Optical Evaluation

Optical Evaluation plots how your coating transmits, reflects and absorbs light across a wavelength range. For every wavelength on the grid you set, it runs a full thin-film calculation on the active design and draws transmittance T(λ), reflectance R(λ) and absorptance A(λ) = 1 − T − R. This is the tool you reach for after every design change to confirm the spectrum still looks the way you expect. All three curves open in percent and can be read in fractions or decibels instead; transmittance can also be read as optical density.

Each quantity is available for s (TE) and p (TM) polarization and for their average. Absorptance is only meaningful when a layer or the substrate has a non-zero extinction coefficient k; for transparent dielectrics it stays at zero and T + R = 100 %.

Settings

Wavelength range and step: the span and sampling resolution of the plot, in nanometres. The default is 400–800 nm. Use a step of 0.5 nm or finer near narrow features such as bandpass notches, where a coarse grid can skip over a deep dip.

Axis units: relabels the horizontal axis in nm, µm, cm⁻¹ (wavenumber), THz or eV. This is a display choice only: the underlying sampling always stays in nanometres, so switching units never changes the computed curves. Wavenumber and energy axes increase in the opposite direction to wavelength.

AOI: the angle of incidence in degrees from the normal. You can list several angles at once; each is drawn as its own set of curves, with later angles slightly more transparent so they read as a family. Click an existing angle to edit it in place, or type a new value to add one.

Curves: toggle which quantities and polarizations are drawn: T, R and A for the average, and the individual s and p curves.

Y-axis: leave on the default fixed 0–100 % range, switch to auto-fit, or enter an explicit minimum and maximum to zoom into a shallow band.

Y-axis units: read the curves as percent, as a fraction of the incident flux (0–1), in decibels, or in optical density. The last two draw the axis logarithmically, which is what a blocking specification needs: a stopband at 0.1 % is a flat line on the bottom edge of a percent axis, and a clear 30 dB below the passband on a dB axis. The relation is T(dB) = 10·log₁₀ T and OD = −log₁₀ T, so full transmittance is 0 in both, and 10 dB of loss is one unit of density.

Decibels apply to all three curves. The decibel is a unit rather than a quantity: ten times the log of a power ratio. T, R and A are each a fraction of the incident flux, so each has a reading in decibels, all three mean the same thing on the axis, and they can share it. Absorption gains the most from it, a loss of 10 ppm being invisible on a percent axis and a clear −50 dB here.

Optical density applies to transmittance only. Density is not a unit but a quantity of its own, D = −log₁₀ T, defined from the incident beam against the beam that got through, so the name is already spoken for. There is no density of a reflectance to quote, and a density of an absorptance reads backwards: an absorptance of 0.001 % would show as density 5, which sounds like heavy blocking and is almost no loss at all. Selecting OD therefore draws the T curves and parks the R and A switches, and the R and A families in the target editor, until you pick another unit; the table and the CSV follow. An exported CSV names its columns with the unit, T_dB or T_OD, so a file opened later is not read as percentages.

Like the horizontal axis units, this is a display choice only: the computed curves and the merit targets keep the values they had. The axis picks its gridlines from the span in front of it, so a blocking filter is ruled one density unit or 10 dB at a time while an antireflection coating is ruled in hundredths, rather than both being forced onto decades. A range that starts at 0 % takes its lower edge from the curves themselves, since a logarithmic axis has no position for zero, while the top keeps the value in the field. Both the fields and the axis stop at OD 12: a thick metal layer transmits far less, and following it down would leave the reflectance in a sliver at the top of the plot.

A curve breaks off where it reaches zero, and on a dB axis the absorptance of a transparent stack disappears entirely: A is calculated as 1 − R − T, so with no absorbing material in the stack what is left is the last digits of that subtraction, not a loss you could measure. Those points are left undrawn rather than plotted somewhere near −150 dB, where they would drag the axis down and squeeze the real curves into the top of the plot. A target written at zero, such as an antireflection target at R = 0, has no position on the axis either: it is not drawn there and cannot be picked up or deleted until you return to percent.

Auto / Calculate: with Auto on, the plot recomputes whenever the design or settings change. Turn it off and use the Calculate button when you want to hold a result while editing.

The evaluation surface (front only, back only, or both sides together) is set in the Design Editor and shown as a badge on the window. Front evaluates the front coating on the substrate, Back the back coating, and Total combines both sides through the substrate. The last is what a spectrophotometer measures for a part coated on both faces. When cone-angle averaging is active, a second badge appears and the curves are averaged over the cone.

Targets on the plot

If the active design has merit-function targets, they are drawn over the spectrum so you can see what the design is being optimized toward. Point targets appear as bold X markers; band and ramp targets appear as a dotted target line with a tinted zone. Targets are color-coded by quantity (R red, T blue, A green) and by polarization through the line style. Turn the Targets button off to compare designs without the markers crowding the chart.

The Edit button lets you build and adjust targets directly on the plot instead of typing them into the Merit Function Editor:

• Draw: drag a line to add a target. Choose the quantity (R/T/A) and polarization, and whether the line is a band average or a continuous per- wavelength ramp. A flat line becomes a band average; a sloped line becomes a ramp.
• Delete: switch to the delete tool and click a target’s line or marker to remove it.
• Snapping: endpoints snap to a grid (in nm and in the vertical unit) and to existing target ends, so near-flat lines settle onto clean levels such as a 50 % band. On a dB or density axis the level grid is in that unit, half a decade by default, so a blocking level lands on −40 dB or OD 4 rather than on a percentage.

Every edit writes straight to the design’s targets, so the Merit Function Editor stays in sync and each change is a single undo step.

How to read it

The curves answer the basic questions about a coating at a glance: an antireflection design drives R toward zero across its band, a mirror drives R toward 100 %, and a filter shows sharp transitions between pass and block regions. A gap between T + R and 100 % is absorption, so check that no layer is accidentally using a metal when you expect a transparent coating.

The collapsible data table lists λ, T, R, A for the curves on screen, and Copy CSV / Save CSV export exactly those curves at the chosen angle and wavelength grid. Curves imported through Measured Spectra appear as dotted lines with open-circle markers, so you can lay a measurement over the prediction.

References

• H. A. Macleod, Thin-Film Optical Filters, 5th ed., §2.4 (Eqs. 2.111, 2.113), §2.6.4 (incoherent substrate), §8.2.1 (Eq. 8.6, transmittance in decibels), Ch. 5 (optical density of neutral-density filters).