A chemical timer approach to delayed chemiluminescence

Significance

Chemical manipulation of the onset time of reactions has been a long-standing challenge. Taking chemiluminescence (CL) as an example, the tight coupling between liquid mixing and the light emission process represents a major obstacle for its applications due to the complicated liquid convection and diffusion. Here, we established a chemical timer approach to decouple the CL emission from liquid mixing for the first time, allowing for precise regulation of the CL onset time through a slow-generation-scavenge mechanism and for dynamic 4-dimensional chemical coding with high information density. This was further proved to be a general approach for engineering the onset time of various types of chemical reactions, with implications for broad fields ranging from bioassays and microfluidics to bioimaging and infochemistry.

Abstract

Although the onset time of chemical reactions can be manipulated by mechanical, electrical, and optical methods, its chemical control remains highly challenging. Herein, we report a chemical timer approach for manipulating the emission onset time of chemiluminescence (CL) reactions. A mixture of Mn²⁺, NaHCO₃, and a luminol analog with H₂O₂ produced reactive oxygen species (ROS) radicals and other superoxo species (superoxide containing complex) with high efficiency, accompanied by strong and immediate CL emission. Surprisingly, the addition of thiourea postponed CL emission in a concentration-dependent manner. The delay was attributed to a slow-generation-scavenging mechanism, which was found to be generally applicable not only to various types of CL reagents and ROS radical scavengers but also to popular chromogenic reactions. The precise regulation of CL kinetics was further utilized in dynamic chemical coding with improved coding density and security. This approach provides a powerful platform for engineering chemical reaction kinetics using chemical timers, which is of application potential in bioassays, biosensors, CL microscopic imaging, microchips, array chips, and informatics.

Chemiluminescence (CL) is a light emission phenomenon accompanying certain chemical reactions (1). The oxidation of certain compounds generates excited state products whose radiative decay to the ground state results in the spontaneous emission of light without irradiation by excitation light. CL not only serves as a natural indicator for studying reaction kinetics and excited states but is also applicable in cold light sources, bioassays, and bioimaging (2–5). For instance, owing to the impressive sensitivity ensured by the complete elimination of the optical background, CL-based immunoassays have largely replaced their colorimetric and fluorometric counterparts to become the most commonly employed and powerful clinical laboratory test platforms for the ultrasensitive detection of a plethora of biomarkers (6, 7).

Existing CL studies are based on the fact that light emission occurs immediately upon the mixing of reactants (CL compounds, oxidants, and catalysts). The CL intensity is a direct indicator of the reaction rate because it quantifies the number of photons (and thus product molecules) released per second. While this feature favors fundamental studies on reaction kinetics and mechanisms, it represents one of the most critical obstacles for improving the performance of CL-based assays and biosensors and for further applications of CL in microscopic imaging, microchips, and microarrays because of the technical challenge of mixing various solutions in a reproducible and controllable manner. Spatiotemporal coupling between the fluid mixing process and reaction process results in significant variations in CL intensity and the kinetic curves generated by parallel experiments, especially in the early, most critical stage (8, 9). Accordingly, auto injectors, flow injection technology, and microfluidics have been utilized to reasonably program the injection and detection processes, relying on motorized devices to minimize variation in the initial physical conditions prior to mixing (10–12). Despite the substantial advancements in relevant technologies, liquid manipulation is still a major concern for newly designed CL-based microfluidic systems when considering the significant complexity and stochasticity during solution operations. For example, microfluidic chips used in bioassays that rely on centrifugal force to manipulate liquid cannot initiate detection before the rotor comes to rest (13). An ideal solution to this problem is to effectively decouple the CL reaction from solution mixing. In other words, the introduction of a timer would postpone the CL reaction to allow sufficient time for detector preparation.

A well-known approach applied for this purpose is electrochemiluminescence (ECL), whereby one of the CL reactants (often the oxidant) is produced at the surface of an electrode upon the application of a suitable potential, which is readily controllable (14, 15). In this way, the ECL reaction can be triggered after the solution reaches stability. ECL remains an active and promising field for the interplay between spectroscopy and electrochemistry (16, 17). However, it is limited to ECL systems and occurs only at the solution–electrode interface, compromising the photon flux because the majority of reactants in the bulk solution are not completely utilized. In addition, the use of electrodes introduces challenges regarding reproducibility due to electrode fouling and inconsistent surface chemistry among different electrodes. Therefore, a chemical timer capable of postponing liquid-phase CL reactions after the mixing process is highly desirable but is yet to be developed.

In this study, we report a chemical timer approach for manipulating the emission onset time of CL reactions. The emission onset time of CL reactions could be precisely regulated in a concentration-dependent manner of the chemical timers via a slow-generation-scavenging mechanism. First, the presence of Mn²⁺ and NaHCO₃ was found to enhance the CL reaction of a CL reagent, 8-amino-5-chloro-2,3-dihydro-7-phenylpyrido-[3,4-d]-pyridazine-1,4-dione (L012), with H₂O₂ by four orders of magnitude at a neutral pH. This enhancement was attributed to the efficient production of reactive oxygen species (ROS) radicals and other superoxo species catalyzed by Mn²⁺ and NaHCO₃. Surprisingly, the addition of ROS radical scavenger thiourea to the mixture resulted in significantly delayed CL emission, and the delay time could be controlled precisely according to the thiourea concentration. A slow-generation-scavenging mechanism is proposed based on further studies. Thiourea acted as a ROS radical scavenger, consuming ROS radicals before they could participate in subsequent reactions. CL emission was only possible after the exhaustion of thiourea, resulting in a dark stage before the CL reactions—i.e., delayed CL. This mechanism was validated to be general for various types of reductive compounds with the scavenging ability of ROS radicals, different types of CL reagents, and even non-CL reactions. Consequently, the CL reaction was isolated from solution mixing, revealing possibilities for engineering CL kinetics. Because the CL reaction of L012 results in blue-light emission, multicolor CL emissions were further realized in the presence of suitable acceptor fluorophores via an indirect CL resonance energy transfer (CRET) mechanism. This feature ultimately allowed for 4-dimensional dynamic chemical coding of messages in terms of position (x and y), emission color, and, most important, delay time.

Results and Discussion

Enhanced CL of L012-H₂O₂ in the Presence of NaHCO₃ and Mn²⁺.

L012 is a well-documented organic molecule that efficiently emits blue light during oxidation by H₂O₂ in the presence of suitable catalysts, such as horseradish peroxidase, hemin, and Co²⁺ (18–20). Here we report the discovery of a catalyst, the combination of HCO₃⁻ and Mn²⁺, capable of enhancing the CL intensity of L012-H₂O₂ by four orders of magnitude under neutral conditions. SI Appendix, Fig. S1 shows representative CL kinetic curves when 100 μL 0.5 M H₂O₂ is injected into a standard 96-well microplate containing 100 μL L012-Mn²⁺-NaHCO₃ solutions with varying Mn²⁺ concentrations. The experiment was performed in a commercial CL microplate reader equipped with an auto injector in which the CL intensity was recorded using a photomultiplier tube detector underneath the well of interest as a function of time. In the control experiment, no enhancement was observed when H₂O₂ was injected into a blank solution without Mn²⁺.

Delayed CL Regulated by Thiourea.

Surprisingly, the presence of thiourea was found to postpone the CL emission while the maximum CL intensity was maintained, leading to the discovery of the delayed CL reaction in homogeneous solution. Furthermore, the delay time could be precisely regulated by adjusting the thiourea concentration. As shown in Fig. 1A, the emission onset time was increasingly delayed with increasing thiourea concentration, exhibiting an approximately linear correlation between their logarithms in the range of 0 to 50 s (Fig. 1B, black solid line). The regression equation was log t_d = 0.657 + 0.872 × log c (unit of c is mM), and the correlation coefficient was 0.989; t_d is the delay time calculated by t–t₀, where t and t₀ are the times at which the CL intensity reaches 100 Arbitrary Unit (a.u.) in the presence and absence of thiourea, respectively. Therefore, the CL emission onset time could be artificially regulated by adjusting the thiourea concentration.

Fig. 1.

Delay effects of thiourea. (A) CL dynamics of L012 system with increasing thiourea concentration (0, 0.2, 1, 5, 10, and 20 mM along the arrow). (B) Linear regression analysis between log delay time (t_d) and log thiourea concentration (THU-1, black solid line), 2-imidazolidinetione (THU-2, red dotted lines), and 2-methyl-3-thiosemicarbazide (THU-3, blue dotted lines). (C) Time-lapse images of 10 wells containing thiourea at concentrations indicated in red on the left. Time = 0 represents the moment when the same amount of H₂O₂ solution was simultaneously injected into the 10 wells. (D and E) CL dynamics of (D) luminol with increasing thiourea concentration (0, 1, 5, 10, 20, 40, 80, 100, 150, and 200 mM along the arrow) and (E) ABEI system with increasing thiourea concentration (0, 0.5, 2, 5, 10, 20, 35, 50, 75, and 100 mM along the arrow). Temperature = 25 °C, pH = 7.5.

The delayed CL was further verified visually. In Fig. 1C, each column displays a time-lapse photograph of 10 wells containing five incremental concentrations of thiourea (two wells for each concentration to demonstrate repeatability). After the simultaneous injection of H₂O₂ into the wells and suitable mixing, the top two wells containing 0.2 mM thiourea began to glow at 2 s and reached maximal intensity at 10 s, which was followed by gradual decay over the next 80 s. However, the bottom two wells containing 40 mM thiourea remained dark until 58 s and reached maximal intensity at 70 s after the injection of H₂O₂. This result clearly illustrated the delay effect of thiourea on CL emissions. Movie S1 comprehensively demonstrates the entire process.

pH dependence on delayed CL was also examined. As shown in SI Appendix, Fig. S2, CL intensity decreased and the lasting time was extended with pH increasing, while the emission onset time only shifted a little. The optimal condition of the delay CL system was pH 7.5.

Mechanistic Studies.

The mechanisms of 1) the NaHCO₃ and Mn²⁺ enhancement effects and 2) the thiourea delay effect on the CL reactions were subsequently investigated. The significant CL enhancement was attributed to the efficient production of ROS radicals and other superoxo species in the presence of H₂O₂, NaHCO₃, and Mn²⁺. First, no CL signal was observed if NaHCO₃ was replaced with phosphate buffer under the same pH (SI Appendix, Fig. S3), which indicated that NaHCO₃ was essential for the CL enhancement rather than simply functioning as a buffer. In addition, according to the cyclic voltammograms (CVs) of Mn(II) oxidation, there was an obvious decrease in the 1-electron oxidation potential of electropositive Mn(II) after the addition of NaHCO₃ (SI Appendix, Fig. S4). Thus, NaHCO₃ was believed to be able to activate the reaction between H₂O₂ and Mn²⁺. Second, the electron paramagnetic resonance (EPR) spin-trapping characterization of radicals in reaction system is shown in SI Appendix, Fig. S5, which was consistent with a combination of standard spectra of the 5,5-dimethyl-1-pyrroline N-oxide (DMPO)-trapped superoxide radical anion (DMPO/O₂^•−) (blue line) and the hydroxyl radical (DMPO/OH^•) (pink line). From the simulation ratio, the signal of DMPO/O₂^•− and DMPO/OH^• took up 78.1% and 21.9%, respectively. Considering that the carbonate radical anion (CO₃^•−), which is commonly generated by the interaction between the hydroxyl radical (OH^•) and HCO₃⁻ in the Mn^II-NaHCO₃-H₂O₂ system (21, 22), does not produce stable radical adducts with DMPO but oxidizes DMPO and also produces the EPR-detectable DMPO/OH^• radical adduct upon water addition (23), it was thus necessary to identify the origin of DMPO/OH^• signals. Dimethyl sulfoxide (DMSO) is often used as a OH^• scavenger and is oxidized by OH^• radicals to produce methyl radicals, which can be captured by DMPO to form an EPR-detectable methyl radical adduct (DMPO/CH₃^•) (24). Bonini et al. (25) found that the addition of DMSO to the acetaldehyde/XO/catalase/bicarbonate system, which only contained CO₃^•−, did not produce the DMPO/CH₃^• radical adduct or changed the yield of the DMPO/OH^• radical adduct. Therefore, DMSO was used as a secondary radical trap in our system to further analyze the origin of the DMPO/OH^• signal. As shown in SI Appendix, Fig. S6, EPR signals for both DMPO/OH^• (●) and DMPO/O₂^•− (▪▪▪) were detected using DMPO as radical trap (black trace) in the Mn^II-NaHCO₃-H₂O₂ system. The addition of DMSO (red trace), however, had little effect on the yield of DMPO/OH^• (●), indicating that the detected DMPO/OH^• was mainly originating from CO₃^•− radicals. In addition, weak DMPO/CH₃^• adduct signals (▪▪▪) were also observed, which possibly derived from the small proportion of OH^• that had not yet been converted to CO₃^•− by HCO₃⁻. Third, after adding H₂O₂ to the mixture of NaHCO₃ and Mn²⁺, an enhanced absorption at 275 nm with time was observed (SI Appendix, Fig. S7A) that corresponded to the superoxide complex Mn^IIO₂(HCO₃)_n as reported previously (26). After adding L012 to the system, the absorbance at 275 nm rapidly disappeared (SI Appendix, Fig. S7 C and D). These results indicated that the formed superoxide radical anion (O₂^•−) would first combine with Mn^II(HCO₃)_n to form Mn^IIO₂(HCO₃)_n and then participate in the oxidation of L012. Furthermore, if sodium pyrophosphate (PP) was added after the addition of H₂O₂, an absorption at 258 nm was observed (SI Appendix, Fig. S7B), which corresponded to a stable Mn(III) complex with PP via coordination (27). Thus, Mn(III) was generated during the decomposition of H₂O₂. Fourth, the CL emission spectra of the L012-Mn^II-NaHCO₃-H₂O₂ system (SI Appendix, Fig. S8) were consistent with the characteristic emission of the L012-H₂O₂ CL system (SI Appendix, Table S1), suggesting that the CL emission resulted solely from the L012-H₂O₂ system throughout the reaction process and that Mn²⁺ and HCO₃⁻ participated as catalysts only. Fifth, because comparable CL intensities were observed in the solutions presaturated with air, N₂, or O₂ (SI Appendix, Fig. S9), the contribution from the dissolved oxygen to the CL reaction was negligible.

Accordingly, a plausible mechanism for the L012-Mn^II-NaHCO₃-H₂O₂ CL system is proposed in Fig. 2, entailing three stages: activation, radical formation, and CL reaction. Because the catalytic activity of Mn²⁺ was highly dependent on the existence of HCO₃⁻, in the activation stage, HCO₃⁻ was considered to interact with Mn²⁺ and H₂O₂ to form the highly active species Mn^II(HCO₃)_n and peroxybicarbonate (HCO₄⁻) (28, 29), respectively. Attiogbe and Francis (29) supposed that negative or neutral Mn(II) species in the solution were more susceptible to electron abstraction by oxidants owing to their higher electron density compared to that of free Mn²⁺ ions. This assumption was also supported by our CV results (SI Appendix, Fig. S4). In addition, the formation of HCO₄⁻ in the system was confirmed by mass spectrometric analysis (SI Appendix, Fig. S10A). Thus, in the radical formation stage, activated Mn^II(HCO₃)_n reduces HCO₄⁻ with relative ease to produce OH^•. This point was also supported by mass spectrometry results. As shown in SI Appendix, Fig. S10B, the signal of hydroxylation products (m/z =153.0177) of 2-hydroxybenzoic acid, which was widely used as an OH^• probe, significantly increased when NaHCO₃ was added to the Mn²⁺-H₂O₂ mixture. After the formation, OH^• will soon interact with HCO₃⁻ in the solution and produce CO₃^•− radicals. Meanwhile, the formed Mn^III(HCO₃)_n will oxidize H₂O₂ to produce O₂^•− and Mn^II(HCO₃)_n, and they will further form the Mn^IIO₂(HCO₃)_n complex. Finally, it has been reported that luminol anions, an analog of L012, could be oxidized to luminol radicals by CO₃^•− radicals (30). Therefore, in the CL reaction stage, the produced CO₃^•− radicals are also considered to oxidize L012 anions (L012⁻) to L012 radicals (L012^•−), and L012^•− will further react with Mn^IIO₂(HCO₃)_n to form the excited-state oxidation product of L012 (L012-ox*), accompanied by light emission.

Fig. 2.

Mechanism of the L012-Mn^II-HCO₃⁻-H₂O₂ CL system and thiourea-induced delay phenomenon.

With the Mn²⁺enhancement effect clarified, the thiourea delay effect, attributed to its strong ROS radical scavenging capability (31, 32), was investigated. Because thiourea rapidly combines with O₂^•− and OH^• generated in the radical formation stage, the conversion of L012⁻ to L012-ox* was completely blocked in its presence (Fig. 2), leading to a dark stage without light emission. This was confirmed by the results of EPR experiments. When thiourea was added, a set of distinct EPR signals was detected at the beginning of the reaction (Fig. 3, black line) and the simulated spectrum (green line) strongly suggested the presence of an S-containing radical species characterized by the following hyperfine splitting parameters: A_N = 14.49 Gauss, A_H^β = 15.63 Gauss (33). This S-containing radical species was found to be short-lived (lifetime of approximately 60 s) and most probably originated from the unstable thiourea radical intermediate (32) upon the reaction between the thiourea and ROS radicals generated by the reaction system. During this process, no obvious DMPO/O₂^•− or DMPO/OH^• signal was detected. Subsequently, after 120 s, a mixture of DMPO/O₂^•− and DMPO/OH^• EPR signals was detected in the second measurement (Fig. 3, gray, red, blue, and pink line), indicating that O₂^•−, OH^• and CO₃^•− (converted from OH^•) would be generated only after thiourea was exhausted. The reaction between thiourea and ROS radicals was also investigated (SI Appendix, Scheme S2). During the early stage, thiourea is oxidized by OH^• to form a radical cation (thiourea^•+) (32). On the one hand, this radical cation can further interact with OH^• radicals to form a thiourea-OH⁺ adduct, which can be converted to a series of oxidation products by HCO₄⁻. On the other hand, thiourea^•+ can also be converted to a dimeric radical cation through a disulfidic linkage, which scavenges O₂^•− radicals in the reaction system (32). More detailed discussion of the reaction of thiourea is presented in SI Appendix, text. Furthermore, the absorption of the system was also monitored during the reaction. As shown in SI Appendix, Fig. S11, the absorption of thiourea at 236 nm decreased rapidly in the first 50 s, indicating the consumption of thiourea via its reaction with O₂^•− and OH^•. The consumption time (50 s) was consistent with the CL delay time under the same conditions. These results suggested that O₂^•− and OH^• were completely scavenged before thiourea exhaustion, implying that their formation rates were low. In addition, a decrease in the concentration of H₂O₂, Mn²⁺, or NaHCO₃ could further delay the initial CL emission time and reduce the CL intensity (SI Appendix, Figs. S12–S14) because these three substances were directly related to the generation of O₂^•− and OH^•. Therefore, when they were present in low concentration, the formation rate of O₂^•− and OH^• was reduced, resulting in thiourea being present for longer. On the other hand, L012, which was not involved in ROS radical formation, had little effect on the delay time as expected (SI Appendix, Fig. S15). Thus, a slow-generation-scavenging mechanism was responsible for the delayed effect of thiourea. If a slow-generated intermediate and an efficient scavenger exist in a multistep reaction system, then this would lead to a delay (SI Appendix, Scheme S1). The delay time increased with the decelerating generation rate of the intermediate, and with increasing scavenger concentration. There are two criteria for this mechanism: 1) the generation rate of the intermediate must be lower than the scavenging rate, and 2) sufficient reagent amounts are necessary for reactions occurring after the delay.

Fig. 3.

EPR spin-trapping characterization of radicals in the presence of thiourea. The spin trap, DMPO, was added to the reaction mixture at the beginning of the reaction (EPR spectrum labeled “Experimental reaction initiation,” black traces) or 120 s after the initiation (EPR spectrum labeled “Experimental 120 s after initiation,” gray traces). Simulations were plotted under the corresponding experimental EPR spectrum.

It was further found that the proposed mechanism for delayed CL was generally applicable to various types of CL reagents and ROS radical scavengers. As shown in Fig. 1 D and E and SI Appendix, Fig. S16, two L012 analogs, luminol and N-(4-aminobutyl)-N-ethylisoluminol (ABEI), and another common CL reagent, 1,10-phenanthroline, were used instead of L012. In each case, CL emission could be delayed by adjusting the thiourea concentration. Furthermore, as shown in Fig. 1C, two other water-soluble analogs of thiourea (THU-1), 2-imidazolidinetione (THU-2), and 2-methyl-3-thiosemicarbazide (THU-3), were also found to exert similar delay effects on CL emission of the L012 system. Detailed kinetic curves are provided in SI Appendix, Fig. S17 A–C. In addition to thiourea derivatives, other compounds with ROS radicals scavenging capabilities were studied. As shown in Fig. 4 A–F and SI Appendix, Fig. S18 A–D, 10 other compounds—catechol, dopamine, o-phenylenediamine, gallic acid, pyrogallic acid, ascorbic acid, catechin, p-phenylenediamine, hydroquinone, and chlorogenic acid—were found to be capable of regulating the CL onset time. The generality of CL reagents and ROS radical scavengers not only provided vivid support for the proposed mechanism but also expanded the chemical timer toolbox for future applications in various fields.

Fig. 4.

Other compounds with delay effects on CL emission. (A) Catechol. (B) Dopamine. (C) o-Phenylenediamine. (D) Gallic acid. (E) Pyrogallic acid. (F) Ascorbic acid. Along the arrow, the concentrations of each compound were 0, 0.1, 0.2, 0.5, 1, and 2 mM. Finally, 2 mM ascorbic acid was not included in 4F, as the delay time was more than 120 s in that case.

More generally, the delay mechanism was further proved to be applicable to a chromogenic reaction that was frequently adopted in colorimetric assays. As shown in Fig. 5, the fading onset time of rhodamine B (RhB) (552 nm; SI Appendix, Fig. S19) in the presence of NaHCO₃-Mn²⁺-H₂O₂ could be also delayed by thiourea in a concentration-dependent manner. This result indicated that the chemical timer approach for regulating reaction onset time is very general to broad types of chemical reactions.

Fig. 5.

Absorbance kinetics of the chromogenic reaction of RhB with NaHCO₃-Mn^II-H₂O₂ at 552 nm (ultraviolet-visible spectra; SI Appendix, Fig. S19) in the presence of increasing concentration of thiourea (0, 1, 2, 5, 10, and 20 mM along the arrow).

Dynamic Chemical Coding.

Temporal regulation of delayed CL reactions would not only be instrumental in eliminating the inconvenience of liquid manipulation in CL-based assays but would also provide opportunities for dynamic chemical coding. The intersection of information science and chemistry, known as infochemistry, has received intensive attention in the past decade. It entails the use of various types of chemical properties and reactions for information encryption as well as anticounterfeiting (34, 35). In their pioneering work, Thomas and Kim (36, 37) created a sequence of dots consisting of specially designed combinations of alkane metal salts (Li, Na, K) in a strip of combustible materials to write, store, and retrieve information via flame reactions. The concept was later expanded to biological systems whereby the chemically induced expression of fluorescent proteins of various colors provided the information carriers (38). Stimuli-responsive hydrogels are another popular system in which the sequence of colors (chromophore, fluorescence, and CL) is used to encode information and transit messages upon suitable chemical treatments (39–41). Although powerful, existing studies often used static coding strategies, in which only position and wavelength (i.e., color) were used to encode information. To improve coding density and security, dynamic coding is desirable, whereby the reaction dynamics can be regulated on demand to carry information along the time dimension (42–44).

Multicolor emission was further achieved by introducing CRET into the delayed CL. Three CRET systems, including one step system that from L012 (blue, emission wavelength: 468 nm) to fluorescein (FL) (green, excitation/emission wavelength: 491/523 nm) (Fig. 6A), one step system that from L012 to RhB (red, excitation/emission wavelength: 550/587 nm) (Fig. 6B), and two-step system that from L012 through FL to RhB (Fig. 6C) (45), were examined. As shown in Fig. 6D, additional intermediate CL colors were produced via hybrid wavelength emissions by adjusting the ratio between FL and RhB.

Fig. 6.

Three CRET systems. (A–C) Time-dependent CL spectra of (A) L012-FL system, (B) L012-RhB system, and (C) L012-FL-RhB system. The inset figures are the peak differentiation imitating analysis of each CRET system, employing OriginLab. (D) Photographs of CL emission color gradient with increasing fluorochrome ratio.

Evidently, the color and delay duration of CL emission in each well can be independently regulated by the acceptor fluorophore and ROS radical scavenger, respectively. Five different colors (red, yellow, green, blue, and purple; see SI Appendix, Table S2) could be well distinguished by the naked eye and a regular color camera. At the same time, temporal regulation of the delayed CL emission allowed us to distinguish six possible time windows with a separation of 15 s. The orthogonal combination of five colors and six time windows generated a code table containing all 26 letters of the alphabet and four frequent symbols, as shown in Fig. 7A. Messages could be intentionally encrypted using a specially designed solution with various types of fluorophore acceptors (color) and varying pyrogallic acid concentrations (delay time). The simultaneous injection of H₂O₂ into a sequence of wells resulted in the delayed appearance of light of a particular color, which was translated to characters in sequence—i.e., a message. As proof of concept, the phrase “hello world” was encrypted using 11 wells. Time-lapse images of these 11 wells after the injection of H₂O₂ were continuously recorded using a smartphone camera (Fig. 7B). A comprehensive movie is provided in Movie S2. Color codes and delay time codes were obtained from each well to report a character. For example, the appearance of green in the second window (25–40 s) codes for “G2” (“green second”), which corresponds to the letter “H” according to the table in Fig. 7A. SI Appendix, Table S3 compares this work with previously reported work of chemical coding. In the existing studies on chemical coding that rely on only one dimension (i.e., color), because of the limited coding density, two or more elements are often required to encrypt one character (36–38, 42, 44, 46). The introduction of the time dimension to dynamic coding greatly increased the coding density (six-fold in this case). Moreover, the coding density can be further improved by simply increasing the observation time as needed. This feature demonstrates the essential value of dynamic coding in improving coding density.

Fig. 7.

Dynamic chemical encoding. (A) Code table for 26 letters of the alphabet and four frequent symbols encrypted with five colors in six time windows. The order of letter “V” and “W” was exchanged in order to perform all five colors within the example in this work. (B) Dynamic chemical coding of a message containing 11 characters: “HELLO WORLD” (the sixth well, without any CL signal, was used to represent “space”). Coding density: one character per well.

Conclusions

In conclusion, we have proposed a chemical timer approach to manipulate the emission onset time of CL reactions. A time-regulated multicolor CL system was achieved via chemical manipulation in neutral solutions. The intense emission of L012-H₂O₂ catalyzed by Mn²⁺/NaHCO₃ could be clearly captured by a smartphone camera or observed by the naked eye. The onset time of CL emission could be precisely regulated by adjusting the thiourea concentration via a slow-generation-scavenging mechanism. This process was confirmed to occur in CL reactions of other CL reagents (luminol, ABEI, and 1,10-phenanthroline), chromogenic reaction (the fading of RhB), and with various ROS radical scavengers such as catechol, dopamine, o-phenylenediamine, gallic acid, pyrogallic acid, ascorbic acid, catechin, p-phenylenediamine, and hydroquinone, offering exciting opportunities for innovating CL assay methods. Furthermore, by combining the time-regulated effects of pyrogallic acid and the CRET principle, a multicolor dynamic CL emission was produced. This emission mode was applied to multitarget analysis with reduced mutual interference because the CL signals of different targets could be detected at different times. It is of great application potential in bioassays, biosensors, CL microscopic imaging, microchips, and array chips. In addition, a straightforward and robust CL encoding method was demonstrated, entailing adjustable space, emission time, and color. The dynamic multicolor CL pattern can store and rapidly display a large number of complex matrices, which are expected to play a critical role in chemical cryptography and signal communication. Finally, the proposed chemical timer approach might be generalized to more chemical reactions with slow-generated intermediate radicals, allowing for broad potential applicability in various fields, such as bioimaging, drug delivery, and photodynamic therapy.

Materials and Methods

Chemicals and Solutions.

We directly dissolved L012 (Wako Pure Chemical Industries, Ltd., Osaka, Japan) and 1,10-phenanthroline (Sinopharm Co., Shanghai, China) in ultrapure water to prepare a 4 mM stock solution. Luminol (Sigma-Aldrich) and ABEI (TCI, Tokyo, Japan) were respectively dissolved in NaOH (pH 12) solution to prepare 4 mM stock solutions. Next, 2-imidazolidinetione, 2-methyl-3-thiosemicarbazide, and pyrogallic acid were purchased from Shanghai Acmec Biochemical Co., Ltd. (Shanghai, China). Catechin was purchased from Solarbio (Beijing, China). Dopamine was purchased from J&K Scientific (Beijing, China); DMPO was purchased from Dojindo Molecular Technologies, Inc. (Kumamoto, Japan); and fluorescein and rhodamine B were purchased from Sangon Biotech (Shanghai, China). H₂O₂ was prepared fresh daily from 30% (vol/vol) H₂O₂ (Xinke Electrochemical Reagent Factory, Bengbu, China). All other reagents were obtained from Sinopharm Co. (Shanghai, China), of analytical grade. Ultrapure water was prepared by a Milli-Q system (Millipore, France) and used throughout.

Preparation of CO₂-Saturated NaHCO₃ Buffer.

NaHCO₃ was directly dissolved in water to prepare a 1 M stock solution. Before use, CO₂ was passed through 5 mL prepared NaHCO₃ solution for more than 40 min to obtain a CO₂-saturated NaHCO₃ buffer. Because NaHCO₃ decomposes under room temperature, CO₂ would ensure that the concentration of NaHCO₃ was stable in a neutral condition. The pH of CO₂-saturated NaHCO₃ buffer used in experiments was 7.5.

CL Kinetic Measurements.

For testing the catalytic ability of Mn²⁺, 25 μL L012 (4 mM) was mixed with 25 μL different concentrations of MnSO₄ in each single well. When initiating the test, the background signal was collected for 2 s first, and then 50 μL CO₂-saturated NaHCO₃ (1 M) and 100 μL H₂O₂ (0.5 M) were automatically injected to the test well. CL signal was collected throughout the whole process. The data analysis and diagramming were done by Excel and OrginLab. For delayed CL experiments, an additional 25 μL ROS radical scavenger was mixed with 25 μL L012 (4 mM) and 25 μL MnSO₄ (800 μM) in each single well. The volume of injected NaHCO₃ (1 M) correspondingly decreased to 25 μL. All other operations were same. A recommended set of test parameters was as follows. The measurement interval time was 0.1 s, gain: 400 (or 3,600 for 1,10-phenanthroline experiment); the injection speed was 220 μL/s. All experiments were conducted at 25 °C, pH 7.5.

CL Imaging.

Similar to the CL kinetic measurements, 25 μL 5 incremental concentrations of thiourea were mixed with 25 μL L012 (4 mM) and 25 μL f MnSO₄ (800 μM) in 10 microwells (2 wells for each concentration to demonstrate repeatability). Then, 25 μL CO₂-saturated NaHCO₃ (1 M) was rapidly added to each well, followed by the injection of 100 μL H₂O₂ (0.5 M) into all wells simultaneously using a multichannel pipette. The time-lapse images were photographed using a digital camera (Panasonic Lumix DMCFZ2000, Japan).

Mass Spectrometric Measurement.

Mass spectra were acquired with an orbitrap mass spectrometer (Exactive Plus; Thermo Fisher Scientific, Waltham, MA). The MS parameters were as follows: capillary temperature = 275 °C, max ion injection time = 10 ms, and 3 microscans for each individual scan.

CL Spectrum Measurements.

As a typical test, 250 μL 4 mM L012 (or with fluorochrome) was mixed with 250 μL 800 μM MnSO₄ (or 250 μL H₂O for control) in the test cell. Then, 500 μL CO₂-saturated NaHCO₃ (1 M) was added to the solution and the cell was put into the instrument within a dark box immediately. After the data acquisition was initiated, 1 mL H₂O₂ (0.5 M) was injected into the system with a micro-injection needle through the injection port. Finally, the time-resolved spectra were plotted to be 3-dimensional CL spectra by OriginLab. A set of referenced parameters was as follows: exposure time = 1 s, electron multiplier gain = 0 (with Mn²⁺) or 800 (without Mn²⁺), wavelength range = 350–700 nm, slit width = 1 mm, shift speed = 4.88 μs, readout rate = 50 kHz at 16 bit.

Dynamic Multicolor Encoding.

The molar ratios of CL mixtures for dynamic multicolor encoding are displayed in SI Appendix, Table S2. We mixed 10 μL different concentrations of pyrogallic acid (2, 10, 20, 35, 50, and 70 mM for the time windows 1, 2, 3, 4, 5, and 6, respectively) with 40 μL CL mixtures in 11 microwells according to the encoded message. Then, 25 μL CO₂-saturated NaHCO₃ (1 M) and 25 μL MnSO₄ (800 μM) were rapidly added to each well, followed by the injection of 100 μL H₂O₂ (0.5 M) into all wells simultaneously using a multichannel pipette. The CL decoding process was continuously photographed using a smartphone (Huawei P30 Pro).

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.