- Email: [email protected]
Mechanism of coloration in copper-stained float glass
Journal of Non-Crystalline Solids 120 (1990) 199-206 North-Holland
199
M E C H A N I S M OF COLORATION IN C O P P E R - S T A I N E D F L O A T G L A S S Hiroyuki N A G A O , M a s a o M I S O N O U and Hideo K A W A H A R A Central Research Laboratory, Nippon Sheet Glass Co. Ltd, Konoike, Itami 664, Japan
Copper staining was applied to a commercial float glass and to a Colburn glass, resulting in green and yellow colors, respectively. Such glass was investigated with optical spectroscopy and X-ray diffraction analysis. It was observed that the colors of copper-stained float glass comprised three kinds of extinction structures in the transmittance spectrum located around the wavelengths of 460 nm, 560 nm and 600 - 660 nm. The first two were attributed to the intrinsic absorption of metallic copper colloids while the last was ascribed to the scattering by the metallic copper colloids. Cuprous oxide was considered to be an ineffective colorant. A coloration model for the copper-stained float glass was presented in terms of the variation in size of metallic copper particles.
1. Introduction Copper-stained glass, as well as c o p p e r - r u b y glass, is a well-known glass exhibiting a clear red color [1]. The staining m e t h o d generally used is the application of a copper c o m p o u n d to a glass surface, followed by baking at a temperature in the glass annealing range to diffuse copper (mainly cuprous) ions which replace alkali ions in its surface [2]. Inside the glass, cuprous ions are subsequently reduced to metallic copper either by reducing agents in the glass or by heat treatment under a reducing gaseous atmosphere [2]. So far, it has been f o u n d that several kinds of color, such as yellow and green, are obtained by the copperstaining m e t h o d [3-7]. The mechanism of this coloration is not fully understood. The color of copper-stained red has been generally ascribed to colloidal particles of metallic copper. R a w s o n [3] and B a m f o r d [8] calculated the extinction spectrum of glasses, on the basis of the Mie theory, in which colloidal particles of metallic copper of various sizes are suspended, and they showed that the calculated curves matched satisfactorily with copper-stained red glasses into which copper ions were chemically a n d / o r electrically diffused. O n the other hand, some researchers [9,10] p r o p o s e d that the red colorant which is actually effective is colloidal cuprous oxide, which exhibits a sharp absorption
edge corresponding to a b a n d - g a p of 2.17 eV [11]. Today, commercial sheet glass is mostly produced by the float process, in which a glass ribbon floats on a molten tin bath in order to obtain optically flat surfaces. This float glass seems to be one of the most appropriate materials for staining, because stannous ions are introduced into glass in the float b a t h and f o r m reducing layers near glass surfaces. In fact, it is k n o w n that the incorporation of c o p p e r into float glass can produce m a n y kinds of color. The objective of this paper is the precise characterization of copper-stained commercial float glass, especially deep-green in color, from the experimental study of optical properties and structure.
2. Experimental 2.1. Glass samples
The glass samples used in the present study were commercial flat glasses of 2 m m in thickness m a d e by the N i p p o n Sheet Glass Co.. Some of them were m a d e by the float-forming process, others by the C o l b u r n (or L i b b e y - O w e n s ) method. T h e y b o t h consist of silicon oxide, sodium oxide, calcium oxide, m a g n e s i u m oxide with a small a m o u n t of a l u m i n u m oxide, iron oxide, potassium
0022-3093/90/$03.50 © 1990 - Elsevier Science Publishers B.V. (North-Holland)
H. Nagao et al. / Mechanism of coloration
200
Table 1 Chemical compositions of glass samples studied (wt%) Glass
SiO 2
A1203
Fe203
CaO
MgO
Na 20
K 2°
SO 3
Float Colburn
71.3 72.5
1.5 1.7
0.1 0,1
9.0 7.1
3.6 4.1
13.5 13.3
0.8 0.9
0.2 0.3
oxide and sulfur oxide. Other elements may be derived from unintentional impurities in raw materials; however, their weight fractions were less than 10 ppm. The chemical compositions are shown in table 1. It is well known that tin is concentrated at the bottom surface of the float glass, which had been in direct contact with molten tin in the float bath. The analytical results of tin at glass surfaces are listed in table 2. Even at the top surface (opposite the tin bath) of float glass, tin was detected by X-ray fluorescence analysis, which was not the case for the Colburn glass.
2.2. Ion-exchange process The staining method we used was the application Of a paste to a glass surface followed by baking. The paste was made of inorganic powders and viscous organic oil. The inorganic powder consisted of cupric sulfate, sodium sulfate, zinc sulfate, and zirconium silicate (zircon). Cupric sulfate is indispensable for coloration. Other sulfates are used for lowering the melting temperature of cupric sulfate down to near the glass annealing temperatures. Powders of zirconium silicate serve as the dispersing and molding agent. The inorganic powders were crushed and mixed in a ball mill with ethanol for about 30 h. After being dried, they were mixed with an organic oil of 2Pa s in viscosity for screen-printing (by Okuno Chemical Industries, Osaka) which resulted in a Table 2 Analytical results of tin at glass surfaces Glass surface
EPMA (wt%)
XF (cps)
Float top Float bottom Colburn
1.9 n.d. n.d.
39100 640 n.d.
n.d.: Not detected.
paste of some 10 Pa s. Table 3 lists compositions of the inorganic powders in two kinds of staining paste used in this study. The weight-fraction of cupric sulfate is 50% in Paste # A whereas that in Paste # B is 12%. The paste coating was applied to the glass surface with an applicator and then baked at 1 5 0 ° C for 15 min to decompose the viscous oil from the paste. In the case of float glass, staining was conducted on the top surface. The thickness of the paste on glass, measured after the baking, was approximately 50 g / c m 2. The glass was then heated to 600 ° C and maintained at this temperature for 6-30 min in an electric furnace. Finally residual paste on the glass was removed by washing and scrubbing. The treatment conditions for all samples are listed in table 4.
2.3. Post heat treatment One of the stained float glass samples (B1) was re-heated at approximately 600 ° C in an electric furnace to investigate the color change. It was annealed for 15 min either in nitrogen gas or in air.
2.4. X-ray diffraction analysis For indentifying crystalline species precipitated in stained glass, we used an X-ray diffractometer with Cu K , radiation from a rotating anode X-ray tube operated at 60 kV and 200 mA. Diffraction Table 3 Constituents of inorganic solids in staining paste used in this study. Values are shown in wt% Paste
CuSO 4
Na2SO 4
ZnSO 4
ZrSiO4
~A #B
50 12
30 48
20
20 20
H. Nagao et aL / Mechanism of coloration
201
Table 4 S u m m a r y of i o n - e x c h a n g e c o n d i t i o n s a n d r e s u l t a n t o p t i c a l p r o p e r t i e s Sample no.
A - 1 A - 2 A - 3 A - 4 A - 5 B- l B-1A B-1N
Paste
# A # A # A # A :~ A ~ B ~B ~B
Glass
float float float Colburn Colburn float float float
Ion-exchange c o n d i t i o n s a)
600 600 600 600 600 600 600 600
o C, oC oC oC oC oC oC oC
6 10 30 10 30 10 10 10
rain rain min min rain min min min
O p t i c a l p r o p e r t i e s b)
Post annealing conditions
-
6 0 0 ° C , 1 5 m i n (air) 600°C, 15rnin (N2)
L (%)
D.W
P~
(nm)
(%)
38 19 7.6 78 69 54 54 54
565 563 569 572 573 571 575 571
42 40 56 56 70 34 31 34
Color
green green green yellow yellow yellowish green brownish green yellowish green
~) In air. h) L, L u m i n o u s t r a n s m i t t a n c e ; D . W , d o m i n a n t w a v e l e n g t h ; Pe, e x c i t a t i o n p u r i t y .
patterns, ranging from 35 to 55 ° (20), were measured by 20 scanning fixing the co axis at 5 ° with a plane-crystal monochromator.
2.5. Optical transmittance spectrum analysis A scanning spectrophotometer was used for the determinations of optical transmission curves in the near ultra-violet, visible and near infrared regions. The normally incident beam illuminated the side of the sample opposite the stained surface. The total transmittance including both parallel (or straight) and diffuse components was measured with an integrating sphere, 60 m m in internal diameter. The parallel transmittance was measured without the integrating sphere. The diffusion transmittance was determined by subtracting the parallel transmittance from the total transmittance.
stannous oxide had been added, were remelted and vitrified as a standard specimen for quantitative estimation of the amount of copper and tin in the glass. An automatic X-ray fluorescence spectrometer was used for the semiquantitative analysis of tin at the glass surface. The incident X-ray beam was obtained from a scandium target operated at 50 kV and 40 mA.
3. Results
3.1. Concentration profile of copper in glass Concentration profiles of copper in the stained float glass are summarized in fig. 1. The profiles
20
2.6. Electron probe X-ray microanalysis and X-ray fluorescence analysis An electron probe X-ray microanalyzer (EPMA) was used for quantitative analysis of copper and tin at the glass surface. An electron beam of about 1.5 pm in diameter was accelerated at 15 kV. In order to obtain a concentration profile of copper in the stained samples, the electron beam was scanned over a section perpendicular to the glass surface. Powders of commercial soda-lime glass, to which a measured reagent of cupric oxide or
.~ io
8 5 o
10
20 30 /.0 50 Depth /jam Fig. 1. Depth profiles o f copper concentration in copper-stained float glass samples.
202
H. N a g a o et aL / M e c h a n i s m o f coloration
1.0
1.0
0.8
0.8
...../.....................................................................
G~
~I 0.6 ~
AI
A2
"
"
A1¢
A5
0.6
0'z v0.2
0
0.2
ZOO
800
/nrn
1200
Wavelength Fig. 2. Parallel transmittance spectra for copper-stained float glass samples. B, G and R indicate three kinds of extinction bands described in the text. Dotted line: unstained sample.
for the stained Colburn glasses are not illustrated as there was only a small difference in the concentration profile between the float glass and the Colburn glass. The concentration of copper decreases smoothly and monotonically from the surfaces in all samples exclusive of surface layers of about 5 ~m in thickness of samples ion-exchanged using Paste # A. A larger amount of copper was diffused into the glass ion-exchanged for 10 min using Paste # A than glass ion-exchanged for the same time using Paste # B.
3.2. Optical properties All stained float glasses are yellowish green or deep green and still they show clearness and transparency. The transmittance spectra of the stained float glass samples are shown in fig. 2. There are two kinds of extinction bands in each transmittance curve located around the wavelengths of 560 nm and 600-660 nm, which are hereafter designated as G (green)- and R (red)-extinctions, respectively. The peak of the R-extinction becomes broadened and shifts to longer wavelengths with increased ion exchange time, while the peak position of the G-extinction seems constant. There is a subtle shoulder-like structure in the transmittance curve of the sample B1 located at about 450 nm, designated as B (blue)-extinction in fig. 2.
o
~6o
soo
600
/nm
76o
800
Wavelength Fig. 3. Parallel transmittance spectra for copper-stained Colburn glass samples. 'C' indicates an absorption edge-like structure described in the text. Dotted line: unstained sample.
In contrast, stained Colburn glasses are bright yellow. Each parallel transmittance spectrum for Colburn glass sample shows only an absorption edge-like structure located around 460 nm, designated as C (Colburn)-extinction (fig. 3). Luminous transmittance is still 63% even after the ion-exchange procedure for 30 min using Paste # A. The remarkable difference in color between stained Co!burn and float glass samples must be associated with the existence of a tin-rich surface layer in }he latter. Diffuse transmittance spectra were measured for all stained samples; however, they were practi0.3
o~ 02
0
' tOO
i i i i 800 ~2'00 Wovelength /nm Fig. 4. Diffuse transmittance spectra for copper-stained float glass samples. In this diagram, transmittance curves for A2 and A3 are shifted 0.05 and 0.1 upward, respectively, in order to avoid intersection of them.
H. Nagao et al. / Mechanism of coloration
203
1.0
Cu=O 111
0.8
c
o.6
o 200
0.~
i0.2
30
35
4~0
Z.~5
50
55
0
28 / deg Fig. 5. X-ray diffraction patterns of copper-stained float glass samples.
cally useless except for the float glass samples shown in fig. 4. One should note a distinct peak in each spectrum curve in fig. 4, the maximum position of which shifts to longer wavelengths with an increase in ion exchange time. The peak positions are rather close to the respective peak positions of the R-extinction seen in fig. 2. The optical properties for all samples are listed in table 4.
3.3. Crystalline species in stained glass The X-ray diffraction patterns for stained samples revealed that crystalline particles of both metallic copper and cuprous oxide were precipitated in copper-stained float glass samples (fig. 5)
Cu=O 111
C
u
=
600
700
800
Wovetength / nrn
and only cuprous oxide in stained Colburn glass samples (fig. 6). Cupric oxide was not detected in any samples. The value of the relative intensity for each diffraction peak is in good agreement with the value which was obtained from a powdered specimen [12]. This result is consistent with the assumption that crystalline colloidal particles are suspended in the glass. Table 5 lists integral intensities after background reduction, and integral breadths of the (111) reflection for both metallic copper and cuprous oxide. The (111) reflection of metallic copper becomes stronger in integral intensity and sharper from sample A1 through A3, which indicates that both the volume and the mean size of Table 5 Integral intensities and integral breadths of 11l reflections in X-ray diffraction for cuprous oxide and metallic copper
O
Sample
"E
35
560
Fig. 7. Effect of annealing for 15 min in nitrogen gas on the transmittance curve of copper-stained float glass. Dotted line: as-stained sample. Solid line: annealed sample.
oJ
30
/,00
~'0
~'5
50
55
20 / deg
Fig. 6. X-ray diffraction patterns of copper-stained Colburn glass samples.
A- 1 A-2 A-3 A-4 A- 5 B- 1
Integral intensity (cps)
Integral width (deg (2 0))
C u 2 0 111
Cu 111
C u 2 0 111
Cu I l l
1.50 × 105 2.06×105 2.82×105 2.63×105 3.80 × 105 7.37 × 104
3.33 × 104 5.00×104 1.02x 105 n.d. n.d. 1.60 X 104
1.182 1.315 1.178 1.719 1.731 0.724
0.567 0.467 0.439 n.d. n.d. 0.547
n.d.: Not detectable.
H. Nagao et al. / Mechanism of coloration
204 1.0
0.8
o~ c o
0.6
o 0.~
i---
0.2
400
500
600
700
800
Wavelength /nm
Fig. 8. Effect of a n n e a l i n g for 15 m i n in the air o n the t r a n s m i t t a n c e curve of c o p p e r - s t a i n e d float glass. D o t t e d line: as-stained sample. Solid line: a n n e a l e d sample.
metal copper colloids gradually increases with increasing ion-exchange time.
3.4. Some changes after post heat treatment The post heat treatment at 600 ° C for 15 min in an atmosphere of nitrogen hardly affected the color of stained yellowish green float glass (fig. 7). Strictly, the B- and R-extinctions became a little stronger after the annealing. On the other hand, the effect of annealing in air of the same sample was rather different (fig. 8). The R-extinction became distinctly enfeebled whereas the G-extinction became a little stronger. As a result, the color turned into brownish green.
4. D i s c u s s i o n
4.1. The colorant in copper-stained float glass Human eyes are, in general, very sensitive to a subtle change in the transmittance curve. The color of stained float glass gets greenish when the R-extinction becomes relatively dominant as compared with the G-extinction, while it becomes reddish in the opposite case. The origin of the extinctions is of great importance for our understanding of the coloration mechanism of copper-stained float glass.
The cuprous ion has the d 1° configuration and, as such, will not impart coloration [13]. A cupric ion shows an absorption band centered at 780 nm in soda-lime silicate glass [13]. The optical properties of such ions cannot explain the optical properties of the copper-stained glass shown in fig. 2. X-ray diffraction measurements suggested that colloidal particles of metallic copper a n d / o r cuprous oxide are possible colorants in the stained float glass. The comparison between float glass samples and Colburn glass samples would be informative concerning the cause for the G- and R-extinctions. The Colburn glass samples show only an absorption edge-like structure located at about 460 nm while the float glass samples exhibit three kinds of extinction structures, inclusive of the G- and the R-extinctions as are shown in figs. 3 and 2, respectively. X-ray diffraction measurements revealed that the integral intensity of the (111) reflection of crystalline cuprous oxide in the former samples is comparable to that in the latter, and that crystalline particles of metallic copper were detected only in the float glass samples (table 4). Thus, it seems unlikely that either the G- or R-extinctions is ascribed to the colloidal particles of cuprous oxide in the glass. It is proposed that only colloidal particles of metallic copper are the effective colorant in copper-stained float glass.
4.2. Nature of metallic copper colloids suspended in glass As a general rule, a colloidal metal particle exhibits both light scattering and absorption attributed to the plasmon oscillations of free-electrons, and such an extinction band, which consists of bands by absorption and scattering, appears in the transmittance spectrum [14]. The transmittance spectra of some glasses in which spherical particles of metallic copper are suspended have been calculated by many researchers [3,8,15,16]. Before we discuss the origin of three extinction structures observed in stained float glass, let us show some calculated curves of the extinction cross-section (Ce×) as well as the scattering crosssection (Csc) and absorption cross-section (Cab)
H. Nagao et aL / Mechanism of coloration 20nm
lb
x
3Onto 40 , 40nm 8C
50nm
~x
20t
o
ex
ex~
E O
O C.>
205
20 nm in radius. However, this band does not appear in the Cex curves for larger particles. It can be concluded from the calculation that the small particles cause extinction bands due to intrinsic absorption at about 430 nm and 580 nm, whereas the larger particles create an extinction band only due to scattering located at about 600 615 nm. -
£c
6 8
04 6 Waveleng h / lOOnm 6
4
6
Fig. 9. Calculated cross-sections for scattering (sc), absorption (ab) and extinction (ex) for spheres of colloidal copper having four different radii. Note different scales for the vertical axis in each figure.
for some colloidal particles of metallic copper based on the Mie theory. We used the subroutine B H M I E [17] to calculate the cross-sections for spherical metal copper particles of 20 - 50 nm in radius. The refractive index of the glass matrix was assumed to be 1.52 and the optical constants of copper were taken from the tabulation in ref. [18]. The calculated curves of Cex, Cs~ and Cab are shown in fig. 9, which demonstrates some informative features. (1) Both Cs~ and Cab become gradually stronger as the particle increases in size, and the enlargement in Csc is much more remarkable than that for Cab. The scattering cross-section is practically negligible as compared with C,b for a particle of 20 nm in radius, but it increases to be roughly twice that of Cab for a particle of 50 nm in radius. (2) As the peak position of Cs~ is always longer in wavelength than Cab, the peak position of Cex is always shorter in wavelength than that of Cs~. (3) The locations of maxima of both C~ and C,b shift gradually to longer wavelengths as the particle size increases, and the change in peak position for C~c is much more rapid than that for Cab. The peak positions for Csc and Cab are about 580 nm and 590 nm in the case of a particle of 20 nm in radius, and they shift to about 585 nm and 615 nm in the case of a particle of 50 nm, respectively. (4) There is another extinction band located around the wavelength of 430 nm for a particle of
4.3. Coloration mechanism of copper-stained float glass The X-ray diffraction analyses demonstrated that the mean particle size of metallic copper colloids gradually increases as the ion-exchange time increases. If the effective colorant in copperstained float glass is regarded as metallic copper colloids, the systematic change in transmittance curve with increasing ion-exchange time should be ascribed to the increase in particle size of colloids (without taking account of the wide distribution of particle size). The positions of the diffuse transmittance maximum shown in fig. 4 are rather close to the locations of the respective minimum in the R-extinction in fig. 2. They are both characterized by the gradual shift to longer wavelengths in peak position with some increase in peak intensity. Table 6, which lists the positions of diffuse transmittance maxima and transmittance minima of the R-extinction on parallel transmittance curves for copper-stained float glass samples, shows that the former position is always longer in wavelength than the latter. All these data are consistent with the assumption that the R-extinction is mainly
Table 6 Positions of the maximum in diffuse transmittance curves (Dm~x) and those of the minimum of the R-extinction in parallel transmittance curves (Train) for copper-stained float glass samples Sample no.
Ion-exchange time (min)
Tmi. (nm)
DmaX (nm)
A-1 A- 2 A- 3
6 10 30
625 630 660
655 700 830
206
H. Nagao et al. / Mechanism of coloration
ascribed to the scattering by relatively large particles of metallic copper colloids in stained glass. O n the left side of each diffuse transmittance curve in fig. 4, there are two kinds of fine structure located a r o u n d 450 n m and 560 n m (arrows in fig. 4), which implies the existence of absorption bands. These structures can be correlated with the B- and the G-extinctions as their positions coincide with positions of the extinctions. These data are consistent with the idea that b o t h the G- and B- extinctions are chiefly attributed to the intrinsic absorption by relatively small particles of metallic copper colloids in stained glass. As a result, extinctions in the transmittance spectra of copper-stained float glass are believed to originate f r o m colloidal particles of metallic copper precipitated in glass surface regions. The change of transmittance curves by post-annealing in air seems to be reasonable on the basis of the assumption that metallic copper colloids decrease in mean size during the annealing. It implicitly results in both the development of the G-extinction and attenuation of the R-extinction. We suppose that the most probable m e c h a n i s m for the decrease in mean size of metallic c o p p e r colloids during post annealing is their oxidation. It is important to mention that the reactions in the glass during staining is severely influenced by reducing agents such as tin in the glass. The effects of reducing agents on the color of copperstained float glass is to be discussed in another paper. The authors are indebted to Dr T. Y o k o of K y o t o University for some valuable discussions,
Mr K. K o m a k i for the E P M A analysis, and Rigaku C o r p o r a t i o n for the X - r a y diffraction analysis.
References [1] W.A. Weyl, Coloured Glasses (Society of Glass Technology, Sheffield, 1951). [2] C.R. Bamford, Color Generation and Control in Glass (Elsevier, Amsterdam 1977). [3] H. Rawson, Phys. Chem. Glass. 6 (1965) 81. [4] O.S. Levi, Bull. Am. Ceram. Soc. 34 (1965) 119. [5] S. Sakka and T. Nishiyuki, J. Non-Cryst. Solids 37 (1980) 139. [6] S. Sakka, T. Nishiyuki, M. Nishitomi and K. Kamiya, Res. Rep. Fac. Eng. Mie Univ. 5 (1980) 85. [7] S. Sakka, K. Kamiya and K. Kato, J. Non-Cryst. Solids 52 (1982) 77. [8] C.R. Bamford, Phys. Chem. Glass. 17 (1976) 209. [9] A. Ram and S.N. Prasad, Advances in Glass Technoogy, Technical Papers of the VI Int. Congress on Glass (Plenum, New York 1962) p. 256. [10] S. Banerjee and A. Paul, J. Am. Ceram. Soc. 57 (1974) 286. [11] P.W. Baumeister, Phys. Rev. 121 (1961) 359. [12] American Society of Testing Materials, X-ray powder data file (Inorganic), Set 1-5 (1960). [13] C.R. Bamford, Phys. Chem. Glass. 3 (1962) 189. [14] H.C. Van de Hulst, Light Scattering by Small Particles. (Wiley, New York, 1957). [15] D.M. Trotter, J.W.H. Schreurs and P.A. Tick, J. Appl. Phys. 53 (1982) 4657. [16] R. Ruppin, J. Appl. Phys. 59 (1986) 1355. [17] C.F. Bohren and D.R. Huffman, Absorption and Scattering of Light by Small Particles (Wiley, New York, 1983) [18] P.B. Johnson and R.W. Christy, Phys. Rev. B6 (1972) 4370.
199
M E C H A N I S M OF COLORATION IN C O P P E R - S T A I N E D F L O A T G L A S S Hiroyuki N A G A O , M a s a o M I S O N O U and Hideo K A W A H A R A Central Research Laboratory, Nippon Sheet Glass Co. Ltd, Konoike, Itami 664, Japan
Copper staining was applied to a commercial float glass and to a Colburn glass, resulting in green and yellow colors, respectively. Such glass was investigated with optical spectroscopy and X-ray diffraction analysis. It was observed that the colors of copper-stained float glass comprised three kinds of extinction structures in the transmittance spectrum located around the wavelengths of 460 nm, 560 nm and 600 - 660 nm. The first two were attributed to the intrinsic absorption of metallic copper colloids while the last was ascribed to the scattering by the metallic copper colloids. Cuprous oxide was considered to be an ineffective colorant. A coloration model for the copper-stained float glass was presented in terms of the variation in size of metallic copper particles.
1. Introduction Copper-stained glass, as well as c o p p e r - r u b y glass, is a well-known glass exhibiting a clear red color [1]. The staining m e t h o d generally used is the application of a copper c o m p o u n d to a glass surface, followed by baking at a temperature in the glass annealing range to diffuse copper (mainly cuprous) ions which replace alkali ions in its surface [2]. Inside the glass, cuprous ions are subsequently reduced to metallic copper either by reducing agents in the glass or by heat treatment under a reducing gaseous atmosphere [2]. So far, it has been f o u n d that several kinds of color, such as yellow and green, are obtained by the copperstaining m e t h o d [3-7]. The mechanism of this coloration is not fully understood. The color of copper-stained red has been generally ascribed to colloidal particles of metallic copper. R a w s o n [3] and B a m f o r d [8] calculated the extinction spectrum of glasses, on the basis of the Mie theory, in which colloidal particles of metallic copper of various sizes are suspended, and they showed that the calculated curves matched satisfactorily with copper-stained red glasses into which copper ions were chemically a n d / o r electrically diffused. O n the other hand, some researchers [9,10] p r o p o s e d that the red colorant which is actually effective is colloidal cuprous oxide, which exhibits a sharp absorption
edge corresponding to a b a n d - g a p of 2.17 eV [11]. Today, commercial sheet glass is mostly produced by the float process, in which a glass ribbon floats on a molten tin bath in order to obtain optically flat surfaces. This float glass seems to be one of the most appropriate materials for staining, because stannous ions are introduced into glass in the float b a t h and f o r m reducing layers near glass surfaces. In fact, it is k n o w n that the incorporation of c o p p e r into float glass can produce m a n y kinds of color. The objective of this paper is the precise characterization of copper-stained commercial float glass, especially deep-green in color, from the experimental study of optical properties and structure.
2. Experimental 2.1. Glass samples
The glass samples used in the present study were commercial flat glasses of 2 m m in thickness m a d e by the N i p p o n Sheet Glass Co.. Some of them were m a d e by the float-forming process, others by the C o l b u r n (or L i b b e y - O w e n s ) method. T h e y b o t h consist of silicon oxide, sodium oxide, calcium oxide, m a g n e s i u m oxide with a small a m o u n t of a l u m i n u m oxide, iron oxide, potassium
0022-3093/90/$03.50 © 1990 - Elsevier Science Publishers B.V. (North-Holland)
H. Nagao et al. / Mechanism of coloration
200
Table 1 Chemical compositions of glass samples studied (wt%) Glass
SiO 2
A1203
Fe203
CaO
MgO
Na 20
K 2°
SO 3
Float Colburn
71.3 72.5
1.5 1.7
0.1 0,1
9.0 7.1
3.6 4.1
13.5 13.3
0.8 0.9
0.2 0.3
oxide and sulfur oxide. Other elements may be derived from unintentional impurities in raw materials; however, their weight fractions were less than 10 ppm. The chemical compositions are shown in table 1. It is well known that tin is concentrated at the bottom surface of the float glass, which had been in direct contact with molten tin in the float bath. The analytical results of tin at glass surfaces are listed in table 2. Even at the top surface (opposite the tin bath) of float glass, tin was detected by X-ray fluorescence analysis, which was not the case for the Colburn glass.
2.2. Ion-exchange process The staining method we used was the application Of a paste to a glass surface followed by baking. The paste was made of inorganic powders and viscous organic oil. The inorganic powder consisted of cupric sulfate, sodium sulfate, zinc sulfate, and zirconium silicate (zircon). Cupric sulfate is indispensable for coloration. Other sulfates are used for lowering the melting temperature of cupric sulfate down to near the glass annealing temperatures. Powders of zirconium silicate serve as the dispersing and molding agent. The inorganic powders were crushed and mixed in a ball mill with ethanol for about 30 h. After being dried, they were mixed with an organic oil of 2Pa s in viscosity for screen-printing (by Okuno Chemical Industries, Osaka) which resulted in a Table 2 Analytical results of tin at glass surfaces Glass surface
EPMA (wt%)
XF (cps)
Float top Float bottom Colburn
1.9 n.d. n.d.
39100 640 n.d.
n.d.: Not detected.
paste of some 10 Pa s. Table 3 lists compositions of the inorganic powders in two kinds of staining paste used in this study. The weight-fraction of cupric sulfate is 50% in Paste # A whereas that in Paste # B is 12%. The paste coating was applied to the glass surface with an applicator and then baked at 1 5 0 ° C for 15 min to decompose the viscous oil from the paste. In the case of float glass, staining was conducted on the top surface. The thickness of the paste on glass, measured after the baking, was approximately 50 g / c m 2. The glass was then heated to 600 ° C and maintained at this temperature for 6-30 min in an electric furnace. Finally residual paste on the glass was removed by washing and scrubbing. The treatment conditions for all samples are listed in table 4.
2.3. Post heat treatment One of the stained float glass samples (B1) was re-heated at approximately 600 ° C in an electric furnace to investigate the color change. It was annealed for 15 min either in nitrogen gas or in air.
2.4. X-ray diffraction analysis For indentifying crystalline species precipitated in stained glass, we used an X-ray diffractometer with Cu K , radiation from a rotating anode X-ray tube operated at 60 kV and 200 mA. Diffraction Table 3 Constituents of inorganic solids in staining paste used in this study. Values are shown in wt% Paste
CuSO 4
Na2SO 4
ZnSO 4
ZrSiO4
~A #B
50 12
30 48
20
20 20
H. Nagao et aL / Mechanism of coloration
201
Table 4 S u m m a r y of i o n - e x c h a n g e c o n d i t i o n s a n d r e s u l t a n t o p t i c a l p r o p e r t i e s Sample no.
A - 1 A - 2 A - 3 A - 4 A - 5 B- l B-1A B-1N
Paste
# A # A # A # A :~ A ~ B ~B ~B
Glass
float float float Colburn Colburn float float float
Ion-exchange c o n d i t i o n s a)
600 600 600 600 600 600 600 600
o C, oC oC oC oC oC oC oC
6 10 30 10 30 10 10 10
rain rain min min rain min min min
O p t i c a l p r o p e r t i e s b)
Post annealing conditions
-
6 0 0 ° C , 1 5 m i n (air) 600°C, 15rnin (N2)
L (%)
D.W
P~
(nm)
(%)
38 19 7.6 78 69 54 54 54
565 563 569 572 573 571 575 571
42 40 56 56 70 34 31 34
Color
green green green yellow yellow yellowish green brownish green yellowish green
~) In air. h) L, L u m i n o u s t r a n s m i t t a n c e ; D . W , d o m i n a n t w a v e l e n g t h ; Pe, e x c i t a t i o n p u r i t y .
patterns, ranging from 35 to 55 ° (20), were measured by 20 scanning fixing the co axis at 5 ° with a plane-crystal monochromator.
2.5. Optical transmittance spectrum analysis A scanning spectrophotometer was used for the determinations of optical transmission curves in the near ultra-violet, visible and near infrared regions. The normally incident beam illuminated the side of the sample opposite the stained surface. The total transmittance including both parallel (or straight) and diffuse components was measured with an integrating sphere, 60 m m in internal diameter. The parallel transmittance was measured without the integrating sphere. The diffusion transmittance was determined by subtracting the parallel transmittance from the total transmittance.
stannous oxide had been added, were remelted and vitrified as a standard specimen for quantitative estimation of the amount of copper and tin in the glass. An automatic X-ray fluorescence spectrometer was used for the semiquantitative analysis of tin at the glass surface. The incident X-ray beam was obtained from a scandium target operated at 50 kV and 40 mA.
3. Results
3.1. Concentration profile of copper in glass Concentration profiles of copper in the stained float glass are summarized in fig. 1. The profiles
20
2.6. Electron probe X-ray microanalysis and X-ray fluorescence analysis An electron probe X-ray microanalyzer (EPMA) was used for quantitative analysis of copper and tin at the glass surface. An electron beam of about 1.5 pm in diameter was accelerated at 15 kV. In order to obtain a concentration profile of copper in the stained samples, the electron beam was scanned over a section perpendicular to the glass surface. Powders of commercial soda-lime glass, to which a measured reagent of cupric oxide or
.~ io
8 5 o
10
20 30 /.0 50 Depth /jam Fig. 1. Depth profiles o f copper concentration in copper-stained float glass samples.
202
H. N a g a o et aL / M e c h a n i s m o f coloration
1.0
1.0
0.8
0.8
...../.....................................................................
G~
~I 0.6 ~
AI
A2
"
"
A1¢
A5
0.6
0'z v0.2
0
0.2
ZOO
800
/nrn
1200
Wavelength Fig. 2. Parallel transmittance spectra for copper-stained float glass samples. B, G and R indicate three kinds of extinction bands described in the text. Dotted line: unstained sample.
for the stained Colburn glasses are not illustrated as there was only a small difference in the concentration profile between the float glass and the Colburn glass. The concentration of copper decreases smoothly and monotonically from the surfaces in all samples exclusive of surface layers of about 5 ~m in thickness of samples ion-exchanged using Paste # A. A larger amount of copper was diffused into the glass ion-exchanged for 10 min using Paste # A than glass ion-exchanged for the same time using Paste # B.
3.2. Optical properties All stained float glasses are yellowish green or deep green and still they show clearness and transparency. The transmittance spectra of the stained float glass samples are shown in fig. 2. There are two kinds of extinction bands in each transmittance curve located around the wavelengths of 560 nm and 600-660 nm, which are hereafter designated as G (green)- and R (red)-extinctions, respectively. The peak of the R-extinction becomes broadened and shifts to longer wavelengths with increased ion exchange time, while the peak position of the G-extinction seems constant. There is a subtle shoulder-like structure in the transmittance curve of the sample B1 located at about 450 nm, designated as B (blue)-extinction in fig. 2.
o
~6o
soo
600
/nm
76o
800
Wavelength Fig. 3. Parallel transmittance spectra for copper-stained Colburn glass samples. 'C' indicates an absorption edge-like structure described in the text. Dotted line: unstained sample.
In contrast, stained Colburn glasses are bright yellow. Each parallel transmittance spectrum for Colburn glass sample shows only an absorption edge-like structure located around 460 nm, designated as C (Colburn)-extinction (fig. 3). Luminous transmittance is still 63% even after the ion-exchange procedure for 30 min using Paste # A. The remarkable difference in color between stained Co!burn and float glass samples must be associated with the existence of a tin-rich surface layer in }he latter. Diffuse transmittance spectra were measured for all stained samples; however, they were practi0.3
o~ 02
0
' tOO
i i i i 800 ~2'00 Wovelength /nm Fig. 4. Diffuse transmittance spectra for copper-stained float glass samples. In this diagram, transmittance curves for A2 and A3 are shifted 0.05 and 0.1 upward, respectively, in order to avoid intersection of them.
H. Nagao et al. / Mechanism of coloration
203
1.0
Cu=O 111
0.8
c
o.6
o 200
0.~
i0.2
30
35
4~0
Z.~5
50
55
0
28 / deg Fig. 5. X-ray diffraction patterns of copper-stained float glass samples.
cally useless except for the float glass samples shown in fig. 4. One should note a distinct peak in each spectrum curve in fig. 4, the maximum position of which shifts to longer wavelengths with an increase in ion exchange time. The peak positions are rather close to the respective peak positions of the R-extinction seen in fig. 2. The optical properties for all samples are listed in table 4.
3.3. Crystalline species in stained glass The X-ray diffraction patterns for stained samples revealed that crystalline particles of both metallic copper and cuprous oxide were precipitated in copper-stained float glass samples (fig. 5)
Cu=O 111
C
u
=
600
700
800
Wovetength / nrn
and only cuprous oxide in stained Colburn glass samples (fig. 6). Cupric oxide was not detected in any samples. The value of the relative intensity for each diffraction peak is in good agreement with the value which was obtained from a powdered specimen [12]. This result is consistent with the assumption that crystalline colloidal particles are suspended in the glass. Table 5 lists integral intensities after background reduction, and integral breadths of the (111) reflection for both metallic copper and cuprous oxide. The (111) reflection of metallic copper becomes stronger in integral intensity and sharper from sample A1 through A3, which indicates that both the volume and the mean size of Table 5 Integral intensities and integral breadths of 11l reflections in X-ray diffraction for cuprous oxide and metallic copper
O
Sample
"E
35
560
Fig. 7. Effect of annealing for 15 min in nitrogen gas on the transmittance curve of copper-stained float glass. Dotted line: as-stained sample. Solid line: annealed sample.
oJ
30
/,00
~'0
~'5
50
55
20 / deg
Fig. 6. X-ray diffraction patterns of copper-stained Colburn glass samples.
A- 1 A-2 A-3 A-4 A- 5 B- 1
Integral intensity (cps)
Integral width (deg (2 0))
C u 2 0 111
Cu 111
C u 2 0 111
Cu I l l
1.50 × 105 2.06×105 2.82×105 2.63×105 3.80 × 105 7.37 × 104
3.33 × 104 5.00×104 1.02x 105 n.d. n.d. 1.60 X 104
1.182 1.315 1.178 1.719 1.731 0.724
0.567 0.467 0.439 n.d. n.d. 0.547
n.d.: Not detectable.
H. Nagao et al. / Mechanism of coloration
204 1.0
0.8
o~ c o
0.6
o 0.~
i---
0.2
400
500
600
700
800
Wavelength /nm
Fig. 8. Effect of a n n e a l i n g for 15 m i n in the air o n the t r a n s m i t t a n c e curve of c o p p e r - s t a i n e d float glass. D o t t e d line: as-stained sample. Solid line: a n n e a l e d sample.
metal copper colloids gradually increases with increasing ion-exchange time.
3.4. Some changes after post heat treatment The post heat treatment at 600 ° C for 15 min in an atmosphere of nitrogen hardly affected the color of stained yellowish green float glass (fig. 7). Strictly, the B- and R-extinctions became a little stronger after the annealing. On the other hand, the effect of annealing in air of the same sample was rather different (fig. 8). The R-extinction became distinctly enfeebled whereas the G-extinction became a little stronger. As a result, the color turned into brownish green.
4. D i s c u s s i o n
4.1. The colorant in copper-stained float glass Human eyes are, in general, very sensitive to a subtle change in the transmittance curve. The color of stained float glass gets greenish when the R-extinction becomes relatively dominant as compared with the G-extinction, while it becomes reddish in the opposite case. The origin of the extinctions is of great importance for our understanding of the coloration mechanism of copper-stained float glass.
The cuprous ion has the d 1° configuration and, as such, will not impart coloration [13]. A cupric ion shows an absorption band centered at 780 nm in soda-lime silicate glass [13]. The optical properties of such ions cannot explain the optical properties of the copper-stained glass shown in fig. 2. X-ray diffraction measurements suggested that colloidal particles of metallic copper a n d / o r cuprous oxide are possible colorants in the stained float glass. The comparison between float glass samples and Colburn glass samples would be informative concerning the cause for the G- and R-extinctions. The Colburn glass samples show only an absorption edge-like structure located at about 460 nm while the float glass samples exhibit three kinds of extinction structures, inclusive of the G- and the R-extinctions as are shown in figs. 3 and 2, respectively. X-ray diffraction measurements revealed that the integral intensity of the (111) reflection of crystalline cuprous oxide in the former samples is comparable to that in the latter, and that crystalline particles of metallic copper were detected only in the float glass samples (table 4). Thus, it seems unlikely that either the G- or R-extinctions is ascribed to the colloidal particles of cuprous oxide in the glass. It is proposed that only colloidal particles of metallic copper are the effective colorant in copper-stained float glass.
4.2. Nature of metallic copper colloids suspended in glass As a general rule, a colloidal metal particle exhibits both light scattering and absorption attributed to the plasmon oscillations of free-electrons, and such an extinction band, which consists of bands by absorption and scattering, appears in the transmittance spectrum [14]. The transmittance spectra of some glasses in which spherical particles of metallic copper are suspended have been calculated by many researchers [3,8,15,16]. Before we discuss the origin of three extinction structures observed in stained float glass, let us show some calculated curves of the extinction cross-section (Ce×) as well as the scattering crosssection (Csc) and absorption cross-section (Cab)
H. Nagao et aL / Mechanism of coloration 20nm
lb
x
3Onto 40 , 40nm 8C
50nm
~x
20t
o
ex
ex~
E O
O C.>
205
20 nm in radius. However, this band does not appear in the Cex curves for larger particles. It can be concluded from the calculation that the small particles cause extinction bands due to intrinsic absorption at about 430 nm and 580 nm, whereas the larger particles create an extinction band only due to scattering located at about 600 615 nm. -
£c
6 8
04 6 Waveleng h / lOOnm 6
4
6
Fig. 9. Calculated cross-sections for scattering (sc), absorption (ab) and extinction (ex) for spheres of colloidal copper having four different radii. Note different scales for the vertical axis in each figure.
for some colloidal particles of metallic copper based on the Mie theory. We used the subroutine B H M I E [17] to calculate the cross-sections for spherical metal copper particles of 20 - 50 nm in radius. The refractive index of the glass matrix was assumed to be 1.52 and the optical constants of copper were taken from the tabulation in ref. [18]. The calculated curves of Cex, Cs~ and Cab are shown in fig. 9, which demonstrates some informative features. (1) Both Cs~ and Cab become gradually stronger as the particle increases in size, and the enlargement in Csc is much more remarkable than that for Cab. The scattering cross-section is practically negligible as compared with C,b for a particle of 20 nm in radius, but it increases to be roughly twice that of Cab for a particle of 50 nm in radius. (2) As the peak position of Cs~ is always longer in wavelength than Cab, the peak position of Cex is always shorter in wavelength than that of Cs~. (3) The locations of maxima of both C~ and C,b shift gradually to longer wavelengths as the particle size increases, and the change in peak position for C~c is much more rapid than that for Cab. The peak positions for Csc and Cab are about 580 nm and 590 nm in the case of a particle of 20 nm in radius, and they shift to about 585 nm and 615 nm in the case of a particle of 50 nm, respectively. (4) There is another extinction band located around the wavelength of 430 nm for a particle of
4.3. Coloration mechanism of copper-stained float glass The X-ray diffraction analyses demonstrated that the mean particle size of metallic copper colloids gradually increases as the ion-exchange time increases. If the effective colorant in copperstained float glass is regarded as metallic copper colloids, the systematic change in transmittance curve with increasing ion-exchange time should be ascribed to the increase in particle size of colloids (without taking account of the wide distribution of particle size). The positions of the diffuse transmittance maximum shown in fig. 4 are rather close to the locations of the respective minimum in the R-extinction in fig. 2. They are both characterized by the gradual shift to longer wavelengths in peak position with some increase in peak intensity. Table 6, which lists the positions of diffuse transmittance maxima and transmittance minima of the R-extinction on parallel transmittance curves for copper-stained float glass samples, shows that the former position is always longer in wavelength than the latter. All these data are consistent with the assumption that the R-extinction is mainly
Table 6 Positions of the maximum in diffuse transmittance curves (Dm~x) and those of the minimum of the R-extinction in parallel transmittance curves (Train) for copper-stained float glass samples Sample no.
Ion-exchange time (min)
Tmi. (nm)
DmaX (nm)
A-1 A- 2 A- 3
6 10 30
625 630 660
655 700 830
206
H. Nagao et al. / Mechanism of coloration
ascribed to the scattering by relatively large particles of metallic copper colloids in stained glass. O n the left side of each diffuse transmittance curve in fig. 4, there are two kinds of fine structure located a r o u n d 450 n m and 560 n m (arrows in fig. 4), which implies the existence of absorption bands. These structures can be correlated with the B- and the G-extinctions as their positions coincide with positions of the extinctions. These data are consistent with the idea that b o t h the G- and B- extinctions are chiefly attributed to the intrinsic absorption by relatively small particles of metallic copper colloids in stained glass. As a result, extinctions in the transmittance spectra of copper-stained float glass are believed to originate f r o m colloidal particles of metallic copper precipitated in glass surface regions. The change of transmittance curves by post-annealing in air seems to be reasonable on the basis of the assumption that metallic copper colloids decrease in mean size during the annealing. It implicitly results in both the development of the G-extinction and attenuation of the R-extinction. We suppose that the most probable m e c h a n i s m for the decrease in mean size of metallic c o p p e r colloids during post annealing is their oxidation. It is important to mention that the reactions in the glass during staining is severely influenced by reducing agents such as tin in the glass. The effects of reducing agents on the color of copperstained float glass is to be discussed in another paper. The authors are indebted to Dr T. Y o k o of K y o t o University for some valuable discussions,
Mr K. K o m a k i for the E P M A analysis, and Rigaku C o r p o r a t i o n for the X - r a y diffraction analysis.
References [1] W.A. Weyl, Coloured Glasses (Society of Glass Technology, Sheffield, 1951). [2] C.R. Bamford, Color Generation and Control in Glass (Elsevier, Amsterdam 1977). [3] H. Rawson, Phys. Chem. Glass. 6 (1965) 81. [4] O.S. Levi, Bull. Am. Ceram. Soc. 34 (1965) 119. [5] S. Sakka and T. Nishiyuki, J. Non-Cryst. Solids 37 (1980) 139. [6] S. Sakka, T. Nishiyuki, M. Nishitomi and K. Kamiya, Res. Rep. Fac. Eng. Mie Univ. 5 (1980) 85. [7] S. Sakka, K. Kamiya and K. Kato, J. Non-Cryst. Solids 52 (1982) 77. [8] C.R. Bamford, Phys. Chem. Glass. 17 (1976) 209. [9] A. Ram and S.N. Prasad, Advances in Glass Technoogy, Technical Papers of the VI Int. Congress on Glass (Plenum, New York 1962) p. 256. [10] S. Banerjee and A. Paul, J. Am. Ceram. Soc. 57 (1974) 286. [11] P.W. Baumeister, Phys. Rev. 121 (1961) 359. [12] American Society of Testing Materials, X-ray powder data file (Inorganic), Set 1-5 (1960). [13] C.R. Bamford, Phys. Chem. Glass. 3 (1962) 189. [14] H.C. Van de Hulst, Light Scattering by Small Particles. (Wiley, New York, 1957). [15] D.M. Trotter, J.W.H. Schreurs and P.A. Tick, J. Appl. Phys. 53 (1982) 4657. [16] R. Ruppin, J. Appl. Phys. 59 (1986) 1355. [17] C.F. Bohren and D.R. Huffman, Absorption and Scattering of Light by Small Particles (Wiley, New York, 1983) [18] P.B. Johnson and R.W. Christy, Phys. Rev. B6 (1972) 4370.