2.1.

Metrics for LSCs

Because an optical concentrator system concentrates an incident photon flux, the ratio of photon fluxes at its incident surface to those at its exiting surface is a relevant metric. Let us call it optical concentration factor $C_{\mathrm{opt}}$. It is expressed as

In an LSC, incident photons are absorbed, and PL photons are collected by its PV cell. Hence, one can express the term ${\eta }_{\mathrm{opt}}$ in Eq. (1) as follows:⁵

The term ${\eta }_{\mathrm{abs}}$ is the absorptance of the luminescent plate. It can be calculated by Lambert-Beer’s law for a photon flux with an arbitrary spectrum $S_{\mathrm{in}}(\lambda )$ and Fresnel equations for an arbitrary incident angle. Denoting the absorption coefficient of the luminescent plate at wavelength $\lambda$ as $\mu (\lambda )$, its thickness as $d$, and the reflectance at the surface as $R$, it is expressed as follows in case of normal incidence:

The term ${\eta }_{\mathrm{col}}$ in Eq. (2) is called the collection efficiency. It is defined as the probability of the absorbed incident photon resulting in a PL photon at the PV cell. Because some PL photons are lost during the collection process, ${\eta }_{\mathrm{col}}$ depends on the position where they are generated. For LSCs with polygonal and curved luminescent plates, one can calculate ${\eta }_{\mathrm{col}}$ as a function of the coordinates of the excitation spot analytically.¹² Because further study is needed to extend this analysis to a leaf LSC, we resort to experiments to determine ${\eta }_{\mathrm{col}}$ in Sec. 3.

2.2.

Metrics for Leaf LSCs

Let us express the optical efficiency of a leaf LSC in terms of ${\eta }_{\mathrm{abs}}$ and ${\eta }_{\mathrm{col}}$. The performance of LSCs with fiber geometry has been studied by ray tracing¹³ and experiments.¹⁴ The configuration of a leaf LSC is equivalent to such a fiber LSC and two luminescent plates with their edge surfaces facing the luminescent fiber. Thus, the photocurrent of a leaf LSC has two origins: the PL photons generated by the fiber directly and those generated by the two-stage photoconversion process. So does its concentration factor. Using Eqs. (1) and (2), it is expressed as

The term ${\eta }_{\mathrm{abs}-\mathrm{P}}$ can be calculated by Eq. (3). One can calculate the term ${\eta }_{\mathrm{abs}-\mathrm{F}}$ by varying the parameter $d$ in Eq. (3) and averaging the results. Hence, if the terms ${\eta }_{\mathrm{col}-\mathrm{F}}$ and ${\eta }_{\mathrm{col}-\mathrm{P}}$ are known, ${\eta }_{\mathrm{opt}}$ can be calculated for an arbitrary incident spectrum.

Note that the term ${\eta }_{\mathrm{col}-\mathrm{P}}$ is defined as the probability of the incident photon absorbed by the plate resulting in a PL photon at the PV cell. It can be expressed as the product of each probability associated with the series of events, namely waveguiding the PL photons in the plate, coupling them to the fiber, converting them to second PL photons, and waveguiding them to the PV cell.⁷ Analysis on each step will fine-tune the leaf design in future. In the experiments described below, we assemble a leaf LSC with off-the-shelf components and measure its collective probability ${\eta }_{\mathrm{col}-\mathrm{P}}$.

3.1.

Sample Preparation

The square samples shown in Fig. 2(a) were manually assembled with off-the-shelf components. The five samples on the right have mirrors on their edge surfaces. The side length of the square samples $L$ varies from 10 to 50 mm in 10 mm steps. The plates emitting in green are 2.0 mm-thick luminescent acrylic plates (Model 1305L, Kanase Inc., Japan). Another plate from Kanase emitting in orange (Model 9989) is used for the tandem structure described in Sec. 4.2. The rods emitting in red are 1.5 mm-diameter luminescent fibers (Model FFOR-60, Plastruct, Inc.).

Fig. 2

Photographs of 2.0 mm-thick square samples. (a) The five samples on the right have reflector films attached to their edge surfaces. (b) A microscope photograph of the edge region before attaching a reflective film.

The samples in Fig. 2(a) are prepared as follows. Two 2.0 mm-thick luminescent plates are bridged with a 10 mm-wide reflective film (bottom mirror). A $25\mu \mathrm{m}$-thick adhesive optical film with a refractive index of 1.48 (Model M3D49, Mecanusa Inc., Japan) is used to attach the plates and the reflective film. The gap between the two plates is 1.8 mm. A 1.5 mm-diameter luminescent fiber is inserted into this gap. The fiber just sits on the adhesive film backed by the bottom mirror. It is not completely straight. For these reasons, there are some regions of the fiber in contact with the plates and the adhesive film. The fiber has a 5 mm-long protruded section for easy handling in the experiment at the expense of the PL photon losses by this extra distance. The same adhesive optical films are used for attaching the reflective film to the edge surfaces of the plates.

As shown in Fig. 2(b), the edge surfaces of the plates appear flat and clear. They are polished by a supplier. The picture also shows that there are some structures inside the fiber and that some PL photons are leaking out from these irregularities. In addition, the bottom mirror is reflecting the light from some surrounding objects (probably the ceiling light). Their images are blurred, indicating that the bottom and edge mirrors reflect light diffusively. As will become clear later, this has both positive and negative effects on the optical efficiency of a leaf LSC.

Because the existing luminescent components are used, the optical design of our samples is not optimized yet. For example, a larger fiber would accept more green PL photons from the edge surfaces of the plates. Concentrations of the luminescent material in the fiber should be chosen to balance the absorption of the green PL photons and the self-absorption of the red PL photons. Advanced luminescent materials with minimum overlaps between absorption and emission spectra would certainly enhance collection efficiency.

3.2.

Measurement

Absorption and emission spectra of the three luminescent components used in our experiments were determined in our previous study.⁷ They are reproduced in Fig. 3(a). The emission spectra are normalized such that their integrals represent their quantum yields.

Fig. 3

Single-spot excitation experiment. (a) Absorption coefficients (solid curves) and emission spectra (dotted curves) of three luminescent components.⁷ (b) Exploded view of the setup for measuring ${\eta }_{\mathrm{opt}}$ and definition of the coordinate system.

The setup for the measurement is shown in Fig. 3(b). In our previous experiment, we used a power meter to measure the optical power exiting the luminescent fiber, resulting in an approximate estimation of optical efficiency.⁹ In this study, spectral photon fluxes are recorded with a spectrometer to evaluate the optical efficiency accurately. A cosine corrector (Model CC-3, Ocean Insight) is attached to the optical fiber head of a spectrometer (Model FLMS13077, Ocean Optics). It is placed near the fiber tip without index-matching oil. Hence, an air gap exists between them. Although this fact decreases the optical efficiency due to Fresnel reflection, we prioritize the ease of handling in this experiment. A mask with an aperture is placed near the edges of the place to prevent the green PL photons from entering the protruded section of the fiber. The coordinate system is defined with $x$ axis on the fiber and $y$ axis along the edge of the two plates. Each sample is placed on a black cloth.

A single spot on the sample is excited by a 1.0 mm-diameter laser beam at 450 nm and the output of the spectrometer is recorded. The excitation spot is stepped on the luminescent plate with fixed intervals in both directions and this recording is repeated. The spectrum of the incident photons is measured by exposing the cosine corrector to the laser beam directly.

For example, the spectra recorded with the largest samples (side length $L=50\mathrm{mm}$) with and without edge mirrors are compared in Fig. 4. The coordinates of the excited spots $(x,y)$ and the integration time set for the spectrometer $t_{\mathrm{int}}$ are indicated in each plot.

Fig. 4

Spectra recorded with the sample of side length 50 mm. There are no mirrors on its edge surfaces for (a) and (b). Mirrors are attached to its edges for (c) and (d).

When the spots on the fiber are excited directly ($y=0$), the PL photons with shorter wavelengths become attenuated at larger $x$, leading to severe redshifts as shown in Fig. 4(a). This is caused by self-absorption in the fiber. As the excitation spot moves away from the origin, this redshift tends to saturate. In addition to self-absorption, PL photons are scattered by some irregularities in the fiber as shown in Fig. 2(b).

When the spots on the plate at $y=17$ mm are excited, as shown in Fig. 4(b), the spectral flux in units of counts/s·nm becomes one order of magnitude smaller. Because the green PL photons spread in the plate, the redshift in these spectra becomes less apparent.

As shown in Figs. 4(c) and 4(d), adding mirrors to the edge surfaces increases the spectral fluxes substantially. The redshift is less apparent because the PL photons reflected by the mirrors are added.

3.3.

Analysis

The number of the PL photons exiting the fiber per second is proportional to $\int S_{\mathrm{PL}}(\lambda )\mathrm{d}\lambda$ where $S_{\mathrm{PL}}$ is the recorded spectrum of the PL photons in units of counts/s·nm. The number of the incident photons is proportional to $\int S_{\mathrm{in}}(\lambda )\mathrm{d}\lambda$ where $S_{\mathrm{in}}$ is the recorded spectrum in units of counts/s·nm. Because the same setup is used, the proportionality constants are the same. Hence, the optical efficiency ${\eta }_{\mathrm{opt}}$ is given by the ratio of these integrals.

Two examples of this analysis are shown in Fig. 5. The optical efficiency in case of the fiber excitation is shown in Figs. 5(a) and 5(b) for the samples with side length $L=50$ and 20 mm, respectively. The results of the plate excitation are shown in Figs. 5(c) and 5(d) for each sample, respectively. The side length of the sample and the coordinates of the excitation spots are indicated in each plot. The solid and empty markers represent the optical efficiencies of the samples with and without edge mirrors, respectively.

Fig. 5

Position-dependent optical efficiency of two samples. The optical efficiency in case of the fiber excitation for the samples with side length $L=50\mathrm{mm}$ (a) and 20 mm (b). The results of the plate excitation for the sample with $L=50\mathrm{mm}$ (c) and 20 mm (d). Solid and empty markers represent the cases with and without edge mirrors, respectively.

In both cases of the fiber and plate excitation, the optical efficiency of the smaller sample is higher than that of the larger sample. The larger sample exhibits greater dependency on the $x$ coordinate of the excitation spot. These observations are consistent with the PL photon losses during the photon collection process. For example, the PL photons generated in the fiber propagate toward the exiting end and the other end. Those emitted in the wrong direction can be collected after reflection at the other end. However, they are more likely to be lost because they need to traverse longer distances in the fiber. Hence, the collection efficiency decreases as the fiber becomes longer. A similar argument applies to the case of the plate excitation. The PL photons reflected at the edge surfaces are less likely to reach the collecting edge due to the extra distances. Direct excitation of the fiber results in ${\eta }_{\mathrm{opt}}$ one order of magnitude larger than the plate excitation. Nevertheless, the two strategies (decreasing the side length and attaching mirrors to the edges) enhance ${\eta }_{\mathrm{opt}}$ substantially.

Note that the optical efficiencies measured here are valid only for excitation at 450 nm. This prompts us to calculate collection efficiency ${\eta }_{\mathrm{col}}$ by Eq. (2). The absorptance of the luminescent plate was measured to be 0.911 at 450 nm. The average absorptance of the fiber at 450 nm is calculated to be 0.212 for partial exposure of the central 1.0 mm-wide region and 0.181 for full exposure of the whole 1.5 mm-wide region. In this calculation, the transmittance of the luminescent fiber is calculated from Lambert-Beer’s law with the absorption coefficients in Fig. 3(a). To account for the reflectance at the fiber-air boundary, a circular fiber cross-section is assumed, and Fresnel equations are used. When a leaf LSC is uniformly irradiated, the collection efficiency is given by averaging the position-dependent collection efficiencies. The results for the cases of the fiber and plate excitation are shown in Figs. 6(a) and 6(b), respectively. Solid and empty markers represent the values with and without the edge mirrors, respectively. Curves are fittings. They will be used for predicting the response of these samples to uniform illumination by a white light source in Sec. 4.1.

Fig. 6

Average collection efficiency calculated for the case of uniform irradiation. (a) Fiber excitation and (b) plate excitation. The solid and empty markers represent the values for the samples with and without the edge mirrors, respectively. Curves are fittings to be used in Sec. 4.1.

Let us regard the $L=50\hspace{0.17em}\mathrm{mm}$ sample without edge mirrors as the reference. Its average collection efficiencies are 0.0052 ($\overline{{\eta }_{\mathrm{col}-\mathrm{P}}}$) and 0.21 ($\overline{{\eta }_{\mathrm{col}-\mathrm{F}}}$). Decreasing $L$ from 50 to 10 mm improves these values to 0.0090 and 0.27, respectively. This is 1.7 times increase in ${\eta }_{\mathrm{col}-\mathrm{P}}$ and 1.3 times increase in ${\eta }_{\mathrm{col}-\mathrm{F}}$. Attaching mirrors at the edges of the luminescent components increases ${\eta }_{\mathrm{col}-\mathrm{P}}$ and ${\eta }_{\mathrm{col}-\mathrm{F}}$ to 0.024 and 0.42, respectively. Therefore, the enhancement factor is 4.4 and 2.0 for the plate and the fiber excitation, respectively. Note that the geometric gain decreases from about 1400 to 57 by decreasing from 50 to 10 mm and that this decrease can be compensated by connecting multiple devices with a clear lightguide.

A theoretical analysis of the trapping efficiency of a PL photon in a luminescent fiber shows that its upper limit is about 0.61 for the fiber with air cladding.¹⁶ The maximum value of ${\eta }_{\mathrm{col}-\mathrm{F}}$ in Fig. 6(a) is 0.47. This discrepancy can be attributed partially to the Fresnel reflection at the air gap between the fiber tip and the cosine corrector and partially to the scattering by the internal and surface irregularities of the fiber used in the experiment.

Regarding the small values of ${\eta }_{\mathrm{col}-\mathrm{P}}$ in Fig. 6(b), the performance of the current samples is limited by the availability of the luminescent components as well as our decision to prioritize the ease of handling. For example, a luminescent fiber with a larger diameter would increase the probability of transferring the green PL photons from the edge of the plates to the fiber. Eliminating the air gap between the fiber tip and the cosine corrector would be effective. The 5 mm-long protruded fiber section should be eliminated for increasing ${\eta }_{\mathrm{col}-\mathrm{P}}$.

4.1.

Measurement

Because white LEDs do not emit in infrared, we can use a Si-based spectrometer to measure their emission spectra. The laser beam in Fig. 3(b) was replaced by a white LED light source. It was placed far away from each sample with edge mirrors to ensure uniform irradiation. A light shield with an aperture surrounded the protruding fiber section so that only the central 1.5 mm-diameter region of the cosine corrector was exposed to the PL photons. This light shield also prevented the protruded fiber section from being exposed to the incident light.

The PL spectra in Fig. 7(a) were acquired when each of the five samples without edge mirrors was irradiated by white LEDs. The dotted curve is the incident spectrum. It was acquired by placing the same module at the irradiated surface and exposing the central 1.5 mm-diameter region of the cosine corrector. The side length $L$ and the integration time set for the spectrometer $t_{\mathrm{int}}$ are indicated in each plot. The measurement was repeated with the five samples with edge mirrors. The results are shown in Fig. 7(b).

Fig. 7

Spectra recorded with the samples in Fig. 2 under uniform excitation by white LEDs. The spectra in (a) are from the samples without edge mirrors and those in (b) are from the samples with edge mirrors. The solid curves are the spectra of the PL photons from the fiber of each sample. The dotted curve is the spectrum of the excitation light incident on each sample directly.

Optical efficiency is calculated from the PL spectra recorded with these 10 samples. As for the number of incident photons, the recorded incident spectra (the dotted curves in Fig. 7) are multiplied by the areal ratio of each sample and the central 1.5 mm-diameter region of the cosine corrector. The solid and empty markers in Fig. 8 represent the values measured with the samples with and without edge mirrors, respectively. For example, attaching edge mirrors to the largest sample ($L=50\hspace{0.17em}\mathrm{mm}$) increases its ${\eta }_{\mathrm{opt}}$ from 0.0022 to 0.0041. Decreasing $L$ from 50 mm to 10 mm increases ${\eta }_{\mathrm{opt}}$ from 0.0041 to 0.013 for the sample with edge mirrors.

Fig. 8

Optical efficiencies of the 10 samples in Fig. 2 under uniform irradiation by white LEDs. The solid and empty markers are the values measured with the samples with and without edge mirrors, respectively. Curves are the values calculated by Eqs. (3) and (5).

The solid and dotted curves in Fig. 8 are the average optical efficiencies calculated from the results of the single-spot excitation measurement. Namely, the fitting equations in Fig. 6 were used for the terms ${\eta }_{\mathrm{col}-\mathrm{F}}$ and ${\eta }_{\mathrm{col}-\mathrm{P}}$ in Eq. (5). The term ${\eta }_{\mathrm{abs}-\mathrm{P}}$ was given by Eq. (3). As for the term ${\eta }_{\mathrm{abs}-\mathrm{F}}$, the distance $d$ in Eq. (3) was varied, and the absorptance was averaged. These values are as indicated in the plot. As shown in Fig. 8, the measured values are reproduced by these curves reasonably well.

The single-spot excitation technique gives more versatile results in the sense that the optical efficiency for the case of non-uniform irradiation by a light source with an arbitrary spectrum can be calculated. However, it takes time to step the excitation spot over the whole incident area of a sample. For this reason, we irradiate various samples uniformly in the following experiments. Nevertheless, the results are only valid for the specific light source used in the measurement.

4.2.

Tandem Leaf LSCs

Stacking one LSC on top of another is a well-known approach for enhancing photon collection.¹⁷ We apply this for leaf LSCs. To assemble a lower sample, we used 2.0 mm-thick luminescent acrylic plates emitting in orange (Model 9989, Kanase Inc., Japan). Its quantum yield was measured to be about 77% of the green plate (Model 1305L). The absorption and emission characteristics of this component are shown in Fig. 3(a). A photograph of the sample with side length $L=50\mathrm{mm}$ and a schematic drawing of its cross-section are shown in the inset of Fig. 9(b).

Fig. 9

Optical efficiency measurement with tandem leaf LSCs. (a) Output spectra of a square tandem LSC with side length $L=50\mathrm{mm}$ without edge mirrors under uniform irradiation by white LEDs. (b) The effect of the side length on the optical efficiency.

The optical efficiency of this square sample was measured under uniform irradiation by white LEDs. The spectra of the PL photons and the incident light are shown in Fig. 9(a). The curve marked as “tandem” is the sum of the spectra from the upper and lower layers.

Samples with different side lengths were assembled and this measurement was repeated. They all had a 10 mm-wide bottom mirror. The result is summarized in Fig. 9(b). For example, in the case of the sample with $L=50\mathrm{mm}$, its upper layer contributes 78% to the total output. The lower layer contributes 22% by converting the transmitted light and the PL photons emitted downward by the upper layer. Hence, the enhancement factor for adding the orange device is 1.28. The contribution from the lower layer decreases to 16% for the sample with $L=20\mathrm{mm}$. This is because the bottom mirror of the upper layer blocks the downward photon flux. This effect is more severe for devices with shorter side lengths.

4.3.

Underlying Surfaces

It is also well known that a reflective surface below an LSC increases its optical efficiency by reflecting the transmitting incident photons. We have avoided this double passing of the excitation light in our measurement so far by placing our samples on a black cloth.

Here, we study how effective an underlying surface is for a leaf LSC. As described in Sec. 3, the bottom mirror in our sample reflects the transmitted excitation light diffusively. Because this complicates the interpretation of the measurement, we have eliminated the bottom mirror in this experiment. The optical adhesive film used for bridging the two plates comes with a thin supporting transparent film. This film is left unremoved in each sample shown in Fig. 10(a). The PL photons generated by the room light enter this supporting film and escape from its edges. Hence, two bright lines are visible near the fiber in these photographs. They are 10 mm apart. No mirrors are attached to the outer edges.

Fig. 10

Square samples without edge mirrors and their optical efficiencies evaluated under uniform irradiation by white LEDs. (a) Photographs of the square samples without bottom mirrors placed on a black cloth. (b) The effects of the underlying surfaces and the side length.

First, each of the samples was placed on the black cloth and irradiated by the white LEDs. The results are shown by the empty squares in Fig. 10(b). For example, the optical efficiency of the sample with $L=10\mathrm{mm}$ is 0.0061.

Second, a mirror replaced the black cloth. The empty triangles in Fig 10(b) are the results. For example, the optical efficiency of the sample with $L=10\mathrm{mm}$ increases to 0.0084. When the reflected light enters the sample, its optical efficiency becomes 1.38 times larger.

Third, a diffuser film and a mirror were stacked, and each sample was placed on it with an air gap. The results are shown by the empty diamond markers in Fig. 10(b). For example, the optical efficiency of the sample $L=10\mathrm{mm}$ is 0.012. The diffuser/mirror configuration is more effective than the simple mirror because the reflected light enters the sample obliquely and it is more likely to be absorbed.

4.4.

Leaf Geometry

The geometry effect of a conventional planar LSC has been studied analytically^18,19 and by ray tracing.²⁰ We assembled leaf LSCs with various shapes by using the same 2.0 mm-thick luminescent plates and the 1.5 mm-diameter luminescent fiber. Photographs of the samples without edge mirrors are shown in Fig. 11(a). They all have the same incident area of $2500{\mathrm{mm}}^2$, 10-mm-wide bottom mirrors, and a 5-mm-long protruded fiber section. The square sample with a side length of 50 mm serves as a reference. Samples with edge mirrors were also prepared. Each of them was placed on a black cloth and the white LED light source irradiated it uniformly. The PL photons exiting the fiber tip from the side of the protruding section were recorded by the spectrometer. The optical efficiency for each case is compared in Fig. 11(b).

Fig. 11

Leaf LSCs with various shapes. (a) The photographs of the samples without edge mirrors. (b) Optical efficiency evaluated under uniform irradiation by white LEDs.

For all the shapes tested in this experiment, ${\eta }_{\mathrm{opt}}$ is almost doubled by attaching mirrors to the edge surfaces. The sample labeled as “circle segment 1” has the lowest value of 0.0022 among the samples without edge mirrors whereas “right triangle 1” has the highest value of 0.0024. Attaching mirrors to the plate edges increases these to 0.0041 and 0.0046, respectively. Because ${\eta }_{\mathrm{opt}}$ is weakly dependent on the shape of the luminescent plate, a wide degree of freedom exists in designing the shape of leaf LSCs.

4.5.

Maple Leaf LSC

We assembled a sample mimicking a Japanese maple leaf. Four pieces of luminescent plates were cut out of the same 2.0 mm-thick plate by laser cutting. The drawing in Fig. 12(a) shows the three lines for this separation. Although we attempted to polish the edge surfaces after laser cutting, this was not very successful. As shown in Fig. 12(b), the edge surface scatters room light. The same adhesive optical films were used to bridge them with gaps in between. These transparent supporting films were left unremoved. The luminescent fibers were inserted to these gaps, and they sat on the adhesive side of the supporting films. The total incident area of this sample is $954{\mathrm{mm}}^2$.

Fig. 12

Maple leaf LSC. (a) A drawing illustrating the three lines for laser cutting. (b) A photograph of the edge of the 2.0 mm-thick luminescent plate after laser cutting and our polishing attempt. (c) Optical efficiency evaluated under uniform irradiation by white LEDs and photographs of the sample without edge mirrors.

The photographs in Fig. 12(c) show the sample placed on three different surfaces. When it is either on a black cloth or a mirror, the edges of the supporting films are visible. The underlying surface changes the appearance of the sample. It looks yellowish when it is on a diffuser film stacked on a mirror. Room light is reflected diffusively and added to the green PL photons emitted by the sample.

Under the illumination by the white LEDs, the PL photons exiting the three fibers were recorded by the spectrometer. The output of the sample was calculated by adding and integrating these spectra. The number of incident photons was measured as before. Optical efficiency ${\eta }_{\mathrm{opt}}$ was evaluated for each underlying surface. After the edge mirrors were attached, the measurement was repeated.

The plot in Fig. 12(c) summarizes the results. The smallest ${\eta }_{\mathrm{opt}}$ value is 0.0049. As shown in Fig. 8(a), the square sample of $L=20\mathrm{mm}$ has ${\eta }_{\mathrm{opt}}$ of 0.0048 under the same conditions (no edge mirrors, black cloth, white LEDs). This is reasonable because the PL photons in the maple leaf configuration propagate shorter distances in the plates and longer distances in the fiber. Attaching edge mirrors increases ${\eta }_{\mathrm{opt}}$ to 0.0062. This enhancement by edge mirrors is noticeably smaller than the cases in Figs. 8(a) and 11(b). This is likely caused by our lack of polishing skill. As shown in Fig. 12(b), the edge surface does not look clear. Regarding the effect of the underlying surface condition, a similar trend is observed in Figs. 12(c) and 11(b). Thus, a diffusive mirror has a negative effect if used as an edge mirror and a positive effect if used as an underlying layer.