S. Eaton-Magana,1,* J. E. Post,2 P. J. Heaney,3 J. A. Freitas, Jr.,1 P. B. Klein,1 R.A. Walters,4 J. E. Butler1
1 Naval Research Laboratory, Washington, D.C. 20375
2 Department of Mineral Sciences, Smithsonian Institution, Washington, D.C. 20560-0119.
3 Department of Geosciences, Penn State University, University Park, PA 16802
4 Ocean Optics, Inc., Dunedin, FL 34698
Sixty-seven natural blue diamonds, including the two largest such gemstones known (the Hope and the Blue Heart), were stimulated by ultraviolet radiation, and their luminescence were analyzed. Prior to this study, the dramatic red phosphorescence of the Hope Diamond has been believed to be quite rare compared to greenish-blue phosphorescence. However, through phosphorescence spectroscopy, we observed that this red phosphorescence is quite common though often obscured. Virtually all specimens showed phosphorescence bands centered at two wavelengths: 500 and 660 nm. The relative intensities of these emissions and their decay kinetics varied significantly between diamonds, perhaps providing the first inexpensive, non-destructive, and reproducible means of discriminating among individual blue diamonds. Treated and synthetic blue diamonds showed behavior distinct from natural stones. Temperature-dependent phosphorescence revealed that the 660 nm emission has an activation energy of 0.4 eV, close to the 0.37 eV acceptor energy for boron providing an indication that the phosphorescence is caused by donor-acceptor pair recombination. INTRODUCTION
Blue diamonds are among the rarest and most valuable of all gems, and the Hope Diamond is its most famous member. At 45.52 ct, the Hope is the largest deep-blue diamond known, and its turbulent history can be traced back nearly four centuries. Today the Hope is the focal point of the Harry Winston Gallery in the United States National Gem Collection. There it attracts millions of visitors each year, exceeding in popularity any other object in the vast repository of the Smithsonian Institution. In fact, it is likely that more people view the Hope Diamond than any other museum piece in the world.
Most visitors, however, do not have an opportunity to observe the fiery red, long-lasting phosphorescence of the Hope after exposure to ultraviolet (UV) light (Fig. 1). This behavior was believed so rare that other blue diamonds with red phosphorescence have been represented as coming from the same parent diamond as the Hope (Fritsch and Scarratt, 1992). Little scientific research exists on the phosphorescence properties of natural blue diamonds, and much is qualitative as only visual observations are provided (e.g., King et al., 1998; King et al., 2003) or focused on the dependence of luminescence with temperature (e.g., Halperin and Chen, 1966). These previous investigations have suggested that the majority of blue diamonds phosphoresce chalky blue to green.
In this study we investigated the luminescence properties of 3 synthetic, 1 treated, and 67 natural blue diamonds, including the Hope; the Blue Heart (at 30.62 ct the second-largest deep-blue diamond known and also housed in the U.S. National Museum of Natural History); and a number of diamonds selected from the Aurora Butterfly and the Aurora Heart collections of colored diamonds. The simultaneous presence of such a large number of rare diamonds in one location provided an extraordinary, and perhaps unique, opportunity to explore the properties of blue diamonds. In addition to providing insights into the cause of luminescence in blue diamonds, this study also has suggested a simple and robust method for identifying individual specimens.
A method that could fingerprint and uniquely identify a diamond, whether loose or mounted in jewelry, would have several applications within the gem industry. As examples, any blue diamond that was stolen, submitted for jewelry repair, or cut into smaller stones would benefit from such a technique. In the last case, fingerprinting would allow the daughter stones to be traced to the parent. Perhaps just as important, the need to distinguish natural blue diamonds from samples that have been treated or synthesized has acquired greater urgency as high-pressure, high-temperature (HPHT) treatments increase in sophistication, and as a range of new techniques to fabricate colored diamonds are showing reproducible success. Any technology for fingerprinting gemstones has to satisfy three criteria: It must be quantitative, non-destructive, and portable for use in a variety of locations.
The blue color in diamonds generally arises from the incorporation of boron at trace (ppb to ppm) levels and relatively low concentrations of nitrogen impurities (below detection using infrared [IR] spectroscopy). Boron impurities are also electrically active with an acceptor state at ~0.37 eV above the valence band. Since natural diamonds are quite rare, most research on the optical properties of boron-containing diamonds has been conducted on synthetic stones (Zaitsev, 2001), which may be fabricated quite easily.
Most of the diamonds used in this study were graded by gemological laboratories and are believed to be natural and untreated. The three synthetic diamonds examined in this study were grown at HPHT conditions and were colored dark blue by doping with boron (B. Feygelson, pers. comm.., 2005). The Aurora Butterfly is a collection of 240 loose colored diamonds (total weight of 166.94 ct) that was on temporary display at the Smithsonian Institution. The Aurora Heart collection consisted of 212 colored diamonds. Since diamonds that were color-graded as gray or having a gray component gave similar phosphorescent results, they were grouped with the blue diamonds in this study. That was not surprising since diamonds with cool colors (such as blue) tend to appear gray when the color is at low saturation. In contrast, diamonds with warm colors (such as pink) tend to appear brown at low saturation (King et al., 1998).
Phosphorescence at room temperature was excited by broad-band UV radiation that was delivered by a portable desktop instrument that included a deuterium lamp coupled to a fiber optic assembly, with a high sensitivity spectrometer to record the spectra. The spectrometer (Ocean Optics USB 2000) measures wavelengths from 340–1020 nm and has a 200 m slit width and were calibrated using the 435.8 and 546.1 nm mercury lines. Due to equipment limitations, we measured the phosphorescence beginning about 1 s after the light source was extinguished in most experiments.
The phosphorescence spectra of a rough natural blue diamond were measured at 10°C temperature intervals ranging from 25°C to 155°C in air. The temperature of the diamond was maintained constant as the phosphorescent decay was recorded following UV excitation.
In addition, two natural blue diamonds and two boron-doped HPHT synthetic diamonds were characterized by steady-state and time-resolved laser-induced photoluminescence (PL) spectroscopy. CL measurements were performed at 10 kV with a beam current of 2 A. The beam diameter was focused to a spot size of either 0.1 mm or 1 mm. The continuous wave PL measurements were performed using the 325 nm line of a He-Cd laser. The diamond samples were positioned (table-side up) on a helium-cooled cold finger. Steady-state PL measurements were recorded using photon counting and a Spex 1404 0.85 m double monochromator fitted with a 1200 grooves mm-1 grating blazed at 350 nm or 1800 grooves mm-1 blazed at 450 nm. The spectra were corrected for the wavelength dependence of the optical system. The pulsed PL measurements were performed at room temperature using 355 nm light from a Q-switched Nd:YAG laser, a 0.22 m double monochrometer, GaAs photomultiplier and time-correlated photon counting.
Our investigations revealed that red phosphorescence is, in fact, a hallmark of most blue diamonds, but the color is often masked by luminescence in the green-blue region of the visible spectrum. Of the 67 natural blue diamonds examined in this study, 62 exhibited two phosphorescence peaks. One peak had a maximum at ~500 nm (greenish blue); the other emission occurred in the red portion of the spectrum, at ~660 nm (Fig. 2A). The observed variation in wavelength for these two peaks was quite small, ±3 nm.
Visual examination of the color change during the phosphorescent decay of the Hope Diamond revealed a distinct reddening with increasing time. The spectra shown in Fig. 2A confirm and quantify this observation. Although both peaks were detectable at the onset of phosphorescence, it can be seen from the figure that for the Hope, the 500 nm peak decayed much more rapidly than the 660 nm band for the Hope, intensifying the red appearance by the fading of the blue-green. Contrasting behavior is illustrated in Fig. 2B which depicts a phosphorescence spectrum for a gray-blue diamond. To the eye, this diamond phosphoresced blue-green to white, and the spectrum demonstrates that the 500 nm peak predominated over the 660 nm band. A secondary peak at ~440 nm also is evident as a shoulder on the 500 nm band. When the 500 nm emission predominated, as was the case for the majority of the natural diamonds examined, the phosphorescence appeared greenish blue, but in the less common cases when the 660 nm emission was stronger (as with the Hope), the phosphorescence appeared orangy red to the eye. Thus, the variation in the relative intensities of these two emission bands accounts for the wide variation in phosphorescence observed in blue diamonds. In some diamonds, the observed phosphorescence changed from greenish blue to red at longer decay times.
The decay of the phosphorescence was characterized in terms of half-lives (i.e., the time for the intensity to diminish by 50%). For the 500 nm peak, the measured half-lives ranged from 0.5 to 6 seconds. For each diamond, the 660 nm peak always exhibited a longer half-life with values ranging from 3 to 30 seconds. The spectral line-shapes and widths did not differ significantly among the diamonds studied. When illuminated by short-wave UV radiation (254 nm) rather than the broad UV source used in most of the experiments, the diamonds emitted at both 500 nm and the 660 nm. Using a long-wave UV source (351 nm), however, only the 660 nm peak was observed.
When excited by the broad UV source, the resulting ratios of the peak intensities of the 500 relative to the 660 nm band may be plotted against the half-life for each specimen and virtually every sample displayed a unique combination of parameters (Fig. 3). Moreover, the spectrometer used in our study was relatively inexpensive and highly portable. Thus, this technique can be easily replicated by other laboratories.
While these results suggest an effective optical technique for identifying individual diamonds, several details need to be standardized for measurement. The sensitivity of the intensity ratios in Fig. 3 to excitation wavelength and intensity has not yet been fully determined. However, such procedures may be easily specified to provide uniformity among laboratories.
In contrast with natural blue diamonds, the 3 boron-doped synthetic diamonds exhibited no phosphorescence band at 660 nm and only phosphoresced at 500 nm. In one deep-blue synthetic diamond, an additional orange peak centered at ~575 nm was apparent (Fig. 2C). The intensity of the emission in these synthetic stones was the highest of any diamonds studied, but the half-lives of the 500 nm peak were comparable to those of natural stones. Additionally, we examined a natural gray diamond that had been treated by HPHT-annealing to turn it blue. Like the synthetics, this diamond phosphoresced only at 500 nm. It was the only blue/gray diamond of natural origin to phosphoresce at 500 nm and not at 660 nm; this result recently was confirmed by visual observation (Breeding et al., 2006). Based on these data, phosphorescence spectroscopy also appears to be an effective tool for discriminating synthetic and HPHT-treated diamonds from natural blue diamonds.
The utilization of the 500 nm and 660 nm luminescence intensities as a fingerprinting technique appears to be applicable for most, but not all blue diamonds: five of the 67 natural blue diamonds differed from the majority. Two showed no phosphorescence and three exhibited yellow fluorescence and green phosphorescence; all other blue diamonds showed negligible fluorescence. While in these few cases the relative intensities of the blue-green and red emission bands are not available to provide a well-defined fingerprint, the specific luminescence behavior in the other spectral regions, and even the lack of an emission signature, may still be employed to help identify a particular stone. Most likely, these few diamonds should not be classified as Type IIb diamonds (i.e., that they did not contain boron) but rather were Type Ia, H-rich diamonds and their bluish color is caused by other impurities (Fritsch and Scarratt, 1992). However, due to experimental restrictions, such a determination could not be made.
Representative data of the steady-state PL experiments performed on a few natural and synthetic diamonds are shown in Fig. 4. The HPHT diamonds show similar characteristics to prior PL experiments on synthetic stones (see, e.g., Klein et al., 1995). There are two broad peaks centered at 1.85 eV (670 nm) and 2.3 eV (540 nm). The natural blue diamond has a single luminescence peak with maximum at ~3.0 eV (413 nm).
Time-resolved PL spectroscopy revealed how the emission spectra evolve as a function of time. Despite the fact that the 500 and 660 nm peaks in phosphorescence spectra of natural diamonds are not observed in steady-state laser-excited PL measurements, both peaks, as well as bands at 440 and 580 nm, were visible in pulsed PL spectra of natural blue diamonds (Fig. 5). At the later times, the intensities of the 440 and the 580 nm peaks weakened significantly, such that by 20 msec, only the phosphorescent 500 and 660 nm peaks could be observed. Boron-doped synthetic diamonds exhibited a significantly different behavior. Unlike natural diamonds, the 575 nm peak in the synthetic diamond persisted sufficiently to be observable in the phosphorescence spectra (Fig. 2C), but the 660 nm luminescence was short-lived and decayed at much shorter time scales. Although this band is clearly seen in steady-state PL for synthetic diamonds (Klein et al., 1995), it was observed to decay after 5 s, and after 1000 s it was not detectable.
In order to monitor the thermal response of the phosphorescence, emission spectra were recorded for a rough natural blue diamond at 10°C intervals between 35°C and 155°C (Fig. 6A). The resulting Arrhenius plot in Fig. 6B (log intensity vs. inverse temperature) reveals exponential behavior for both bands. Fitting the data using I(T)[1+ Aexp(-Et/kT)]-1, where T is temperature, A is the Arrhenius constant, Et is the thermal activation energy, and k is the Boltzmann constant yields an activation energy of about 0.4 eV for the 660 nm peak, close to the 0.37 eV binding energy of the neutral boron acceptor.
What is the origin of the phosphorescence in blue diamonds? The variability in the intensities and decay times of the 500 and 660 nm peaks implies some differences in the concentrations of the impurities and other defects that create the excited states. We have insufficient evidence to completely describe the defect states, impurities, or energy transfer mechanisms responsible for the observed phosphorescence in this work. Nevertheless, our results do constrain the possible mechanisms. Donor-acceptor pair recombination (DAPR) is reviewed in depth by Dean et al. (1965), and it supposes that holes bound to acceptors recombine with electrons that are bound to donors and emit light with an energy that is approximately equal to the difference in energy level between the donor and acceptor.
Numerous authors have hypothesized that DAPR is the mechanism responsible for a variety of bands observed in phosphorescence and laser-induced PL in synthetic diamonds produced by chemical vapor deposition (Dischler et al., 1994) and HPHT conditions (Klein et al., 1995; Watanabe et al., 1997). To our knowledge, this is the first analogous study on natural blue diamonds.
The presence of boron, and therefore a likely candidate as the acceptor in the DAPR mechanism, was confirmed in some of the natural blue diamonds by the presence of boron-related peaks (not shown; Zaitsev, 2001). Prior tests of the Hope Diamond using FTIR spectroscopy confirmed the presence of boron-related peaks (King et al., 2003).
The temperature dependence of DAPR depends on whether the emission itself or the formation of the neutral donor-acceptor pair is rate limiting; in the latter case, the thermal ionization of the weaker of the donor or acceptor will be the rate-limiting step. The observation that the 660 nm band yields an activation energy of approximately 0.4 eV (Fig. 6B), close to the 0.37 eV acceptor energy of boron, suggests that this band is caused by DAPR. The band at 500 nm is also likely boron-related (Watanabe et al., 1997; Won et al., 1996), but the high activation energy of 0.8 eV (for reference, the bandgap of diamond is 5.5 eV) suggests that thermal activation of boron is not the rate-limiting step. At long time scales (order of seconds), the excitation mechanism of carrier migration between distant neutral and ionized boron acceptors is likely (Watanabe et al., 1997). In Fig. 6, the observation that the intensity appears to decrease more rapidly at the highest two temperatures is consistent with the beginning of thermalization of the much deeper donor level involved in DAPR.
Examination of the luminescence spectra taken under steady-state conditions (Fig. 4), intermediate time scales (order of milliseconds; Fig. 5) and long time scales (order of seconds; Fig. 1) shows that the dominant peaks change with time. For example, in natural diamonds, the band centered at 660 nm is not observed in Fig. 4, seen to develop in Fig. 5 as more dominant but faster-decaying peaks, and may still be observed by phosphorescence. Conversely, a similar peak at 670 nm is observed in synthetic diamonds under steady-state PL conditions (Fig. 4), decays on the μsec–msec time scale and is not observed by phosphorescence (Fig. 2C).
In addition, a band at ~575 nm is observed in synthetic diamond (Fig. 1C) by phosphorescence, which has been observed by other researchers (Klein et al., 1995; Watanabe et al., 1997). A similar band is seen in natural diamond at ~580 nm in natural diamond at shorter times scales (Fig. 5). It is tempting to speculate that the peaks at ~660 and ~580 nm observed by the two measurements and at different time scales have an identical origin, and similar values for FWHM seem to suggest that fact. The difference in decay times between natural and synthetic diamonds might be due to different concentrations of defects.
Although the 660 nm emission may be attributed to DAPR, the specific form of the donor is unknown. Due to the energy of the observed luminescence, the donor energy associated with a DAPR mechanism with boron as an acceptor is constrained to the approximate range of 2.9–3.7 eV (see Klein et al., 1995 for additional information of calculation). The donor energy of substitutional nitrogen is 1.7 eV, which is located too high in the bandgap to be accountable, and the nitrogen A aggregate has a donor energy of 4 eV which is too low in the bandgap. However, prior research has indicated that substitutional nitrogen may undergo strong lattice relaxation, with an energy of about 1 eV, which would bring its recombination energy much closer to the experimental range (Nazare et al., 1995).
Breeding C.M., Wang W., Shen A.H., McClure S.F., Shigley J.E., and DeGhionno D., 2006, High-energy ultraviolet luminescence imaging: Applications of the DTC DiamondView for gem identification: Gems & Gemology v. 42, no. 3, p. 88.
Dean P.J., Lightowlers E.C., and Wright D.R., 1965, Intrinsic and extrinsic recombination radiation from natural and synthetic aluminum-doped diamond: Physical Review, v. 140, no. 1, p. A353–A386.
Dischler B., Rothemund W., Maier K., Wild C., Biebl H., and Koidl P., 1994, Diamond luminescence – resolved donor-acceptor pair recombination lines: Diamond and Related Materials, v. 3, p. 825–830.
Fritsch E., and Scarratt K., 1992, Natural-color nonconductive gray-to-blue diamonds: Gems & Gemology, v. 28, no. 1, p. 35–42.
Halperin A., and Chen R., 1966, Thermoluminescence of semiconducting diamonds: Physical Review, v. 148, no. 2, p. 839–845.
King J.M., Moses T.M., Shigley J.E., Welbourn C.M., Lawson S.C., and Cooper M., 1998, Characterizing natural-color type IIb blue diamonds. Gems & Gemology, v. 34, no. 4, p. 246–268.
King J.M., Johnson E.A., and Post J.E., 2003, A comparison of three historic blue diamonds. Gems & Gemology, v. 39, no. 4, p. 322–325.
Klein P.B., Crossfield M.D., Freitas J.A. Jr., and Collins A.T., 1995, Donor-acceptor pair recombination in synthetic type IIb semiconducting diamond: Physical Review B, v. 51, no. 15, p. 9634–9642.
Nazare M.H., Mason P.W., Watkins G.D., and Kanda H., 1995, Optical detection of magnetic resonance of nitrogen and nickel in high-pressure synthetic diamond: Physical Review B, v. 51, no. 23, p. 16741–16745.
Watanabe K., Lawson S.C., Isoya J., Kanda H., and Sato Y., 1997, Phosphorescence in high-pressure synthetic diamond: Diamond and Related Materials, v. 6, no. 1, p. 99–106.
Won J., Hatta A., Ito T., Sasaki T., and Hiraki A., 1996, The investigation of defect formation in chemical vapor deposited diamond using photoluminescence by ultraviolet synchrotron radiation: Applied Physics Letters, v. 69, no. 27, p. 4179–4181.
Zaitsev A.M. 2001, Optical Properties of Diamond: A Data Handbook. Springer-Verlag, Berlin.
We thank Alan Bronstein of Aurora Gems for allowing us to perform these measurements on the diamonds in his collections, Tom Moses and Wuyi Wang of the Gemological Institute of America in New York City for the use of their DiamondView microscope, and Russell Feather, Gem Collection Manager at the Smithsonian Institution for his assistance. We are grateful to the Steinmetz Group for providing the rough natural blue diamond used in the PL experiments. Financial support was provided by the Office of Naval Research and the Naval Research Laboratory. S. E.-M. is a recipient of an NRC Fellowship Award.
Figure Captions Fig. 1. The 45.52 ct Hope Diamond is the most famous example of a blue diamond showing red phosphorescence. The blue color of the Hope Diamond is caused by boron impurities (King et al., 2003). Photo by John Nels Hatleberg.
Fig. 2. Phosphorescence spectra for the Hope Diamond (A), a gray-blue diamond from the Aurora Butterfly collection, B29 (B), and a dark blue synthetic diamond, HPHT-C, (C). In (A), the large peak at 660 nm accounts for the strong red phosphorescence that is observed. The peak at 500 nm plays a lesser role for this particular blue diamond. The half-life, , for the 660 nm peak is 9.2 sec and 1.8 sec for the 500 nm peak. In (B), is 12.7 sec and 2.5 sec for the 500 nm peak. In (C), is 0.82 sec for the 500 nm peak and 1.4 sec for the 575 nm peak. For all samples, the diamond was illuminated for 20 seconds. The data were taken at 1 second integration times.
Fig. 3. The ratio of initial intensities of the 500 nm and 660 nm bands are plotted against the measured half-life of the 660 nm emission for the natural gray-to-blue diamonds. Therefore, for y-axis values greater than 1, the greenish-blue band dominated and conversely for values less than 1, the red band dominated. For diamonds that had ratios of ~1, the observed phosphorescence would appear whitish.
Fig. 4. Photoluminescence at 5 K from a boron-doped synthetic diamond and a natural blue diamond using a 325 nm (3.82 eV) laser and an 1800 mm-1 grating blazed at 350 nm. In addition to the luminescence bands, there are also the first and second order, one, two, and three phonon Raman bands of diamond. The sharp peaks at 2.54 and 2.56 eV in the synthetic diamond are likely related to nickel incorporation during growth (Zaitsev, 2001).
Fig. 5. The PL decay of natural blue diamond is excited by a 355 nm laser at room temperature in the time range of 2–150 msec. At shorter delay times (e.g., 5 s, not shown), the dominant feature is a broad peak centered at about 580 nm. In this time frame shown, this peak appears to shift to 500 nm. The shift is only a consequence of the 580 nm peak decays at a faster rate than the other emissions, which leaves only the 500 nm and 660 nm peaks at the later times. The 440 nm emission is seen to decay on the millisecond time scale.
Fig. 6. A natural blue diamond shows phosphorescence at elevated temperatures. (A) Heating the diamond reduces the phosphorescence of the broad band centered at about 660 nm. (B) The integrated intensity of the phosphorescent decay for the 660 nm band shown in (A) along with similar data for the 500 nm band are plotted against inverse temperature on semi-logarithmic coordinates. The straight-line approximations yield activation energies of approximately 0.4 eV and 0.8 eV for the 660 nm and 500 nm bands, respectively.