Professional Documents
Culture Documents
net/publication/338520535
CITATIONS READS
9 1,679
2 authors, including:
James Harvey
University of Central Florida
205 PUBLICATIONS 3,319 CITATIONS
SEE PROFILE
All content following this page was uploaded by James Harvey on 17 January 2020.
James E. Harvey
Richard N. Pfisterer
James E. Harvey, Richard N. Pfisterer, “Understanding diffraction grating behavior, part II:
parametric diffraction efficiency of sinusoidal reflection (holographic) gratings,” Opt. Eng. 59(1),
017103 (2020), doi: 10.1117/1.OE.59.1.017103.
Abstract. With the widespread availability of electromagnetic (vector) analysis codes for describing the diffrac-
tion of electromagnetic waves by periodic grating structures, the insight and understanding of nonparaxial
parametric diffraction grating behavior afforded by approximate methods (i.e., scalar diffraction theory) is being
ignored in the education of most optical engineers today. We show that the linear systems formulation of non-
paraxial scalar diffraction theory enables the development of a scalar parametric diffraction grating model [for
transverse electric (TE) polarization] for sinusoidal reflection gratings with arbitrary groove depths and arbitrary
nonparaxial incident and diffracted angles. This scalar parametric analysis is remarkably accurate as it includes
the ability to redistribute the energy from evanescent orders into the propagating ones, thus allowing the cal-
culation of nonparaxial diffraction efficiencies to be predicted with an accuracy usually thought to require rigorous
electromagnetic theory. These scalar parametric predictions of diffraction efficiency are compared to paraxial
scalar and rigorous electromagnetic (vector) predictions for a variety of nonparaxial diffraction grating configu-
rations, thus providing quantitative limits of applicability of nonparaxial scalar diffraction theory to sinusoidal
reflection gratings as a function of the grating period-to-wavelength ratio (λ∕d ). © 2020 Society of Photo-Optical
Instrumentation Engineers (SPIE) [DOI: 10.1117/1.OE.59.1.017103]
Keywords: sinusoidal phase gratings; scalar parametric nonparaxial diffraction grating efficiency model; generalization of the classical
paraxial expression for diffraction efficiency of sinusoidal reflection gratings.
Paper 191193 received Aug. 26, 2019; accepted for publication Dec. 6, 2019; published online Jan. 9, 2020.
this paper: if one does not thoroughly understand the non- pffiffiffiffiffiffi r a
paraxial diffraction behavior of a single sinusoidal reflection tA ðx1 ; y1 Þ ¼
EQ-TARGET;temp:intralink-;e001;326;752 E0 Gaus 1 exp i sinð2πx1 ∕dÞ ; (1)
b 2
grating, then there is little chance of understanding the wide-
angle scatter behavior of random rough surfaces. where
In Sec. 2 of this paper, we derive the well-known classical
expression of the (monochromatic) diffraction efficiency of 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r r
a coarse (d ≫ λ) sinusoidal reflection grating with arbitrary Gaus 1 ¼ exp −π 1
EQ-TARGET;temp:intralink-;e002;326;705 and r1 ¼ x21 þ y21 ;
b b
groove depth as predicted by classical scalar diffraction
theory for paraxial incident and diffracted angles.15 This (2)
detailed derivation was not included in Refs. 1, 7, or even
15. We also look at the well-known approximation of that where E0 is the peak irradiance of the incident Gaussian
expression for “shallow” (smooth) sinusoidal reflection gra- beam. We have ignored a factor representing the average
tings and quantitatively evaluate how shallow is shallow phase delay through the grating. The parameter a represents
enough for that approximation to be valid. In addition we the “peak-to-peak excursion of the sinusoidal phase variation
illustrate the diffracted intensity profile as a function of of the wavefront emerging from the grating.”15
both groove depth and diffraction angle as predicted by the The irradiance distribution produced by a Gaussian beam
paraxial model for a sinusoidal reflection grating of period in an arbitrary plane perpendicular to the propagation direc-
d ¼ 20λ operating at normal incidence and discuss its tion is given by the squared modulus of the complex ampli-
limitations. tude distribution in that plane
In Sec. 3, we review the linear systems formulation of
r
nonparaxial scalar diffraction theory,2–4 which suggests a Eðr; zÞ ¼ jtA ðx; yÞj2 ¼ E0 Gaus2
EQ-TARGET;temp:intralink-;e003;326;547: (3)
simple generalization of the classical paraxial parametric b
expression for the diffraction efficiency of sinusoidal reflec-
tion gratings. This nonparaxial scalar diffraction theory The total radiant power in the Gaussian beam is obtained by
enables the formulation of a parametric diffraction grating integrating the irradiance over that perpendicular plane
model for sinusoidal reflection (phase) gratings with arbi- Z Z
trary groove depths and arbitrary nonparaxial incident and 2π ∞ r
PT ¼ E0 Gaus2
rdr dφ: (4)
diffracted angles.
EQ-TARGET;temp:intralink-;e004;326;478
φ¼0 r¼0 b
In Sec. 4, we generalize the classical paraxial parametric
expression for diffraction efficiency to include arbitrary As described in Ref. 18 (p. 421),
groove depth and nonparaxial incident and diffraction
pffiffiffi
angles. This exercise will provide insight into the nature r 2r
of the proper obliquity factor not only for nonparaxial dif- Gaus2 ¼ Gaus :
EQ-TARGET;temp:intralink-;e005;326;421 (5)
b b
fraction grating phenomena but also for a general wide-angle
surface scatter theory.13,14,16,17 Likewise, as shown in Ref. 18 (p. 74), the volume under
In Sec. 5, we compare the diffraction efficiencies pre- Gauss(r∕d) is equal to d2 .
dicted by the classical paraxial expression, the generalized The relationship between the total radiant power of the
scalar nonparaxial parametric expression, and rigorous laser beam and the peak irradiance in a given plane
electromagnetic (vector) diffraction theory for a variety of perpendicular to its propagation is thus given as
nonparaxial diffraction grating configurations. A summary,
statement of conclusions, and an extensive set of references b2
then completes the paper. PT ¼ E : (6)
2 0
EQ-TARGET;temp:intralink-;e006;326;320
2 Derivation of the Classical Expression for Making use of the Bessel function identity15 provided by
the Paraxial Efficiency of Sinusoidal Eq. (7)
Phase Gratings
X∞
We will first derive the well-known classical paraxial expres- a a
exp i sinð2πx1 ∕dÞ ¼ Jm expði2πmx1 ∕dÞ;
sion of the (monochromatic) diffraction efficiency of para-
EQ-TARGET;temp:intralink-;e007;326;255
2 m¼−∞
2
xial (d ≫ λ) sinusoidal reflection gratings with arbitrary
groove depth by using classical scalar diffraction theory. (7)
We will assume that the grating is illuminated by a small where Jm is a Bessel function of the first kind, order m, and
two-dimensional radially symmetric Gaussian laser beam the fact that the exponential function Fourier transforms into
of width b (radius at which the field drops to e−π of its peak a shifted delta function,18 we can apply the convolution theo-
value) that underfills the oversized physical grating. rem and write the Fourier transform of Eq. (1) as
X ∞
2.1 For Arbitrary Groove Depth, Normal Incidence, a
and Paraxial Diffraction Angles F ftA ðx1 ; y1 Þg ¼ Jm δðξ − m∕d; ηÞ
EQ-TARGET;temp:intralink-;e008;326;152
m¼−∞
2
Using the symbolic notation for special functions popular- pffiffiffiffiffiffi
ized by Goodman15 and Gaskill,18 the complex amplitude E0 b2 GausðbρÞ; (8)
transmittance (reflectance) function of a perfectly reflecting pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sinusoidal phase grating illuminated as described above can where ρ ¼ ξ2 þ η2 , and the spatial frequencies ξ and η are
be expressed as the reciprocal variables to x and y, respectively.
Owing to the replicating property of the delta function and the diffracted orders are approximately rotationally
under convolution,18 this can be written as an infinite array symmetric and centered. Hence we can go back to radial
of scaled and shifted Gaussian functions: coordinates
pffiffiffiffiffiffi X ∞
Eo b4 X
∞
EQ-TARGET;temp:intralink-;e009;63;719
a 2 a 2 br2
F ftA ðx1 ; y1 Þg ¼ E0 b 2 Jm Gaus½bðξ − m∕d; ηÞ: E2 ðx2 ; y2 Þ ≈ 2 2 Jm Gaus : (14)
2 λ f m¼−∞ λf
EQ-TARGET;temp:intralink-;e014;326;719
m¼−∞ 2
(9)
Again from Eq. (5), this can be written as
If the grating is placed immediately behind an aberration- pffiffiffi
Eo b4 X
∞
2 a 2br2
free positive lens of focal length f with the beam waist E2 ðx2 ; y2 Þ ≈ 2 2 Jm Gaus : (15)
λ f m¼−∞ λf
EQ-TARGET;temp:intralink-;e015;326;661
2
Z Z ∞ pffiffiffi
Eo b4 2 a 2π 2br2
Pm ¼ 2 2 J m Gaus rdr dϕ:
λ f λf
EQ-TARGET;temp:intralink-;e016;326;579
From Eq. (10) the irradiance distribution on the focal 2 ϕ¼0 r¼0
plane can be written as (16)
E2 ðx2 ; y2 Þ
EQ-TARGET;temp:intralink-;e011;63;529 But as shown in Ref. 18, this integral is just equal to
2 λ2 f 2 ∕2b2 , and hence
∞
Eo b 4 X a
¼ 2 2 Jm Gaus½bðξ − m∕d; ηÞj ; b2
λ f m¼−∞ 2 ξ ¼ x2 ∕λf Pm ¼ Eo Jm 2 a
: (17)
η ¼ y2 ∕λf
EQ-TARGET;temp:intralink-;e017;326;497
2 2
(11)
Dividing Eq. (17) by Eq. (6), the diffraction efficiency of
or, evaluating this function at spatial frequencies ξ ¼ x2 ∕λf the m’th diffracted order of a sinusoidal phase grating oper-
and η ¼ y2 ∕λf, we obtain ating with normal incidence and paraxial diffracted angles
(θi ¼ 0, d ≫ λ) is given as
∞ 2
Eo b4 X a b mλf
E2 ðx2 ;y2 Þ¼ 2 2 Jm Gaus x2 − ;y2 : Pm 2 a
λ f m¼−∞ λf ηm ¼ ¼ RJ m
EQ-TARGET;temp:intralink-;e012;63;415
2 d ;
EQ-TARGET;temp:intralink-;e018;326;407 (18)
PT 2
(12)
where R is the reflectance of the reflection grating material
Since the size of the incident beam is large compared to the and a represents the peak-to-peak excursion of the sinusoidal
grating period (b ≫ d), there is a negligible overlap between phase variation of the wavefront emerging from the grating.
the individual Gaussian functions (diffracted orders). Thus, The phase variation induced upon the reflected wavefront
there are no cross terms in the squared modulus of the above by the surface variation of the grating is given by 2π∕λ times
summation, hence the optical path difference (OPD) of the reflected wavefront
from one reflected by the mean surface shown in Fig. 2.7 The
E b4 X ∞
EQ-TARGET;temp:intralink-;e013;63;299
a b mλf
E2 ðx2 ;y2 Þ ¼ 2o 2 J2m Gaus2 x2 − ;y2 : total groove depth (peak to peak) of the sinusoidal surface is
λ f m¼−∞ 2 λf d equal to h. The surface height variation and the correspond-
ing reflected wavefront are illustrated as a function of x when
(13)
the grating grooves are aligned with the y axis:
For the current situation of normal incidence and paraxial
2π nh 2π
diffracted orders (θi ¼ 0, d ≫ λ), the shift parameter in φðxÞ ¼ OPDðxÞ; OPDðxÞ ¼ sin x ; (19)
λ
EQ-TARGET;temp:intralink-;e019;326;241
Fig. 1 Geometry for producing a Fraunhofer diffraction pattern of an Fig. 2 Illustration of a normally incident plane wavefront reflected
aperture (or grating) in the back focal plane of a lens. from a sinusoidal reflection grating.
where n is the refractive index of the media adjacent to the shallow grating approximation expressed in Eq. (22) is accu-
grating surface. If this media is air (n ¼ 1), the relative phase rate to within 1% for groove depths h < 0.03λ, or accurate to
variation is thus given as within 5% for h < 0.07λ but has an error >10% for h > 0.1λ.
It also illustrates that the maximum diffraction efficiency that
hπ 2π can be achieved in the first order with a paraxial sinusoidal
φðxÞ ¼ sin x : (20)
λ phase grating is equal to η1 ≡ 0.3386.
EQ-TARGET;temp:intralink-;e020;63;719
incidence. m¼−∞
The conservation of energy is easily shown for a perfectly (23)
conducting paraxial (d ≫ λ) reflection grating at normal
incidence because the sum over m from −∞ to ∞ of the where a is the peak-to-peak phase excursion and b is the
squared Bessel function in Eq. (18) is equal to unity. radius of the Gaussian beam (radius at which the field drops
Thus, we get the same expression for the diffracted effi- to e−π of its peak value).7
ciency of the m’th diffracted order from a sinusoidal phase Figure 4 illustrates this diffracted intensity profile plotted
grating as Goodman reported for normal incidence and para- as a function of groove depth along the x axis and diffraction
xial diffracted angles in the case of a perfectly conducting angle along the y axis. The maximum value of J21 ða∕2Þ is
(R ¼ 1) finite square grating illuminated by a unit amplitude 0.3386 and occurs for a ¼ 3.68, corresponding to a groove
normally incident plane wave.15 depth of h ¼ 0.293λ. The diffraction efficiency of the first
few orders for this value of a are tabulated in Table 1.
Note that the energy falls off rapidly, with 99.88% of the
2.2 Approximation for Diffraction Efficiency of diffracted radiant power contained in diffracted orders
Shallow Sinusoidal Phase Gratings jmj ≤ 3. This paraxial model is accurate only for very coarse
In addition to being a paraxial (d ≫ λ) grating, if the sinus- gratings (d ≫ λ). This paraxial model, thus, leads to the
oidal reflection grating is also shallow (i.e., the groove depth common misconception that it is impossible to get more than
is much less than a wavelength of the incident light), then the 33.86% of the incident energy into the first diffracted order
diffraction efficiency of the first orders of the sinusoidal of a sinusoidal phase grating.
reflection grating can be approximated as A simple thought experiment can prove this misconcep-
tion to be false. Suppose that we have a perfectly conducting
η1 ≡ J21 ða∕2Þ ≈ a2 ∕16:
EQ-TARGET;temp:intralink-;e022;63;331 (22) grating with a normally incident beam. If you decrease the
grating period, the diffracted angles increase and the higher
In Sec. 5.3 of Part I of this two-part paper,1 we provided orders eventually go evanescent. When only the 0 and 1
the expression in Eq. (18) without a detailed derivation and orders remain, changing the incident angle will cause the
graphically compared the diffraction efficiency of the first −1 order to go evanescent. Then one can vary the groove
diffracted orders as predicted by Eqs. (18) and (22). That depth to squelch the energy in the zero order. For a perfectly
curve is duplicated here as Fig. 3, and it establishes that the conducting sinusoidal reflectance grating, we can thus get
100% of the incident energy in the þ1 diffracted order.
Since there are now only two propagating orders, and there
is no energy in the 0’th order, “all” of the energy has to be
contained in the þ1 order. The maximum diffraction effi-
ciency of the first order for a perfectly conducting sinusoidal
reflection grating is therefore equal to unity, not 0.3386 as
predicted by the paraxial model.
If we use this paraxial model to describe the behavior of
fine (small spatial period) gratings whose diffracted orders
do not fall within the paraxial region, we will obtain incorrect
results that lead to a variety of other misconceptions. For
example: (1) the small angle approximation to the grating
equation inherent in Eq. (23) leads to the erroneous prediction
Fig. 3 Comparison of diffracted efficiency of a sinusoidal phase of equally spaced diffracted orders (in diffraction angle);1
grating as predicted by Eq. (18) and the common approximation for (2) when a diffracted order of finite width is located near
shallow (smooth) gratings expressed in Eq. (22). 90 deg, the predicted diffracted intensity (or irradiance)
Fig. 4 Diffracted intensity profile as a function of groove depth and diffraction angle as predicted for
a sinusoidal reflection grating of period d ¼ 20λ operating at normal incidence.
where K is a renormalization factor to be defined in Eq. (28) For large incident and diffraction angles, a portion of the
below, γ o is the cosine of the incident angle, and As is the differ- diffracted radiance distribution function will fall outside of
ential source area. the unit circle in direction cosine space. Those evanescent
Fig. 5 The profile of a broad Gaussian radiance distribution in direction cosine space, shifted due to a
64-deg incident angle is illustrated. The renormalized propagating portion of the radiance distribution is
also illustrated.
orders are imaginary and do not diffract any radiant energy a function of the incident and/or the diffracted angles in addi-
away from the grating surface. Parseval’s theorem from tion to the wavelength and the groove depth as expressed in
Fourier transform theory then requires that a renormalization Eq. (21). This general situation is depicted in Fig. 6.
factor, K, be applied to the diffracted radiance distribution The relative phase variation clearly depends upon both
remaining inside the unit circle in direction cosine space, the incident and the diffracted angles and is now given as
as illustrated in Fig. 5.2
The normalization factor, K, is given by the ratio of the 2π
ϕð^x; y^ Þ ¼ OPD; OPD ≈ ðγ i þ γ m Þhð^x; y^ Þ∕2; (28)
integral of the radiance distribution function over infinite λ
EQ-TARGET;temp:intralink-;e028;326;547
1−α 2
1
α¼−1 2 β¼− 1−α The peak-to-peak excursion of the sinusoidal phase varia-
tion is again equal to twice the peak amplitude of the relative
≡ Renormalization factor: (27)
phase variation
This renormalization factor K differs from unity only if a ¼ 2πðγ i þ γ m Þh∕λ: (30)
the diffracted radiance distribution function extends beyond
EQ-TARGET;temp:intralink-;e030;326;428
the unit circle in direction cosine space (i.e., only if evanes- Recall that Eq. (18) was an expression for diffraction effi-
cent waves are produced). The well-known Wood’s anoma- ciency for a paraxial application where radiant intensity and
lies that occur in diffraction-grating efficiency measurements radiance are essentially equivalent. However, the primary
are entirely consistent with this predicted renormalization in premise of the linear systems formulation of nonparaxial sca-
the presence of evanescent waves.2,4 lar diffraction theory described in Ref. 2 is that diffracted
radiance (not irradiance or intensity) is the fundamental
4 Generalizing the Classical Scalar Parametric radiometric quantity that exhibits shift invariance with
Expression for Diffraction Efficiency respect to changes in the incident and diffracted angles only
We will now show that the classical paraxial parametric when formulated in direction cosine space. Hence, for non-
expression for diffraction efficiency given in Eq. (18) can paraxial diffraction angles, we must use the following pro-
be generalized to accurately model nonparaxial situations. cedure for calculating diffraction grating efficiencies with
We will do this by modifying the parameter a (the peak- our scalar-based nonparaxial parametric expression for dif-
to-peak phase excursion of the diffracted wavefront) to be fraction efficiencies of sinusoidal phase gratings:
Fig. 6 Illustration of the wavefront diffracted at an arbitrary angle for an arbitrary incident angle.
the efficiencies of all propagating orders slightly larger than Fig. 9 Diffraction grating efficiency of the first order of a perfectly con-
unity. By applying the renormalization factor in Eq. (32), the ducting sinusoidal reflection grating (h∕d ¼ 0.20) in Littrow condition
sum of the efficiencies of all propagating orders is precisely as predicted by the Beckmann theory, the paraxial scalar theory,
the nonparaxial scalar diffraction theory presented in this paper, and
unity as required. a rigorous integral vector theory.
the %error of the paraxial prediction exceeds 86%, whereas Thus, as λ∕d varies from 0.1 to 1.9, as shown in Fig. 9, the
the %error of the scalar parametric prediction remains <4%.7 incident angle (and θ1 ) varies from ∼3 deg to 65 deg.
Figures 7 and 8 demonstrate that the classical paraxial It is quite remarkable that our nonparaxial scalar parametric
expression of Eq. (18) is indeed very restrictive with regard diffraction grating efficiency model is so accurate over this
to diffraction angle, and the generalized scalar parametric wide a range of diffraction angles.
expression greatly relaxes those restrictions, at least up to It is of practical interest to continue to test the angular
moderate diffraction angles exceeding 30 deg.7 range of validity of this generalized scalar parametric model
of diffraction grating efficiencies for various diffracted
5.3 Diffraction Efficiency of the +1 Order Satisfying orders at increasingly larger diffracted angles.
the Littrow Condition for a Reflection Grating
Many practical diffraction grating applications are not per-
formed at normal incidence but instead involve maintaining 5.4 For Normal Incidence and Large Nonparaxial
a given diffracted order in the Littrow condition (θm ¼ θi for Diffraction Angles
reflection gratings). Also, substantially more insight into dif- Recall from Sec. 3 of Ref. 1 that λ∕d is the separation of the
fraction grating behavior, and our scalar parametric model, equally spaced diffracted orders when displayed in a direc-
can be demonstrated by calculating and displaying diffrac- tion cosine diagram. Figure 10 illustrates the direction cosine
tion grating efficiencies as a function of the dimensionless diagrams for three different values of Δα ¼ 0.2, 0.25, and
parameter λ∕d. 0.333 where the fifth, fourth, and third diffracted orders are
Figure 9, therefore, illustrates the diffraction efficiency of just going evanescent, i.e., reaching a diffraction angle of
a perfectly conducting (R ¼ Q ¼ 1) sinusoidal reflection 90 deg.
grating with groove depth-to-period ratio (h∕d) equal to Figure 11 graphically illustrates the predicted diffracted
0.20 when the þ1 order is maintained in the Littrow condi- efficiency (paraxial, scalar intensity parametric, and rigorous
tion. The above nonparaxial scalar diffraction theory pro- vector calculations) versus the quantity λ∕d for diffracted
vides remarkably good agreement with rigorous integral orders 3, 4 and 5 at normal incidence.
electromagnetic theory, not merely in the paraxial regime and We have again set the wavelength at 0.6328 μm and the
the smooth surface (shallow grating) regime, but over the groove depth at 0.293λ. Clearly, from the simple planar gra-
entire range of λ∕d, including at λ∕d ¼ 0.667 where there ting equation, the fifth, fourth, and third orders will reach
is a major Rayleigh anomaly when the −1 and þ2 diffracted 90 deg and go evanescent at λ∕d equals 0.2, 0.25, and
orders go evanescent simultaneously (see discussion of 0.33, respectively. The dashed black lines in Fig. 11 are the
Fig. 34 in Ref. 1). paraxial predictions from Eq. (18) for a perfectly conducting
Figure 9 was previously published in Ref. 4, and it should sinusoidal reflection grating with the parameter a given by
be noted that to maintain the þ1 order in the Littrow con- Eq. (21). The black dotted lines are the rigorous predictions
dition, each data point on these curves requires a different from the Waterman integral method. The lower solid curve
incident angle. It is readily shown by substituting the for each case is the intensity parametric prediction from
Fig. 10 Direction cosine diagrams for normal incidence and (a) Δα ¼ 0.2, (b) Δα ¼ 0.25, and
(c) Δα ¼ 0.333. For these three cases, the fifth, fourth, and third diffracted orders are just going
evanescent.
Fig. 11 Comparison of predicted paraxial, rigorous, and intensity Fig. 12 Predicted paraxial, rigorous, parametric intensities and para-
parametric diffracted efficiencies as a function of λ∕d for the third, metric radiance diffracted efficiencies as a function of λ∕d for the third,
fourth, and fifth diffracted orders for normal incidence. fourth, and fifth orders for normal incidence.
PT 2
orders to the propagating ones. Finally, we substituted the
Finally, Fig. 14 shows the angular range of validity (given polarization reflectance, Q,13 for the scalar reflectance, R,
an allowable 5% error from the rigorous prediction) of the in Eq. (18) to obtain the generalized parametric expression
empirically modified scalar radiance parametric model for for diffraction efficiency of the m’th diffracted order for a
diffraction efficiency of sinusoidal reflection gratings. Note sinusoidal reflection grating with arbitrary groove depth and
that these empirically modified scalar radiance parametric arbitrary incident and diffracted angles given by Eq. (31).
predictions of diffraction efficiency are accurate (to within This generalized scalar nonparaxial parametric model for
5%) out to diffraction angles of 56 deg, 65 deg, and 77 deg predicting diffraction grating efficiencies is thus capable of
for the third, fourth, and fifth diffracted orders (at normal predicting (1) Rayleigh (Wood’s) anomalies and (2) polariza-
incidence), respectfully. tion effects when other than TE polarized light is present.
Table 3 shows the comparisons of the %errors of the vari- These two effects are usually thought to require rigorous
ous scalar diffraction models from the rigorous EM predic- electromagnetic theory.
tions. The second column of Table 3 tabulates the angular In Sec. 5, we graphically compared the predictions of this
range over which the %error of the empirically modified sca- generalized scalar nonparaxial parametric model of diffrac-
lar radiance parametric model remains less than 5% for the tion efficiency for sinusoidal reflection gratings to rigorous
third, fourth and fifth diffracted orders. The third and fourth calculations using the well-known vector Waterman method10
columns of Table 3 tabulates the corresponding %errors of and determined the angular range of validity (for a 5% error
the classical scalar paraxial model and the scalar intensity tolerance) for a variety of grating configurations.
parametric model respectively. These comparisons demon- An empirical modification was made by raising the γ m in
strate a dramatic improvement in the angular range of Eq. (31) to an arbitrary power in order to achieve a better fit
validity of simple intuitive nonparaxial scalar parametric to the rigorous predictions for the third, fourth, and fifth dif-
diffraction grating efficiency predictions. fracted orders over the entire range of diffracted angles illus-
trated in Figs. 12 and 13. Table 3 summarizes the results of
raising γ m to the 0.7 power, as shown bypEq. 7
ffiffiffi (35). Note that
6 Summary, Discussion, and Conclusions 0.7 is very close in value to both 2/3 and 2∕2. This suggests
We have developed a scalar nonparaxial parametric model of that there might be a theoretical basis for this empirical
diffraction efficiency for sinusoidal phase gratings that is modification.
useful for making accurate engineering calculations and vali- Finally, it was mentioned in Sec. 1 of this paper that ran-
dated it by rigorous vector calculations. This intuitive para- dom scattering surfaces are routinely modeled as a superpo-
metric model provides insight and understanding concerning sition of sinusoidal surfaces of different amplitudes, periods,
orientations, and phases.11–14 The results of this paper may 11. P. Beckman and A. Spizzichino, The Scattering of Electromagnetic
Waves from Rough Surfaces, Pergamon Press, New York (1963).
thus be of substantial interest to the surface scattering com- 12. J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and
munity. In particular, due to an ongoing debate concerning Scattering, Optical Society of America, Washington, DC (1989).
13. J. C. Stover, Optical Scattering, Measurement and Analysis, McGraw-
the difference in the obliquity factor of the classical Hill, New York (1990).
Rayleigh–Rice and the GHSSmooth surface scatter theories,14 14. J. E. Harvey, Understanding Surface Scatter Phenomena: A Linear
the generalized expression for the parameter a (peak-to-peak Systems Formulation, SPIE Press, Bellingham, Washington (2019).
15. J. W. Goodman, Introduction to Fourier Optics, p. 69, McGraw-Hill,
excursion of the sinusoidal phase variation) of Eq. (30) New York (1968).
clearly shows that the obliquity factor in the GHSSmooth sur- 16. S. O. Rice, “Reflection of electromagnetic waves from slightly rough
face scatter theory is rooted in the fundamental diffraction surfaces,” Commun. Pure Appl. Math. 4, 351–378 (1951).
17. J. E. Harvey and R. N. Pfisterer, “Comparison of the GHSSmooth and the
behavior of sinusoidal phase gratings, and this gives cre- Rayleigh-rice surface scatter theories,” Proc. SPIE 9961, 996103
dence to the GHSSmooth obliquity factor. (2016).
18. J. D. Gaskill, Linear Systems, Fourier Transforms, and Optics, Wiley,
New York (1978).
References 19. J. E. Harvey and E. A. Nevis, “Angular grating anomalies: effects of
1. J. E. Harvey and R. N. Pfisterer, “Understanding diffraction grating finite beam size upon wide-angle diffraction phenomena,” Appl. Opt.
behavior: including conical diffraction and Rayleigh anomalies from 31, 6783–6788 (1992).
transmission gratings,” Opt. Eng. 58(8), 087105 (2019). 20. J. A. Ratcliff, “Some aspects of diffraction theory and their application
2. J. E. Harvey et al., “Diffracted radiance: a fundamental quantity in to the ionosphere,” Rep. Prog. Phys. 19, 188–267 (1956).
a non-paraxial scalar diffraction theory,” Appl. Opt. 38, 6469–6481
(1999). James E. Harvey is a retired associate professor from CREOL, the
3. J. E. Harvey et al., “Diffracted radiance: a fundamental quantity in non- College of Optics and Photonics at the University of Central Florida,
paraxial scalar diffraction theory: errata,” Appl. Opt. 39(34), 6374–6375 and currently is a senior optical engineer with Photon Engineering,
(2000).
4. J. E. Harvey, A. Krywonos, and D. Bogunovic, “Non-paraxial scalar LLC in Tucson, Arizona. He has a PhD in optical sciences from the
treatment of sinusoidal phase gratings,” J. Opt. Soc. Am. A 23, 858– University of Arizona and is credited with over 220 publications and
865 (2006). conference presentations in diverse areas of applied optics. He is
5. R. W. Wood, “On a remarkable case of uneven distribution of light in a member of OSA and a fellow and past board member of SPIE.
a diffraction grating spectrum,” Philos. Mag. 4, 396–402 (1902).
6. E. G. Loewen and E. Popov, Diffraction Gratings and Applications, Richard N. Pfisterer is a cofounder and president of Photon
p. 376, Marcel Dekker, New York (1997). Engineering, LLC. He received his bachelor’s and master’s degrees
7. J. E. Harvey and R. N. Pfisterer, “Parametric diffraction efficiency of in optical engineering from the Institute of Optics at the University of
non-paraxial sinusoidal reflection gratings,” Proc. SPIE 10375,
103750B (2017). Rochester in 1979 and 1980, respectively. Previously, he was the
8. D. Rudolph and G. Schmahl, “Verfaren zur Herstellung von head of optical design at TRW (now Northrop-Grumman) and senior
Rongtenlinsen und Beugungsgittern (Method for producing x-ray lenses optical engineer at Breault Research Organization. He is credited with
and diffraction gratings),” Umsch. Wiss. Tech. 67, 225 (1967). over 20 articles and conference presentations in the areas of optical
9. A. Labeyrie and J. Flamand, “Spectrographic performance of holo- design, stray light analysis, and phenomenology. He is a member of
graphically made diffraction gratings,” Opt. Commun. 1, 5–8 (1969). OSA and SPIE.
10. R. Petit, Electromagnetic Theory of Gratings, p. 98, Springer-Verlag,
Berlin (1980).