monomer limits the rate of
decomposition [274,309,329]. The formation of char will limit
heat transfer and the outward diffusion of product species [320].
Localized effects, such as those occurring in laser-induced thermal
decomposition [339], are not explored here.
The atmosphere present during decomposition is indicated
in Table 6. Relative to an N2 ambient, O2 was found to reduce
degradation at low temperatures by suppressing regimes I and
II [137,317,319,330]. This was attributed to the neutralization/
alteration of radical species by O2, eliminating the head-to-head/
weak-site and chain-initiated damage mechanisms. Conversely,
O2 was found to enhance degradation as temperature was
increased. The Td for regime III is decreased by 60 1C, evidenced
by alteration of the Mw distribution [318]. With O2, the degrada-
tion process is described as a thermal oxidation [318], with
corresponding evidence of different peroxide and acid product
species [330,331]. Regarding the production of monomer,
decreased zip length at high temperature when O2 was present
was attributed to radical quenching [137].
Several studies suggest that a lesser Mw enhances Ea (increase
in Td) [306,313,314,322,323,333]. This assertion is supported in
TGA profiles, where the data profiles at lower tempera-
tures (regimes I and II) are more prominent for high Mw
PMMA [314,333]. The Td for regime III (where the majority of
the degradation may occur in a short time span) is decreased in
both radical- and anionically initiated PMMA [314], suggesting
that high Mw is universally vulnerable to decomposition.
The effects of stereochemistry are examined in Refs. [137,312].
Below a threshold Mw, the Td for isotactic PMMA was less than
that of syndiotactic PMMA [137]. This is attributed to the less
stable isotactic structure, making it more prone to unzip. Above
the threshold Mw, the trend is reversed. Here, the more reactive
isotactic structure is more prone to terminate during depropaga-
tion, limiting the zip length.
3.2.4. Use of co-polymers
The use of co-polymers may improve the thermal stability (Td)
of PMMA [137,302,307,314,321]. Commercial PMMA may contain
methyl acrylate [137], found to improve thermal stabi-
lity [137,314]. Additives such as styrene [307,314] may enhance
thermal stability at the expense of photodegradation. Other
additives (such as propyl ester phosphazene [302]) may improve
thermal stability, but are added primarily to promote char
formation (reducing flammability). An alternate approach is to
promote cross-linking to stabilize PMMA [328]. Here, Ea is
increased because other bonds in the network must be strained
to liberate MMA. The literature related to the thermal stabiliza-
tion of PMMA is reviewed in Ref. [339].
3.2.5. Synergy between photo- and thermal degradation
From the analysis method in Ref. [15], the optical flux of
107 W m  2 is expected to be absorbed from the AM1.5 terrestrial
global solar spectrum [68] in water-saturated PMMA sheet
(3.2 mm-thick). (water-saturated acrylic is more optically absorb-
ing than dessicated PMMA [60]). Empirically, this renders a
temperature rise on the order of 5–15 1C. Relative to the much
greater Td values in Table 6 and Fig. 8, this limited temperature
rise prompts the question: is thermal decomposition relevant to
the CPV application? To answer, the synergy between photo- and
thermal degradations has been determined to render chain
scission at temperatures as low as 54 1C [256], 110 1C [274],
130 1C[66,251], and 140 1C [54]. The synergistic nature is
2058 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
demonstrated explicitly in TVA measurements, where Td was
decreased by 200 1C when light was present [274]. In such case,
combined photo- and thermal degradation may become a possi-
bility within a 30-year service life in a high-desert location.
Thermal decomposition becomes even more likely if PMMA is
used as an optical component (such as a lens or homogenizer) or
encapsulation layer within a CPV system. In such applications the
temperature within PMMA may rise sufficiently to enable direct
thermal decomposition, particularly if thermal management is
not adequately addressed. Use of PMMA in unconcentrated sun-
light, however, is not expected to result in sufficient temperature
rise to invoke thermal decomposition without photo-catalysis.
From the measured decrease in Mw, the synergistic degrada-
tion is identified to be facilitated by chain scission (regime III
in Fig. 8) [54,251,256,259]. The zip length for PMMA at less-
elevated temperatures has been estimated to be 0.2 [268],
5 [255], and 30 [50], where Z increases with temperature [50].
The F can correspondingly increase with temperature [259]
(particularly near Ta, e.g., 75 1C [53,66]) or the presence of
ozone [50]. In addition to the reduction in kinetic activity as T is
decreased, Z may be decreased if monomer becomes repolymer-
ized at To165 1C [251,266]. Another study separately suggested
F was decreased with the incident l, whereas unzipping became
increasingly prevalent with l [52].
3.2.6. Monomer-enabled catalysis
The effect of monomer on the thermal degradation of PMMA
has been examined at low temperatures, i.e., 50–125 1C [280].
Here, the number of chain scissions was inversely proportional to
the concentration of residual monomer. Over time, the concen-
tration of monomer was found to assume a steady-state value
(0th-order reaction) in an oxygenated environment. Degradation
was therefore attributed to radicals generated from oxidized
monomer (which generates a peroxide [36,276]). Investigation
examining monomer-catalyzed decomposition (possibly syner-
gistically aided by light) below the thermally activated energy
barrier is sparse. One study, however, indicates that MMA itself is
stable below 250 K [289].
Monomer, whether present during photodegradation or ther-
mal decomposition, may affect the mechanical characteristics of
PMMA. To explain, MMA may act as a plasticizer, decreasing Ta
and affecting visco-elastic flow [135]. Monomer-aided degrada-
tion may explain the chain scission identified in a long-term
physical-aging experiment conducted at 80 1C [118].
3.3. Degradation mechanisms: future study
From the literature, the primary degradation mechanism
expected for PMMA lenses is UV-enabled photolysis. Regarding
photolysis, the literature lacks comprehensive validation using
applicable light sources, i.e., outdoor exposure and Xe lamps.
Additional examination might corroborate the specific damaging
wavelengths of light by mapping the wavelength-specific action
spectrum for PMMA. Examination of co-monomers and residuals
(including unpolymerized monomer) is important, since trace
content may catalyze photodegradation. In contrast to the exam-
ination of purified PMMA, examination of UV absorbers and UV
stabilizers has historically remained proprietary to PMMA man-
ufacturers. The influence of moisture (as a synergistic factor) is
implied in the literature, but likewise lacks adequate study. Lastly,
the exact mechanism of initiation (the carbonyl vs. methylene
sites) could benefit from further examination. This largely aca-
demic topic may ultimately apply specifically to high-energy
irradiation outside of the terrestrial solar spectrum.
Synergy occurring between photolysis and thermal decompo-
sition is explicitly identified in the literature, but has not been
examined relative to the CPV application. Separately, the nature
and origin of the degradation at low temperatures (identified as
regime I in Fig. 8) is not widely agreed upon. Such a minimal rate
of decomposition, however, could become noticeable over a 30-
year service life. Thermal decomposition may also be catalyzed by
the same factors as in photodegradation, including additives,
residual content (monomer), and ambient conditions (humidity).
It may turn out that thermal decomposition cannot be entirely
decoupled from photodegradation within the CPV application
environment.
4. Silicone-on-glass or PMMA/glass composite lenses
The use of silicone, i.e., PDMS, facets patterned onto a glass
superstrate [3,5,340–343] is an alternative to a monolithic PMMA
Fresnel lens. PMMA facets patterned onto a glass superstrate
(PMMA/glass) can be used similarly and is likely subject to many
of the same concerns as SOG. Not much public literature exists,
despite early research programs developing SOG composite
lenses [344–346]. Therefore, the goal of this section is to identify
what is known as well as to speculate what characteristics may be
critical to the durability of composite lenses. This includes the
topics of optical durability (including transmittance of the lens, as
well as solarization and corrosion of the glass superstrate), mechan-
ical durability (including fracture/fatigue, delamination, physical
aging, lens shape, and wear and pitting), soiling, chemical stability
(photo- and thermal degradations), and the advantages inherent to
the glass/glass module construction. A summary of the advantages
and disadvantages of SOG and PMMA lenses concludes the section.
4.1. Optical performance and durability
One advantage of silicone relative to PMMA is its greater
optical transmittance, particularly in the wavelengths specific to
PV application [15,347,348]. As in Table 1, the lens facets for SOG
are relatively thin, favoring minimal absorption. PDMS has a
broader spectral bandwidth and generally lesser optical absorp-
tance than PMMA [15], favoring additional energy production. In
contrast, the transmittance and optical bandwidths of glass can
vary with its composition. In practice, the transmittance for SOG
will be limited by the glass superstrate. At the expense of UV
bandwidth, glass may be doped with CeO2, making SOG radiation
resistant for aerospace applications. To explain, the AM0 extra-
terrestrial solar spectrum [272] includes radiation at wavelengths
prone to readily degradable polymeric materials, including
PMMA. The transmittance for glass, however, may be improved
using an AR coating and tempering (which also improves its
mechanical strength). The loss (reflectance of 0.2% for the n
mismatch of 0.13 [15]) occurring at the glass/silicone interface
can therefore be overcome by tempering and/or an AR coating.
The optical efficiency of a SOG lens can exceed that of a PMMA
lens [15,341,343].
4.2. Optical durability
4.2.1. Optical durability of silicone
The optical durability of silicone is expected to be excel-
lent [347], as long as the aromatic phenyl ring structure is not
present [350–352]. This vulnerability is unfortunate, because
phenyl improves the optical performance by better matching
the refractive indexes to glass. Its UV vulnerability likely prohibits
poly(phenyl-methyl silane) (PPMS) from use in CPV. Although
slight optical degradation is observed for PDMS [353,354], the
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068 2059

majority of changes would be expected for the glass superstrate, When this layer is porous, it may temporarily act as an AR
which is subject to the processes of solarization and corrosion. coating. However, the depleted surface layer may crack, delami-
nate, spall, and pit, particularly when it becomes depolymerized
in an alkaline environment (pH49) [224,244]. The change in
4.2.2. Solarization of the glass superstrate
volume associated with ion exchange (H + protons are smaller in
For the iron-containing soda-lime float glass typically used in
diameter than alkali ions) or hydrolysis motivates such mechan-
the PV industry, the term ‘‘solarization’’ refers to the redox
ical damage. Similar to pH, the factors of temperature and
balance between Fe2 + (ferrous iron) and Fe3 + (ferric
surface-area to volume (SA/V) were found to affect the proclivity
iron) [355–362]. Fe2 + has a broad absorption region from 1000
for leaching or dissolution. For example, conditions including
to 1100 nm [360–362], which renders a blue appearance. Fe3 + has
To30 1C and SA/Vo0.07 cm  1 were found to favor leaching of
a narrower absorption band centered about 380 nm [359,362],
the alkali species, whereas T480 1C and SA/V4 0.77 cm  1
which renders a ‘‘yellow’’ appearance. The redox balance there-
favored bulk dissolution [224]. In the absence of material trans-
fore favors either infrared or UV absorptance, with no affect on
port at the surface, an electric field will develop, limiting corro-
light scattering (haze) [356]. Solarization occurs strictly in
sion. The presence of an external electric field in PV, however,
response to optical radiation, for either natural or artificial
additionally facilitates the diffusion of oxygen at its subsequent
sources. Using polished solarized specimens [355–357], the effect
evolution at the anode surface [367].
was confirmed to occur throughout the bulk of glass. For a lamp
Beyond mechanically motivated degradation, transmittance is
operated indoors at 10  the nominal solar irradiance, glass was
adversely affected by the accumulation of Mg and Mg(OH)2 at the
stabilized within 500 h [356,357], suggesting solarization will
corroded surface [372]. For sufficiently corroded glass, a powdery
complete within the early service life of a PV module (less than
white coating or sticky paste forms at the surface [366]. Reaction
1 year in Albuquerque, NM). In comparison, the process of
between the leached alkali species and reactive atmospheric gases
solarization was determined to occur between 2 and 11 years of
(including water vapor, carbon dioxide, and sulfur dioxide) will
field deployment in Odeillo, France [356,357].
produce a salt at the surface [224]. The corrosion of the glass is
Several studies describe solarization occurring as the result of
detrimental to its use in PV applications, motivating its prevention.
the oxidation of iron, i.e., Fe2 + - Fe3 + . The occurrence and
The addition of alkali species (e.g., those containing Na)
magnitude of solarization may depend on the initial state of the
reduces the melting temperature of glass, aiding factory proces-
iron, which is present in glass to facilitate its processing via
sing. As quartz is itself robust, the alkali additives inherently
radiative heat transfer [360]. In particular, the bubbling of
render the glass vulnerable to corrosion. Lime (CaO) will yield a
O2 [355,360,362] and/or low-temperature processing (as opposed
chemically protective surface layer and will also couple Na to the
to rapid cooling) [361] favors the oxidized state, i.e., Fe3 + . Other
glass network [224]. CaO prevents corrosion in concentrations up
trace components present in glass may facilitate solarization by
to 20 wt%, with the nominal mixture of 20:10:70wt% (NaO2:CaO:-
acting as oxidizing agents, e.g., Ce4 + - Ce3 + , Sb5 + - Sb3 + , or
SiO2) being common [224,374]. The doping of the surface with
Mn3 + - Mn2 + . The thermodynamically favored state for iron and
alkaline-earth (group II) ions was found to also inhibit the
the corresponding effect on optical transmittance therefore
diffusion of Na (which is 41000  more mobile than hydro-
depend on the additives present, as well as their initial state of
gen [366,369]), further mitigating corrosion [366]. Similarly, the
charge. For example, because Ce3 + is absorbing at 314 nm (as
mixing of alkali (group I) species, such as K, may also decrease the
opposed to Ce4 + , which absorbs strongly at 240 nm [363]), cerium
rate of corrosion [244]. Al2O3 (up to 2 wt%) may be added to glass
may be added as Ce3 + for the purpose of UV absorption. In this
to reduce water absorption and form an additional corrosion-
application, Ce3 + acts as a reducing agent, favoring the production
resistant layer at the surface [224,371,373]. The aforementioned
of Fe2 + . Tin (present in float glass) may also act as a reducing
additives are those most commonly used to mitigate corrosion;
agent according to the reaction Sn2 + - Sn4 + [355]. Iron was also
however, other additives may be present in glass to improve its
found to be reduced to Fe2 + by annealing between 500 and
durability. Of the general types of glass, durability of borosilicate
610 1C [356–359]. Here, the restoration of glass to its pre-
and alumino-silicate was found to be superior to soda-lime
solarized state is attributed to a diffusion-limited reduction with
glass [185,191,192,196,372]. Treatment with gaseous SO2 or an
hydrogen [358,359]. The additives and redox balance for glass
aqueous KHCO3 may also be used to prevent corrosion [224,245].
superstrates in SOG lenses is not currently well-established;
These chemicals react with the alkali species present near the
therefore, the expected behavior trend remains to be identified.
surface, rendering a durable silica surface layer. After gaseous
From the data in Refs. [15,364], where the oxidation of cerium
treatment, the glass must then be rinsed of the sodium sulfate
and reduction of iron occurs in vintage FP-PV glass, solarization is
powder produced at its surface. Lastly, the use of an AR coating
expected to render a 2% or 3% reduction in transmittance (i.e.,
may help prevent glass corrosion, particularly for ceramic oxide
energy production) for Si and III–V multijunction cell technologies,
films deposited at the surface.
respectively. UV light was found to degrade polymeric components
Regarding field use, the water vapor in air may facilitate the
in the Carissa Plains incident [365], making the use of Ce-doped
leaching of alkali species within minutes when the relative
glass popular for at least a decade. The PV industry has largely
humidity exceeds 50% [224,374]. The rate of attack in humid
discontinued the use of Ce-doped glass in recent years in lieu of
air increases exponentially with relative humidity [374]. Regard-
improved UV stabilization schemes within polymeric materials.
ing precipitation or the formation of condensation at the surface,
there are significant differences between static and dynamic
4.2.3. Corrosion of the glass superstrate conditions. Na and Ca render an alkaline chemistry if they are
The ‘‘corrosion of glass’’ refers to chemical degradation not removed quickly. (The interaction between precipitation and
occurring in the outdoor environment, enabled through factors particulate matter can also affect the pH at the surface.) Static
including moisture, temperature, and contamination [224, moisture layers, such as dew, may produce a highly basic
244,366–374]. In particular, the leaching of Na and Ca (thought chemistry, where glass is dissolved directly. Run-off of moisture
to be facilitated by replacement with H + or H3O + from water) is therefore preferable. Similarly, direct stacking of glass compo-
creates a silica-rich layer at the surface of soda-lime nents is to be avoided in a humid environment. Here, the pH of
glass [244,370,372]. The thickness of the depleted layer typically the monolayers of water formed between adjacent surfaces of
ranges from nanometers to a few micrometers [224,371,373]. glass can also quickly exceed the threshold for network
2060 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
dissolution, where corrosion becomes difficult to prevent.
A similar situation may occur at scratches and surface defects,
where concentration gradient corrosion will facilitate localized
etching.
Regarding float glass, the tin-rich side was found to be more
durable (resistant to corrosion) than the tin-poor side [372]. In
that study, the Ca content of the tin-poor side was found to be less
than that of the tin-rich side (to explain, the tin-poor side is
exposed to the atmosphere during processing). Furthermore, Sn
may exchange with Na, improving the network connectivity of
the glass and its durability [375]. The tin-poor side of float glass is
also more prone to moisture absorption, facilitating corrosion and
stress at that surface [372].The tin-rich layer may extend up to
10–40 mm into the surface of float glass [375].
4.3. Fracture and fatigue
As indicated in Table 2, the mechanical modulus of PDMS can
vary considerably, depending on the Mw and the extent of
molecular cross-linking. The ‘‘encapsulation’’ formulations of
PDMS tend to be brittle, rigid, and hard (where ‘‘resins’’ are even
more so). The ‘‘elastomeric’’ formulations of PDMS tend to be
tough, compliant, and soft (where ‘‘gels’’ are even more so). The
harder formulations of PDMS are more prone to cracking, facili-
tated by the sharp valley features (Fig. 1). Once initiated, such
cracks may readily propagate through the thickness and also
laterally across the surface of the lens. In contrast, the softer
formulations of PDMS can demonstrate self-healing characteris-
tics, preventing the optical efficiency of the lens from being
compromised. PMMA is more brittle than many PDMS formula-
tions, favoring stress relief through cracking (rather than mechan-
ical straining) in PMMA/glass lenses.
Also evident in Table 2 is the significant CTE mismatch
between glass and PDMS. This misfit may motivate delamination,
in addition to fracture. The analysis of delamination for SOG or
PMMA/glass would be the same as that of a compliant film on a
rigid substrate [376,377]. Delamination may be prevented by
applying a primer to the glass or an adhesive agent within the
formulation. An interfacial adhesive, however, must be rugged
against environmental factors including temperature, humidity,
oxygen/ozone, and UV radiation. The CTE of PMMA is more
similar to glass; however, PMMA/glass lenses may require an
adhesive interlayer because PMMA is much more rigid. Further-
more, PMMA is prone to absorb a greater amount of moisture
(Table 2), which can also motivate cyclic stress variation.
4.4. Physical aging and shape change
Many of the elastomeric, encapsulation, and resin formula-
tions of PDMS cured from two-part systems should not creep,
because they exist as cross-linked networks. The oil and gel
formulations of PDMS, however, may exhibit flow. Physical aging
would not be expected for PDMS, because it exists above its a-
relaxation (Tg   125 1C) and melt (Tm  40 1C) temperatures in
most PV application sites. The lack of creep should always be
verified empirically rather than being assumed.
Dimensional stability of the lens facets is important for SOG,
because it would adversely affect the focal length, flux uniformity,
and chromatic distribution of the lens. Although it is not expected
to flow, PDMS may change shape based on its substantial CTE. The
change in optical performance associated with CTE-motivated
strain (which may occur entirely within the elastic regime) has
received limited study [378,379], but may be compensated using
a secondary optic. The focusing error caused by CTE misfit should
be distinguished from that associated with the change in refrac-
tive index with temperature [378,379].
In addition to the shape of the facet elements, the overall
shape of the lens is important; curvature, asymmetry, or other
deviations will render a defocused/non-uniform spot, decreasing
module efficiency. For SOG, the lens facets rest on a significantly
stiffer glass superstrate (but correspondingly heavier, Table 2).
Gas contained within a hermetically sealed module will expand or
contract in response to the ambient temperature. The added
stiffness of a glass superstrate protects against the bellowing of
the lens, if the module is required to be sealed from the
environment.
4.5. Solid erosion and wear
Optical hazing facilitated by microcracking of the surface of
PMMA or PDMS [350] may be understood through fracture
mechanics. Similar optical loss, facilitated by externally generated
surface pitting, depends on the tribological properties of the
superstrate. Because glass is much harder than PMMA, the rate
of surface wear due to sand abrasion should be greatly reduced.
One study, however, indicated a PMMA surface coating actually
improved the abrasion resistance for glass [164]. Its greater wear
resistance may give SOG an advantage in aerospace applications,
where the velocity of impacting particles is substantially greater.
Glass and PMMA are both brittle materials, therefore, neither will
survive kinetic events exceeding a critical threshold (the mass
and/or velocity limit governed by hs).
4.6. Soiling
Regarding the accumulation of particulate matter, the perfor-
mance of glass and PMMA is compared in Section 2.3.8. In
contrast to glass and PMMA, PDMS is much more prone to
soiling [185,191,192,196,350] (also examined in Section 2.3.8).
This immediately suggests that a SOG lens must be protected
from particulate matter. If sealed through a filter, heated air and
humidity may be exchanged to the environment. If sealed
hermetically, the lens may bellow, because a pressure difference
can occur between the interior of the module and the environ-
ment. Further, the pressure in the module and the corresponding
change in physical shape will vary with time as the module
becomes heated by the sun. The use of a hardcoat layer on the
surface of PDMS would prevent the soiling of a non-hermetic
module. Such a hardcoat could either be deposited separately or
perhaps achieved through processing, i.e., additional cross-linking
at the surface.
4.7. Chemical stability
The photodegradation of PDMS in artificial, terrestrial, and
extraterrestrial environments is examined in Refs. [347–354,
380–392]. Most elastomeric and encapsulation formulations are
expected to be robust, i.e., more durable than PMMA [349]. For
example, uncured PDMS tends to become cross-linked and
incorporated into the bulk, rather than catalyzing degradation
as in PMMA. Its greater robustness against high energy radiation
also makes PDMS suitable for use in the extraterrestrial environ-
ment [383,389]. The products of its degradation include broken
cross-links and hydroperoxides (caused by irradiation), methane
and cyclic oligomers (from thermal decomposition), and 2,4
dichlorobenzoic acid (for combined optical and thermal stress).
Visually, degradation may be manifested as discoloration
(whitening or yellowing) as well as hazing of the surface [390].
Trace amounts (ppm) of residual catalyst, additives, or impurities
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
are identified as sources of degradation for PDMS [349], facilitat-
ing its degradation above 150 1C [390]. Compromising impurities
may also diffuse into PDMS at its interface with other materi-
als [390]. The integrity of an external primer or an adhesive agent
present in the formulation could likewise compromise PDMS,
because a primer may be more vulnerable to environmental
factors than the bulk. Alternately, a primer may filter wavelength
bands of optical radiation, improving the durability of PDMS.
4.8. The glass/glass module
The effects of differential thermal expansion on a SOG lens
may be minimized if the module body is also constructed of glass.
For an ‘‘all-glass’’ PV module, a small-aperture-area implementa-
tion may be used to achieve cost-effective thermal management.
The advantages of small-area modules (modules using a small
sized cell) include: low ohmic losses, higher utilization of wafers
(during cell manufacture), ‘‘automatic’’ compensation between
cell and heat sink, reduced module thickness, reduced chromatic
aberration, reduced use of module-housing and heat-sink materi-
als, and improved electrical insulation and environmental isola-
tion (glass module walls) [342,343]. An all-glass module may
prove the most robust implementation of a refractive CPV design,
but may require hermetic packaging to prevent soiling of internal
components [343]. In an all-glass module, all vulnerable compo-
nents (including lens facets, as well as the PV cells) are favorably
located inside a robust glass housing.
4.9. Merits/benefits
Early on, PMMA was identified as a leading candidate for applica-
tion in the outdoor environment because of its excellent resistance to
oxidative photodegradation [28,152]. A SOG lens may provide the
advantages of greater optical bandwidth, greater optical transmit-
tance, additional surface functionality (e.g., AR or anti-soiling coatings
on glass), radiation hardening, lack of creep or physical aging of facets,
more permanent overall shape, greater impact durability, greater
tribological robustness (in the environment and during subsequent
cleaning), and greater chemical stability. Potential disadvantages
include dependence on the quality of the glass superstrate, suscept-
ibility to solarization or corrosion of glass, possible propensity for
delamination, possible propensity for localized dimensional instability
(CTE-motivated shape change of the lens facets), increased likelihood
of soiling of internal components, added cost of materials, and added
weight of the superstrate. Although not without drawbacks, the
advantages make SOG lenses the uniquely capable refractive optic
in aerospace applications. The long-term durability of SOG relative to
PMMA, however, remains to be proven through field deployment.
5. Summary of recommendations for future study
The literature related to the durability of Fresnel lenses used in
the concentrating photovoltaic (CPV) application has been
reviewed. The majority of the examination here concerned
monolithic lenses constructed of poly(methyl methacrylate),
followed by a brief examination of silicone-on-glass (SOG) com-
posite lenses. A basic foundation of understanding exists in the
research conducted at academic and laboratory institutions. Some
important aspects, however, have not yet been examined in the
context of the CPV application. Those considerations warranting
further examination include: the optical durability (transmit-
tance) of PMMA has historically ranked among the best of
polymeric materials. Optical durability should be validated for
contemporary formulations of PMMA, through the use of indoor
accelerated testing and outdoor examination in benchmark
2061
locations (such as Miami and Phoenix). The mechanical durability
of PMMA relative to the CPV application is largely unexamined.
This includes: embrittlement/loss of toughness over time; poten-
tial crazing of the surface/bulk of PMMA; creep and dimensional
stability of the prism elements and the lens; and the solid erosion
of the first surface. The effect of particulate matter on optical
transmittance documented in the literature clearly identifies its
importance (i.e., critical for concentrating optics), warranting
further study of soiling. Future study should include: investiga-
tion in prospective locations (future markets); examination of
modules tracking the sun (as opposed to the stationary specimens
in the literature); the prediction and prescription of cleaning, and
antisoiling technology (which may simultaneously improve trans-
mittance and chemical durability).
Ultraviolet-enabled photolysis is the dominant degradation
mechanism implied in the literature. The literature lacks: greater
validation using applicable (indoor and outdoor) light sources;
greater validation of the specific damaging wavelengths of light;
and comprehensive understanding of the effects of co-monomers
and residual content (unpolymerized methyl methacrylate). In
contrast to the examination of purified PMMA, examination of UV
absorbers and UV stabilizers has remained proprietary to PMMA
manufacturers. The extent of the synergy occurring between
photolysis and thermal decomposition has not been examined
relative to the CPV application. Perhaps catalyzed by the same
additives, residual content, and external factors (such as humid-
ity), a minimal rate of thermal decomposition could become
noticeable over a 30-year service life.
SOG lenses may ultimately demonstrate superior performance
and durability relative to PMMA. However, literature examining SOG
technology is scarce. Noteworthy topics include: durability (delami-
nation) of the silicone/glass interface; coefficient of thermal expan-
sion-motivated straining of the lens facets; and soiling of the lens.
Fortunately, the existing understanding of the solarization of the glass
superstrate, corrosion of the glass superstrate, and durability of
silicone may prove adequately advanced for the CPV application.
Acknowledgements
The authors are grateful for feedback provided by Ralf Leutz of
Concentrator Optics GmbH, Andy Hartzell of the 3M Co., as well
as Peter Colburn and Peter Marks of Evonik Cyro LLC.
Appendix
See Table 7
Table 7
Table of acronyms, nomenclature, and symbols.
Item Meaning
60
Co Cobalt-60, the radioactive isotope
III–V High efficiency multi-junction cell technology, which may include
the materials: As, Ga, In, and P (possibly with a Ge substrate)
+ Diameter (from technical drawing)
a The crack half-length (from fracture mechanics)
qa qN  1 Extension of the crack-length per unit cycle (from fracture
mechanics)
at The time ‘‘multiplier’’ (in the Williams–Landel–Ferry relationship)
Al2O3 Alumina, i.e., aluminum oxide
AM Air mass (which may vary with aerosol concentration, angle of the
sun ,etc.)
AM0 Zero air mass
AM1.5 The air mass of 1.5
AR Antireflection
2062 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068

Table 7 (continued ) Table 7 (continued )

Item Meaning Item Meaning

ASTM American Society for Testing And Materials H+ The hydrogen ion
Aw Maximum water absorption H2 Hydrogen
avg. Average H3O Hydronium, protonated water
C_ Carbon containing a free radical (unbonded electron) HALS Hindered amine light stabilizers
C–H The carbon–hydrogen bond HCOOCH3 Methyl formate
–CH2– The methylene structure, within poly(methyl methacrylate) HFOV Half field of view
–CH3 The methyl side group within the structure of poly(methyl hv The energy of electromagnetic radiation (photon energy)
methacrylate) J Creep compliance, a parameter that may be used to describe visco-
QCH2 The unsaturated vinylidene end group, which may be observed in elastic flow
poly(methyl methacrylate) k Extinction coefficient (the complex portion of the refractive index)
CH3OH Methanol kr The rate of chemical reaction
CH4 Methane kT Thermal conductivity
C–O The carbon–oxygen bond K Potassium
CQO The carbonyl structure (carbon–oxygen double bond) K Stress intensity (from fracture mechanics)
CO Carbon monoxide KIC Mode I critical fracture toughness (from fracture mechanics)
CO2 Carbon dioxide Kmax Maximum stress intensity (from fracture mechanics)
–COOCH3 The ester side group within the structure of poly(methyl Kmean Mean stress intensity (from fracture mechanics)
methacrylate) Kmin Minimum stress intensity (from fracture mechanics)
–CO2CH3 The ester side group (alternate notation) DK The range of stress intensity factor (from fracture mechanics)
c A coefficient KHCO3 Potassium bicarbonate
c1 A coefficient m The coefficient in the Paris Law relationship (from fracture mechanics)
c2 A coefficient m_ Rate of mass loss
C Carbon MA Methyl acrylate (C4H6O2)
C1 A parameter-specific (often empirically derived) coefficient MMA Methyl methacrylate, the monomer form of poly(methyl
C2 A parameter-specific (often empirically derived) coefficient methacrylate), C5H8O2
C3 A parameter-specific (often empirically derived) coefficient MDC McDonnell Douglass Astronautics (historic research division)
CaO Calcium oxide (‘‘lime’’), a common ingredient used to improve the Mc Critical molecular weight
chemical stability of glass Mw Molecular weight
Ce Cerium Mg Magnesium
CeO2 Cerium oxide Mg(OH)2 Magnesium hydroxide
Cg Geometric concentration Mn Manganese
CI Carbonyl index, in the FTIR of poly(methyl methacrylate), the ratio n The real component of the refractive index
of the intensity at 1734 cm  1 to the intensity at 753 cm  1 N The f-number (from optics)
Cmax Maximum geometric concentration N2 Nitrogen
CL Centerline (from technical drawing) Na Sodium
CPV Concentrating photovoltaic NaO2 Sodium superoxide
c-Si Crystalline silicon, a common photovoltaic cell technology NBSR National Bureau of Standards (now NIST)
CSP Concentrating solar power, the industry concentrating solar flux NIST National Institute of Standards and Technology
for non-photovoltaic energy generation NMR Nuclear magnetic resonance, a compositional analysis technique
CTE (The mechanical) coefficient of thermal expansion NREL National Renewable Energy Laboratory (historic research institution)
d1 A coefficient O2 Oxygen
d2 A coefficient –OCH3 Methoxy, a methyl functional group bound to oxygen
d3 A coefficient PC Polycarbonate
D The dose PDMS Poly(dimethylsiloxane)
DSET Desert Sunshine Exposure Testing , later Heraeus, currently Atlas pH The acid/base balance
Material Testing Technology, LLC (historic test facility) PIMs Common name for annual photovoltaic project integration
E The spectral irradiance or elastic modulus meeting reports at Sandia National Laboratories
Ea Activation energy (as in an Arrhenius relationship) PM2.5 Particulate matter, with a diameter r 2.5 mm
Ea1 The first activation energy PM10 Particulate matter, with a diameter between 2.5 and 10 mm
Ea2 The second activation energy ppm Parts per million
Ec Creep modulus, a parameter that may be used to describe visco- PMMA Poly(methyl methacrylate)
elastic flow PPMS Poly(phenyl-methyl silane)
Edp The activation energy of chain depropagation (in PS Polystyrene
depolymerization) PTFE Polytetrafluoroethylene
Eg Band gap (energy gap) PV Photovoltaic
Eir The activation energy of the initiation reaction (in PVC Polyvinyl chloride
depolymerization) r The rate of volumetric relaxation
Et The activation energy of the termination reaction (in R Optical reflectance or the ideal gas constant
depolymerization) Rb Beam radius (in optics)
EPA Environmental Protection Agency Rp Prism peak radius
eV Electron-volt Rv Prism valley radius
Fe2 + Ferrous iron, with the oxidation state of + 2 SAND Common name for general reports authored at Sandia National
Fe3 + Ferric iron, with the oxidation state of +3 Laboratories
FEA Finite-element analysis (a common numerical method for the SA/V Ratio of surface area to volume
analysis of mechanical systems) Sb Antimony
FP-PV Flat-panel photovoltaic (the historic photovoltaic technology) SD Standard deviation
FTIR Fourier transform infrared spectroscopy SI System international
G Energy release rate (from fracture mechanics) SiO2 Glass, silicon dioxide
Gc Critical energy release rate (from fracture mechanics) Sn Tin
Gr Yield of radiolysis, the number of scissions per 100 eV of total SnO2 Tin dioxide
energy absorbed SERI Solar Research Institute, now NREL (historic research institution)
h Thickness SNL Sandia National Laboratories (historic research institution)
hf Facet thickness SOG Silicone-on-glass
hs Superstrate thickness SPF Institut für Solartechnik Prüfung Forschung (historic research
ht Total lens thickness institution)
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068 2063

Table 7 (continued ) [11] R. Winston, W. Zhang, Pushing concentration of stationary solar concen-
trators to the limit, Opt. Exp. 18 (2010) A64–A72.
Item Meaning [12] L.W. James, Lens design considerations, SAND 81-1240, 1981, pp. 144–146.
[13] E. Lorenzo, A. Luque, Fresnel lens analysis for solar energy applications,
t Time Appl. Opt. 20 (1981) 2941–2945.
T Optical transmittance or temperature [14] E. Lorenzo, Chromatic aberration effect on solar energy systems using
Fresnel lenses, Appl. Opt. 20 (1981) 3729–3732.
Td The decomposition temperature
[15] D.C. Miller, M.D. Kempe, C.E. Kennedy, S.R. Kurtz, Analysis of Transmitted
Tf Final optical transmittance
optical spectrum enabling accelerated testing of multi-junction CPV
Tg The glass transition temperature (more formerly known as Ta)
designs, Opt. Eng., 50 (1) (2010) 013003.
Ti Initial optical transmittance [16] S.R. Kurtz, M.J. O’Neill, Estimating and controlling chromatic aberration
Tm The melting temperature losses for two-junction, two-terminal devices in refractive concentrator
Tr The reference temperature systems, Proc. IEEE PVSC (1996) 361–364.
kT Reduction in optical transmittance [17] L. Fu, R. Leutz, H.P. Annen, Secondary optics for Fresnel lens solar
kTmax Maximum reduction in optical transmittance concentrators, Proc. SPIE 7785 (2010) 7785-09.
kTtyp Typical reduction in optical transmittance [18] R. Leutz, A. Suzuki, A. Akisawa, T. Kashiwagi, Nonideal concentration of
TGA Thermogravimetric analysis nonimaging Fresnel lenses, Proc. SPIE 4446 (1999) 379–387.
TVA Thermal volatilization analysis [19] P.E. Glaser, M.M. Chen, J.B. Berkowitz-Mattuck, The flux redistributor, Solar
Ta The temperature of the glass transition , also known as the a- Energy 7 (1963) 12–17.
relaxation [20] M.M. Chen, J.B. Berkowitz-Mattuck, P.E. Glaser, The use of a kaleidoscope to
Tb The temperature of the b-relaxation obtain uniform flux over a large area in a solar or arc imaging furnace, Appl.
Opt., 2 (3) (1963) 265–271.
tan d The dissipation coefficient, commonly used to assess the damping
[21] H. Ries, J.M. Gordon, M. Lasken, High-flux photovoltaic solar concen-
of dynamic systems
trators with kaleidoscope-based optical designs, Solar Energy 60 (1997)
tan db The dissipation coefficient, here in the (temperature) region of the
11–16.
b-relaxation [22] K. Araki, M. Kondo H. Uozumi, M. Yamaguchi, Experimental proof and
TTS Time–temperature superposition, a principle which may be used theoretical analysis on effectiveness of passive homogenizers to 3 J con-
to analyze visco-elastic flow centrator solar cells, in: Proceedings of the World Conference on PV Energy
UV Ultraviolet Conversion, 2003, pp. 853–856.
V The volume [23] K. Araki, R. Leutz, M. Kondo, A. Akisawa, T. Kashiwagi, M. Yamaguchi,
wl Lens width Development of a metal homogenizer for concentrator monolithic multi-
wp Prism width junction cells, Proc. IEEE PVSC (2002) 1572–1575.
W The weight [24] S. Cobb, Development and evaluation of an improved efficiency polymeric
WLF Williams–Landel–Ferry, a representation that may be used to web point-focus Fresnel lens, SAND 86-7046, 1986, 1–32.
analyze visco-elastic flow [25] L.W. James, Use of imaging refractive secondaries in photovoltaic concen-
Z Zip length, in poly(methyl methacrylate) the number of successive trators, SAND 89-7029, 1989, pp. 1–54.
[26] www.matweb.com.
decompositions of monomer from the main chain
[27] Z. Mathys, P.J. Burchill, Influence of location on the weathering of acrylic
a The absorption coefficient
sheet materials, Polym. Deg. Stab. 55 (1997) 45–54.
b- The breaking of a carbon-carbon bond, typically resulting in the [28] J.E. Greivenkamp, Field Guide to Geometrical Optics, SPIE Press, Bellingham,
scission formation of free radicals 2004.
g-rays Gamma rays, the high frequency electromagnetic radiation [29] S.N. Kasarova, N.G. Sultanova, C.D. Ivanov, I.D. Nikolov, Analysis of disper-
e Mechanical strain sion of optical plastic materials, Opt. Mater. 29 (2007) 1481–1490.
em Maximum mechanical strain [30] A.L. Andrady, Wavelength sensitivity in polymer photodegradation, Adv.
Z The dynamic (absolute) viscosity Polym. Sci. 128 (1997) 47–94.
Ze The optical efficiency [31] A.L. Andrady, S.H. Hamid, X. Hu, A. Torikai, Effects of increased solar ultraviolet
Zo A reference viscosity, typically the zero-shear viscosity radiation on materials, J. Photochem. Photobiol. B 46 (1998) 96–103.
ya The acceptance angle, more specifically the acceptance angle along [32] N.D. Searle, Activation spectra: techniques and applications to stabilization
the plane of the azimuth and stability, in: ASTM STP 1202, 1994, pp. 41–58.
yc Contact angle (from surface science) [33] A. Factor, W.V. Ligon, R.J. May, The role of oxygen in photoaging of
bisphenol A polycarbonate. 2. GC/GC/ high-resolution MS analysis of
yd Draft angle
Florida-weathered polycarbonate, Macromolecules 20 (1987) 2461–2468.
yf Facet angle
[34] A. Factor, Degradation of bisphenol A polycarbonate by light and g-ray
l (Optical) wavelength radiation, in: D.G. LeGrand, J.T. Bendler (Eds.), Handbook of Polycarbonate
n Poisson’s ratio Science and Technology, Marcel Dekker, New York, 2000, pp. 267–293.
r Density [35] M. Diepens, P. Gijsman, Photodegradation of bisphenol A polycarbonate,
F The quantum yield Polym. Deg. Stab. 92 (2007) 397–406.
ca The acceptance angle of the altitude (perpendicular to the plane of [36] H. Zweifel, R.D. Maier, M. Schiller, Plastics Additives Handbook, Carl Hanser
the azimuth) Verlag, Munich, 2009, pp. 176–180.
[37] L.G. Rainhart, W.P. Schimmel, Effect of outdoor aging on acrylic sheet, SAND
74-0241, 1974, pp. 1–41.
[38] L.G. Rainhart, W.P. Schimmel, Effect of outdoor aging on acrylic sheet, Solar
Energy 17 (1975) 259–264.
References [39] E. Boes, Lens transmission, SAND 79-1791, 1979, pp. 105–114.
[40] J.L. Chevalier, A. Balme, H. Sallee, Durabilité des matériaux plastiques
transparent utilizes en couverture de capteurs solaires plans, J. de Phys.
[1] A. Luque, S. Hegedus (Eds.), Handbook of Photovoltaic Science and Engi-
42 (1981) 401–419.
neering, John Wiley & Sons Inc, Hoboken, 2003.
[41] L.C. Beavis, Unique material challenges in photovoltaic concentrator mod-
[2] R. Leutz, A. Suzuki, Nonimaging Fresnel Lenses, Design and Performance of
ules, Proc. IEEE PVSC (1984) 820–826.
Solar Concentrators, Springer-Verlag, New York, 2001.
[42] G.A. Zerlaut, T.E. Anderson, Commercial solar materials exposure studies,
[3] J.R. Egger, Manufacturing Fresnel lens master tooling for solar photovoltaic
DSET-R2658, 1989, pp. 1–67.
concentrators, SAND 79-1791, 1979, pp. 149–154.
[43] J. Metzger, COSTAR I: correlation of solar and thermal aspects of weath-
[4] R. Leutz, L. Fu, H.P. Annen, Stress in large-area optics for solar concentrators, ering, Heraeus DSET R2600-23, 1992, pp. 1–107.
Proc. SPIE 7412 (2009) 741206. [44] J.E. Brzuskiewicz, Correlation of solar and thermal aspects of weathering
[5] G. Sala, E. Lorenzo, Hybrid silicone glass Fresnel lens as concentrator for COSTAR I, Heraeus DSET R2600-28,1993, pp. 1–28.
photovoltaic applications, J. Appl. Metalwork. (1979) 1004–1010. [45] A.R. Mahoney, J.E. Cannon, J.R. Woodworth, Accelerated UV-aging of acrylic
[6] M. Collares-Pereria, High temperature solar collector with optimal concen- materials used in PV concentrator systems, Proc. IEEE PVSC (1993)
tration: non-focusing Fresnel lens with secondary concentrator, Solar 1216–1221.
Energy 23 (1979) 409–420. [46] J. Metzger, COSTAR I., Correlation of solar and thermal aspects of weath-
[7] L.W. James, J.K. Williams, Fresnel optics for solar concentration on photo- ering, Heraeus DSET R2600-30, 1993, pp. 1–156.
voltaic cells, Proc. IEEE PVSC (1978) 673–679. [47] J.W. Summers, E.B. Rabinovitch, Weatherability of vinyl and other plastics,
[8] D.Y. Goswami (Ed.), Taylor and Francis, Philadelphia, 1999. in: G. Wypch (Ed.), Weathering of Plastics: Testing to Mirror Real Life
[9] N. Shatz, J. Bortz, R. Winston, Thermodynamic efficiency of solar concen- Performance, William Andrew Publishing, Norwich, 1999.
trators, Opt. Exp. 18 (2010) A5–A16. [48] F. Ruesch, S. Brunold, Langzeitalterungsuntersuchung an Abdeckungsma-
[10] R. Leutz, A. Suzuki, A. Akisawa, T. Kashiwagi, Design of a nonimaging Fresnel terialien für ThermischeSonnenkollektoren. BfE Schlussbericht. Projekt
lens for solar concentrators, Solar Energy 65 (1999) 379–387. 102349, Publikation 280247. /http://www.solarenergy.chS.
2064 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
[49] F. Ruesch, S. Brunold, Ageing performance of collector glazing
materials—results from 20 years of outdoor weathering, in: Proceedings
of the Eurosun Conference, 2008.
[50] A. Charlesbey, D.K. Thomas, A comparison of the effects of ultra-violet and
gamma radiation in polymethylmethacrylate, Proc. Roy. Soc. A 269 (1962)
104–124.
[51] M. Abouelezz, P.F. Waters, Studies on the photodegradation of poly(methyl
methacrylate), NBSIR 78-1463, 1978, pp. 1–55.
[52] M. Abouelezz, P.F. Waters, Studies on the photodegradation of poly(methyl
methacrylate), NBSIR 79-1766, 1979, pp. 1–46.
[53] B. Dickens, J.W. Martin, D. Waksman, Analysis of damage profiles in
poly(methyl methacrylate) in terms of oxygen diffusion and consumption,
Polym. Deg. Stab. 15 (1986) 265–279.
[54] I. Mita, K. Obata, K. Horie, Photoinitated thermal degradation of polymers II:
poly(methyl methacrylate), Polym. J. 22 (1990) 397–410.
[55] A. Torikai, M. Ohno, K. Fueki, Photodegradation of poly(methyl methacry-
late) by monochromatic light: quantum yield, effect of wavelengths, and
light intensity, J. Appl. Polym. Sci. 41 (1990) 1023–1032.
[56] A. Torikai, S. Hiraga, K. Fueki, Photodegradation of blends of poly(methyl
methacrylate) and poly(styrene-co-methyl methacrylate), Polym. Deg. Stab.
37 (1992) 73–76.
[57] T. Mitsuoka, A. Torikai, K. Fueki, Wavelength sensitivity of the photode-
gradation of poly(methyl methacrylate), J. Appl. Polym. Sci. 47 (1993)
1027–1032.
[58] A. Torikai, T. Hattori, T. Eguchi, Wavelength effect on the photoinduced
reaction of polymethylmethacrylate, J. Polymer Sci. A 33 (1995) 1867–1871.
[59] X. Colom, T. Garcı́a, J.J. Suñol, J. Saurina, F. Carrasco, Properties of PMMA
artificially aged, J. Non-Cryst. Solids 287 (2001) 308–312.
[60] D.C. Miller, L.M. Gedvilas, B. To, C.E. Kennedy, S.R. Kurtz, Durability of
poly(methyl methacrylate) lenses used in concentrating photovoltaics, Proc.
SPIE (2010) 7702–7773.
[61] A.R. Shultz, Degradation of polymethyl methacrylate by ultraviolet light,
J. Phys. Chem. 65 (1961) 967–972.
[62] R.B. Fox, L.G. Isaacs, S. Stokes, Photolytic degradation of poly(methyl
methacrylate), J. Polymer Sci. A 1 (1963) 1079–1086.
[63] R.B. Fox, L.G. Isaacs, S. Stokes, R.E. Kagarise, Photodegradation of poly
(methyl methacrylate), J. Polymer Sci. A 2 (1964) 2085–2092.
[64] O.V. Startsev, V.P. Rudney, Yu.N. Ivonin, Ye.F. Nikishin, Ye.A. Barbashev,
V.A. Bogatov, B.V. Perov, Effect of exposure to outer space on some physical
properties or organic glass, Polym. Sci. USSR 29 (1987) 2836–2845.
[65] T. Caykara, O. Güven, UV degradation of poly(methyl methacrylate) and its
vinyltriethoxysilane containing copolymers, Polym. Deg. Stab. 65 (1999)
225–229.
[66] M. Shirai, T. Yamamoto, M. Tsunooka, Ablative photodegradation of poly(-
methyl methacrylate) and its homologues by 185 nm light, Polym. Deg.
Stab. 63 (1999) 481–487.
[67] N. Siampiringue, J.-P. Leca, J. Lemaire, Mecanismes de photo-oxydation do
poly(methacrylate de methyl), Eur. Polym. J 27 (1991) 633–663.
[68] ASTM G173 standard tables for reference solar spectral irradiances, ASTM
International: West Conshohocken, 2003.
[69] G.A. Zerlaut, Solar ultraviolet radiation: aspects of importance to the
weathering of materials, in accelerated and outdoor durability testing of
organic materials, ASTM STP 1202 (1994) 3–26.
[70] S. Karabanov, Y. Kukmistrov, Study of PV module methylmethacrylate
encapsulation, Proc. Mater. Res. Soc. 945 (2006).
[71] R.M. Bethea, M.T. Barriger, P.F. Williams, S. Chin, Environmental effects on
solar concentrator mirrors, Solar Energy 27 (1981) 497–511.
[72] R.W. Hertzberg, Deformation and Fracture Mechanics of Engineering
Materials, Fourth Edition, Wiley, Hoboken, 1996.
[73] P.J. Burchill, R.H. Stacewicz, Influence of location on the weathering of
acrylic sheet materials, Polym. Deg. Stab. 55 (1997) 45–54.
[74] J. Hennig, Spannungsrissverhalten von PMMA nach freibewitterung, Angew.
Makromol. Chem. 114 (1983) 131–139.
[75] M.M. Qayyum, J.R. White, Influence of location on the weathering of acrylic
sheet materials, Polym. Deg. Stab. 55 (1997) 45–54.
[76] J.P. Berry, Fracture processes in polymeric materials. I. The surface energy of
poly(methyl methacrylate), J. Polym. Sci. 1 (1961) 107–115.
[77] J.P. Berry, Fracture processes in polymeric materials. V. Dependence of the
ultimate properties of poly(methyl methacrylate) on molecular weight,
J. Polym. Sci. A 2 (1964) 4069–4076.
[78] R.P. Kusy, D.T. Turner, Influence of the molecular weight of poly(methyl
methacrylate) on fracture surface energy in notched tension, Polymer 17
(1976) 161–166.
[79] R.P. Kusy, Ductile–brittle transition of acrylic, J. Non-Cryst. Solids 24 (1977)
141–144.
[80] V.T. Truong, P.E.M. Allen, D.R.G. Williams, The effect of molecular-properties
on mechanical-behavior of poly(methyl methacrylate: 1. Stress cracking in
air, Eur. Polym. J. 22 (1986) 903–910.
[81] P.J. Burchill, G. Mathhys, R.H. Stacewics, Analysis and properties of some
commercial poly(methyl methacrylate) based materials, J. Mater. Sci. 22
(1987) 483–487.
[82] J.P. Berry, Surface characteristics of fractured poly(methyl methacrylate),
Nature 185 (1960) 91–92.
[83] H.B. Lee, D.T. Turner, Influence of crosslinking on the tensile-strength and
fractography of poly(methyl methacrylate), Polym. Eng. Sci. 19 (1979)
95–98.
[84] M.Y. He, H.C. Cao, A.G. Evans, Mixed-mode fracture: the four-point shear
specimen, Act. Metall. Mater. 38 (1990) 839–846.
[85] P.J. Burchill, R.H. Stacewicz, Effect of water on the crazing of a crosslinked
poly(methylmethacrylate), J. Mater. Sci. Lett. 1 (1982) 448–450.
[86] D.T. Turner, Water sorption of poly(methyl methacrylate): 1. effects of
molecular weight, Polymer 28 (1987) 293–296.
[87] Plexiglass will not yellow! Guarantee Statement, Evonik Röhm GnBH,
Darmstadt, 2007.
[88] R.S. Yamasaki, A. Blaga, Effect of weathering at different exposure angles on
the tensile impact resistance of thermoplastics, J. Test. Eval. 10 (1982)
156–161.
[89] A. Blaga, R.S. Yamasaki, Weatherability of acrylic-based plastics in long-
term outdoor exposure, Durability Build. Mater. 4 (1986) 21–34.
[90] N.E. Waters, Crack-growth phenomena in fatigued poly)methyl methacry-
late), J. Mater. Sci. 1 (1966) 354–361.
[91] N.H. Watts, D.J. Burns, Fatigue crack propagation in polymethylmethacry-
late, Polym. Eng. Sci. 7 (1967) 90–93.
[92] H.F. Borduas, L.E. Culver, D.J. Burns, Fracture-mechanics analysis of fatigue-
crack propagation in polymethylmethacrylate, J. Strain Anal. 3 (1968) 193–199.
[93] B. Mukherjee, D.J. Burns, Fatigue-crack growth in polymethylmathacrylate,
Exp. Mech. 11 (1971) 433–439.
[94] S.L. Kim, M. Skibo, J.A. Manson, R.W. Hertzberg, Fatigue crack-propagation
in poly(methyl methacrylate)—effect of molecular-weight and internal
plasticization, Polym. Eng. Sci. 17 (1977) 194–203.
[95] S.L. Kim, J. Janiszewski, M.D. Skibo, J.A. Manson, R.W. Hertzberg, Effect of
molecular-weight distribution on fatigue crack-propagation in poly(methyl
methacrylate), Polym. Eng. Sci. 19 (1979) 145–150.
[96] J. Michel, J.A. Manson, R.W. Hertzberg, A simple viscoelastic model for
fatigue crack propagation in polymers as a function of molecular weight,
Polymer 25 (1984) 1657–1665.
[97] T.R. Clark, R.W. Hertzberg, N. Mohammadi, Fatigue mechanisms in poly
(methyl-methacrylate) at threshold-effects of molecular-weight and mean
stress, J. Mater. Sci. 28 (1993) 5161–5168.
[98] P.C. Paris, M.P. Gomez, W.E. Anderson, A rational analytic theory of fatigue,
Trends Eng. 13 (1961) 9–14.
[99] G.L. Cloud, Optical Methods of Engineering Analysis, Cambridge University
Press, Cambridge, 1995.
[100] K. Ramesh, Digital Photoelasticity: Advanced Techniques and Applications,
Springer, New York, 2000.
[101] S.-D. Hong, S.Y. Chung, R.F. Fedours, Study of molecular deformation
mechanisms in the glassy state. I. Temperature effect on stress-birefrin-
gence and strain-birefringence responses of poly(methylmethacrylate),
J. Polym. Sci. 21 (1983) 1647–1660.
[102] R.P. Kusy, A.R. Greenberg, Influence of molecular-weight on the dynamic
mechanical-properties of poly(methyl methacrylate), J. Therm. Anal. 18
(1980) 117–126.
[103] D. Hull, Fractography: Observing, Measuring and Interpreting Fracture
Surface Topography, Cambridge University Press, Cambridge, 1999.
[104] R.P. Kusy, Effect of molecular-weight on fracture morphology of poly
(methylmethacrylate) in cleavage, J. Mater. Sci. 11 (1976) 1381–1384.
[105] A. Takimoto, Y. Yoshimura, An analytical expression of fracture surface
marks in PMMA and the variation of marks due to primary crack propaga-
tion, Trans. J. Soc. Mech. Eng. A 53 (1987) 1754–1761.
[106] N. Murphy, A. Ivankovic, The prediction of dynamic fracture evolution in
PMMA using a cohesive zone model, Eng. Fract. Mech. 72 (2005) 861–875.
[107] K.M. Sinnott, Shear modulus and internal friction of polymethyl methacry-
late and polyethyl methacrylate between 4.2 and 100 K, J. Polym. Sci. 35
(1959) 273–275.
[108] W.G. Gall, N.G. McCrum, Internal friction in stereoregular polymethyl
methacrylate, J. Polym. Sci. 50 (1961) 489–495.
[109] K.C. Rusch, Time–temperature superposition and relaxation behavior in
polymeric glasses, J. Macromol. Sci. B2 (1968) 179–204.
[110] P.P. Benham, D. McCammond, Studies of creep and contraction ratio in
thermoplastics, Plast. Polym. 39 (1971) 130–136.
[111] J.M. Hutchinson, C.B. Bucknall, The effects of thermal history on the density
and mechanical properties of poly(methyl methacrylate), Polym. Eng. Sci.
20 (1980) 173–181.
[112] M. Cizmecioglu, R.F. Fedors, S.D. Hong, J. Moacanin, Effect of physical aging
on stress relaxation of poly(methyl methacrylate), Polym. Sci. Eng. 21
(1981) 940–942.
[113] B.E. Read, G.D. Dean, Time-dependent deformation and craze iniaition in
PMMA: volume effects, Polymer 25 (1984) 1679–1686.
[114] G.B. McKenna, A.J. Kovacs, Physical aging of poly(methyl methacrylate) in
the nonlinear range: torque and normal force measurements, Polym. Eng.
Sci. 24 (1984) 1138–1141.
[115] J.M. Crissman, G.B. McKenna, Relating creep and creep rupture in PMMA
using a reduced variable approach, J. Polym. Sci. B25 (1987) 1667–1677.
[116] G.D. Dean, B.E. Read, G.D. Small, Influence of physical aging on creep in the
a- and b-relaxation regions in PMMA, Plast. Rub. Proc. Appl. 9 (1988)
173–179.
[117] R. Diaz-Calleja, A. Ribes-Greus, J.L. Gomez-Ribelles, Study of structural
relaxation by dynamic-mechanical methods in poly(methyl methacrylate),
Polymer 30 (1989) 433–1438.
[118] J.M. Crissman, G.B. McKenna, Physical and chemical aging in PMMA and
their effects on creep and creep rupture, J. Polym. Sci. B28 (1990)
1463–1473.
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
[119] J. Pérez, J.Y. Cavaille, R. Diaz-Calleja, J.L. Gomez-Ribelles, Physical aging of
amorphous polymers. Theoretical analysis and experiments on poly(methyl
methacrylate), Makromol. Chem. 192 (1991) 2141–2161.
[120] B.E. Read, Analysis of creep and physical aging in glassy polymers, J. Non-
Cryst. Solids 131–133 (1991) 408–419.
[121] S. Yoshioka, H. Usada, Y. Nanzai, Variation of dynamic viscoelasticity during
yielding and post-yield relaxation in a glassy polymer, J. Non-Cryst. Solids
172–174 (1994) 765–770.
[122] E. Muzeau, G. Vigier, R. Vassoille, J. Pérez, Changes of thermodynamic and
dynamic mechanical properties of poly(methyl methacrylate) due to structural
relaxation: low-temperature ageing and modeling, Polymer 36 (1995) 611–620.
[123] Y. Nanzai, A. Miwa, S.Z. Cui, Aging in fully annealed and subsequently
strained poly(methyl methacrylate), Polym. J. 32 (2000) 51–56.
[124] P.R. Higgenbotham-bertolucci, H. Gao, J.P. Harmon, Creep and stress
relaxation in methacrylate polymers: two mechanisms of relaxation beha-
vior across the glass transition region, Polym. Eng. Sci. 41 (2001) 873–880.
[125] J. Bischoff, E. Catsiff, A.V. Tobolsky, Elatoviscous properties of amorphous
polymers in the transition region, J. Am. Chem. Soc. 74 (1951) 3378–3381.
[126] L.C.E. Struik, Physical aging in plastics and other glassy materials, Polym.
Eng. Sci. 17 (1977) 165–173.
[127] R. Greiner, F.R. Schwarzl, Thermal contraction and volume relaxation of
amorphous polymers, Rheol. Acta 23 (1984) 378–395.
[128] L.C.E. Struik, Effect of thermal history on secondary relaxation process in
amorphous polymers, Polymer 28 (1987) 57–68.
[129] L.C.E. Struik, Volume relaxation and secondary transitions in amorphous
polymers, Polymer 28 (1987) 1869–1875.
[130] R. Greiner, F.R. Schwarzl, Volume relaxation and physical aging of amorhpus
polymers I. Theory of volume relaxation after single temperature jumps,
Colloid Polym. Sci. 267 (1989) 39–47.
[131] G.B. McKenna, On the physics required for prediction of long term
performance of polymers and their composites, J. Res. Natl. Inst. Stand.
Technol. 99 (1994) 169–189.
[132] J.M. Hutchinson, Physical aging of polymers, Prog. Polym. Sci. 20 (1995)
703–760.
[133] A.S. Wineman, K.R. Rajagopal, Mechanical Response of Polymers, Cambridge
University Press, Cambridge, 2000.
[134] K. Chen, K.S. Schweizer, Theory of physical aging in polymer glasses, Phys.
Rev. E78 (2008) 031802.
[135] S.R. Tatro, G.R. Baker, K. Bisht, J.P. Marmon, A. MALDI, TGA, TG/MS, and DEA
study of the irradiation effects on PMMA.
[136] S. Havriliak, Dielectric relaxation and other properties of syndiotactic and
isotactic poly(methyl methacrylate), Polymer 9 (1968) 289–324.
[137] K. Hatada, T. Kitayama, N. Fujimoto, T. Nishiura, Stability and degradation of
polymethacrylates with controlled structure, J.M.S.—Pure Appl. Chem. A30
(1993) 645–667.
[138] H.-A. Schneider, M.-J. Brekner, H.-J. Cantow, Some remarks concerning the
WLF-relation, Polym. Bull. 14 (1985) 479–485.
[139] A.Q. Tool, Viscosity and the extraordinary heat effects in glass, J. Res. Natl.
Inst. Stand. Technol. 37 (1946) 73–90.
[140] O.S. Narayanaswamy, A model of structural relaxation in glass, J. Am.
Ceram. Soc. 54 (1971) 491–498.
[141] C.T. Moynihan, P.B. Macedo, C.J. Montrose, P.K. Gupta, M.A. DeBolt, J.F. Dill,
B.E. Dom, P.W. Drake, A.J. Easteal, P.B. Elterman, R.P. Moeller, H. Sasabe,
J.A. Wilder, Structural relaxation in vitreous materials, Ann. New York Acad.
Sci. 279 (1976) 15–35.
[142] T. Chen, Determining a Prony series for a viscoelastic material from time
strain data, NASA/TM-2000-210123/ARL-TR-2206, 2000, pp. 1–26.
[143] J.-S. Wang, R.S. Porter, On the viscosity-temperature behavior of polymer
melts, Rheol. Acta 34 (1995) 496–503.
[144] S.R. Kurtz, K. Whitfield, D.C. Miller, J. Joyce, J.H. Wohlgemuth, M.D. Kempe,
N. Dhere, N.S. Bosco, T. Zgonena, Evaluation of high-temperature exposure
of rack-mounted photovoltaic modules, Proc. IEEE PVSC (2009) 2399–2404.
[145] D.C. Miller, M.D. Kempe, S.H. Glick, S.R. Kurtz, Creep in photovoltaic
modules: examining the stability of polymeric materials and components,
Proc. IEEE PVSC (2010) 262–268.
[146] W. Whitney, R.D. Andrews, Yielding of glassy polymers: volume effects,
J. Polym. Sci. C16 (1967) 2981–2990.
[147] N. Tanio, Structural relaxation and refractive index of low-loss poly(methyl
methacrylate) glass, Polym. J. 34 (2002) 466–470.
[148] N. Tanio, T. Nakanishi, Physical aging and refractive index of poly(methyl
methacrylate) glass, Polym. J. 38 (2006) 814–818.
[149] A. Wypch, E. Duval, A. Mermet, G. Boiteux, L. David, J. Ulanski, S. Etienne,
Kovacs effect in PMMA observed by low-frequency Raman scattering (boson
peak), J. Non-Cryst. Solids 352 (2006) 4562–4567.
[150] H. Cheng, R.B. Bannerot, On the weathering of thin plastic films, J. Solar Eng.
105 (1983) 149–156.
[151] S. Cobb, Polymeric web lens evaluation, SAND 83-7023, 1984, pp. 1–25.
[152] H.L. Hampton, M.A. Lind, Weathering characteristics of potential solar
reflector materials: a survey of the literature, PNL-2824, 1978, pp. 1–17.
[153] T.H. Kosel, Solid particle erosion, in: P.J. Blau, S.D. Henry, G.M. Davidson,
T.B. Zorc, D.R. Levicki (Eds.), ASM Handbook: Volume 18, Friction, Lubrica-
tion, and Wear Technology, ASM International, Materials Park, 1998.
[154] J.H. Nielson, A. Gilchrist, Erosion by a stream of solid particles, Wear 11
(1968) 111–122.
[155] H. Hojo, K. Tsuda, C.M. Thai, Erosion damage of polymethyl methacrylate,
J. Soc. Mater. Sci. 29 (1980) 731–735.
2065
[156] H. Hojo, K. Tsuda, C.M. Thai, Erosion mechanism of polymethyl methacry-
late at high angle of attack, J. Soc. Mater. Sci. 29 (1980) 1215–1220.
[157] P.V. Rao, S.G. Young, D.H. Buckley, Solid spherical glass particle impinge-
ment studies of plastic materials, NASA 2161 (1983) 1–18.
[158] P.V. Rao, S.G. Young, D.H. Buckley, Solid microglass particle impingement
studies of thermoplastic materials at normal incidence, ASLE Trans. 27
(1984) 373–379.
[159] P.V. Rao, D.H. Buckley, Angular particle impingement studies of thermo-
plastic materials at normal incidence, ASLE Trans. 29 (1986) 283–298.
[160] R.H. Telling, J.H. Jilbert, J.E. Field, The erosion of aerospace materials by solid
particle impact, Proc. SPIE 3060 (1997) 56–67.
[161] H. Getu, A. Ghobeity, J.K. Spelt, M. Papini, Abrasive jet micromachining of
polymethylmethacrylate, Wear 263 (2007) 1008–1015.
[162] H. Getu, A. Ghobeity, J.K. Spelt, M. Papini, Abrasive jet micromachining of
acrylic and polycarbonate polymers at oblique angles of attack, Wear 265
(2008) 888–901.
[163] J.P. Thornton, The effect of sandstorms on PV arrays and components, Proc.
Solar (1992) 81–85.
[164] B. Baum, M. Binette, Prevention of soiling of heliostat surfaces, SAND
79-8182, 1979, pp. 1–112.
[165] A.R. Mahoney, R.B. Pettit, G.S. Kinoshita, Accelerated aging of aluminized
acrylic solar mirrors, SAND 84-0532, 1984, pp. 1–20.
[166] D. Goosens, E. Van Kerschaver, Aeolian dust deposition on photovoltaic
solar cells: the effects of wind velocity and airborne dust concentration on
cell performance, Solar Energy 66 (1999) 277–289.
[167] D.C. Jordan, R.M. Smith, C.R. Osterwald, E. Gelak, S.R. Kurtz, Outdoor PV
degradation comparison, Proc. IEEE PVSC (2010) 2694–2697.
[168] M.C. Alonso-Garcia, W. Hermann, W. Böhmer, B. Proisy, Thermal and
electrical effects caused by outdoor hot-spot testing in associations of
photovoltaic cells, Prog. Photovolt.: Res. Appl. 11 (2003) 293–307.
[169] W. Herrmann, M.C. Alonso, N. Martin, W. Böhmer, B. Proisy, Improvement of
photovoltaic modules—measures for withstanding electrical and thermal
effects caused by reverse biasing of cells, TÜV Rheinland Rep. ERK5-CT1999-
00005, 2002.
[170] O. Hamberg, F.D. Tomlinson, Sensitivity of thermal surface solar absorp-
tance to particulate contamination, Air Force Report SAMSO-TR-71-264,
1971, pp. 1–8.
[171] H.P. Garg, Effect of dirt on transparent covers in flat-plate solar energy
collectors, Solar Energy 15 (1974) 299–302.
[172] J.E. Gilligan, J. Brzuskiewicz, Outdoor testing of solar energy utilization
materials, Proc. Inst. Environ. Sci. (1976) 303–307.
[173] R.B. Pettit, B.L. Butler, Semiannual review ERDA thermal power systems,
dispersed power systems, distributed collectors, and research and develop-
ment. Mirror materials and selective coatings. SAND 77-0111, 1977, pp. 1–11.
[174] R.S. Berg, Heliostat dust buildup and cleaning studies, SAND 78-0510, 1978,
pp. 1–34.
[175] J.B. Blackmon, M. Curcija, Heliostat reflectivity variations due to dust
buildup under desert conditions, Proc. Inst. Environ. Sci. (1978) 169–183.
[176] J.B. Blackmon, H.H. Dixon, Dust buildup tests of heliostats and mirror
specimens, MDC-G-7543, 1978, pp. 1–82.
[177] J.M. Freese, Effects of outdoor exposure on the solar reflectance properties
of silvered glass mirrors, SAND 78-1649, 1978, pp. 1–31.
[178] R.B. Pettit, J.M. Freese, D.E. Arvizu, Specular reflectance loss of solar mirrors
due to dust accumulation, in: Proceedings of the Seminar Testing Solar
Energy Materials and Systems, 1978, pp. 164–168.
[179] J.M. Freese, Effects of outdoor exposure on the solar reflectance properties
of silvered glass mirrors, in: Proceedings of the International Solar Energy
Meeting, 1979, pp. 1340–1344.
[180] B. Nimmo, S.A.M. Seid, Effects of dust on the performance of thermal and
photovoltaic flat plate collectors in Saudi Arabia—preliminary results, in:
Proceedings of the International Conference on Alternative Energy Sources,
1979, pp. 223–225.
[181] M.B. Sheratte, Cleaning agents and technologies for concentrating solar
collectors, SAND 79-7052, 1979, pp. 1–96.
[182] P.J. Call, Summary of Solar Experience with the soiling of optical surfaces,
SERI/TP-334-478, 1980, pp. 1–12.
[183] E.F. Cuddihy, Theoretical considerations of soil retention, Solar Energy
Mater. 3 (1980) 21–33.
[184] R.B. Pettit, J.M. Freese, Wavelength dependent scattering caused by dust
accumulation on solar mirrors, Solar Energy Mater. 3 (1980) 1–20.
[185] A.R. Hoffman, C.R. Maag, Photovoltaic module soiling studies May 1978
–October 1980, DOE/JPL-1012-49, 1980, pp. 1–57.
[186] E.P. Roth, A.J. Anaya, The effect of natural soiling and cleaning on the size
distribution of particles deposited on glass mirrors, Trans. ASME 102 (1980)
248–256.
[187] E.P. Roth, R.B. Pettit, The effect of soiling on solar mirrors and techniques
used to maintain high reflectivity, Solar Materials Science, Academic Press,
San Francisco, 1980, pp. 199–227.
]188] R.M. Bethea, M.T. Barriger, P.F. Williams, S. Chin, Environmental effects on
solar concentrator mirrors, Solar Energy 27 (1981) 497–511.
[189] R.M. Bethea, E.G. Collier, J.D. Reichert, Dust storm simulation for accelerated
life testing of solar collector mirrors, J. Solar Eng. (1983) 329–335.
[190] D.E. Randall, V.L. Morris, Initial experience and preliminary results—solar
collector materials exposure to the IPH site environment, SAND 81-0290,
1981, pp. 1–196.
2066 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
[191] E.F. Cuddihy, Surface soiling: theoretical mechanisms and evaluation of
low-soiling coatings, in: Proceedings of the Flat-Plate Solar Array Project
Research Forum, 1983, pp. 379–396.
[192] E.F. Cuddihy, P.B. Willis, Antisoiling technology: theories of surface soiling
and performance of antisoiling surface coatings, DOE/JPL-1012-102, 1984,
pp. 1–32.
[193] B. Khoshaim, F. Huraib, A. Al-Sani, A.A. Salim, M.S. Imamura, Performance of
350 kW photovoltaic power system for Saudi Arabian villages after 30
months, Proc. IEEE PVSC (1984) 1426–1432.
[194] M.S. El-Shobokshy, A. Mujahid, A.K.M. Zakzouk, Effects of dust on the
performance of concentrator photovoltaic cells, IEE Proc. 132 (1985) 5–8.
[195] A.A.M. Sayigh, S. Al-Jandal, H. Ahmed, Dust effect on solar flat surfaces
devices in Kuwait, in: Proceedings of the Workshop on Physics, 1985,
pp. 353–367.
[196] E. Cuddihy, C. Coulbert, A. Gupta, R. Liang, Flate-plate solar array project
final report volume VII: module encapsulation, DOE/JPL-1012-125, 1986,
pp. 1–63.
[197] H. Al-Busairi, A. Al-Kandari, Performance evaluation of photovoltaic mod-
ules in Kuwait, Proc. Intl. PVSEC (1987) 323–326.
[198] N.M. Nahar, J.P. Gupta, Effect of dust on transmittance of glazing materials
for solar collectors under arid zone conditions of India, Solar Wind Technol.
7 (1990) 237–243.
[199] S.A.M. Said, Effects of dust accumulation on performance of thermal and
photovoltaic flat-plate collectors, Appl. Energ. 37 (1990) 73–84.
[200] A. Hasan, A.A. Sayigh, The effect of sand dust accumulation on the light
transmittance, reflectance, and absorbance of the PV glazing, Renewable
Energy: Technology and the Environment, Pergamon Press, New York, 1992,
pp. 461–466.
[201] P.C. Pande, Effect of dust on the performance of PV panels, Proc. Intl. PVSEC
(1992) 539–542.
[202] M.S. El-Shobokshy, F.M. Hussein, Effect of dust with different physical
properties on the performance of photovoltaic cells, Solar Energy 51
(1993) 505–511.
[203] J. Bonvin, Dirt deposit level measurements on different glass type in various
surroundings, Proc. Eur. PVSEC (1995) 740–742.
[204] H. Becker, W. Vaassen, W. Hermann, Reduced output of solar generators due
to pollution, Proc. Eur. PVSEC (1997) 251–255.
[205] V.L. Morris, Final report solar collector materials exposure to the IPH site
environment, SAND 81-7028, 1981, pp. 1–287.
[206] R. Hammond, D. Srinivasan, A. Harris, K. Whitfield, Effects of soiling on PV
module and radiometer performance, Proc. IEEE. PVSC (1997) 1121–1124.
[207] A.Y. Al-Hasan, A new correlation for direct beam solar radiation received by
photovoltaic panel with sand dust accumulated on its surface, Solar Energy
63 (1998) 323–333.
[208] H. Haeberlin, J.D. Graf, Gradual reduction of PV generator yield due to
pollution, in: World Conference and Exhibition on PV Solar Energy Genera-
tion, 1998, pp. 2764–2767.
[209] D. Goossens, E. Van Kerschaever, Aeolian dust deposition on photovoltaic
solar cells: the effects of wind velocity and airborne dust concentration on
cell performance, Solar Energy 66 (1999) 277–289.
[210] A.A. Hegazy, Effect of dust accumulation on solar transmittance through
glass covers of plate-type collectors, Renew. Energy 22 (2001) 525–540.
[211] M. Vivar, R. Herrero, R. Moretón, F. Martinez-Moreno, G. Sala, Effect of
soiling on PV concentrators: comparison with flat modules, Proc. Eur. PVSEC
(2001) 142–145.
[212] A.M. El-Nashar, Effect of dust deposition on the performance of a solar
desalination plant operating in an arid desert area, Solar Energy 75 (2003)
421–431.
[213] A. Sahm, A. Gray, R. Boehm, Cleanliness maintenance for an Amonix lens
system, Proc. ASME ISEC (2005) 817–822.
[214] A. Kimber, L. Mitchell, S. Nogradi, H. Wenger, The effect of soiling on large
grid-connected photovoltaic systems in California and the southwest region
of the United States, Proc. IEEE WCPVC (2006) 2391–2395.
[215] A. Kimber, The effect of soiling on photovoltaic systems located in arid
climates, Proc. Eur. PVSEC (2007) 3023–3026.
[216] L.D. Banchik, A. Sahm, R. Boehm, Fresnel lens soiling characterization and
effect upon performance, Proc. Energy Sustainability (2009) 1–5.
[217] W. Lam, P. Nielson, A. Ravitz, D. Cocosa, Getting the most energy out of
Google’s solar panels, Google Inc.: Mountain View, 2009 (internal
presentation).
[218] E.B. Murphy, S.E. Forman, Measuring dirt on photovoltaic modules, COO-
4094-37, 1979, pp. 1–16.
[219] D. Goossens, Z.Y. Offer, Comparisons of day-time and night-time dust
accumulation in a desert region, J. Arid Environ. 31 (1995) 253–281.
[220] R.P. Young, Low-scatter mirror degradation by particle contamination, Opt.
Eng. 15 (1976) 516–520.
[221] R.S. Berg, A survey of mirror-dust interactions, SAND 77-1461C, 1977, pp. 1–17.
[222] D. Goossens, The effect of surface curvature on the deposition of loess: a
physical model, CATENA 15 (1988) 179–194.
[223] K. Kidney, Solar reflector soiling pattern distributions and reflectance
measurements requirements, SERI/TP-255-3369, 1980, pp. 1–36.
[224] C.G. Pantano, D.E. Clark, L.L. Hench, Corrosion of Glass, Books for Industry
and the Glass Industry, New York, 1979.
[225] R.L. Patrick, The effect of water on the adhesive bond, in: Proceedings of the
Symposium on Adhesion Principles and Practice, 1980.
[226] D. Goossens, Z.Y. Offer, A. Zangvil, Wind tunnel experiments and field
investigations of eolian dust deposition on photovoltaic solar collectors,
Solar Energy 50 (1993) 75–84.
[227] D. Goossens, Z. Offer, Eolian deposition of dust over symmetrical hills: an
evaluation of wind tunnel data by means of terrain measurements,
Z. Geomorph. N.F. 37 (1993) 103–111.
[228] R.M. Bradway, F.A. Record, National assessment of the urban particulate
problem: volume II particle characterization, EPA-450/3-76-025, 1976,
1–288.
[229] F.H. Haynie, Size distribution of particles that may contribute to soiling of
material surfaces, J. Air Pollut. Contr. Assess. 35 (1985) 552–554.
[230] E.B. Murphy, S.E. Forman, Measuring dirt on photovoltaic modules, CO
O-4094-37, 1979, pp. 1–20.
[231] H. Yoon, D.E. Joslin, D.C. Law, D. Krut, R.R. King, P. Vijayakumar, D. Peterson,
J. Hanley, N. Karam, Application of infrared reflecting (IRR) coverglass on
multijunction III–V solar cells, Proc. WPEC (2007) 1861–1864.
[232] K. Puckett, W. Russo, Canada—United States Transboundary Particulate
Matter Science Assessment, Public Works and Government Services of
Canada, Toronto, 2004, pp. 1–150.
[233] Our nation’s air: status and trends through 2008, Report EPA-454/R-09-002,
2010, pp. 1–54.
[234] http://www.airnow.gov/.
[235] K. Awazu, H. Onuki, Photo-induced synthesis of amorphous SiO2 film from
tetramethoxy-silane on polymethylmethacrylate at room temperature,
J. Non-Cryst. Solids, 215 (1997) 176–181.
[236] A. Gombert, W. Glaubitt, K. Rose, J. Driebholz, B. Blasi, A. Heinzel, D. Sporn,
W. Döll, V. Wittwer, Antireflective transparent covers for solar devices, Solar
Energy 68 (2000) 357–360.
[237] U. Schultz, H. Jänchen, N. Kaiser, Plasma pre-treatment and coating of
PMMA Fresnel lenses, Proc. SPIE 3738 (1999) 511–516.
[238] C. Neinhaus, W. Barthlott, Characterization and distribution of water-
repellent, self-cleaning plant surfaces, Ann. Bot. 79 (1997) 667–677.
[239] M. Nosonovksy, B. Bhushan, Multiscale friction mechanisms and hierarchal
surfaces in nano- and bio-tribology, Mater. Sci. Eng. 58 (2007) 162–193.
[240] S.J. Wilson, M.C. Hutley, The optical properties of ‘moth eye’ antireflection
surfaces, Opt. Acta 29 (1982) 933–1009.
[241] F. Burmeister, C. Kohn, R. Kuebler, G. Kleer, B. Bläsi, A. Gombert, Applica-
tions for TiAIN- and TiO2-coatings with nanoscale surface topographies,
Surf. Coat. Techol. 200 (2005) 1555–1559.
[242] I.P. Parkin, R.G. Palgrave, Self-cleaning coatings, J. Mater. Chem. 15 (2005)
1689–1695.
[243] T. Guang-Li, H. Hong-Bo, S. Jian-Da, Effect of microstructure of TiO2 thin
films on optical band gap energy, Chin. Phys. Lett. 22 (2005) 1787–1789.
[244] L.L. Hench, D.E. Clark, Physical chemistry of glass surfaces, J. Non-Cryst.
Solids 28 (1978) 83–105.
[245] P.R. Anderson, F.R. Bacon, B.W. Byrum, Effect of surface treatments on the
chemical durability and surface composition of soda-lime glass bottles,
J. Non-Cryst. Solids 19 (1975) 251–262.
[246] D.L. King, M.A. Quintana, J.A. Kratochvil, D.E. Ellibee, B.R. Hansen, Photo-
voltaic module performance and durability following long-term field
exposure, Prog. Photovolt. Res. Appl. 8 (2000) 241–256.
[247] D.V. Rosato, R.T. Schwartz (Eds.), John Wiley and Sons, New York, , 1968.
[248] J.E. Guillet, Polymer Photophysics and Photochemistry, Cambridge Univer-
sity Press, Cambridge, 1987.
[249] S.H. Hamid, Handbook of Polymer Degradation, Marcel Dekker, New York, 2000.
[250] J.W. Martin, J.A. Lechner, R.N. Varner, Quantitative characterization of
photodegradation effects of polymeric materials exposed in weathering
environments, in: W.D. Ketola, D. Grossman (Eds.), Accelerated and Outdoor
Durability Testing of Organic Materials, ASTM STP 1202, American Society
for Testing Materials, Philadelphia, 1984, pp. 12–51.
[251] P.R.E.J. Cowley, H.W. Melville, The photo-degradation of polymethylmetha-
crylate. I: the mechanism of degradation, Proc. Roy. Soc. Lond. A210 (1952)
461–481.
[252] R.B. Fox, L.G. Isaacs, Photodegradation of high polymers. Part III—photolysis
of poly(methyl methacrylate) in vacuum and in air, NRL 5720 (1961) 1–19.
[253] D.G. Gardner, L.H. Epstein, Protection against radiation damage in poly-
methylmethacrylate by high-energy electrons and by ultraviolet light,
J. Chem. Phys. 34 (1961) 1653–1660.
[254] A.R. Shultz, Degradation of polymethyl methacrylate by ultraviolet light,
J. Phys. Chem. B65 (1961) 967–972.
[255] R.B. Fox, T.R. Price, Indirect photodegradation of poly(methyl methacrylate)
and poly(a-methylstyrene) in solution, in: Proceedings of the Meeting of
the American Chemical Society (1964) 390–395.
[256] H.H.G. Jellinek, I.C. Wang, Photodegradation of polymethylmethacrylate in
solution, Kol. Zeit. Zeit. Polym. 202 (1965) 1–10.
[257] J.P. Allison, Photodegradation of poly(methyl methacrylate) J. Polym. Sci.
Part A-1 4 (1966) 1209–1221.
[258] D.G. Gardner, G.A. Henry, Sensitivity of mixtures of syndiotactic and
isotactic poly(methyl methacrylate) to chain scission induced by ultraviolet
light, Polym. Lett. 5 (1967) 101–104.
[259] J.R. MacCallum, C.K. Schoff, Photolytic decomposition of polymethacrylates.
Part 2—poly(methyl methacrylate); quantum yields and molecular weight
changes, Trans. Faraday Soc. 16 (1971) 2383–2388.
[260] K. Morimoto, S. Suzuki, Ultraviolet irradiation of poly(alkyl acrylates) and
poly(alkyl methacrylates), J. Appl. Polym. Sci. 16 (1972) 2947–2961.
 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
[261] N. Grassie, Photodegradation of methylmethacrylate copolymers,
J.M.S.—Pure Appl. Chem. 34 (1973) 247–257.
[262] M. Kato, Y. Yoneshige, Photodegradation behavior of some vinyl ketone
copolymers, Die Makromil. Chem. 164 (1973) 159–169.
[263] V.Y. Shlyapintokh, V.I. Gol’denberg, Effect of photostabilizers on the rate of
photodegradation of polymethylmethacrylate, Eur. Polym. J. 10 (1974)
679–684.
[264] H. Kambe, H. Watanabe, S. Nagatomo, Y. Itoh, Degradation of poly(methyl
methacrylate) by flash irradiation, Polymer 18 (1977) 1063–1067.
[265] M. Abouelezz, P.F. Waters, A study of photodegradation of poly(methyl
methacrylate), Proc. 1978.
[266] D. Panke, W. Wunderlich, Mechanism of the photo-oxidative degradation of
PMMA, J. Appl. Polym. Sci. 35 (1979) 321–328.
[267] N. Grassie, A.J. Davidson, Photodegradation in solution of copolymers of
methyl vinyl ketone and methyl isopropenyl ketone with methyl metha-
crylate, Polym. Deg. Stab. 3 (1980) 45–52.
[268] A. Gupta, R. Liang, F.-D. Tsay, J. Moacanin, Characterization of a dissociative
excited state in the solid state: photochemistry of poly(methyl methacry-
late). Photochemical processes in polymeric systems, Macromolecules 13
(1980) 1696–1700.
[269] A. Torikai, K. Fueki, Photodegradation of poly(methyl methacrylate) studied
by ESR and viscosity measurements, Polym. Photochem. 2 (1982) 297–308.
[270] B. Dickens, J.W. Martin, D. Waksman, Thermal and photolytic degradation of
plates of poly(methyl methacrylate) containing monomer, Polymer 25
(1984) 706–715.
[271] S. Beauvois, D. Renault, R. Lazzaroni, L.D. Laude, J.L. Bredas, Physico-
chemical characterization of the effect of excimer laser irradiation on
PMMA thin films, Appl. Surf. Sci. 109/110 (1997) 218–221.
[272] ASTM E490, Standard solar constant and zero air mass solar spectral
irradiance tables, ASTM International, West Conshohocken, 2006.
[273] A.R. Shultz, P.I. Roth, G.B. Rathman, Light scattering and viscosity study of
electron-irradiated polystyrene and polymethacrylates, J. Polym. Sci. 22
(1956) 495–506.
[274] N. Grassie, A. Scotney, L. Mackinnon, Observations of the photothermal
degradation of polymethacrylates by using thermal volatilization analysis, J.
Polym. Sci. 15 (1977) 251–254.
[275] J.W. Martin, A Stochastic model for predicting service life of photolytically
degraded poly(methyl methacrylate), J. Appl. Polym. Sci. 29 (1984) 777–794.
[276] M.J. Bowden, E.A. Chandross, I.P. Kaminow, Mechanism of refractive index
increase in photosensitized poly(methyl methacrylate), Polym. Eng. Sci. 14
(1974) 494–497.
[277] J.W. Martin, B. Dickens, D. Waksman, D.P. Bentz, W.E. Byrd, E. Embree,
W.E. Roberts, Thermal degradation of poly(methyl methacrylate) at 50 1C to
125 1C, J. Appl. Polym. Sci. 34 (1987) 377–393.
[278] P. Alexander, A. Charlesby, M. Ross, The degradation of solid polymethyl-
methacrylate by ionizning radiation, Proc. Roy. Soc. (Lond.) A223 (1954)
392–404.
[279] L.A. Wall, D.W. Brown, Chemical activity of gamma-irradiated polymethyl
methacrylate, J. Res. Nat. Bur. Stan. 57 (1956) 131–136.
[280] A.R. Shultz, High-effect radiation effects on polyacrylates and polymetha-
crylates, J. Polym. Sci. 35 (1959) 369–379.
[281] A. Todd, The mechanisms of radiation-induced changes in vinyl polymers, J.
Polym. Sci. 42 (1960) 223–247.
[282] Y. Hajimoto, N. Tamura, S. Okamoto, Electron spin resonance spectra of
poly(methyl methacrylate) irradiated at 77 K, J. Polym. Sci. A3 (1965)
255–263.
[283] J.F. Kircher, F.A. Sliemers, R.A. Markle, W.B. Gager, R.I. Leininger, On the
degradation and electron spin resonance spectra of irradiated methacrylate
polymers, J. Phys. Chem. 69 (1965) 189–193.
[284] C. David, D. Fuld, G. Geuskens, A. Charlesby, Degradation thermique du
polymethacrylate de methyl irradie des rayons ultra-vilets ou gamme, Eur.
Polym. J. 5 (1969) 641–648.
[285] C. David, D. Fuld, G. Geuskens, Radiolysis of polymethylmethacrylate in the
prescence of ethylmercaptan, Makromol. Chem. 139 (1970) 269–276.
[286] T. Ichikawa, H. Yoshida, Mechanism of radiation-induced degradation of
poly(methyl methacrylate) as studied by ESR and electron spin echo
methods, J. Polym. Sci. 28 (1990) 1185–1196.
[287] N. Ueno, Y. Kobayashi, T. Sekiguchi, H. Ikeura, K. Sugita, K. Honma,
K. Tanaka, E. Ortı́, R. Viruela, Valence bands of poly(methylmethacrylate)
and photoion emission in vacuum ultraviolet region, J. Appl. Phys. 72 (1992)
5423–5428.
[288] J. Gaynor, G. Schueneman, P. Schuman, J.P. Harmon, Effects of fluorinated
substituents on the refractive index and optical radiation resistance of
methacrylates, J. Appl. Polym. Sci. 50 (1993) 1645–1653.
[289] T. Ichikawa, K.-I. Oyama, T. Kondoh, H. Yoshida, Efficiency of radiation-
induced main-chain scission of poly(methyl methacrylate) depends on the
irradiation temperature because of coexisting monomer, J. Polym. Sci. A32
(1994) 2487–2492.
[290] H. Kudoh, T. Sasuga, T. Seguchi, Y. Katsumura, High-energy-ion-irradiation
effects on polymer materials: 3. The sensitivity of cellulose triacetate and
poly(methyl methacrylate), Polymer 37 (1996) 2903–2908.
[291] S.N. Goyanes, G.M. Benites, J.J. Gonzalez, G.H. Rubiolo, A.J. Marzocca,
Dynamical mechanical properties of polymethylmethacrylate after expo-
sure to 60Co gamma radiation, Polym. Test. 16 (1997) 7–18.
[292] S.M. Sayyah, I.A. Sabbah, M.M. Ayoub, B.N. Barsoum, E. Elwy, Molecular
weight and TLC studies on the gamma degradation of some poly(methyl
2067
methacrylate) in the absence and presence of N,N-dioxalyl p-aminobenzene
sulphonic acid and its iron (II) complex, Polym. Deg. Stab. 58 (1997) 1–9.
[293] P.R.H. Bertolucci, J.P. Harmon, E. Biagtan, G. Schueneman, E. Goldberg,
P. Schuman, W. Schuman, Radiation effects on low refractive index,
fluorinated methacrylate polymers for fiber cladding, Polym. Eng. Sci. 38
(1998) 699–705.
[294] A.N. Pravednikov, E.N. Teleshov, I. Shen-Kan, S.S. Medvedev, Investigation of
the mechanism of radiolysis of polymers containing quaternary atoms of
carbon, J. Polym. Sci. 58 (1962) 1039–1048.
[295] K.K. Okudaira, E. Morikawa, S. Hasegawa, P.T. Sprunger, V. Saile, K. Seki,
Y. Harada, N. Ueno, Radiation damage of poly(methylmethacrylate) thin
films analyzed by UPS, J. Electron Spectrosc. 88–91 (1998) 913–917.
[296] H. Nagai, Infrared spectra of stereoregular polymethyl methacrylate, J. Appl.
Polym. Sci. 7 (1963) 1697–1714.
[297] K.E. Isakson, Use of infrared specular reflectance in study of ultraviolet
degradation of polymer films, J. Paint Technol. 44 (1972) 41–62.
[298] M. Celina, D.K. Otteson, K.T. Gillen, R.L. Clough, FTIR emission spectroscopy
applied to polymer degradation, Polym. Deg. Stab. 58 (1997) 15–31.
[299] K. Aparecia, S. Aquino, E.S. Araujo, Effects of a hindered amine stabilizer
(HAS) on radiolytic and thermal stability of poly(methyl methacrylate), J.
Appl. Polym. Sci. 110 (2008) 401–407.
[300] B. Smith, Infrared Spectral Interpretation: A Systematic Approach, CRC
Press, Washington D.C, 1995.
[301] ASTM E313, Standard practice for calculating yellowness and whiteness
indices from instrumentally measured color coordinates, ASTM Interna-
tional, West Conshohocken, 2005.
[302] B.-L. Denq, Y.-S. Hu, W.-Y. Chiu, L.-W. Chen, Y.-S. Chiu, Thermal degradation
behavior and physical properties for poly(methyl methacrylate) blended
with propyl ester phosphazene, Polym. Deg. Stab. 57 (1997) 269–278.
[303] J.A. Esfahani, Oxygen sensitive thermal degradation of PMMA: a numerical
study, Combust. Sci. Technol. 174 (2002) 183–198.
[304] N. Grassie, H.W. Melville, The mechanism of thermal degradation of
polymethyl methacrylate, Disc. Faraday Soc. 2 (1947) 378–383.
[305] N. Grassie, H.W. Melville, The thermal degradation of polyvinyl compounds.
I. A new type of molecular still, Proc. Roy. Soc. Lond., A 199 (1949) 1–13.
[306] S. Bywater, Thermal degradation of polymethyl methacrylates, J. Phys.
Chem. 57 (1953) 879–882.
[307] S.L. Madorsky, Rates and activation energies of thermal degradation of
styrene and acrylate polymers in a vacuum, J. Polym. Sci. 11 (1953) 491–506.
[308] V.A. Brockhaus, E. Jenkel, U  ber die kinetic des thermischen abbaues von
polymethacrylsäuremethylester, Die Makromol. Chem. 18/19 (1956)
263–293.
[309] G. Bagby, R.S. Lehre, J.C. Robb, Kinetic measurements by micropyrolysis-
GLC. Thermal degradation of polymethylmethacrylate possessing lauryl–
mercaptyl end groups, Polymer 10 (1960) 683–690.
[310] M. Gordon, Probability model theory of chain-end activated polymer
degradation. III. Statistical kinetics of the degradation of polymethyl
methacrylate, Trans. Faraday Soc. 64 (1960) 19–29.
[311] J.E. Clark, H.H.G. Jellinek, Thermal degradation of poly(methyl methacrylate)
in a closed system, J. Polym. Sci. A3 (1965) 1171–1184.
[312] H.H.G. Jellinek, M.D. Luh, Thermal degradation of isotactic and syndiotactic
poly(methyl methacrylate), J. Phys. Chem. 70 (1966) 3672–3680.
[313] H.H.G. Jellinek, M.D. Luh, Thermal degradation of polymethylmethacrylate:
energies of activation, Die makromol. Chem. 115 (1968) 89–102.
[314] I.C. McNeil, A study of the thermal degradation of methyl methacrylate
polymers and copolymers by thermal volatilization analysis, Eur. Polym. J. 4
(1968) 21–30.
[315] A. Ballistreri, G. Montaudo, C. Puglisi, Reliability of the volatilization method
for determination of the activation energy in the thermal decomposition of
polymers, J. Thermal Anal. 29 (1984) 237–241.
[316] T. Hirata, T. Kashiwagi, J.E. Brown, Thermal oxidation of poly(methyl
methacrylate), Am. Polym. Preprints 25 (1984) 176–177.
[317] T. Hirata, T. Kashiwagi, J.E. Brown, Thermal and oxidative degradation of
poly(methyl methacrylate): weight loss, Macromolecules 18 (1985)
1410–1418.
[318] T. Kashiwagi, T. Hirata, J.E. Brown, Thermal and oxidative degradation of
poly(methyl methacrylate): molecular weight, Macromolecules 18 (1985)
131–138.
[319] T. Kashiwagi, A. Inaba, J.E. Brown, K. Hatada, T. Kitayama, E. Masuda, Effects
of weak linkages on the thermal and oxidative degradation of poly(methyl
methacrylates), Macromolecules 19 (1986) 2160–2168.
[320] N.A. Khalturniskii, High-temperature pyrolysis of polymers, J. Therm. Anal.
32 (1987) 1675–1682.
[321] A.Z. El-Sonbati, M.A. Diab, Thermal stability of poly(acryloyl chloride)
homopolymer and copolymers of acryloyl chloride with methyl methacry-
late, J. Therm. Anal. 34 (1988) 769–776.
[322] A. Inaba, T. Kashiwagi, J.E. Brown, Effects of initial molecular weight on
thermal degradation of poly(methyl methacrylate): Part 1—model 1, Polym.
Deg. Stab. 21 (1988) 1–20.
[323] L.E. Manring, Thermal degradation of saturated poly(methyl methacrylate),
Macromolecules 21 (1988) 528–530.
[324] T. Kashiwagi, A. Inaba, Behavior of primary radicals during thermal
degradation of poly(methyl methacrylate), Polym. Deg. Stab. 26 (1989)
161–184.
[325] L.E. Manring, Thermal degradation of saturated poly(methyl methacrylate).
2. Vinyl-terminated polymer, Macromolecules 22 (1989) 2673–2677.
2068 D.C. Miller, S.R. Kurtz / Solar Energy Materials & Solar Cells 95 (2011) 2037–2068
[326] J. Rychlý, J. Pavlinec, Thermal degradation of free radically prepared
poly(methyl methacrylate). A nonisothermal weight loss study, Polym.
Deg. Stab. 28 (1990) 1–15.
[327] L.E. Manring, Thermal degradation of poly(methyl methacrylate). 4. Ran-
dom side-group scission, Macromolecules 24 (1991) 3304–3309.
[328] S.M. Lomakin, J.E. Brown, R.S. Breese, M.R. Nyden, An investigation of the
thermal stability and char-forming tendency of corss-linked poly(methyl
methacrylate), Polym. Deg. Stab. 41 (1993) 229–243.
[329] H. Arisawa, T.B. Brill, Kinetics and mechanisms of flash pyrolysis of
poly(methyl methacrylate) (PMMA), Combust. Flame 109 (1997) 415–426.
[330] J.D. Peterson, S. Vyazokin, C.A. Wight, Kinetic study of stabilizing effect of
oxygen on thermal degradation of poly(methyl methacrylate), J. Phys.
Chem. 103 (1999) 8087–8092.
[331] B.J. Holland, J.N. Hay, The kinetics and mechanisms of the thermal
degradation of poly(methyl methacrylate) studied by thermal analysis-
Fourier transform infrared spectroscopy, Polymer 42 (2001) 4825–4835.
[332] B.J. Holland, J.N. Hay, The effect of polymerization conditions on the kinetics
and mechanisms of thermal degradation of PMMA, Polym. Deg. Stab. 77
(2002) 435–439.
[333] M. Ferriol, A. Gentilhomme, M. Cochez, N. Oget, J.L. Mieloszynski, Thermal
degradation of poly(methyl methacrylate) (PMMA): modeling of DTG and
TG curves, Polym. Deg. Stab. 79 (2003) 271–281.
[334] Y.-H. Hu, C.-Y. Chen, The effect of end groups on the thermal degradation of
poly(methyl methacrylate), Polym. Deg. Stab. 82 (2003) 81–88.
[335] K. Wondraczek, J. Adams, J. Fuhrman, Effect of thermal degradation on glass
transition temperature of PMMA, Macromol. Chem. Phys. 205 (2004)
1858–1862.
[336] K. Aparecida, S. Aquino, E.S. Araujo, Effects of a hindered amine stabilizer
(HAS) on radiolytic and thermal stability of poly(methyl methacrylate), J.
Appl. Polym. Sci. 110 (2008) 401–407.
[337] J.A. Esfahani, Oxygen-senstive thermal degradation of PMMA: a numerical
study,, Combust. Sci. Technol. 174 (2002) 183–198.
[338] S.I. Stoliarov, P.R. Westmoreland, M.R. Nyden, G.P. Forney, A reactive
molecular dynamics model of thermal decomposition in polymers: I.
poly(methyl methacrylate), Polymer 44 (2003) 883–894.
[339] R. Chandra, R. Saini, New developments in the degradation, stabilization,
and sensitization of poly(methyl methacrylate), JMS Rev. Macromol. Chem.
Phys. C30 (1990) 155–208.
[340] V.D. Rumyantsev, O.I. Chosta, V.A. Grilikhes, N.A. Sadchikov, A.A. Soluyanov,
M.Z. Shvarts, V.M. Andreev, Terrestrial and space concentrator PV modules
with composite (glass–silicone) Fresnel lenses, Proc. IEEE PVSC (2002)
1596–1599.
[341] V.M. Andreev, E.A. Ionova, V.D. Rumyantsev, Concentrator PV modules of
‘‘all-glass’’ design with modified structure, in: Proceedings of the World
Conference on Photovoltaic Energy Conversion, 2003, 3O-D9-05.
[342] V.D. Rumyantsev, A.E. Chalov, N.Yu. Davidyuk, E.A. Ionova, N.A. Sadchikov,
V.M. Andreev, Solar concentrator modules with Fresnel lens panels, in:
Proceedings of the International Conference on Solar Concentrators, 2007,
pp. 33–36.
[343] V.D. Rumyantsev, Concentrator photovoltaics, in: A. Luque, V. Andreev
(Eds.), Springer Series in Optical Sciences, vol. 130, 2007, pp. 151–174.
[344] R.W. Jebens, Laminated Fresnel lens, SAND 78-0988, 1978, pp. 1–8.
[345] R.C. Hodge, Design and development of a laminated Fresnel lens for point
focus systems, SAND 80-2373, 1980, pp. 1–7.
[346] L. Matalon, Development of a laminated glass-plastic Fresnel lens, SAND 80-
0852, 1980, p.1.
[347] B. Ketola, K.R. McIntosh, A. Norris, M.K. Tomalia, Silicones for photovoltaic
encapsulation, Proc. Eur. PVSEC (2008) 2969–2973.
[348] K.R. McIntosh, J.N. Cotsell, J.S. Cumpston, The effect of accelerated aging
tests on the optical properties of silicone and EVA encapsulants, Proc. Eur.
PVSEC (2008) 3475–3482.
[349] B. Riegler, R. Elgin, R. Thomaier, A characterization of UV effects on optical
silicones used in opto-electronic devices and new developments in resistant
materials, Proc. SPIE 6478 (2007) 64780F.
[350] Dow Corning Corp., Develop silicone encapsulation systems for terrestrial
silicon solar arrays, DOE/JPL-954995-80, 1979, pp. 1–57.
[351] Y. Israëli, J. Lacoste, J. Cavezzan, J. Lemaire, Photooxidation of polydimethyl-
siloxane oils and resins. IV—effect of phenyl groups, Polym. Deg. Stab. 47
(1995) 357–362.
[352] G. Deshpande, M.E. Rezac, The effect of phenyl content on the degradation
of poly(dimethyl diphenyl) siloxane copolymers, Polym. Deg. Stab. 74
(2001) 363–370.
[353] C.G. Zimmerman, On the kinetics of photodegradation in transparent
silicone, J. Appl. Phys. 103 (2008) 083547.
[354] C.G. Zimmerman, Time dependent degradation of photovoltaic modules by
ultraviolet light, Appl. Phys. Lett. 92 (2008) 241110.
[355] J.F. White, W.B. Silverman, Some studies on the solarization of glass, J. Am.
Ceram. Soc. 33 (1950) 252–257.
[356] J. Vitko, J.E. Shelby, Solarization of heliostat glasses, SAND 79-8754, 1979,
pp. 1–27.
[357] J. Vitko, J.E. Shelby, Solarization of heliostat glasses, Solar Energy Mater. 3
(1980) 69–80.
[358] J.E. Shelby, J. Vitko, Hydrogen transport in a machinable glass-ceramic, J.
Non-Cryst. Solids 45 (1981) 83–92.
[359] J.E. Shelby, J. Vitko, The reduction of iron in soda-lime-silicate glasses by
reaction with hydrogen, J. Non-Cryst. Solids 53 (1982) 155–163.
[360] A.J. Faber, Optical properties and redox state of silicate glass melts, C.R.
Chim. 5 (2002) 705–712.
[361] S.B. Donald, A.M. Swink, H.D. Schreiber, High-iron ferric glass, J. Non-Cryst.
Solids 342 (2006) 539–543.
[362] W. Thiemsorn, K. Keowkamnerd, S. Phanichphant, P. Suwannathada,
H. Hessenkemper, Influence of glass basicity on redox interactions of
iron–manganese–copper ion pairs in soda-lime-silica glass, Glass Phys.
Chem. 34 (2008) 19–29.
[363] J.S. Stroud, Photoionization of Ce3 + in glass, J. Chem. Phys. 35 (3) (1961)
844–850.
[364] M.D. Kempe, M. Kilkenny, T.J. Moricone, J.Z. Zhang, Accelerated stress
testing of hydrocarbon-based encapsulants for medium-concentration CPV
applications, Proc. IEEE PVSC (2009) 1826–1831.
[365] P. Klemchuck, M. Ezrin, G. Lavigne, W. Holley, J. Galica, S. Agro, Investigation
of the degradation and stabilization of EVA-based encapsulant in field-aged
solar energy modules, Polym. Deg. Stab. 55 (1997) 347–365.
[366] R.H. Doremus, Weathering and internal friction in glass, J Non-Cryst. Solids
3 (1970) 369–374.
[367] D.E. Carlson, Ion depletion of glass at a blocking anode: I, theory and
experimental results for alkali silicate glasses, J. Am. Ceram. Soc. 57 (1974)
291–294.
[368] L.L. Hench, Characterization of glass corrosion and durability, J. Non-Cryst.
Solids 19 (1975) 27–39.
[369] W.A. Lanford, K. Davis, P. Lamarche, T. Lauren, R. Groleau, R.H. Doremus,
Hydration of soda-lime glass, J. Non-Cryst. Solids 33 (1979) 249–266.
[370] J.E. Shelby, J. Vitko, Weathering of low iron float and CGW-0317 glasses,
SAND 79-8225, 1979, pp. 1–23.
[371] D.E. Clark, E.L. Yen-Bower, Corrosion of glass surfaces, Surf. Sci. 100 (1980)
53–70.
[372] J.E. Shelby, J. Vitko, C.G. Pantano, Weathering of glasses for solar applica-
tions, Solar Energy Mater. 3 (1980) 97–110.
[373] H. Scholze, Chemical durability of glasses, J. Non-Cryst. Solids 52 (1982)
91–103.
[374] R.G. Newton, The durability of glass—a review, Glass Technol. 26 (1985) 21–38.
[375] M.H. Krohn, J.R. Hellmann, B. Mahieu, C.G. Pantano, Effect of tin-oxide on
the physical properties of soda-lime-silica glass, J. Non-Cryst. Solids 351
(2005) 455–465.
[376] J.W. Hutchinson, Z. Suo, Mixed mode cracking in layered materials, in:
J.W. Hutchinson, T.Y. Wu (Eds.), Advances in Applied Mechanics, vol. 29,
Academic Press, San Diego, 1992, pp. 63–191.
[377] L.B. Freund, S. Suresh, Delamination and fracture, Thin Film Materials:
Stress, Defect Formation and Surface Evolution, Cambridge University Press,
Cambridge, 2003.
[378] T. Schult, M. Neubauer, Y. Bessler, P. Nitz, A. Gombert, Temperature
dependence of Fresnel lenses for concentrating photovoltaics, in: Proceed-
ings of the Concentrating Photovoltaic Workshop, 2009.
[379] T. Hornung, A. Bachmaier, P. Nitz, A. Gombert, Temperature dependent
measurement and simulation of Fresnel lenses for concentrating photovoltaics,
in: Proceedings of the International Conference on CPV Systems, 2010.
[380] A. Gupta, Effect of photodegradation on chemical structure and surface
characteristics of silicone pottants used in solar cell modules, DOE/JPL-
5101-79, 1978, pp. 1–15.
[381] S.W. Addy, D.W. Clegg, A.A. Collyer, G.C. Corfield, P. Crum, Chemorheological
studies of a g-irradiated dimethylsiloxane elastomer, J. Rheol. 31 (1987)
297–315.
[382] Y. Israëli, J.-L. Philippart, J. Cavezzan, J. Lacoste, J. Lemaire, Photo-oxidation
of polydimethylsiloxane oils: I—effect of silicon hydride groups, Polym.
Deg. Stab. 36 (1992) 179–185.
[383] K.K. de Groh, T.A. McCollum, Low Earth orbit durability of protected silicone
for refractive photovoltaic concentrator arrays, J. Spacecr. Rockets 32 (1995)
103–109.
[384] S.B. Lin, High-temperature stability of silicone polymers and related
pressure-sensitive adhesives, ACS Symp. Ser. 603 (1995) 37–51.
[385] G. Camino, S.M. Lomakin, M. Lazzari, Polydimethylsiloxane thermal degra-
dation Part 1. Kinetic aspects, Polymer 42 (2001) 2395–2402.
[386] G. Camino, S.M. Lomakin, M. Lazzari, Polydimethylsiloxane thermal degra-
dation Part 2. Kinetic aspects, Polymer 42 (2001) 2011–2015.
[387] W. Huang, Y. Fu, C. Wang, Y. Xu, Y. Chen, Radiation effects on methyl vinyl
polysilicone foam, J. Appl. Polym. Sci. 89 (2003) 3437–3441.
[388] A. Ghanbari-Siahkali, S. Mitra, P. Kingshott, K. Almdal, C. Bloch,
H.K. Rehmeier, Investigation of the hydrothermal stability of cross-linked
liquid silicone rubber (LSR), Polym. Deg. Stab. 90 (2005) 471–480.
[389] J.A. Dever, B.A. Banks, L. Yan, Effects of vacuum ultraviolet radiation on
DC93-500 silicone, J. Spacecr. Rockets 43 (2006) 386–392.
[390] R. Elgin, B. Riegler, R. Thomaier, How temperature effects transmission of
silicone encapsulants, in: Proceedings of the IMAPS Advanced Technology
Workshop on Packaging and Assembly of Power LEDs, 2006.
[391] A. Labouriau, J.D. Cox, J.R. Schoonover, B.M. Patterson, G.J. Havrilla,
T. Stephens, D. Taylor, N.M.R. Mössbauer, ATR-FTIR spectroscopic investiga-
tion of degradation in RTV siloxane foams, Polym. Deg. Stab. 92 (2007)
414–424.
[392] F. Delor-Justin, N.S. Tomer, R.P. Singh, J. Lacoste, Durability of crosslinked
polydimethylsyloxanes: the case of composite insulators, Sci. Technol. Adv.
Mater. 9 (2008) 02440 6.
[393] P.J. Burchill, R.H. Stacewicz, Effect of tropical exposure on the
craze resistance of some acrylic sheet materials, Mater. Forum 12 (1988) 3–7.