A Glow-in-the-Dark Butterfly

And just in time for Halloween! Papilio zalmoxis fluoresces nicely under long wave ultraviolet light. It couples a fluorescent pigment with the butterfly's photonic crystal structure to enhance light emission.

Mouse over the photo at right and you'll see how the butterfly looks under white light.

Referring in particular to its white-light appearance, P. zalmoxis is a perfect demonstration of how a blend of pigment and structure works to produce pastel colors that neither component alone can. The butterfly specimen sitting on my table looks like it was painted by an incredibly gifted artist!

(Please pay particularly close attention to the blue color in its wings seen under white light. That coloration will be the focus of this page.)


Commonly known as the Giant Blue Swallowtail, P. zalmoxis is a member in good standing of the Papilionidae family. There are around 560 species of butterflies in the family.

I first read about the iridescence and fluorescence of P. zalmoxis in a paper by Prum et al. The authors discuss their development of a 2D Fourier analysis program that allows them to determine scattering properties of materials based on index of refraction, structure spacing, etc.

Prior to their work, P. zalmoxis was believed to produce its pastel blue color by incoherent Tyndall scattering. (Refer to the paper by J. Huxley in References.) Tyndall scattering is the phenomenon that causes cigarette or engine smoke to look blue. This type of scattering is applicable to randomly arranged colloidal suspensions. (Often this type of scattering is referred to as Rayleigh scattering, but that designation should be limited to molecular-sized particles. However, the terms are now so entangled that it's hopeless. Poor Tyndall.)

The authors present the case that the butterfly's blue wing color is due in part to the presence of a fluorescent pigment that has been deposited on tubular nanostructures within the butterfly's scales. Incident light is horizontally scattered by vertically arranged tubes running through the scales. The scattered light irradiates adjacent alveolar structures within the scales with a resulting increase in brilliance. They go on to predict coherent scattering at wavelengths of 370nm and 740nm. The scattering is considered coherent rather than incoherent based on calculations of the interdependence of vertical tube locations one on another. If the tubes were randomly distributed, then the scattering of light would be random, and Tyndall scattering could then explain the butterfly's blue wing color.

Prum and his group also describe the yellow coloration visible when the butterfly is considerably tilted as being structural in nature. This is evidenced by the disappearance of the yellow color when isopropanol is dropped on to the wing. We have here a butterfly that exhibits coherent scattering, structural color, and UV fluorescence. (Please refer to the referenced paper for full details.)

This is probably a good time to mention that structural color in butterfly wings will vanish temporarily when organic liquids, such as acetone, hexane, or isopropanol, are applied to the wings. When that occurs, air located between layers of chitin that make up the scales and photonic crystal structures is temporarily replaced with a liquid whose refractive index is closer to that of chitin (1.56). When the applied liquid completely evaporates, structural color returns. It is conceivable that organic pigments may dissolve away too—though ammonia water is a better solvent for many pigments—but then the observed color loss is permanent. I will discuss chitin in more detail at a later time.

P. zalmoxis scales

The photo at top left shows three types of scales from P. zalmoxis. You can see blue, white, and black scales. Note the different margin shapes. That photo was taken using reflected darkfield illumination employing white light. This type of lighting often produces more accurate color renditions of specimens in comparison to reflected brightfield.

The lower photo at left was taken of the butterfly's fluorescent scales using the FluorEVER microscope, described on page 3 of my LED Fluorescence section. The FluorEVER was equipped with a 5-watt, royal-blue LED with a peak wavelength of 455nm. I used a 530nm longpass filter for the emitted light.


Below is a composite of two SEM photos of a scale that exhibits fluorescence. Compare these images to those taken of a different species, P. nireus, seen further down the page.

Some views from the inside

With proper mishandling of butterfly wing mounts, it is not too difficult to fracture some of the delicate scales. When scales are broken open, a treasure trove of components comes into view.

Below are three more photos of scales of P. zalmoxis.

(A) A broken scale showing the inside of the top or dorsal surface of a scale. The oblong projection at the base of the topmost scale in the photo is called a pedicel. It fits within inscribed pockets or pits in the butterfly's wing in order to hold on to the scale. When the butterfly dies, the wing's grasp on the pedicel weakens, allowing the scales to easily brush off.

(B) Greater detail is visible, clearly showing projections (pillar or columnar trabeculae) and the bottom side of vertically arranged tubes. Light is channeled along those air-filled cylinders inside the scale. The tubes are heavily infused with a fluorescent pigment, and horizontal scattering with optical enhancement occurs.

To my eye, the cylinders or vertical tubes seem to exhibit a triangular symmetry in their distribution. They do not seem to be randomly distributed and may act as coherent scatterers.

The trabeculae scattered among the tubes are also present in the scales of P. nireus and will be further discussed below.

The underside of the scale (as viewed in the picture) shows hints of ridges that are more clearly seen in the next photo. Those ridges actually are on the topmost (dorsal) surface of the scale as the butterfly flies.

(C) The lower magnification dorsal view of the type of scale is shown in A and B. Compare this picture with the composite of two photos above.


Due to the physical similarities between P. zalmoxis and P. nireus (to be described next) much of what is said of one species probably applies to the other, though that has yet to be confirmed.

Papilio nireus—another fluorescent butterfly

The butterfly hindwing shown below belongs to Papilio nireus. This beauty is commonly called a Narrow Green-Banded Swallowtail. P. nireus is also a member of the Papilionidae family. (It occurred to me that if the Papilionidae family is anything like my family, P. zalmoxis and P. nireus probably don't get along too well.)

As with P. zalmoxis (top of page), the P. nireus wing (dorsal view) was photographed under long wave UV light with a yellow-orange filter placed over the camera lens.


Three different microscopy illumination techniques were used to image the dorsal side of scales from the butterfly's forewing.

At left below, two fluorescent scales and one black scale are shown using a combined reflection/transmission Normarski DIC technique. (I'll describe this technique in a future update to my Special Methods section.) In the center are several fluorescing scales viewed with my FluorEVER LED Fluorescent microscope. At right are different types of scales viewed with a reflecting darkfield microscope. Though the appearance of that image is reminiscent of one taken using polarizing filters, the picture was taken with unpolarized light. Color diffraction effects are evident.

Though not detectable here, looking at the scales adhering to the glass microscope slide reveals a pale blue coloration depending upon how I tilt the slide.

When I mount the scales flat on a piece of double-stick tape, it is pure luck as to whether some scales will be damaged in the "right" way. Other mounts I make involve using two pieces of single-sided, transparent Scotch tape with segments of the wings sandwiched between the sticky surfaces. These mounts are cut with dissecting scissors to fit my fabricated holder on a SEM stub. Scissors, no matter how sharp, tend to break and macerate specimens. This often is exactly what I want to do.

Looking under the hood

In the top photo at right, we see the inside of the upper surface of the scale (i.e., the surface that faces upwards as the butterfly zips along, but seen here on the bottom) covered with columnar or pillar trabeculae and holes; however, the holes are not easily visible here. The projections may act solely as structural supports or possibly as light scatterers too.

The lower surface of the scale (seen here at the top) is broken open and bent back. It reveals a smooth inner surface. Wrapped over that are several segments of the true upper surface of the scale with ridges and holes. Whew!

The lower photo at right shows a scale again with its top facing down. Here the ridges are clearly visible. Also, longitudinal views or sections of the tubes that penetrate the scale can be seen. The thin sheet (basement lamella), seen here supported by trabeculae at a small distance from the other surface, is believed to act as a distributed Bragg reflector (DBR) to bounce penetrating and enhanced light back out, further enhancing the emission.

We'll discuss in greater detail the optics of light reflection at the very end of this entire section. However, I will point out some of the possible causes of structural colors along the way.



Here is a closeup of a P. nireus scale showing a frontal view of the ridges. Now the holes that represent the tops of vertically positioned tubes that penetrate the scale are clearly visible. The holes are really the tops of vertically arranged tubes that form the alveolar structure of the fluorescent scale.

Since these scales are physically similar to those belonging to P. zalmoxis, the positions of the tubes are probably not independent or random either. If that is the case, then coherent scattering is occurring in this species as well. If so, that would add to the photonic crystal behavior of these scales described in the References.

I found a P. nireus scale broken into several pieces that show the vertical tubes. The broad edges of the pieces demonstrate longitudinal sections of the hollow tubes. The ridges are somewhat flattened due to handling. The triangular symmetry referred to previously seems more obvious in this photo.

Analysis by Prum et al of P. zalmoxis scales demonstrated that the air-filled cylinders or tubes are not randomly distributed. Considering that the scales of both species are practically identical, coherent scattering and multilayer thin-film reflection may also apply to both.

So where does the blue color come from?

Prum and his group have presented the case that the Tyndall effect is not applicable in explaining the blue color in the wings of P. zalmoxis. They posit that, because of the non-random distribution of the vertically arranged air-filled cylinders in its wings, the yellow cast seen when the wings are placed at a low angle relative to incident light is due to coherent scattering. However, they expect that the blue is most likely a fluorescent pigment. Both Prum et al (studying P. zalmoxis) and Vukusic et al (working with P. nireus) see the tubes as increasing the emission of a fluorescent pigment by their ingenious structure.

Since blue is not usually considered a pigment-produced color in nature, I ran six additional experiments that I list below.

(1) I placed a wing from each butterfly between two sheets of circular filter paper to which a few drops of 10% ammonium hydroxide had been added. The idea was to produce a paper chromatographic separation of any butterfly wing pigments present. The wings were left in separate Petri dishes overnight to allow the pigments to dissolve and transport themselves through the paper disks.

The next day, I removed the disks and allowed them to dry. I checked the disks under white light and with UV light. For each species of butterfly, in addition to the expected yellow and brown colors, a strongly fluorescent blue-white spot was observed under long wave UV. There was no sign of a white-light visible blue pigment. (Of course, there's a host of reasons why any blue pigment might not show up with a quick extraction like this. The pigment may have been altered or destroyed by the ammonium hydroxide. The eluted quantity of pigment may be too small to be seen. The blue pigment migrated to a spot that is masked by the presence of another, highly colored pigment. Or there simply is no blue pigment present.) The wings, now devoid of any color, were discarded.

(2) Once again, I selected a wing from each butterfly. Each wing was illuminated with white light passed through a 441nm longpass filter that blocks all light below that wavelength. Then a 455nm longpass filter was tried. In both cases, the wings still appeared blue or, at least, blue-green. Many scientists believe that the 420nm wavelength present in sunlight is capable of exciting fluorescence in some pigments present in butterfly wings. However, both of the filters I used blocked that wavelength, as well as the UV band. This seems to rule out UV fluorescence as a significant cause of the blue color. Also, the results indicate that scattering is a possible cause of the blue color.

Below is a series of photos taken of the two Papilio species, along with light-source spectra of the metal-halide lamp illuminator. These pictures were photographed with and without the 441nm longpass filter. Since some of the deep blue spectrum and all of the UV spectrum were filtered out, the wing colors shifted to the blue-green. I could barely see a difference viewing the original photos, so the color shift may not be noticeable over the Internet. It was even harder to detect using a tungsten lamp illuminator.

(3) A 5-watt royal-blue LED was equipped with a 467nm narrow-band filter. A wing from P. zalmoxis was illuminated with the filtered light. The light reflected from the wing peaked at 467nm with no visible fluorescence. The bandwidth of the filter + LED allowed light from 450 to 475nm to irradiate the wing. The lack of fluorescence seems to indicate that none was excited by this band of wavelengths. This seems to boost the likelihood that scattered light may be the root cause of the blue color.

(4) A hindwing from P. zalmoxis was illuminated at a glancing angle with a beam of intense white light using a fiber-optic tungsten illuminator. A polarizing filter was inserted in the light path. The wing was placed horizontally on a rotatable platform. Pictures were taken with the polarizer axis vertical and then with the axis horizontal. In some areas of the wing, the blue color vanished with a change in the axis position. This is similar to the effects expected from Tyndall scattering rather than from a pigment. (Photos of this experiment will be seen below.)

(5) Close examination of the P. zalmoxis wing by eye revealed yellow, orange, and red coloration as the angle of the wing was positioned nearer and nearer to a glancing angle of illumination. This reminded me of the chemical sunset experiment used to demonstrate Tyndall scattering. More about this coming up.

(6) Drops of hexane or isopropanol placed on the wing temporarily eliminate the yellow coloring, thus indicating a structural cause. The drops eliminate the blue colors as well. This calls into question the presence of a blue pigment and hints at a structural source.

As promised, we'll now take a look at Tyndall scattering and how that may be involved in the production of the butterflies' blue wing colors.

The Tyndall effect: incoherent scattering by colloids

John Tyndall was one of the most eminent nineteenth-century physicists. His work encompassed investigations of sound, glaciers, electricity, magnetism, heat, and light that included the scattering of light by colloidal particles. Lord Rayleigh followed up later with theoretical studies of scattering by, initially, molecular-sized gas particles that accounts for the production of the blue color in blue skies.

The best known "modern" demonstration of the Tyndall effect may be seen in the chemical sunset experiment so beloved by science teachers. It's easy to do and involves the production of colloidal sulfur particles generated by adding a dilute acid (sulfuric, hydrochloric, or even white distilled vinegar) to a solution of sodium thiosulfate in water. The experiment produces an image on a screen of a glowing sphere that changes in color from white through yellow and finally orange-red as the sulfur particles grow in size. It has been found that the smaller the particle, the shorter the wavelength likely to be scattered.

When I do this experiment, I like to kick off my shoes, sit back, sip an ice tea, and watch the sun setting in my lab. No question about it—I have got to get a life!

These are the particulars for my version of the experiment:

(1) I used a rectangular, molded-glass staining dish as my glass tank. It holds 300 milliliters of liquid. It's seamless glass, so there is no worry about leaking aquarium seals.

(2) A solution of 15 grams of sodium thiosulfate in one gallon of tap water was prepared. I filtered the solution just before use to eliminate particulate contamination. (If you plan to use regular tap water instead of distilled water, make sure it's not acidic.) I added 250 ml or so to the glass tank.

(3) I prepared a 0.6M solution of hydrochloric acid. One ml was added to the solution in the glass tank and stirred just before the effect was required.

(4) My light source was situated to the right of the tank (see pictures at right). An ordinary polarizing filter was placed in front of the light source.

At right (top) is a view of the tank after a few minutes. I allowed the sulfur particles to reach a size that scatters white light instead of the mixed blue that first appears with smaller particles. For this photo, the polarizing element was positioned with its axis vertical to avoid blocking light polarized by scattering (strongest at 90 degrees from the light path).

At right (middle), I rotated the polarizer axis to its horizontal position. This blocks most of the light polarized in scattering, but it also allows the beautiful Tyndall residual blue color to show up. The blue light is linearly polarized and can be completely blocked by a second polarizer (with its axis vertical) that is held near the eye.

At right (bottom) I positioned the camera at an angle closer to the light source. You'll notice the additional colors now visible in the light beam and the image of the setting sun (skewed because of camera perspective and the poor optical quality of the molded glass tank).


To instantly halt the growth of sulfur particles at any point, just add a few milliliters of dilute ammonium hydroxide to the glass tank and stir.

For this image, I switched to a Glan-Thompson prism in order to eliminate the polarizer as a potential source of color error. Other than the change in the polarizing element, the setup was identical to that used above. The Glan-Thompson was set for horizontal polarization.

Once again, I angled the camera from a vantage point closer to the light source and sighted down the light beam. Here we can see a yellow coloration all along the light path until the light strikes the glass wall on the far side where a blue spot is visible.

In the lab setup, if you had moved to the left and looked squarely at the front of the tank, you would have seen the Tyndall residual blue shown above.

Let's take a look at P. zalmoxis in a different light

With the Tyndall scattering experiment in mind, let us reexamine a wing from P. zalmoxis under polarized light. Scattering light from colloidal particles can produce a substantial degree of polarization as we just saw. Does something like this happen with a butterfly wing?

The setup for the two pictures shown above was as follows:

The hindwing of P. zalmoxis was placed on a tiltable platform sitting on a copy stand. The camera was pointing straight down at the wing. Off to one side, I placed an intense, fiber-optic tungsten light source (of the type used with stereo microscopes) with the light grazing the wing at an angle of 10 degrees above horizontal. A polarizing filter was placed over the light source with the filter's polarizing axis arranged either vertically or horizontally, as depicted with arrows in the photos. With the light arranged as described, any light scattered by extremely minute particles or structures in the wing would be horizontally polarized to some degree. Therefore, with the filter's axis set horizontally, scattered and reflected light—polarized or not— would not be blocked. But with the filter's axis set vertically, horizontally polarized light would be blocked. (Strictly speaking, what I just said applies to a setup where the polarizing filter is mounted on the camera lens rather than over the light source. However, the end results are exactly the same.)

Notice that in the photos, the picture on the right expectedly shows the wing with virtually all the scattered (and reflected) light visible to the camera when the polarizing filter's axis is horizontal. However, the photo on the left was taken with the filter's axis set vertically. Note that much of the blue (or blue-green) color has vanished. This seems to indicate that the blue light was horizontally polarized and probably produced by incoherent, or Tyndall, scattering rather than by a fluorescing pigment.

I should mention that producing the above pair of photos that shows such a color disparity associated with different polarization directions took a fair number of attempts. The lighting direction and wing tilt have to be just right for such an obvious color distinction to be recorded.

At right is another picture of the hindwing, this one set at an angle of nearly 90 degrees relative to the camera. The light source was aimed almost perpendicularly to the wing. Note that the blue color is completely gone while the yellow color becomes quite visible. This seems odd if the presence of blue pigment is involved. Compare this photo with the fourth picture in the section on Tyndall scattering above.

In structured wings, such as Morpho rhetenor's wing discussed previously, even a steep angle as this still demonstrates an intense blue-violet color. It is true that the yellow color seen on P. zalmoxis' wing—and that is believed to be structural in origin—does vanish temporarily when a suitable organic liquid is dropped on it. However, the blue color also vanishes rather substantially when the liquid is applied. This also seems strange for a pigment-produced coloration, although some darkening is to be expected as when one moistens a piece of colored paper. It appears that structures are involved in producing the blue color.

A look at the coherent structures in Papilio's scales

If Tyndall scattering is really present, how does one explain the fact that there are clearly coherent structures producing coherent scattering? What could be the source of incoherent scattering in Papilio scales?

Below are two images taken of P. zalmoxis and P. nireus scales showing the tube structures as viewed from above. To produce the photos, I cropped segments of wing scale pictures. These cropped images were subjected to FFT processing. Using suitable masks, the FFT images were manipulated to emphasize periodic patterns and then they had an inverted FFT performed. The FFT-enhanced images were finally colorized in Photoshop. (The FFT techniques I used are simplifications of those used by Prum and his group. Their software is certainly more powerful and specific.)

For each photo, I overlaid a number of triangular and hexagonal outlines to demonstrate periodic structures in the scales. (Mouse over each photo to see the outlines.) In the center of each hexagon, there is a central spot or column. This means that each hexagon is really just made up of six triangles. Though I could have drawn more hexagons and triangles than shown, I am unable to group each and every column (shown in yellow) with one of the two regular geometric outlines. This may be important in discussions of coherent versus incoherent scattering as the ultimate cause of the blue wing color.

Also, I measured some of the adjacent inter-tube distances as 350 to 400 nanometers. The tube diameters measured some 275 nanometers. Obviously, tubes that could not be associated within geometric shapes may be located at distances > 400 nanometers from other such tubes or from coherent structures.

P. zalmoxis

P. nireus

We will revisit the coherence/incoherence scattering question one more time in the summary of this page. But first, one more look at UV fluorescence in a different butterfly family.

Morpho rhetenor revisited

As I mentioned earlier, Morpho rhetenor helena fluoresces in the UV as well as in the NIR.

Note: The UV fluorescence colors are exaggerated for display. To the naked eye, the actual fluorescent colors are more subdued.




Compared to the scales from P. nireus and P. zalmoxis, Morpho rhetenor scales that are located within the white spots on its wings fluoresce rather weakly. The scales that yield the strong blue color appear black when viewed with the FluorEVER microscope.

However, the pedicels do fluoresce intensely, as seen here. I didn't notice this behavior with the scales from Papilionidae. I don't know whether this increased brightness is due to an actual difference in distribution of the fluorescent pigment or whether the entire scale is acting like a light pipe, concentrating the emitted fluorescent light in the pedicels.

In summary

We have looked at three butterflies so far: Morpho rhetenor helena (previously), Papilio zalmoxis, and Papilio nireus. The three are from two different families, Nymphalidae and Papilionidae. It's not too surprising that there are differences in the color-producing structures between the two families, while there are extremely strong similarities between the two Papilio species.

Both families contain members that possess fluorescence. In the case of P. nireus and P. zalmoxis, the nature of the pigment-coated tube structures within their scales seems tailor-made to produce enhanced fluorescence. Morpho rhetenor, on the other hand, contains a weakly fluorescent pigment that may or may not be influenced by the structures within its scales. Morpho's pigment may just be painted on, so to speak, with no intention of producing an enhanced emission.

Both Morpho and Papilio demonstrate blue coloration in their wings. Generally, blue is not often associated with pigments in the world of insects (and many other living organisms). It may be the case that structures are responsible for the coloration in both groups that we've looked at. Morpho adopted a diffraction grating and multilayer thin-film interference approach to producing its blues. Papilio uses vertical tubes as coherent scatterers and fluorescence enhancers for part of its coloration, whether visible or not. P. nireus—and, likely, P. zalmoxis—also use a thin DBR along the lines of the Morpho's multilayer thin-film interference approach. The Papilio butterflies also demonstrate a yellowish to reddish coloration when strongly tilted. This is due to the vertical tubes and DBR as well.

Vukusic and his group described transmission electron micrographs of P. nireus scales. Their TEM work determined that the DBRs are three-layered. Mathematical models were used to calculate the expected wavelengths with that type of reflector.

The results of my experiments—along with some speculations about those results— are as follows:

(1) Based on a simple chromatographic extraction, there does not seem to be any evidence of a blue pigment (i.e., visible under white light) in the Papilio wings. However, there is a small amount of a highly fluorescent blue-white compound, probably a pigment, that can be seen under long wave UV. As the photos at the top and elsewhere on this page demonstrate, the butterflies are clearly fluorescent for whatever survival value that offers. That may be reason enough for the presence of the fluorescent pigment.

(2) Using cutoff filters demonstrates that neither UV nor deep-blue light is required to produce a blue wing coloration. This seems to eliminate the fluorescent pigment as the cause of the blue color. On the other hand, the presence of the fluorescent pigment may enhance the blue intensity along the lines of optical brighteners in "whiter than white" laundry. So, if there is UV or deep-blue light present, the wings may have a "bluer than blue" appearance. The optical-brightening pigment intensified by the presence of highly reflective DBRs at the base of the scales may serve as a luminous background tapestry for incoherent and coherent scattering. This may be the source of the observed sheen in these butterflies' wings.

(3) Referring to the P. zalmoxis species, the diminution or outright disappearance of the blue color with a 90-degree change in polarization direction seems to indicate that incoherent scattering is very much involved in the production of wing color.

I've discussed what happens with incoherent scattering and the polarization of light. This brings up the question as to whether coherent scattering also can produce polarization identical to that encountered in my experiments. However, Prum et al calculated that the peak coherent-scattering wavelengths from the wings of P. zalmoxis should be located at 370nm and 740nm. This means that I need not concern myself with the possibility of coherent scattering causing the observed polarization since the broad peak of the polarized light that I detected was located between 480-520nm in the blue. Therefore, the blue scatter is most likely incoherent in origin.
(4) The disappearance of the blue color and the formation of yellow and orange when the wing is highly tilted also lend credence to Tyndall scattering as a possible cause of the blue wing color. Take a look at the series of photos above demonstrating the Tyndall effect for support.

Due to these results, I believe that Tyndall scattering cannot be ruled out as a major contributing factor to the blue coloration in the Papilio species discussed above. If that's true, it would mean that Prum et al and Vukusic et al, as well as Huxley, are jointly correct in their investigations. That is, both coherent and incoherent scattering are taking place in the wings of these two butterfly species to produce the colors we see—including the blue.

The tubes in the Papilio wing scales that are not associated with coherent structures may be distributed incoherently. And there could be enough of these incoherent scatterers situated among the coherent scatterers to produce the Tyndall effect with the resulting blue wing coloration. The fluorescent pigment may augment the Tyndall scattering by enriching the blue. As has already been reported by Prum et al, the coherent scatterers are involved with producing the yellow shades visible when tilting the wings. And if Tyndall scattering is present, it may contribute to the shift in color as well. I hope that this area of research will be revisited by others.

Next time we'll look at an iridescent moth that probably thinks it's a butterfly.


Views From Science ©2009