A Thesis Presented to

Faculty of Alfred University

PHOTOCHROMISM

IN RARE-EARTH OXIDE GLASSES

by

Charles H. Bellows

In Partial Fulfillment of

the Requirements for

The Alfred University Honors Program

May 2016

Under the Supervision of:

Chair: Alexis G. Clare, Ph.D.

Committee Members:

Danielle D. Gagne, Ph.D.

Matthew M. Hall, Ph.D.

 SUMMARY

The following thesis was performed, in part, to provide glass artists with a succinct listing
of colors that may be achieved by lighting rare-earth oxide glasses in a variety of sources.
While examined through scientific experimentation, the hope is that the information
enclosed will allow artists new opportunities for creative experimentation.

Introduction

Oxides of transition and rare-earth metals can produce a multitude of colors in glass
through a process called doping. When doping, the powdered oxides are mixed with
premade pieces of glass called frit, or with glass-forming raw materials. When melted
together, ions from the oxides insert themselves into the glass, imparting a variety of
properties including color. The color is produced when the electrons within the ions move
between energy levels, releasing energy. The amount of energy released equates to a
specific wavelength, which in turn determines the color emitted. Because the
arrangement of electron energy levels is different for rare-earth ions compared to
transition metal ions, some interesting color effects can arise. Some glasses doped with
rare-earth oxides fluoresce under a UV “black light”, while others can express
photochromic properties. Photochromism, simply put, is the apparent color change of an
object as a function of light; similar to transition sunglasses.

This thesis explores such photochromism in rare-earth oxide glass samples. I

hypothesized that the reason for this is small ranges of color the glass absorbs, combined
with gaps in the ranges of color certain light sources emit.

[ii]
Procedure

The samples were made by mixing various rare-earth oxides with the glass frit used in the
art department, allowing for the most compatible data for the artists, and then melting the
mixture. Eight samples, each with different dopants, were produced.

The samples were then observed under isolated lighting conditions to see what color they
were. The light sources included: sunlight, incandescent, fluorescent, and LED. A
photograph of each sample, under each light source was taken.

As anyone skilled in photography knows, the colors you observe don’t always match the
colors in the photo. Cameras register color and light differently than the human eye,
adding discrepancies in color between the two devices. Additionally, a picture viewed on
a computer screen can look completely different than the same picture after it has been
printed out. To correct for these discrepancies, a color standard was used to normalize the
observed pigments. The color standard assigns numbers and coordinates to colors to
ensure exact replication. Determining the standard color of each sample under each light
source was done by a RAL color standard wheel. RAL color scaling is a German color
standard that assigns a 4-digit number to each color. The number can then be converted to
other color standard systems, allowing one to choose the best system that fits their needs.
The RAL color standard wheel itself resembles a paint swatch book; but the colors are
guaranteed standards. It should be mentioned that, while standardized, the color
assignments are still only approximations of the true color of the samples. At first glance,
the RAL color standard containing colors seems like a lot; but it does not account for
every possible color. For example, the praseodymium oxide sample appeared a yellow-
green color. The RAL color scaling did not even have an adequate approximation, and
therefore another color standard, Hex, had to be used for that particular sample.

[iii]
Through color leveling in Photoshop, colors were balanced similar to setting a white
balance on a camera. This allowed the photographs to appear closer to what was observed
with the naked eye. Correction was based on using the terbium oxide sample, which was
clear under all lighting conditions, as the standard. Additionally, a white sheet of paper
was used as a standard, neutral background. All photographs, raw and color corrected,
can be seen in the Appendix. While the corrected images are more accurate to the
observed colors, the raw images were included to showcase the colors and variations in
“white light” emitted by the different light sources.

Using a special type of film over the camera lens, spectra of the different light sources
were photographed. The film, known as a diffraction grating, acts similarly to a crystal
prism; separating the white light into the component rainbow colors. Thus, one could see
what colors combined to make the white light of each source.

Results

Most of the samples did not exhibit photochromism, staying relatively the same color
under every lighting condition. However, the samples containing samarium ions (Sm3+)
and dysprosium ions (Dy3+) fluoresced under black light. The Sm3+ sample glowed a
bright red orange and the Dy3+ sample glowed a pale ivory color. These, however, were
not investigated further, because the focus was placed upon reasons why certain glasses
changed color with variations in visible light.

The only two samples to produce any photochromic qualities were the two samples
containing neodymium ions (Nd3+) and holmium ions (Ho3+). In every light source except
fluorescent, the Nd3+ sample appeared violet; while appearing blue under fluorescent
lighting. The Ho3+ sample exhibited a similar trait; salmon pink under fluorescent, yellow
under all others.

[iv]
Because one of the samples was doped with both Nd3+ and Ho3+, it was expected that that
sample would also show photochromic qualities; however, the sample, for the most part,
remained a neutral blue-grey without any drastic color changes.

Spectroscopy was then conducted to determine what colors Nd3+ and Ho3+ absorbed, and
what colors were transmitted. It turned out that the two ranges of wavelengths Nd3+
absorbs correspond to a blue and reddish-purple color. Similarly, Ho3+ possesses
wavelength ranges that have complementary colors of yellows, reds, and oranges.

When looking at the photographs of the light spectra, most showed the standard,
continuous rainbow gradient from red to violet. Interestingly, fluorescent lighting did not.
Instead, its spectra showed black bars separating the colors. This indicated that there were
gaps of wavelength ranges where the fluorescent lighting did not emit that color of light.

Conclusions

As expected, the different arrangements of electron energy levels cause the rare-earth
ions to absorb only very small ranges of wavelengths. When photochromism is exhibited,
it is due to the fact that these narrow ranges coincide with small gaps in the emission
spectra of various lights. If the light does not emit that wavelength range, it cannot be
absorbed by the glass sample. If it is not absorbed, its complementary color is not
observed. When under fluorescent lights, the light does not emit the wavelengths
necessary for the reddish-purple of the Nd3+. Therefore only the blue can be seen.
Likewise, the light does not emit the wavelengths corresponding to Ho3+’s yellow,
leaving some red and orange that appears as salmon pink. Under other light sources, the
reds and oranges are usually canceled out, leaving only the yellow.

[v]
Further work

As the sample containing both Nd3+ and Ho3+ contained the same amount of 2g of dopant
as the other samples, it only contained 1g of each of the individual oxides. Due to the
lower dopant concentration, there may have not been enough rare-earth oxide to produce
a noticeable photochromic effect. The two ion types may have also negated each other’s
coloration, resulting in the greyer tones. It would be an interesting study to further
explore the potential photochromism in samples containing both Nd3+ and Ho3+.

[vi]
 ACKNOWLEDGEMENTS

I would like to extend my thanks to:

Dr. Alexis Clare, my thesis advisor, for her guidance and expertise in the subject matter,

Drs. Danielle Gagne and Matthew Hall, my Honors committee members, for their
support and suggestions,

Professor Angus Powers and the Alfred University Hot Glass Studio for the System 96®
glass frit used in the fabrication of the samples,

The Kazuo Inamori School of Engineering, NYS College of Ceramics at Alfred

University, for access to laboratory equipment and the rare-earth oxides,

As well as everyone else who helped in this process.

[vii]
 TABLE OF CONTENTS

Page

I. Introduction 1
II. Experimental Procedure 3
A. Fabrication 3
B. Photography 3
C. Color Analysis 5
D. Spectroscopy 6
III. Results 6
3+ 3+ 3+
A. Tb , Sm & Dy 7
B. Pr3+ & Er3+ 8
C. Nd3+, Ho3+ & Nd3+-Ho3+ 9
D. Emission Spectra 11
IV. Discussion 12
3+
A. Nd 14
3+
B. Ho 16
V. Conclusion 17
VI. Future Work 18
VII. Appendix 19
VIII. References 37

[viii]
 LIST OF TABLES

Page

Table I: Spectral colors and their complements 2

Table II: Coloration of each sample under all lighting conditions 7

Table III: All colors observed with CIELab and RGB conversions 19

Table IV: Specifications of all light sources utilized 20

[ix]
 LIST OF FIGURES

Page

Figure 1: Visual representation of CIELab coordinates 2

Figure 2: Setup for photographing samples under various lighting 4

conditions

Figure 3: Tb3+ sample under soft white incandescent a) raw and b) 8

corrected

Figure 4: Nd3+ sample under a) soft white CFL and b) soft white 9
incandescent

Figure 5: Absorption/Transmittance spectrum for the Nd3+ sample 10

Figure 6: Ho3+ sample under a) soft white CFL and b) soft white 10
incandescent

Figure 7: Absorption/Transmittance spectrum for the Ho3+ sample 11

Figure 8: Emission spectra of a) fluorescent b) daylight CFL c) soft white 12

CFL d) daylight incandescent e) soft white incandescent f)
daylight LED g) soft white LED

Figure 9: A graphical representation of emission spectra from a variety of 13

lightings

Figure 10: Comparison of the absorption/transmittance spectrum of Nd3+ 15

and the emission spectrum of a CFL

Figure 11: Comparison of the absorption/transmittance spectrum of Ho3+ 17

and the emission spectrum of a CFL

Figure 12: Corrected Pr3+ sample under a) sunlight b) daylight 21

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

[x]
Figure 13: Corrected Nd3+ sample under a) sunlight b) daylight 22
incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

Figure 14: Corrected Sm3+ sample under a) sunlight b) daylight 23

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

Figure 15: Corrected Tb3+ sample under a) sunlight b) daylight 24

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

Figure 16: Corrected Dy3+ sample under a) sunlight b) daylight 25

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

Figure 17: Corrected Ho3+ sample under a) sunlight b) daylight 26

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

Figure 18: Corrected Er3+ sample under a) sunlight b) daylight 27

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

Figure 19: Corrected Pr3+-Ho3+ sample under a) sunlight b) daylight 28

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

Figure 20: Raw Pr3+ sample under a) sunlight b) daylight incandescent c) 29

soft white incandescent d) fluorescent e) daylight CFL f) soft
white CFL g) UV LED h) daylight LED i) soft white LED

Figure 21: Raw Nd3+ sample under a) sunlight b) daylight incandescent c) 30

soft white incandescent d) fluorescent e) daylight CFL f) soft
white CFL g) UV LED h) daylight LED i) soft white LED

[xi]
Figure 22: Raw Sm3+ sample under a) sunlight b) daylight incandescent c) 31
soft white incandescent d) fluorescent e) daylight CFL f) soft
white CFL g) UV LED h) daylight LED i) soft white LED

Figure 23: Raw Tb3+ sample under a) sunlight b) daylight incandescent c) 32

soft white incandescent d) fluorescent e) daylight CFL f) soft
white CFL g) UV LED h) daylight LED i) soft white LED

Figure 24: Raw Dy3+ sample under a) sunlight b) daylight incandescent c) 33

soft white incandescent d) fluorescent e) daylight CFL f) soft
white CFL g) UV LED h) daylight LED i) soft white LED

Figure 25: Raw Ho3+ sample under a) sunlight b) daylight incandescent c) 34

soft white incandescent d) fluorescent e) daylight CFL f) soft
white CFL g) UV LED h) daylight LED i) soft white LED

Figure 26: Raw Er3+ sample under a) sunlight b) daylight incandescent c) 35

soft white incandescent d) fluorescent e) daylight CFL f) soft
white CFL g) UV LED h) daylight LED i) soft white LED

Figure 27: Raw Nd3+-Ho3+ sample under a) sunlight b) daylight 36

incandescent c) soft white incandescent d) fluorescent e)
daylight CFL f) soft white CFL g) UV LED h) daylight LED i)
soft white LED

[xii]
 ABSTRACT

Oxides of transition and rare-earth metals are commonly used as colorants, adding
pigment to glasses. However, some rare-earth oxides can cause the glass to exhibit
photochromism, changing color under different lighting conditions. This thesis explores
the extent of this photochromism. To accomplish this, glasses were produced with a
variety of select rare-earth oxides added. These samples were then observed and
photographed under different lighting. Light sources utilized include: fluorescent, LED,
incandescent, and sunlight. Further spectral analysis was conducted on samples of note.
While slight variations in color were observed across several of the samples, only
holmium and neodymium oxide doped samples exhibited strong photochromism.
Additionally, the samples containing samarium and dysprosium oxide fluoresced when
exposed to ultraviolet “black light”. Through research into technical literature, it was
found that some light sources, such as fluorescent, possess gaps across certain
wavelengths of the visible spectrum. This, in conjunction with very narrow discrete
absorption spectra of some rare-earth elements, is the cause the photochromic qualities of
the glasses.

[xiii]
 I. Introduction

Glass makers have both knowingly and unknowingly used oxides of transition metals and
rare-earth metals give color to glass for millennia; from intentionally adding cobalt oxide
to produce beautiful blues in vases and bottles to iron oxide contamination giving panes
glass a green tint that one sees along the edge. While any artist knows the importance of
lighting a piece, how much can the lighting actually change its appearance? In rare
instances, oxides added to the glass can change its color depending on the light the glass
is exposed to; this is what is known as photochromism. In this thesis, the potential for
photochromic properties of rare-earth oxide glasses is explored. However, before this can
happen, one must first consider color’s relation to light.

Observed color is the result of the colored object absorbing select portions of
wavelengths within the visible spectrum. The color one sees is a combination of all the
wavelengths transmitted or reflected by the material, i.e. not absorbed. This unabsorbed
light appears as the color complementary to the color associated with the wavelengths
absorbed. For example, an object which absorbs orange light appears to be blue. Table I
details ranges of visible light wavelengths, as well as the color associated with them and
their complements. Color, however, is rarely monochromatic, being composed of a
singular color at full intensity. Therefore, there are three main criteria for which a color
can be defined: hue, saturation, and value. Hue denotes what spectral range the color is
contained within; this is also commonly known as the name of the color (red, blue,
green…). Saturation indicates the amount of white within the color. The less white in a
color, the more saturated it is. For example, there is no range of wavelengths that denote
pink; it is instead an unsaturated red. Value describes how dim or bright a color is;
characterized by the amount of grey that makes up the color1.

[1]
Table I: Spectral colors and their complements1

Wavelength (nm) Spectral Color Complementary color

400-420 Violet Green-Yellow
420-460 Violet-Blue Yellow
460-480 Blue Orange
480-495 Blue-Green Red
495-530 Green Purple
530-570 Green-Yellow Violet
570-590 Yellow Violet-Blue
590-630 Orange Blue
630-700 Red Blue-Green

Because color is so dependent on lighting, several systems have been created in attempts
to standardize the observed color. One such system is the CIELab color system. Fully
developed in 1976, colors are defined based on a three coordinate system; L, a, and b.
The L coordinate describes the lightness of a color from a scale of 0-100. 0 equates to
black, while 100 represents white. Other systems, such as RGB, lightness is determined
based on relative ratios of three defining colors. In the case of the aforementioned system,
these colors are: red, green, and blue. The a and b coordinates are related to
complementary colors. The a coordinate represents how red or green the color appears;
red on the positive end, green on the negative. The b coordinate does the same with
yellow and blue, respectively. This system is illustrated by the graph in Figure 1. Neutral
grey is located at a: 0, b: 02.

Figure 1: Visual representation of CIELab

coordinates2

[2]
 II. Experimental Procedure

A. Fabrication
Samples were created using roughly 100g of System 96® glass frit and about 2g of oxides
of the following rare-earth elements: Praseodymium (Pr3+), Neodymium (Nd3+),
Samarium (Sm3+), Terbium (Tm3+), Dysprosium (Dy3+), Holmium (Ho3+), Erbium (Er3+),
and a 1:1 wt. % composition of Neodymium and Holmium (Nd3+-Ho3+). Lanthanum,
Cerium, Gadolinium, Thulium, Ytterbium, and Lutetium were not tested, as they were
not likely to produce any color. Europium was not tested due to an insufficient quantity
of the oxide to produce a viable sample; however, it was previously known that Eu3+-
doped glass fluoresces a bright scarlet under UV “black light” and is mostly colorless
under all other sources. Promethium was excluded because of its radioactive nature.

Because of the size of the frit obtained from the Alfred University Hot Glass Studio, it
was necessary to crush the glass to a finer size with a mortar and pestle; allowing for a
better melt. The frit was then mixed with one of the rare-earth oxides and melted in a
ceramic crucible. All samples were melted in Carbolite furnace at a temperature of
1250°C for an hour, poured into a cylindrical graphite mold, removed and annealed at
530°C for an hour in a Thermolyne furnace. At the end of the aforementioned duration,
the annealer was shut off and the samples were allowed to cool overnight.

B. Photography
Samples were photographed under nine lighting conditions using a Nikon D50 digital
camera. The lighting conditions were as follows: direct sunlight, daylight incandescent,
soft white incandescent, fluorescent tubing, daylight CFL, soft white CFL, LED UV
“black light”, daylight LED, and soft white LED. With the obvious exception of sunlight,
photographs were taken under all lighting conditions indoors at night. This was done to
prevent light pollution from even indirect sunlight. Furthermore, all sources of artificial

[3]
light were shut off save the source of the light being tested at the time. Specifications for
the bulbs used can be seen in Table IV in the Appendix.

Samples were photographed using the setup shown in Figure 2 below. The
aforementioned camera was positioned on a tripod over the sample; an architect’s lamp
was positioned just outside of the field of view of the camera. This was done so the
sample would be exposed to the light directly, while also preventing the camera from
casting a shadow. To ensure a neutral background for side-to-side comparisons, a piece
of Mohawk Color Copy 98, 28lb Bright White paper was placed beneath all samples.
With the exception of sunlight, the fluorescent tubing, and the UV “black light”, samples
under all lighting types were photographed in this manner. The fluorescent tubing was the
ambient light source of the room, and therefore did not require the use of the architect’s
lamp. Similarly, the UV “black light” condition was satisfied using a UV flashlight
consisting of nine “black light” LEDs. Sunlight conditions were conducted in an open
area around noon, when the Sun was directly overhead, and there was little to no cloud
coverage. In each of the three cases, the tripod and sheet of paper background were still
utilized.

Figure 2: Setup for photographing samples

under various lighting conditions

[4]
Photographs of the emission spectra from most of the light sources were also taken. UV
“black light” and sunlight were not included due to an inability to be photographed. The
UV “black light” did not provide bright enough spectral lines for them to show up in a
photograph. Likewise, there was too much ambient sunlight for the spectral lines to show
up in a photogenic manner, and looking directly at the Sun is not advised for any reason.
In order to photograph the other sources, a black piece of card stock with a narrow slit in
it was placed between the camera and the light source. The emission spectra were then
photographed by placing an Edmund Scientific Co., 1000 lines/mm linear diffraction
grating over the lens of the camera.

C. Color Analysis
In addition to photographing each sample, samples were matched to a similar color using
a RAL color standard wheel, for all light sources. Due to a lack of similar colors within
the RAL scaling, Pr3+ samples were categorized using Hex scaling3. These colors and
their corresponding RAL number were recorded and also converted to CIELab
coordinates4, as well as RGB values5. Table III in the Appendix lists these colors and
their respective values.

Due to discrepancies between observations made by the naked eye and what is imaged by
the camera, some color correction was employed. Initially, this was intended to be
achieved via white balancing on the camera itself, using the white sheet of paper as the
reference. However, due to unknown malfunctions within the camera this was not
possible and alternative measures were taken. Instead, images were level balanced in
Adobe® Photoshop® Elements 9. The corrected images can be seen in Figures 12-19 in
the Appendix; while the raw images can be seen in Figures 20-27 in the Appendix. The
corrected images depict colors closer to those observed with the naked eye. Correcting
the images also generates a relatively standard white background, and allows one to
compare the colors side-by-side with little interference from the color of the cast light.

[5]
Raw images are included to showcase the color given off by the light source, which can
be seen clearly when looking at the color of the “white” background.

D. Spectroscopy
Select samples that exhibited photochromism were them prepared for further spectral
analysis. Samples were cut to have two flat, parallel sides using a Leco ® Corp.
Vari/CutTM VC-50 saw. They were then polished to a higher optical clarity using a
Buehler MetaServ® 250 grinder-polisher. Silicon carbide pads of varying grades ranging
from 120-1200 were utilized.

Spectral data was collected using a PerkinElmer Lambda 950 UV-Vis spectrometer and
the PerkinElmer UV WinLab software. Analysis was conducted from 200-1000nm at a
2.00nm interval. The data was then used to generate spectral graphs using Microsoft ®
Excel 2010.

III. Results
Table II provides a summary of the colors observed for each sample under the various
light sources. Only two of the samples exhibited strong photochromic properties, while
two others exhibited fluorescence with UV exposure. Considering the fluorescent tube
source was the ambient light, there was a concern about the distance between the source
and the sample. However, distance seemed to have no effect on the color of the sample,
so long as it was the only source of light the sample was being exposed to at the time.

[6]
Table II: Coloration of each sample under all lighting conditions

Dopant
Light
Nd3+-
Source Pr3+ Nd3+ Sm3+ Tb3+ Dy3+ Ho3+ Er3+
Ho3+
Green- Oyster Zinc Light Silver
Sunlight Blue lilac Colorless Colorless
yellow white yellow pink grey
Daylight Bitter Oyster Zinc Light Platinum
Blue lilac Colorless Colorless
Incan. lemon white yellow pink grey
Soft
Bitter Oyster Rape Light Pastel
White Blue lilac Colorless Colorless
lemon white yellow pink violet
Incan.
Fluor. Green- Pastel Salmon Light Pigeon
Colorless Colorless Colorless
Tube yellow blue pink pink blue
Daylight Green- Pastel Salmon Light Agate
Colorless Colorless Colorless
CFL yellow blue pink pink grey
Soft
Bitter Pastel Oyster Light Antique Pigeon
White Colorless Colorless
lime blue white pink pink blue
CFL
Daylight Bitter Zinc Light Oyster
Blue lilac Colorless Colorless Colorless
LED lime yellow pink white
Soft
Bitter Zinc Light Platinum
White Blue lilac Colorless Colorless Colorless
lemon yellow pink grey
LED
Bright
UV Light
N/A N/A red N/A N/A N/A N/A
(LED) ivory
orange

A. Tb3+, Sm3+ & Dy3+

For all nine light sources, the Tb3+ sample yielded no coloration. While this did not
provide further information regarding photochromism, it did help to reinforce the color
standard. In the raw Tb3+ images depicted in Figure 23, it is clear that the coloring of the
sample appears to be different under each lighting condition. However, the sample was
colorless to the naked eye when observed under every source. Therefore, in the raw
images, any color observed is due to the camera’s detection of the color cast by the light
itself. Comparatively, the corrected images in Figure 15 are more accurate to the
observed coloration of the sample. While there is still a defined difference in colors,
especially with the warmer-toned soft white bulbs, it is lessened with the correction. A
side-by-side comparison of the correction can be seen in Figure 3 below.

[7]
 a b
3+
Figure 3: Tb sample under soft white incandescent a) raw
and b) corrected

The Sm3+ sample, most notably, exhibited strong fluorescence under UV “black light”.
Under this condition, the mostly colorless sample fluoresced a bright red-orange.
Additionally, the sample appeared off-white to pale yellow under select lighting
conditions. Coloration was only observed under sunlight, soft white CFL, and both
incandescent sources. The coloration is so slight that, on its own, it is hard to distinguish
whether the sample itself is a yellowish color or if it is colorless with a yellowish cast
from the aforementioned sources. However, if compared with the Tb3+ sample, it
becomes apparent there is indeed a difference in coloration of the two samples. While
easy to see with the naked eye, this is difficult to determine based on the corrected
images; but near impossible to gather from the raw images.

The Dy3+ sample, similar to Tb3+, was colorless under nearly all circumstances. However,
under UV “black light” it fluoresced a pale yellow, ivory color. Dy3+ amd Sm3+ were the
only two samples to exhibit any fluorescence under UV light.

B. Pr3+ & Er3+

Both the Pr3+ and Er3+ samples did not exhibit any interesting characteristics; remaining
consistently green-yellow and pink, respectively. While there was some variation in color

[8]
from source to source, there was not a significant enough change to warrant
photochromism.

C. Nd3+, Ho3+ & Nd3+-Ho3+

For sunlight, both incandescent, and both LED, the Nd3+ sample was a light shade of
violet. However, under all three fluorescent sources, the lilac shifted to a more blue color.
This shift can be seen in Figure 4.

a b
Figure 4: Nd3+ sample under a) soft white CFL and b) soft
white incandescent

Spectral analysis of the sample revealed three major absorption peaks within the visible
spectrum (Figure 5). Peaks at 574nm and 586nm show absorption of yellow light,
resulting in the observation of a violet-blue color. The third peak, at 524nm, absorbs
green light, resulting in a purple color. These peaks provide a clear color profile for Nd3+
that agrees with the observed colorations.

[9]
 Nd
45
40

% Transmittance
35
30 524
25
20
15 574
10 586
5
0
380 480 580 680 780
Wavelength (nm)

Figure 5: Absorption/Transmittance spectrum for the Nd3+

sample

The Ho3+ sample exhibited photochromism similar to that of the Nd3+ sample. Likewise,
the sample only changed under fluorescent lighting; appearing as a goldenrod color in all
other lightings, then changing to a salmon pink. An example of this is provided in Figure
6.

a b
Figure 6: Ho3+ sample under a) soft white CFL and b) soft
white incandescent

The absorption/transmittance spectrum of Ho3+ (Figure 7) proved to be more complex

and less clear than the Nd3+ spectral graph. There are two absorption peaks in the violet-
blue range, 446nm and 460nm, adding yellow to the observed color. The 488nm peak

[10]
absorbs in the blue-green, providing red. All these peaks are to be expected considering
the two-tone Ho3+ sample. Surprisingly, there are also two green-yellow peaks, 520nm
and 536, resulting in purple and violet contributions; as well as a red peak at 638nm,
adding blue-green to the mix.

Ho
25
% Transmittance

20 520
488 638
536
15

10 460

5
446
0
380 480 580 680 780
Wavelength (nm)

Figure 7: Absorption/Transmittance spectrum for the Ho3+

sample

Interestingly, the Nd3+-Ho3+ sample did not seem to produce any strong photochromic
qualities. Instead it produced varying shades of blue-grey; with Neodymium’s cool tones
slightly overpowering the warm tones of the Holmium. Additionally, there was no
discernable pattern to the shifting tones as there was for the two constituent dopants.

D. Emission Spectra
Photographs of the emission spectra for fluorescent, incandescent and LED sources are
compiled below in Figure 8. The incandescent sources appear to be the most complete,
having wholly continuous spectra from red to violet-blue. The LED sources, for the most
part, also have continuous spectra. However, there appears to be a slight gap in the blue-
green to blue range. Arguably the most significant of all were the spectra for the

[11]
fluorescent sources. The spectra are more discrete than the other two types of light;
possessing substantial gaps in the range of visible light.

b c

d e

f g
Figure 8: Emission spectra of a) fluorescent b) daylight CFL
c) soft white CFL d) daylight incandescent e) soft white
incandescent f) daylight LED g) soft white LED

IV. Discussion
If the emission photographs are compared to a spectral graph of each source (Figure 9),
the graph confirms the initial observations. The incandescent graph shows an even and
continuous spectrum, inclining towards the redder tones; agreeing with the yellowish cast
given off by the light source. The LED graph indeed has a decrease in the blue-green to

[12]
blue range that matches the gap in the emission spectra collected. The fluorescent graph
also agrees with the photographed spectra; exhibiting narrow spikes of light at
wavelengths that match the discrete emission spectra.

Figure 9: A graphical representation of emission spectra

from a variety of lightings6

Due to their location on the periodic table, rare-earth elements are dominated by valence
electrons in the 4f shell. Because the filled 5s and 5p shells shield the 4f orbital, the
environment a rare-earth ion is placed in has very little effect on the spectrum the ion
generates. In fact, spectra closely resemble the spectra of elemental rare-earth atoms. This
means that the type of glass used should not affect the color the ion imparts on the glass.
Interactions between 4f orbitals also generate narrower bands in absorption spectra than
d-group interactions. This makes rare-earth oxide glasses more susceptible to gaps in
emission spectra than glasses doped with transition metal oxides. While rare-earth ions
have d-orbitals as well as f-orbitals, 4f-4f transitions are the only ones necessary to study
the cause of color. The 4f-5d transitions possess sufficiently high energy to produce
wavelengths of 200nm or less; well within the ultraviolet range and outside the visible
spectrum1.

[13]
4f-4f transitions possess weaker intensities compared to other color producing transitions.
Therefore, a glass must be doped with more rare-earth oxide to achieve the same
saturation as if it were doped with a transition metal oxide. This explains why all colored
samples exhibited pale and pastel shades of color, and may be a potential explanation as
to why the Nd3+-Ho3+ sample did not exhibit photochromism. Another possible
explanation is a cancellation of wavelengths; seen in other rare-earth ions, such as Er3+.
Within its absorption spectra, Er3+ has a green peak at 518nm which comprises the
majority of its pink color. There is another peak in the blue-green region at 483nm;
however it is neutralized by a red peak at 647nm, the first peak’s complementary color.
Without either of these peaks, the sample remains a pink color due to the 518nm peak1.
The combination of the Nd3+ spectrum and Ho3+ spectrum may have caused a similar
cancellation in the resulting Nd3+-Ho3+ spectrum.

A. Nd3+
Research into technical literature confirmed that two main peaks are responsible for the
coloration of the Nd3+ sample. The first is a peak at 521nm in the green region of the
spectrum that matches well with the experimental peak at524nm; resulting in the reddish-
purple portion of the sample’s coloration. The other peak is located within the region at
575nm, giving a violet-blue tint. However, when doped in a glass, the Nd3+ ion can
interact with other ions through the nephelauxetic effect. This translates to a shift in that
peak to 589nm at most. This explains the two experimental peaks at 574nm and 586nm.
Interestingly, when the shift occurs the new peak position complements the yellow
emission peak of Na+ ions. Due to their complementary nature, the two peaks cancel each
other. Thus, Nd3+ doped glass is used in glassmaker’s goggles to negate the yellow light
given off by molten glass, protecting the wearer’s eyes1.

In Figure 10, the aforementioned peaks are extrapolated down to compare with the
emission peaks of fluorescent light. While the peak responsible for the violet-blue
coloration overlaps a small emission peak, the peak resulting in reddish-purple coincides

[14]
with a gap in the spectrum. Therefore, none of the purple color is present while the
sample is under fluorescent lighting.

Nd
45
40
35
% Transmittance

30
25
20
15
10
5
0
380 480 580 680 780
Wavelength (nm)

Figure 10: Comparison of the absorption/transmittance spectrum of Nd3+

and the emission spectrum of a CFL7

[15]
B. Ho3+
The Ho3+ spectrum consists of two main peaks at 535nm and 447nm, corresponding to
the experimental peaks at 536nm and 446nm. The peak at 447nm generates the Ho3+
sample’s yellow color, while the violet of the 535nm peak is cancelled by a portion of the
447nm peak. The spectrum also contains minor peaks at 641nm, 481nm, 471nm, and
465nm, corresponding to peaks at 638nm, 488nm, and 460nm. The blue-green color of
the 641nm peak is normally neutralized by the reds, oranges, and yellows of the complex
band around 465-481nm1. Though, under fluorescent lighting some of the peaks coincide
with emission spectrum gaps (Figure 11). The yellow contribution from the 447nm and
465nm peaks overlap a gap in the spectrum, as does the blue-green contribution at
641nm. The remaining peaks at 481nm (red) and 471nm (orange) and some of the 535nm
peak (violet) combine to produce a salmon pink sample under fluorescent lighting.

[16]
 Ho
25

20

% Transmittance 15

10

0
380 480 580 680 780
Wavelength (nm)

Figure 11: Comparison of the absorption/transmittance spectrum of Ho3+

and the emission spectrum of a CFL7

V. Conclusion
It was determined that, as hypothesized, the photochromism in select rare-earth oxide
glasses was due to a combination of narrow absorption bands and gaps in emission
spectra of select light sources. The absorption peak that provides Nd3+ glass with a purple
tint coincides with a gap in the emission spectrum of fluorescent light; causing the glass
to appear blue due to Nd3+ only other significant peak. Likewise, the Ho3+ peaks that
produce yellow, as well as the blue-green peak that cancels the reds and oranges overlap

[17]
with fluorescent gaps. The resulting combination of reds and oranges yield a salmon pink
color.

Possible sources of error arise from quality of the sample preparation for spectral
analysis. Bubbles and suboptimal polishing due to human error may have caused some of
the absorption peaks to appear broader and shallower. Additionally, the RAL color
scaling wheel contains a finite variety of colors; the standards found for each sample the
best approximations and not the exact color observed. Adding another level of difficulty
was the fact that the colors were being matched to transparent colored glass samples,
making it hard to determine the exact color if there were full opacity. Due to the RAL
scaling’s lack of a decent approximation of the Pr3+ sample’s color, Hex scaling was
needed. However, as the Hex color standard was observed online and not in a tangible
form, such as the RAL color scaling wheel, it may vary from a physical Hex color.

VI. Future Work

Further research can be conducted into the potentials of photochromism in the Nd3+-Ho3+
sample. Increasing the dopant level may provide a stronger color, allowing a more
noticeable shifting in color. Additionally, changing the ratio of the two dopants may yield
interesting results. Spectral analysis should also be conducted to test the viability of the
wavelength cancellation theory. In that respect, how would combining non-photochromic
colorants affect the photochromism of either Nd3+ or Ho3+ doped glass?

[18]
 VII. Appendix
Table III: All colors observed with CIELab and RGB conversions

Color RAL CIE Lab Scaling RGB Scaling

Sample
Name Number L a b R G B
N/A
Bitter lemon 84.94 -28.58 82.60 202 224 13
Hex #cae00d
N/A
Bitter lime 92.84 -46.88 89.36 191 255 0
Hex #bfff000
Green- N/A
91.96 -52.48 81.87 173 255 47
yellow Hex #adff2f
Oyster
1013 88.13 0.19 9.67 227 217 198
white
Light ivory 1015 86.40 2.06 15.48 230 210 181

Zinc yellow 1018 84.83 3.05 69.19 250 202 48

Rape yellow 1021 78.88 10.03 82.04 246 186 0

Bright red
2008 61.99 44.64 51.72 237 107 33
orange
Antique
3014 60.17 32.49 12.58 203 115 117
pink
Light pink 3015 72.73 20.48 3.96 216 160 166

Salmon pink 3022 58.10 36.44 27.34 207 105 85

Blue lilac 4005 50.92 15.58 -23.06 118 104 154

Pastel violet 4009 60.59 10.38 -2.88 157 134 146

Pigeon blue 5014 53.79 -2.64 -15.59 99 125 150

Pastel blue 5024 60.50 -9.53 -17.38 96 147 172

Silver grey 7001 63.81 -2.22 -4.05 140 150 157

Platinum
7036 63.49 1.27 0.78 151 147 146
grey
Agate grey 7038 72.97 -1.50 2.97 176 176 169

[19]
Table IV: Specifications of all light sources utilized

Lighting Bulb Brightness Light

Brand Energy Usage
Type Type (lumens) Appearance
Sunlight - - - - -
Daylight GE 43 W
A19 565 2900K
Incan. Reveal® (60 W replacement)
Soft White
GE A15 560 60 W 2700K
Incan.
Fluor. GE
F32T8 2800 32 W 4100K
Tube Ecolux®
Daylight 9W
EcosmartTM GP19 450 5000K
CFL (40 W replacement)
Soft White 9W
EcosmartTM GP19 550 2700K
CFL (40 W replacement)
Daylight GE 11 W
A19 680 2700K
LED Reveal® (60 W replacement)
Soft White 7W
GE A-shape 470 2700K
LED (40W replacement)
UV (LED) Ultra-Light - - - -

[20]
 a b c

d e f

g h i
Figure 12: Corrected Pr3+ sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[21]
 a b c

d e f

g h i
3+
Figure 13: Corrected Nd sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[22]
 a b c

d e f

g h i
3+
Figure 14: Corrected Sm sample under a) sunlight b) daylight incandescent c) soft
white incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h)
daylight LED i) soft white LED

[23]
 a b c

d e f

g h i
Figure 15: Corrected Tb3+ sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[24]
 a b c

d e f

g h i
3+
Figure 16: Corrected Dy sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[25]
 a b c

d e f

g h i
3+
Figure 17: Corrected Ho sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[26]
 a b c

d e f

g h i
Figure 18: Corrected Er3+ sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[27]
 a b c

d e f

g h i
3+ 3+
Figure 19: Corrected Pr -Ho sample under a) sunlight b) daylight incandescent c) soft
white incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h)
daylight LED i) soft white LED

[28]
 a b c

d e f

g h i
3+
Figure 20: Raw Pr sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[29]
 a b c

d e f

g h i
3+
Figure 21: Raw Nd sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[30]
 a b c

d e f

g h i
3+
Figure 22: Raw Sm sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[31]
 a b c

d e f

g h i
3+
Figure 23: Raw Tb sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[32]
 a b c

d e f

g h i
3+
Figure 24: Raw Dy sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[33]
 a b c

d e f

g h i
3+
Figure 25: Raw Ho sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[34]
 a b c

d e f

g h i
3+
Figure 26: Raw Er sample under a) sunlight b) daylight incandescent c) soft white
incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h) daylight
LED i) soft white LED

[35]
 a b c

d e f

g h i
3+ 3+
Figure 27: Raw Nd -Ho sample under a) sunlight b) daylight incandescent c) soft
white incandescent d) fluorescent e) daylight CFL f) soft white CFL g) UV LED h)
daylight LED i) soft white LED

[36]
 VIII. References

[1] K. Binnemans and C. Görller-Walrand, "On the color of the trivalent lanthanide
ions," Chem. Phys. Lett., 235 [3-4] 163-74 (1995).

[2] "Technical Guides: Color Models, CIELAB" (2000) Adobe Systems Inc. Accessed
on: March 2016. Available at
<http://dba.med.sc.edu/price/irf/Adobe_tg/models/cielab.html>

[3] M. Gallagher, Encycolorpedia. Accessed on: March 2016. Available at

<http://encycolorpedia.com>

[4] "List of RAL colors" (2016) Wikimedia Foundation. Accessed on: March 2016.
Available at <https://en.wikipedia.org/wiki/List_of_RAL_colors>

[5] C. Cabo, "RAL Colours Classic" (2014). Accessed on: March 2016. Available at
<http://rgb.to/ral>

[6] J. Herrman, "Ultimate Light Bulb Test: Incandescent vs. Compact Fluorescent vs.
LED" (2011) Popular Mechanics. Accessed on: April 2016. Available at
<http://www.popularmechanics.com/technology/gadgets/reviews/g164/incandescent-
vs-compact-fluorescent-vs-led-ultimate-light-bulb-test/?slide=1>

[7] S. Watson, "Basics when Designing and Picking Light Sources" (2016) LEDinside.
Accessed on: April 2016. Available at
<http://www.ledinside.com/knowledge/2012/9/basics_when_designing_and_picking_
light_sources_20120926>

[37]