The spectral features, i.e., wavelength and intensity, of fluorescence generated from semiconductor nanocrystals (quantum dots) can be used for coding information. Unlike the 1-D and 2-D barcodes, the information carrier is applied to a very small area and hardly visible. The information retrieving by a fluorospectrometer is not subjected to the changes of rotation and scale. A de-convolution-based algorithm is used to separate the overlapped spectral profiles. This technology can be applied to small products labeling, document security and object identification.
©2004 Optical Society of America
The barcodes and the associated readers are the most prevailing technologies for the object identification. Since the barcode needs space to arrange the ordered data either in 1D bar-sequence or 2D image, and thus the barcode reader has to scan the 1D bar-sequence or register the 2D image, the information retrieving systems are bulky and complicated. In addition, the visible printed pattern of a barcode is vulnerable to counterfeiting. In the viewpoint of security and simplicity, an information carrier with a tiny spatial dimension and invisibility to human eyes will be of great advantage. If the information can be encoded inside the carrier in a way that is totally position- and rotation-insensitive, i.e. neither sequence nor pixel based, it will simplify the design of decoder or reader.
It is proposed that the wavelength and intensity of fluorescence generated from multiple fluorescent substances be used for coding information. Organic dyes and metal complexes are commonly used as fluorescent materials in various application areas. In principle, they are applicable to the proposed multiplexed spectral coding technology (e.g., using multiple wavelengths and multiple intensities). However, they generally have inadequate adsorption and emission properties. Different exciting light wavelengths are needed to excite a mix of multiple fluorescent molecules and the emission spectra are often broad and asymmetrical, making the information retrieving difficult. Among other problems, there are also possible interactions between two different fluorescent molecules and the immiscibility of the multiple fluorescent molecules in a common matrix material or solvent. From the technical point of view, an ideal set of luminescent substances should have the following properties in order to meet the multiplexed coding.
• Strong, single wavelength or mono-dispersed narrow emission for each individual luminescent substance;
• Emission spectrum independent of the exciting light in certain range of the exciting wavelength;
• Single light source for all luminescent substances used in the mixture;
• No interaction (energy transfer) among different fluorescent molecules, i.e., each luminescent substance responds to the exciting light independently;
• No influence of matrix material on the emission features;
• Good miscibility of all luminescent substances in the selected matrix material/solvent systems;
• Easy to modify both the intensity and wavelength of fluorescent spectrum.
Due to their size-tunable optical properties originated from the quantum confinement effect, the recently available quantum dots (QDs)  have demonstrated many of the above-mentioned characteristics and thus offer applicability to the multiplexed optical coding technology as demonstrated in bioanalytical application [2–4]. QDs are semiconductor nanocrystals of 1–10 nm in size. The semiconducting materials are selected from, preferably but not limited to, the Group IIB and Group VIA in the periodic table of the elements, such as cadmium selenide (CdSe), cadmium sulfide (CdS), zinc selenide (ZnSe) and zinc sulfide (ZnS). QDs can be made from a single compound, such as CdSe or ZnS, or from multiple compounds in a specific manner such as CdSe-ZnS core-shell configuration [5,6]. For the same materials system, the smaller the QD particles, the shorter the emitted fluorescent wavelength. For example, CdSe nanoparticles with a nominal diameter of 2.8 nm show the fluorescence at 535 nm, while CdSe QDs of 5.6 nm diameter have an emission centered at 640 nm . QDs of lead selenide (PbSe) of various diameters can emit fluorescence in the near-infrared range. A dedicatedly designed mixture of QDs with different emission wavelengths can emit light with spectral features that represent a set of data.
The idea of using QDs for spectral coding has been demonstrated in the bioanalytical area [2–4]. Here we report how to expand this technology to non-biological domains. Particularly, we show a prototype system capable of retrieving the information coded with QDs on the surface of an object, including the passport page, ID card and even a nail of human finger.
2. Preparation of info-ink
QDs are generally prepared via sophisticated solution chemical processes and stored in specific solvents to prevent the aggregation and precipitation [1,5–7]. The commercial availability of QDs with different wavelengths provides a great number of combinations of wavelength and intensity. For example, an encoder using 6-wavelength and 10-intensity scheme has a theoretical coding capacity of about one million discrimination codes . The coding capacity can be even expanded by utilizing a third parameter (e.g., 1D sequence or 2D array of QD beans). To be able to use QDs for the spectral coding of non-biological objects such as banknotes, passport, certificate and other valuable documents, a paintable or printable QDs/polymer/solvent system, namely info-ink, is needed. The info-inks, consisting of polymer(s), solvent, multiple QDs and other additives are applied on the objects that need to be coded. A hybrid optic-electronic-digital system is used to extract the data from the emitted spectra. The detailed description is given in Section 3.
Polymers are used in the info-ink as matrix materials, in which a mixture of QDs with predefined emission features are distributed homogeneously  after the info-ink is applied to a target surface and become dried. Polymers should not have quenching effect on the fluorescence of QDs and must meet other requirements such as solubility in selected solvents, long-term environmental stability, good compatibility and miscibility with QDs.
The commercial CdSe nanocrystals (Core Evidots from Evident Technologies ) have sizes in the nanometer range and are well dispersed in toluene, which is also a good solvent for polystyrene (PS). Therefore, the info-ink prepared in this preliminary work consisted of CdSe nanocrystals, high molecular weight PS and toluene (both from Aldrich) only. CdSe QDs with five wavelengths, i.e., 535±10, 560±10, 585±10, 610±10, 640±10 nm, were selected for creating the predetermined spectral patterns that represent the coded data. According to their individual emission intensity, different ratios of QDs were mixed with certain amount of PS and toluene to form the info-inks with required emission spectral features and viscosity for testing.
Figure 1 illustrates schematically the designed samples of info-inks consisting of three different QDs with different emission wavelengths. Adjusting the amount of the QDs can produce a series of 3-digital codes.
3. The configuration of the hybrid retrieving system
A system for retrieving the information hidden in a tiny spot of dried info-ink is shown in Fig. 2. The exciting light is provided by a 370 nm LED light source. A bunch of optical fiber guides the exciting light to the info-ink spot applied on the surface of any object. The fluorescence emitted from the QDs is collected by the detecting fiber in the fiber bunch and fed to a spectrometer. The data generated by the spectrometer is further delivered to an intelligent instrument, e.g., a micro-processor or a PC, which eventually extracts the information originally coded in the info-ink. In order to obtain an even exciting light, the exciting fibers are arranged to surround the detecting fiber evenly to form an optical fiber bundle, as shown in Fig. 2. A rubber cup is connected at the end of the fiber bundle to ensure that only the excited fluorescent light can enter the detecting fiber.
4. Spectrum signal processing
The raw data collected from spectrometer are processed by the intelligent instrument. The processing consists of the following steps.
1) removing the noise by a digital filter;
2) separating the spectral center lines emitted by info-inks from the overlapped spectra;
3) finding the wavelengths (Ws) and intensities (Is) of all the spectral center lines,
4) calibrating these Ws and Is, and retrieving the original data according to a prior known code-book.
Figure 3 shows a fluorescence spectrum measured from an info-ink containing only QDs with emission peak at 535 nm. Because of the Gaussian-like profile, the neighboring spectral profiles may mutually affect the intensity of each other in an info-ink emitting multiple wavelengths, i.e., introduce spectral alias, as shown in the example of Fig. 4. The acquired spectrum is the top black curve, which is actually composed of spectrums from two different QDs, represented by the light and dark gray curves in the figure. As the dark one is only about 1/5 of the light one in intensity, its peak could not be distinguished from the input spectrum. This effect will eventually result in a decoding error if no measure is taken.
A spectrum function of info-ink can be described as
where δ(λ) represents an impulse function, physically, a spectral line. ki is the intensity of a δ(λ) at λi, p(λi) denotes the profile function centered at λi. ⊗ represents a convolution operation. As described above, the broad wavelength profile of the info-ink is the main reason for the spectrum alias. To get rid of the alias effect, the spectrum line must be separated.
The Fourier transform, FT[ ], of Eq. (1) is given by
Because the f(λ) or F(u) is the measured input data, FT[f(λ)]=F(u), and all the p(λi)s or Pi s are known functions, the accurate non-profile spectral line at λm can be restored by
When all the Pi s are the same, i.e., Pi=Pm, Eq. (4) becomes
The above Eq. (5) yields a serial of δ(λ)s, indicating that all the spectral lines are extracted and separated as individual impulses. However, as each spectrum profile of info-ink is actually different from others, the de-convolution operation can only extract one narrow sharp impulse, like the km δ(λ-λm) in Eq. (4). To find all spectral lines, ki δ(λ-λi) i=1.. N, N times operations are needed.
An experimental result of this procedure is illustrated in Fig. 4, in which the info-ink has center wavelengths at 611 and 632 nm, and intensities 1 and 0.2, respectively. The spectrum of the info-ink is presented by the black bold line in the figure. Because the emission at 632 nm is weak, there is no noticeable peak at 632 nm in the spectrum. After two operations of spectrum line extraction, two spectral lines are obtained as shown in Fig. 4. With all the extracted λis and kis, the original data sequence can be retrieved eventually based on a prior decode book. A procedure for calibrating the λis and kis may be required before the decoding.
In our experiments, we used a fiber optic spectrometer made by Ocean Optics . Comparing with conventional spectrometers, this device dramatically reduces the size and cost and is easy-of-use.
We have described a method for encoding information using fluorescent info-inks containing fluorescent substances with well-defined emission spectra. A key to the successful application is the emission features and the long-term stability of the fluorescent substances. We found the excitation conditions and the emission features of QDs meet our requirement very well. However, the long-term stability of the QDs, particularly, their emission features as a function of time and environment is yet to be established. A possible factor that leads to the change of the fluorescence spectra with time is the slow aggregation process of QD particles in polymer matrix. However, it is possible to prevent this process by cross-linking the polymer (e.g., by UV) after the info-ink is applied to the targeted surfaces. A mixture of solvents and other additives may be used to improve the properties such as solubility, viscosity, volatility, storage stability and adhesion etc., so that the info-ink can be applied to surfaces with different chemical natures.
The de-convolution procedure is adopted basically due to its ability of narrowing and separating the overlapped neighboring spectrums. However, if the spectrum shape changes, this procedure will introduce some measurement error. Fortunately, our experiments and computer simulations show that the spectrum profiles of info-ink have little change before and after the blend with others. Although many algorithms of signal detection and spectrum evaluation are available , they are basically designed for the purpose of improving the signal to noise ratio, which is put at the lower priority comparing to the spectrum overlap problem encountered in this application. The de-convolution based spectral line extraction is actually a correlation method used for pattern recognition. The known info-ink profiles are the reference patterns to be found inside the spectra. However, when a useful signal in the input is weak, its auto-correlation peak may not be able to provide accurate intensity information of a spectral line. In this case, a calibrating procedure should be introduced. The convolution-based algorithm has the ability of smoothing noise . However, if the average energy of noise exceeds the minimum resolution of the intensity, the decoding will yield wrong data. The present paper provides only basic idea and primary experimental results about the use of semiconductor nanocrystals, or QDs, for security applications. More experiments, such as repeatability, sensitivity to nonuniform illumination, detection limit for a given excitation laser power, measurement time and so on, will be conducted in our future works.
References and links
1. A. P. Alivisatos, “Perspectives on the physical chemistry of semiconductor nanocrystals,” J. Phys. Chem. 100, 13226–13239 (1996). [CrossRef]
2. M. Han, X. Gao, J. Z. Su, and S. Nie, “Quantum-dot-tagged microbeads for multiplexed optical coding of biomoleclues,” Nature Biotechnol. 19, 631–635 (2001). [CrossRef]
3. W. C. W. Chan, D. J. Maxwell, X. Gao, R. E. Bailey, M. Han, and S. Nie, “Luminescent quantum dots for multiplexed biological detection and imaging,” Current Opinion in Biotechnol. 13, 40–46 (2002). [CrossRef]
4. X. Gao, W. C. W. Chan, and S. Nie, “Quantum-dot nanocrystals for ultrasensitive biological labeling and multicolor optical encoding,” J. Biomed. Opt. 74, 532–537 (2002). [CrossRef]
5. M. A. Hines and P. Guyot-Sionnest, “Synthesis and characterization of strongly luminescing ZnS-capped CdSe nanocrystals,” J. Phys. Chem. 100, 468–471 (1996). [CrossRef]
6. B. O. Dabbousi, J. Rodriguez-Viejo, F. V. Mikulec, J. R. Heine, H. Mattoussi, R. Ober, K. F. Jensen, and M. G. Bawendi, “(CdSe)ZnS core-shell quantum dots: synthesis and characterization of a size series of highly luminescent nanocrystallites,” J. Phys. Chem. B , 101, 9463–9475 (1997). [CrossRef]
8. J. Lee, V. C. Sundar, J. R. Heine, M. G. Bawendi, and K. F. Jensen, “Full color emission from II–VI semiconductor quantum dot-polymer composites,” Adv. Mater. 12, 1102–1105 (2000). [CrossRef]
10. C. C. Chan, W. Jin, and M. S. Demokan, “Enhancement of measurement accuracy in fiber Bragg grating sensors by using digital signal processing,” Opt. & Laser Technol. 31, 299–307 (1999). [CrossRef]
11. S. Chang and C. P. Grover, “Centroid detection based on optical correlation,” Opt. Eng. 41, 2479–2486 (2002). [CrossRef]